USF Libraries
USF Digital Collections

Low temperature polymer electrolyte fuel cell performance degradation

MISSING IMAGE

Material Information

Title:
Low temperature polymer electrolyte fuel cell performance degradation
Physical Description:
Book
Language:
English
Creator:
Fedock, John Andrew
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Subzero
Alternative energy
Catalytic reactor
Cryogenic
Ice formations
Dissertations, Academic -- Mechanical Engineering -- Masters -- USF   ( lcsh )
Genre:
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: The goal of this research was to quantify the degradation experienced by a polymer electrolyte fuel cell after storage at subzero temperatures ranging from 0 to -40°C. The performance loss was determined by comparing the polarization and other applicable power curves before and after the subzero storage cycle. The causes of this performance degradation were investigated by the use of Scanning Electron Microscope, Energy Dispersive x-ray Spectroscopy, and porosity scanning technologies. It was found that there are two distinct types of degradation experienced by the membrane. The first type was identified as a variance of the actual voltage - current relationship of the cell. The membrane experienced a 2 - 15% power reduction depending on the load applied to the cell. This mode of degradation only pertained to the initial freeze/thaw cycle and was not observed after any number of subsequent cycles.The cause of this type of degradation has been hypothesized to be related to the delamination of the proton exchange, gas diffusion, and micro porous layers. The second type of degradation was only observed during the subsequent cycles, and mainly affected the high power regions of the operating range. A 5% reduction in current density and power output was observed as a result of further freeze/thaw cycles. Mass transport limitations may have been caused by the destruction of the meso-porous gas diffusion and micro-porous layers. The pore size, volume, and membrane surface area were quantified using a B.E.T. porosity scanner. The results showed that the pore diameter of the catalyst and proton exchange layer did not increase significantly. The porosity scanner did indicate that a pore volume increased by a factor of ten and was confirmed by the surface area measurements of the membrane. The S.E.M. investigations allowed visual inspection of the membrane's structural integrity.Physical separation of the catalyst and gas diffusion layers was observed in the experimental sample, while a more homogeneous assembly was seen in the control sample.
Thesis:
Thesis (M.S.M.E.)--University of South Florida, 2008.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by John Andrew Fedock.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 106 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002001089
oclc - 319537141
usfldc doi - E14-SFE0002565
usfldc handle - e14.2565
System ID:
SFS0026882:00001


This item is only available as the following downloads:


Full Text

PAGE 1

Low Temperature Polymer Electrolyte Fuel Cell Performance Degradation by John Andrew Fedock A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Mechanical Engineering Department of Mechanical Engineering College of Engineering University of South Florida Major Professor: D. Yogi Goswami, Ph.D. Elias Stefanakos, Ph.D., P.E. Muhammad Rahman, Ph.D. Date of Approval: July 2, 2008 Keywords: Subzero, Alternative Energy Catalytic Reactor, Cryogenic, Ice Formations Copyright, 2008 John Andrew Fedock

PAGE 2

Dedication I would like to dedicate this work to my family, friends and my God whom has inspired and guided me throughout my entire life. Their constant encouragement and wisdom has helped me to realize the eter nal and real truth, on my own I am weak, but by pure faith everything is possible.

PAGE 3

Acknowledgments I would like to thank Dr. Yogi Gosw ami for giving me the opportunity to work under his guidance at the Clean Ener gy Research Center. Secondly, I would like to thank Chuck Garreston, Nikh il Kothurkar, and Sesha Srinivasan for their patience, guidance, and helpful suggestions Thirdly, I would like to thank Refrigeration for their donati on of cryogenic freezer materials. Finally, I would like to thank my father, John Robert Fedo ck, for providing the valuable insight, guidance and encouragement wit hout which this work would not have come to fruition.

PAGE 4

Note to Reader The original of this document contains color that is necessary for understanding the data. The orig inal dissertation is on file with the USF library in Tampa, Florida.

PAGE 5

i Table of contents List of t ables ......................................................................................................... iiiList of fi gures ........................................................................................................ivNomenclatu re .......................................................................................................viAlphabet s .....................................................................................................viGreek symbol s............................................................................................. viiSubscripts .................................................................................................... viiAbbreviati ons ............................................................................................... viiABSTRACT ........................................................................................................ viiiChapter 1: In troduction ........................................................................................ 1Chapter 2: Background and literature review ...................................................... 3Sulfonated tetrafluoroethyl ene copolymer history ......................................... 7General fabr ication ............................................................................... 8Nafion water balance .......................................................................... 10Water balance in a sub zero climate ................................................... 11Effects of ice formation ............................................................................... 13Catalyst layer ...................................................................................... 13Gas diffusion layer .............................................................................. 16Mitigation strategies .................................................................................... 18Patented mitigation techniques .......................................................... 21P.E.M. fuel cell me mbrane mate rials .......................................................... 22Finite element model ing simula tion ............................................................ 26Chapter 3: Experimental fac ility and testing procedures .................................... 31Test bed configuratio n phase: bu ild 1 ......................................................... 32Purpose .............................................................................................. 32Overview ............................................................................................ 32Data acquisi tion .................................................................................. 34Test bed configuratio n phase: bu ild 2 ......................................................... 36Purpose .............................................................................................. 36Overview ............................................................................................ 36Cryogenic expansion c limate cham ber ....................................................... 37Design upg rades ................................................................................ 41Testing proc edures ..................................................................................... 42Baseline te sts ..................................................................................... 42Cold storage inve stigations ................................................................ 43Methodolog y ....................................................................................... 44Investigated membrane el ectrode asse mbly ...................................... 46Chapter 4: Result s ............................................................................................. 47Cold storage in vestigat ion .......................................................................... 47B.E.T. invest igations ................................................................................... 69

PAGE 6

ii S.E.M. invest igations .................................................................................. 74Meso-porous inve stigations ................................................................ 76Gas diffusion layer in vestigati ons ....................................................... 81Chapter 5: C onclusion s ..................................................................................... 86One hour freeze/thaw cycl e investigat ions ................................................. 86Extended duration in vestigati ons ................................................................ 87Twenty-four hour freeze/thaw duration invest igations ................................ 87B.E.T. invest igations ................................................................................... 88S.E.M. – optical investigat ion ..................................................................... 88Delaminat ion ...................................................................................... 88G.D.L. micro-st ructure ........................................................................ 89Referenc es ......................................................................................................... 90Appendice s ......................................................................................................... 98Appendix A: Ca libration s ............................................................................ 99Appendix B: Uncert ainty anal ysis ............................................................. 101S.E.M. Uncertainty ............................................................................ 101Appendix C: Scanning electron microscope (S .E.M.) ............................... 102Energy dispersive x-ray spectroscopy (EDS) ................................... 104Burnauer, Emmett and Teller (B.E.T.) surface area and pore size distribution meas urements ............................................................... 105

PAGE 7

iii List of tables Table 1: Instrument calibrations ........................................................................ 100 Table 2: Unce rtaint y ......................................................................................... 101

PAGE 8

iv List of figures Figure 1: Electro-c hemical energy balance........................................................ 4 Figure 2: Basi c fuel cell operation ...................................................................... 5 Figure 3: Standard f uel cell a ssembly ................................................................ 7 Figure 4: Micro-sized ion passage illu stration .................................................... 9 Figure 5: Proton exchange layer model ........................................................... 10 Figure 6: Ice formati on and subsequent po wer losses .................................... 12 Figure 7: Ice formation effects on agglomer ate pores ..................................... 15 Figure 8: Interdig inated flow patterns .............................................................. 28 Figure 9: Fuel ce ll test bed build 1 ................................................................... 33 Figure 10: Data ac quisition bui ld 1 .................................................................... 33 Figure 11: Physical la yout of bu ild 1 .................................................................. 34 Figure 12: Test bed data acquisition sc hematic................................................. 35 Figure 13: Climate cham ber heat exc hanger ..................................................... 38 Figure 14: C.A.D. representati on of the climat e chamber .................................. 39 Figure 15: Completed build 2 test bed ............................................................... 40 Figure 16: Polarization curve com parison ......................................................... 47 Figure 17: One hour freeze/thaw cycle investigation polarization graph 60% relative hu midity ............................................................................... 49 Figure 18: One hour freeze/thaw cycle in vestigation – power v. current graph 60% relative humidit y ....................................................................... 50 Figure 19: One hour freeze/thaw cycle in vestigation – voltage v. power graph 60% relative humidit y ....................................................................... 51 Figure 20: One hour freeze/thaw cycle investigation – polarization graph 70% relative hu midity ............................................................................... 52 Figure 21: One hour freeze/thaw cycle in vestigation – power v. current graph 70% relative humidit y ....................................................................... 53 Figure 22: One hour freeze/thaw cycle in vestigation – voltage v. power graph 70% relative humidit y ....................................................................... 54 Figure 23: One hour freeze/thaw cycle investigation – polarization graph 80% relative hu midity ............................................................................... 55 Figure 24: One hour freeze/thaw cycle in vestigation – power v. current graph 80% relative humidit y ....................................................................... 56 Figure 25: One hour freeze/thaw cycle in vestigation – voltage v. power graph 80% relative humidit y ....................................................................... 57 Figure 26: Freeze/thaw cycle duration in vestigation – polarization graph ......... 58 Figure 27: Freeze/thaw cycle duration inve stigation – power v. current graph .. 59 Figure 28: Freeze/thaw cycle duration inve stigation – voltage v. power graph .. 60

PAGE 9

v Figure 29: Freeze/thaw cycle duration in vestigation – voltage v. power (high load) ................................................................................................. 61 Figure 30: Twenty four hour freez e/thaw cycle – polarization graph .................. 62 Figure 31: Polarizati on graph tuto rial ................................................................. 63 Figure 32: Post twenty four hour freeze/thaw cycle area specific resistance .. 64 Figure 33: NM4 power v. curr ent ....................................................................... 65 Figure 34: Twenty four hour freeze/t haw cycle voltage v. power graph ........... 67 Figure 35: Twenty four hour freeze/thaw cycle – polarization graph (full load region) ............................................................................................. 68 Figure 36: Pore si ze distri bution ....................................................................... 70 Figure 37: Surface area: adsorbed ni trogen v. relative pressure ...................... 71 Figure 38: Pore volume v. pore di ameter ......................................................... 72 Figure 39: Proton exchange membrane crosssecti on ..................................... 75 Figure 40: Control (OS1) proton ex change – M.P.L. inte rface x1100 ............... 76 Figure 41: NM3 proton exchange – M.P.L. inte rface x 2500 ............................. 77 Figure 42: Control (OS1) proton ex change – M.P.L. inte rface x10,000 ............ 78 Figure 43: NM3 proton exchange – M.P.L. inte rface x5 ,000 ............................ 79 Figure 44: Control (OS1) proton ex change – M.P.L. inte rface x70,000 ............ 80 Figure 45: Proton exchange – M. P.L. interfac e x35,000 ................................... 81 Figure 46: Control (OS1) gas diffusion la yer x100 ............................................ 82 Figure 47: Control (OS1) gas diffusion la yer x2000 .......................................... 83 Figure 48: Control (OS1) gas diffusion laye r x12,000 ....................................... 84 Figure 49: Comparison of NM3 and control (OS1) G.D.L. layers (macro sized) ....................................................................... 85 Figure 50: Comparison of NM3 and control (OS1) G.D.L. layers (micro sized) .......................................................................... 85 Figure 51: Sample electron scatt ering ............................................................ 103 Figure 52: S.E. M. lay out ................................................................................. 104 Figure 53: Nitrogen ads orption pr ocess .......................................................... 106

PAGE 10

vi Nomenclature Alphabets A active area of the fuel cell (m2) C concentration, constant (mol) D constant, diffusion coefficient E ohmic voltage loss (V) E voltage (V) F Faraday constant (96485 Cmol-1) G Gibbs energy (kJmol-1) H enthalpy (kJmol-1) I current (A) R resistance, gas constant ( ), (8.314 Jmol-1K-1) T temperature (K) V volume flow rate (NTP) (m3s-1) V0 standard molar volume (0.002 241 m3mol-1) i current density (Am-2) i0 exchange current density (Am-2) k gas permeability (m2) n mass flow rate (mols-1) p pressure (Pa) p0 atmospheric pressure (101325 Pa) s line slope v flow velocity (ms-1) x molar fraction z charge number of the electrochemical reaction

PAGE 11

vii Greek symbols change charge transfer coefficient thickness of the diffusion layer potential difference, loss (kgm-1s-1,V) stoichiometry density (kgm-3) kinetic viscosity (m2s-1) Subscripts 0 reference a anode b bulk c cathode d diffusion L limiting m measured s surface real actual gas flow H2 hydrogen O2 oxygen Abbreviations ASR Area Specific Resistance BET Burnauer, Emmett and Teller porosity scanning technologies GDL Gas Diffusion Layer MEA Membrane Electrode Assembly MPL Micro Porous Layer ORR Oxygen Reduction Reaction PEMFC Polymer Electrol yte Membrane Fuel Cell PTFE Poly-Tetra-Fluoro-Ethylene SEM Scanning Electron Microscope

PAGE 12

viii Low Temperature Polymer Electrolyte Fuel Cell Performance Degradation John Andrew Fedock ABSTRACT The goal of this research was to quantify the degradatio n experienced by a polymer electrolyte fuel cell after stor age at subzero temperatures ranging from 0 to -40C. The performance loss was det ermined by comparing the polarization and other applicable power curves before and after the subzero storage cycle. The causes of this performance degradatio n were investigated by the use of Scanning Electron Microscope, Energy Dispersive x-ray Spectroscopy, and porosity scanning technologies. It was found that there are two distinct types of degradation experienced by the membrane. The first type was identified as a variance of the actual voltage – current re lationship of the cell. The membrane experienced a 2 15% power reduction dep ending on the load applied to the cell. This mode of degradation only pertained to the initial freeze/thaw cycle and was not observed after any number of subsequent cycles. The cause of this type of degradation has been hypothesized to be re lated to the delam ination of the proton exchange, gas diffusion, and micro porous layers. The second type of degradation was only observed during the subsequent cycles, and mainly affected the high power regions of the operating range. A 5% reduction in current density and power output was observed as a result of further freeze/thaw cycles. Mass transport limitations may have been caused by the destruction of the meso-porous gas diffusion and microporous layers. The pore size, volume,

PAGE 13

ix and membrane surface area were quantifie d using a B.E.T. porosity scanner. The results showed that the pore diameter of the cata lyst and proton exchange layer did not increase significantly. The porosity scanner did indicate that a pore volume increased by a factor of ten and was confirmed by the surface area measurements of the membr ane. The S.E.M. investigations allowed visual inspection of the membrane’s structural integrity. Physical separation of the catalyst and gas diffusion layers was observed in the experimental sample, while a more homogeneous assembly was seen in the control sample.

PAGE 14

1 Chapter 1: Introduction In the 2003 State of the Union Addr ess, President George W. Bush pledged to pursue alternative energy technol ogies such as hydrogen fuel cells. The Department of Energy developed a goa l to “Develop and demonstrate fuel cell power system technologies for tr ansportation, stationary and portable applications” [1] The following ob jectives were outlined as par t of this initiative[1]: • By 2010, develop a 60% peak-efficient, durable, direct hydrogen fuel cell power system for transportation at a cost of $45/kW; by 2015, at a cost of $30/kW. • By 2010, develop a distributed generati on PEM fuel cell system operating on natural gas or LPG that achieves 40% electrical efficiency and 40,000 hours durability at $400-$750/kW. • By 2010, develop a fuel cell system for consumer electronics with (<50 W) an energy density of 1,000 Wh/L. • By 2010, develop a fuel cell system fo r auxiliary power units (3-30 kW) with a specific power of 100 W/kg. In order for fuel cells to be a viabl e option for the transportation industry, the Department of Energy requires not onl y durable, but reliable operation for at least a 5000 hours (150,000 mile equivalent ) lifespan under heavy loading, and a power density of 100 W/L[2]. In or der for a fuel cell system to be termed reliable, the cell must be able to operate e fficiently in sub zero climates. The Department of Energy determi ned that the following mini mal temperature criteria should be obtained [1]:

PAGE 15

2 • Operating range: –40C to 150C • Response time: in the –40–100C range <0.5 sec with 1.5% full-scale accuracy; in the 100–150C range, a response time <1 sec with 2% full-scale accuracy • Gas environment: high humidity reformer /partial oxidation gas: H2 30%–75%, CO2, N2, H2O, CO at 1–3 atm total pressure • Insensitive to flow velocity Many problems arise when a Polyme r Electrolyte Membrane (PEM) fuel cell is frozen, most of thes e are primarily caused by internal ice formation [2]. Internal ice crystallization causes physi cal tearing of the polymer membrane, clogging of internal fuel passages, dehydr ation of the membrane, and separation of the Gas Diffusion Layer (GDL) fr om the Membrane Electrode Assembly (MEA). This research was conducted to ex perimentally study effect of the subfreezing conditions down to -40C for periods up to twenty four hours at a time, and for up to twenty cycles. The ex perimental procedures and results are described in the following chapters. Chapter 2 provides the background, fundamentals and a review of the literature.

PAGE 16

3 Chapter 2: Background and literature review The first demonstration of the elec trochemical cell principle was produced by William Grove in 1839 using the experi ment illustrated below [3]. In figure 1, water is being electrolyzed into hydrogen and oxygen by facilitating an electric current around the circuit. If the power supply of figur e 1 is replaced with a light bulb, as in figure 2, the bulb would be di mly lit. The low intens ity of the bulb is due to the extremely small contact ar ea between the catalyst, hydrogen, and oxygen gases. In order to increase this current density the electrodes are commonly thin, flat, and porous. This design helps to facilitate the largest available contact area between the electr ode, the electrolyt e, and the gases. The chemical reactions that take place within the acidic cell are as follows: At the anode e H H 4 4 22, and at the cathode O H H e O2 22 4 4 The governing chemical equation for this reaction is then O H O H2 2 22 2 [3].

PAGE 17

4 Figure 1: Electro-chemical energy balance It cannot be assumed that because the system’s potential energy is lowered during the chemical reaction that the reaction proceeds at an acceptable rate to suit the desired application. T he amount of energy available to do work can be defined as the difference between the Gibb’s free energy (G) of formation of the products and the react ants. If there were no losses in the cell this value would represent the output of the cell. The change in Gibb’s free energy equals the charge of one electron times the number of moles of electrons released per one mole of hydrogen multiplied by Faraday’s constant (F) [3]. The theoretical electro-moti ve force is defined as: F G Ef2 Where F = 96,485 coulombs/mole. Due to the activation energy of the r eaction, the probability of a molecule having enough energy to overcome this barri er is relatively low. Therefore, in

PAGE 18

5 order to reduce the activation over-vol tage more effective catalysts, higher temperatures, larger electrode surf ace area, and incr easing reactant concentration can all be utilized. The fi rst two techniques are common to any chemical reaction, but the thir d is specific to fuel cells[ 3]. In order for the cell to operate correctly there must be a three phase contact on the electrode, so that an increase in the electrode area will proporti onally increase the reaction rate. Figure 2: Basic fuel cell operation A PEM fuel cell usually consists of an MEA, two graphite bipolar plates and two organic glass end plates. The graphite polar plates are used as collectors machined with parallel flow (f uel on one side, cooling on the other) channels. A silicon O-ring seals each fuel cell within the stack so the single cells

PAGE 19

6 operate in series. The operation of a fuel ce ll is similar to that of a battery except that the anode and cathode reac tants are constantly fed from a remote reservoir instead of an internal one. Of course, the “reactants” of a PEM fuel cell (usually hydrogen and oxygen) are quite different fr om that a of nickel cadmium battery, yet the principle is similar. A chemical reaction occurs in the MEA releasing an electron at the anode; this a llows the positive hydronium ion to pass through the semi-permeable membrane while the elec trons travel through and external loop eventually returning to the cathode. The el ectrons join with the positive hydrogen ion and oxygen molecules to form water. Water is essential to the operation of a PEM fuel cell because it facilitates the passage of hydrogen ions through the pol ymer electrolyte membrane. Figure 3 illustrates how critical water is to the operation of the cell, thus confirming that complete removal of water from a PEM f uel cell membrane is not possible. The complete dehydration of the membrane w ould not only limit the performance of the cell due to decreased proton conducti vity, but actually cause permanent deterioration. Since water is vital to t he cell’s operation alternate ways of limiting the ice formation must be considered. Fuel cells operate at a lower effici ency when the ambien t temperature is below 0C. There are many probable c auses for this performance loss such as decreased cell reaction rate, increased cell resistance, increased activation over potential, and possible mass diffusion over pot ential [4]. Once the ice formation begins, an additional inefficiency (primar ily due to the blocking of fuel from reaching the reaction sites) is now pres ent along with the other inefficiencies previously noted, thus rest ricting power production even further [5]. In P.E.M. fuel cells the reactant gases must be able to access the ca talytically active sites, protons and electrons must conduct thr ough the electrode, and the water product must be removed from the cathode to prev ent flooding of the gas diffusion layer [6].

PAGE 20

7 Figure 3: Standard fuel cell assembly Sulfonated tetrafluoroethylene copolymer history Nafion was developed in the 1960’s by Walther Grot of DuPont de Nemours, but the numerous other uses of organic ion exchange resins (i.e. a micro encapsulation film, the rmoplastic, and a coating) stirred research interest as early as the 1940’s [3]. One type of ion exchange membrane consists of a linear chain fluorocarbon with a small perce ntage of sulfonic and carboxylic acid groups. These side chains are where hy drogen ions can be partially or fully exchanged with a number of differing cations. Many differing configurations were explored, but Grot eventually settled on a polytetrafluorol backbone with polyvinyl ether pendant side groups terminated wit h a sulfonate ion group [7]. The chemical formula is:

PAGE 21

8 [(CF2-CF2)n-CF-CF2-]m l O-CF-CF2-O-CF2-SO-aM+ I CF3 The M+ can represent any number of count er ions such as H+, Li+, and Na+. The membrane contains two phases of components which are the fluorocarbon and ionic phases. These phases are separat ed by the covalent bonds that tie them together. The structure of the me mbrane is of a cluster with the aqueous ions imbedded in the continuous fluorocarbon phase. Each ionic phase region is connected to the other regions by sm all channels determining the transport properties[8]. Figures 4 and 5 illustrate t he layout of the ionic region of the membrane. General fabrication Since Nafion could be considered an industry standar d when dealing with PEM fuel cells a review of the fabrication process is beneficial. The base of the membrane is polyethylene. This polym er is modified by exchanging the hydrogen molecules with fluorine in a pr ocess called perflourination. The resulting polymer is polytetrafluoroethy lene (PTFE) also known as Teflon (a registered trade mark of IC I). Next, the PTFE is sulphonated by adding HSO3 side chains. Since these side chains are ionically bonded, the SO3ion resides at the end of this side chain produc ing an ionomer. The areas around these ionomers become highly hydrophilic wit hin a generally hydr ophobic structure, thus allowing for large quantities of wate r to be absorbed. In these hydrated

PAGE 22

9 regions the H+ ions are weakly bonded to the SO3groups and therefore are able to propagate [9]. Figure 4: Micro-sized ion passage illustration The amount of water present in the memb rane is determined by the type of polymer and the selection of the counter ion.

PAGE 23

10 Figure 5: Proton exchange layer model Nafion water balance Since the development of the Na fion membrane material there has been an intricate relationship between the me mbrane’s water content and the cell’s power output. Since the water facilit ates the diffusion of the hydrogen ion through the membrane it would be logica l that the more water in the cell the higher the efficiency. Experience has s hown that there is an optimal amount of hydration that the membrane should main tain. There are many theories to why over-hydration effects the power output, but the most widely accepted speculates that Gas Diffusion Layer on the cathode side of the cell becomes flooded with water [5, 9-13] .This would block the diffu sion of the oxygen to the reaction site within the membrane. This dictates that a delicate water balance must be maintained in order to efficiently operat e the cell. Evenly hydrating the membrane has been identified as a critical dev elopment area and has become the main focus of a current field research [10, 11] Without the proper hydration levels the ion transfer (ohmic) resistanc e of the cell increases, thus limiting power output.

PAGE 24

11 Research has shown that as the cell operates the anode tends to dry out while the cathode has the propensity to flood. A water balance of a standard PEM fuel cell has been outlined as follows: Water may enter the anode via the humid ified hydrogen gas or by diffusion through the Nafion membrane driven by a la rge concentration gradient. Water exits the anode through the anode exhaust outle t or osmotic drag, caused by the hydrogen ion – water affinity, through t he membrane. The cathode side has a similar balance except for the addition of water that is produced by the cells chemical reactionO H O H2 2 22 2 It is this extra water produced on the cathode side which is the logical source of the cathode’s tendency to flood, while on the other hand the osmotic force produced by the ion transfer through the cell tends to drag water molecules through the semi-permeable membrane causing the anode to dry out. It has been shown that there can be dry and flooded areas co-existing in the same electrode region [12]. These inter nal variations can be related to both the membrane and GDL design parameters such as chemical structure, composition, and graphite plate fuel c hannel design. Sundar esan et al. [12] found that membranes tend to be more hy drated near the inlet and drier near the exhaust. Therefore, optimized fuel patte rns must be utilized to evenly distribute the fuel gas over the electr ode and minimize over hydration in a localized area. With a basic understanding of the continuo us water cycle occurring within the cell, the ice formation can now be analyz ed during sub freezing conditions. Water balance in a sub zero climate In order to achieve the performanc e and longevity that the Department of Energy has defined, an understanding of t he destructive mechanisms must be

PAGE 25

12 obtained. In order to determine the water distribution within the cell, techniques such as neutron radiography, optical visualization, thermograph, and thermocouple investigations were performed [11]. Y. Ishikawa et al. developed a process using special mirrors, therma l imaging equipment and a peltier cooling element in order to simultaneously obtai n visual and thermal images of the operating cell. It was found that liquid water was still generated on the surface of cathode when the cell was operated below fr eezing. The water was initially super-cooled, only warming to zero degrees Celsius upon freezing. It can be seen from as the generated wa ter freezes the cell’s performance is negatively affected. Figure 6: Ice formation and subsequent power losses If the water could be kept in a supe r-cooled liquid state, the cell could be operated at sub-freezing conditions without negative degradation effects of ice formation [11]. It can also be noted that the freezing of the super cooled water droplets occurs once the small water dr oplets, approximately 10m in diameter,

PAGE 26

13 combine to form larger droplets of approx imately 100m in diameter. The cause of the transformation, from super-cooled liquid to ice, has been theorized by Shichiri et al. [14] to be the result of bursting air bubbles, but further investigation is needed to properly explain this phenomenon. According to Cappadonia and Saito [15, 16], the water located withi n the Nafion membrane partially freezes near -20 degrees C. They speculate that there is still super-cooled water still remaining un-crystallized even at -20 Celcius. Furthermore, experiments conducted by Ishikawa et al. [11] show that water behaves differently during freezing if it is located in the ion phas e of the membrane in stead of being part of the cell’s bulk water. Effects of ice formation Catalyst layer The volumetric expansion during the phase change of water can cause destructive tearing of the porous catalyst layers [4]. Guo et al. [17] found that some areas of the catalyst were torn away from the membrane. This was confirmed after these areas showed no El ectron Dispersion Spectroscopy signals which means they were initially beneath the surface layer. Gou [17] inferred this type of destruction could cause delaminat ion of the catalyst layer from the membrane or the GDL, theref ore increasing the contact resistance. According to Wannabe et al. [18] these membranes ar e physically made up of two distinct pore size distributions. The small pores (called primary pores sized from .01nm to 1micrometer) are located within t he primary particles, thus forming agglomerates (approximately one micrometer), while the secondary pores were located among the agglomerates. The micros tructure of this catalyst layer can be viewed as carbon particles flooded with the electrolyte which will form agglomerates. The reactant gases diffu se through the ionomer thin film, the

PAGE 27

14 electrolyte, and then through the agglomerates to reac h the reaction site. Therefore, since there is a porous electrode and a hydrated ionomer layer; the catalyst layer can only hold a certain amount of water. As the amount of residual water increases the pores are filled in the following order: small hydrophilic pores, large hydrophilic pores, large hydrophobic pores, and then the small hydrophobic pores. Once the water freezes the volumetric expansion of the ice enlarges the pores between the agglomer ates and thus compresses the pores within the agglomerates. This sequence becomes evident after a few freeze thaw cycles because the number of sma ll pores (<20nm) decreases while the number of larger pores (>20nm) increases [19]. Figure 7 illustrates the flooded agglomerates (a) before and (b) after t he freeze degradation. Experiments conducted by Kim et al. suggest ed that the catalyst layer destruction is related to irregular pressure uniformity of the diffusion media upon the catalyst layer. It has been postulated that extra stress is in curred by the stress concentrations experienced at the channels and lands(raised area) that bipolar plates impose on membrane [20]. An example of these channels and lands can been seen in figure 3. Physical damage was observed to occur in regions directly in contact with these channel regions, while the regions located under the lands were not destroyed due to the stabilizing pressure provided by the land. Kim and Mench [20] proposed that a stiffer gas diffusion layer would help mitigate these stress concentrations during freezing. Although it was found that less physical damage resulted from flow patterns with hi gh land/channel ratios, there was still performance losses experienced that coul d not be corrected unless all the water was removed from the membrane prior to freezing [20].

PAGE 28

15 Figure 7: Ice formation ef fects on agglomerate pores Hou [21] found that there was significant destruction of the catalyst layer’s morphology which can be directly correla ted to the amount of residual water contained in the membrane. The memb ranes investigated included water amounts of .9 mg/cm2 with no significant performance loss at 1 A/cm2, but when the residual water weight was incr eased to 3.6 mg/cm2 the performance degradation was about 16% when operated at the same temperature and pressure. Hou [19] confirmed the assump tions made by Watanabe [17] by considering the increase in resistance as seen in the Tafel slope after sub-zero climate exposure. Hou observed that at higher current densities the Tafel slope became steeper indicating that there is an agglomerate diffusion effect within the catalyst layer. The Tafel slope increases after the freezing process especially for membranes with a large amount of residual water content. This suggests that the pores in the agglomerates have been compressed, so the agglomerate diffusion becomes greater thus increasing the Tafel slope. This experiment elucidates the fact that vo lumetric expansion that occu rs within the agglomerates can be neglected when compared to t he expansion of t he pores among the agglomerates.

PAGE 29

16 By measuring the amount of nitrogen absorption with a BET surface area analysis, the number of pores with a size greater then 25nm increased while the number of smaller sized pores decreased [9]. Cho et al. [22] found by means of cyclic voltammetry that a significant loss of electrochemical active surface was the result of the volumetric expansion. Ice can form on the gas diffusion layer blocking the reactants from reaching the membrane [23]. This result was confirmed by repetitive fr eeze/ thaw cycles on both a hydrated and de-hydrated membranes [24]. Zhang et al. [24] f ound that the current density at .6 volts dropped from 570 2cm mA to 420 2cm mA after nine freeze/thaw cycles. In comparison to the de-hydrated membrane which had almost indistinguishable polarization curves before and after the ni ne freeze cycles. It is known that the density of water and ice at 0C are .9998 2cm g and .9168 2cm g respectively, so the effective volume of ice can increas e as much as 9%. This volumetric increase induces a mechanical stress in the hydrated regions causing the porous structures to be altered. These mechani cal stresses account for most of the irreversible degradation of the membrane. Gas diffusion layer The gas diffusion layer serves several essential functions such as reactant transport, water transport, electronic c onductivity, thermal conductivity, and mechanical support. It is obvious that the G. D.L. has a critical role in facilitating maximum power output. The gas diffusion layer is typically constructed of carbon fiber paper treated with a hydr ophobic polymer coating such as polyteterfluoroethylene (P.T.F.E.). This c oating allows for efficient water removal and helps to prevent flooding at higher curr ent densities. Gas diffusion layers are often treated with a micro porous layer (M.P.L.) on the catalyst side to help ensure sufficient water transport and elec trical contact [25 ]. Some recent publications have focused on utilizing ex -situ methods to quantify the G.D.L.’s

PAGE 30

17 durability. A decrease in sessile drop c ontact angle was reported by Wood et al. [26] when the GDL was submersed in deionized water at 60C and 80C. This process of submersing the GDL in cons tant temperature solution is known as aging and will be referenced later. Wood observed a decrease in hydrophobicity within in the first 100 – 150 hours of s ubmersion, and when the experiment was repeated in air the degradation was found to be even greater during the same duration of time. A possible cause of this degradation was not discussed and could be attributed to various processe s. Another study [27] conducted examined the weight loss and MPL contac t angle of the GDL while being aged in a hydrogen peroxide solution at 82C. Kangasniemi [28] found that both weight loss and contact angle increased with prolon ged exposure time. This effect has been attributed to the oxidation of car bon within the MPL. The performance of the altered G.D.L.s was evaluated by inse rtion in to control M.E.A.s. The MEA performance showed increased mass trans port losses, and was in agreement with results given for long term fuel cell te sts conducted. The postulate that the MPL surface oxidation and the accompany ing decrease in the sessile drop contact angle could affect the water tr ansport and therefore the mass transport has also been investigated and confirmed by Williams et al. [29]. Lee et al. [30] investigated the effect of freezing condi tions on the G.D.L.’s in-plane electrical resistivity, bending stiffne ss, plate-side surface cont act angle, catalyst side surface, contact angle porosity, water v apor diffusion, In-plane air permeability, and through-plane air permeability. The only parameter which experienced any sort of degradation was air permeability of the membrane which increased by 18% and 80% (depending on if the G.D.L. wa s aged before freezing). The dry un-aged GDL did not exhibit any increase in permeability, thus concluding that this degradation is caused by the volume tric expansion experienced during the water/ice phase change. The dry aged G.D .L. (w/ 18% permeability increase) has been theorized to be the result of M.P. L. material loss and not alteration of the pore structure during the ageing process.

PAGE 31

18 Mitigation strategies Before fuel cells will be able to compete with internal combustion technology for market viabi lity, cold temperature st art up of the cell must be accomplished quickly and efficiently. Some of the problems fa ced during a cold start include slower chemical reacti on rates, poor humid ification of the membrane, and ice blockage of fuel lines. These problems have been addressed by using both passive and acti ve heating schemes. A passive heating or start up scheme uses the irreversible heat produced by the chemical reactions within the fuel cell to raise the cell to the optimum temperature. An active heating scheme uses a secondary heat source (i.e. resistive or waste heat recovery) to quickly raise the fuel cell to the optimum operation temperature. An example of such a scheme would be to use the cool ing loop of cell for heating during cold startup. It has also been proposed that the catalytic combustion of hydrogen/oxygen mixtures could also be a fo rm of internal active heating [4, 31, and 32]. Oszcipok et al. found that when t he efficiency of the passive and active start up processes are compar ed; the active start up met hod is at least twice as efficient. The active start up scheme (f rom -20C to 0C) uses 4.5 Watt hours (includes an 85% battery load efficien cy and 52% fuel cell efficiency) as compared to 10 Watt hours required with the passive scheme [4]. Another mitigation technique involves the reducti on of the membrane’s water content in attempt to minimize the internal ice formation. In order to minimize the effects, many differing start up and shut down strategies have been developed. The most common type of shut down strategy is to purge the cell with an inert gas (C02 or N2) so as to dehumidify the cell and minimize the cell’s bulk water content before freezing [4]. A PEM fuel cell membrane fabricated from Nafion-117 was found to operat e at temperatures as low as -35C at the Indian Sc ientific Research Station in Antarctica [17]. After the experiments were completed t he membrane was purged with nitrogen gas before shut down in order to avoid t he devastating effects of internal ice

PAGE 32

19 formation. While these experiments st ill had setbacks, as the cooling unit and humidification bubblers had a tendency to fr eeze, the successful operation of the fuel cells gives hope for the future for this technolog y. A variation of this principle involves reducing the humidity of the fuel gases (H2 and O2) between 50-60% in order to dry out the cell slowly to prevent overdrying [9]. Hou et al. conducted seve ral cold start experiments at -5C and found while the cell did not have any problems upon startup there was a decrease in performance observed after ea ch freeze/thaw cycle was completed. By using several membranes with differing amounts of hydration, Hou et al. were able to determine the effect that wate r has upon the performance. They found that as the water content increased, so did the amount of degradation seen after each cycle. Hou et al. also completed investigations concerning the physical structure of the membrane by com parison of SEM photographs [13]. Another variation of the concept presented above would be to manufacture thinner membranes which would be able to operate properly with dry reactant gases. Since the membrane is thinner, the back diffusion resistance is decreased, thus allowing more wate r to propagate across the membrane for the same concentration gradient. Vengates an et al. [33] found these thinner membranes (~25m), increased cell’s perfo rmance values at lower temperatures (from 35C 50C), but experienced ex cessive performance losses at higher operating temperatures gr eater then 60C due to t he excessive drying of membrane and decreased prot on conductivity [33]. Another strategy was to use an ant i-freeze solution like methanol to prevent ice formation. Palan et al. found that the anti-freeze solution reduced the formation temperature, but did not reduc e the amount of freezable water in the membrane [34]. Further studies conclude that the amount of freezable water actually increased from 23% to 63% after the methanol addition. This phenomenon can be attributed to the fact that super cooled methanol could

PAGE 33

20 replace water in the solvation of the ioni c groups, thus driving more water out of the polymer and into the bulk volume [4]. Before fuel cells will be able to compete with internal combustion technology for market viability, cold temperature start up of the cell must not only be accomplished without performance degradation, but also quickly and efficiently. Some of the problems fa ced during a cold start include slower chemical reaction rate (caused by lower te mperatures), poor hum idification of the membrane, and ice blockage of fuel lines. These problems have been addressed by using both passive and acti ve heating schemes. A passive heating or start up scheme uses the irreversible heat produced by the chemical reactions within the fuel cell to raise the cell to the optimum temperature so that maximum power output can be obtained. The proce ss of the catalytic combustion of hydrogen/oxygen mixtures could also be a form of internal heating [4]. This mitigation technique will be further illustrated in the following Patented Techniques section. An active heating scheme uses a sec ondary heat source (i.e. resistive or waste heat recovery) to quickly raise t he fuel cell to the optimum operation temperature. An example of such a sc heme would be to use the cooling loop of the cell for heating during cold startup. Wang et al. in vestigated another type of active startup scheme which utilizes an external current source during subfreezing startup. This technique transfo rms the fuel cell into a hydrogen pump, which not only helps to clear the pores of water, but actually helps increase the amount of reactant gases at the cata lytic regions. Since one of the major problems with subzero startup is the re striction of fuel from reaching these catalytic sites, this technique showed a cell could warm from -15C to 0C within 100 seconds [35]. Oszcipok et al. found t hat when the efficiency of the passive vs. active start up processes is compared t he active start up method is more than twice as efficient. The active start up scheme (from -20C to 0C) uses 4.5 Watt

PAGE 34

21 hours (includes an 85% battery load effici ency and 52% fuel cell efficiency) as compared to 10 Watt hours required with the passive scheme [4]. Patented mitigation techniques Although there are still many questi ons regarding sub-zero operation and the inherent degradation of fuel cells, there has been some progress in this area. United States patents are a preview of cutting edge technologies, and techniques soon to be employed. The following patents are focused upon solving the inherent problems associated with sub-ze ro storage and operation, although this is not a comprehensive list of available technologies a conscientious effort was made to include one patent from each mitigation type surveyed. Patent number 6,777,115 describes a ba ttery assisted fuel cell which effectively uses a battery or a capacitor wir ed in series with the cell to aid in the heating of the cell [31]. According to the invention, the additional current provided by the battery forces the weak ce lls in the fuel cell stack to a negative cell voltage which produces heat as a consequence of polarizations within the cell. The performance of the weak cells quickly approaches the typical performance of good cells. Furthermore, while the battery is connected in series with the fuel cell, the excess fuel, whic h may be hydrogen or hydrogen-containing fuel, is supplied to the anodes of the fuel cell stack to further support the stack’s efficient startup. Patent number 6,103,410 illus trates how dilute hydrogen and air mixtures can be introduced into one of the electrodes in order to react with the noble metal catalyst to produce heat at sub-flame temp eratures [32]. An embodiment of this concept includes a catalyst structure to be located in between the cathode and oxidant channels. The invention further explains that if the cell is not

PAGE 35

22 hydrophobic enough to prevent the freezing ov er of the pores, wh ich facilitate the transportation of fuel to the catalyst regi on, that larger hydrophobic pores can be designed into the M.E.A. in order to ensure unrestrict ed flow of the hydrogen/air mixture. Patent number 6,797,421 describes a pr ocess which uses the available hydrogen to fuel an external catalytic burner in order to keep the fuel cell and the encapsulation device above freezing temperat ures [36]. This invention describes a fuel cell climate chamber which would regu late the cells internal temperature to protect from ice formation. This conc ept, further investigated by Sun et al., proved that the catalytic hydrogen reacti on is an effective way to heat a frozen fuel cell. Sun proposes that the fuel channels are small enough to operate similar to micro-channel reactors, thus confirming that a lean mixture will not explode when hydrogen concent rations are kept below 20 volume percent [37]. Sun reports the catalytic hydrogen does not have any effects on the performance of the cell according to cyclic volt ammetry measurements and polarization curves. P.E.M. fuel cell membrane materials One plausible solution to the free ze degradation problem is to develop a material which can withstand the stresses and strains imposed by the formation of the ice crystals. There are many curr ent endeavors which seek to mitigate this problem. Both organic and inorganic hy brid membranes have been considered and investigated. An example would be membranes manufactured from materials such as biphenyl sulfone [ 38] and polyphenylene sulfone copolymers [39]. These polymers may provide t he delicate balance between conductivity, thermal and chemical stability, and mechanica l integrity to make a stronger, more resilient membrane. Great strides have been made in the ion conduction ability

PAGE 36

23 of these membranes, but further devel opments in mechanical strength and chemical stability are still desired [2]. Vengatesan et al. developed a compos ite membrane which has the ability to retain water allowing for non humidifi ed operation. This is accomplished by adding inorganic particles to the membr ane such as SiO2, TiO2, and ZrO2 which keep the membrane well hydrated [40]. T he actual composite material added to the membrane and electrodes is silica. Vengatesan found that low equivalent weight ionomers exhibited a greater ab ility to retain water then their heavier counter parts. The results of the init ial performance evaluation showed that cell actually operated at a higher efficiency than the normal Nafion membrane. This is due to the fact that thin membranes allo wed more water to diffuse to the anode side of the fuel cell, thus keeping the cell more evenly hydrated. This membrane may be able to withstand the effects of freeze degradation because of its low bulk water content which would minimize t he effects of the internal ice formation. Di Vona et al. [41] investigated the addi tion of the same inorganic fillers for the membrane’s improvement. They descr ibe similar results as Vengatesan et al., but concluded that there would eventually be porosity problems (i.e. increased membrane ohmic resistance) due to the delamination of the catalyst and GDL layers [41]. Other avenues cu rrently being investigated include the addition of acids such as heteropoly acid s, but they tend to leach out during operation. Polymer or organic-inorganic hybrid blends ar e another plausible solution because these materials allo w for advanced control over material properties at the molecular level, thus strengthening the membrane while still maintaining low ion conductivity. In organic/inorganic polymers, car bon forms the base chain and the inorganics comprise the side gr oups. These polymers c ould allow for the precise control of the number of hydrophilic and hydrophobic regions which would optimize proton conductivity, material st rength, and morphological stability [36].

PAGE 37

24 One of these materials is called pol yetheretherketone (PEEK), which when compared to Nafion is less acidic, has a less hydrophobic backbone, allows for wide separation of the SO3H groups, la rger number of na rrow channels, and a lower humidity dependence. DI vona concluded after completing a PEEK – Nafion performance comparison that PEEK improved the membrane’s mechanical properties, but still had a s lightly higher resistance to proton transport. Sgreccia and Licoccia [42, 43] found that in order to achieve improvement required. As a result of the sulfonation the mechani cal properties were degraded because the acid groups attached to the hydrophobic backbone cannot arrange in separate phase domains (as illustrated in figure 3) as well as the perfluorinated systems with hydrophobic side chains [44] Since the membrane’s mechanical properties have deteriorated the membr ane tends to swell and eventually become water soluble and useless as a PEM membrane [45]. A proposed mitigation technique was investigated by using siliated polymers which promise to enhance thermal stability, conductivity, and minimize membrane swelling [42, 43] Sgreccia et al. suggested that the hardening of the polymer could be attributed to the large phenylsilanol side chains. These side chains would fortify the S-PEEK polymer chains during shearing [42]. Anot her postulation presented by Sgreccia is that the in crease of polymer strength co uld be related to hydrogen bond interactions between the silanol and su lfonate groups. Both investigations of siliated S-PEEK concl ude that there must be a delicate balance between the membrane’s mechanical integrity and impedanc e which is dictated by the amount of membrane hydration. Another interesting approach to so lve the hydration problem was outlined by Ho Jung [46]. SiO2 was added to the anode catalyst layer in attempt to improve the water management. These m odifications were conducted on both naturally aspirated (free br eathing fuel cell) and the st andard forced air cell. Test results showed that both cells which had SiO2 added to the cathode had

PAGE 38

25 improved performance through th e entire current range [46]. The air blowing cell showed a 26.9% increase in performance, while the natural air breathing system had a 44.4% increase. This performanc e increase has been attributed to the increased hydrophilic properties of the anode caused by t he SiO2 addition. The membrane is not the only fuel ce ll component which is in need of a water management scheme. The Gas Diffusion Layer (GDL) and the Micro Porous Layer (MPL) have been known to become flooded, therefore restricting the flow of fuel to the reaction sites. A solution proposed by Na kajima et al. [47] involves the treatment of these two structures with hy drophobic PTFE. It was found that the hydrophobic treatment water accumulates at the electrode and in the flow channel, while decreas ing the amount of water present in the GDL layer. Nakajima et al. proposed that the decr ease of the performance was due to the accumulated water blocking the transport of oxygen to the electrode [47]. In this case, the hydrophobic treatment of GDL laye r simply moved the oxygen transport (i.e. cathodic flooding) problem to another area of the cell. The last method of membrane alteration involves the dispersion of multiwalled carbon nano-tubes throughout the Na fion membrane [43]. This type of modification has been proposed to aid in the reduction of methanol cross-over in Direct Methanol Fuel Cells (DMFC), but also could improve in the membrane’s strength and ion conductivity at low humidit y levels. Thomassin et al. found that with the addition of the carbon nano-t ubes the methanol permeability was significantly decreased (by approximately 60%), while the Young’s modulus of the Nafion was increased 160%. The increase of the Young’s modulus could help to minimize the membrane degradation that occurs after the internal ice crystallizes. The meshing of Nafi on and carbon nano-tube technology has shown promise, but the full potential of this hy bridization has yet to be realized.

PAGE 39

26 Finite element modeling simulation Thermal management plays an important role in the optimization of sub zero storage and operation of the cell. Modeling approximations provide an opportunity to gain insight about comple x processes occurring within the cell. Mathematical models use f undamental equations to simu late the operation of a fuel cell. This allows for investigati on of key processes (i.e. electro-chemical reactions in the catalyst layer, pr oton mitigation in the semi-permeable membrane, and mass transport in all regi ons) to be explored and optimized. The thermal, pressure, and humidity maps gene rated greatly aid design engineer’s ability to understand and improve on the curr ent technology. The first fuel cell models were adapted from the battery i ndustry models. These models tended to be one dimensional and inaccurate when compared with the actual results [4851]. Most basic modeling techniques consi der the fuel cell stack as a lumped parameter heat transfer probl em which uses the exiti ng coolant temperature to approximate the stack temper ature. In order to mo re accurately model the internal temperature gradi ent during the transient st art up period, a layered experimental model was proposed [11] The experimental cell would have thermocouples throughout a layered memb rane in order to get a better idea of the temperature gradient. As a result of this experiment Sundaresan et al. recommends using an active heating scheme, circulating the cooling loop during warm up, minimizing the thermal mass of the cooling loop heating of the end plates, and using metal based bipolar plates to facilitate the su ccessful startup of a cell that has been frozen. The cell’s MEA tends to dry out in some areas, while flooded in others suggesting that a one dimens ional lumped capacitance model will not provide enough resolution to reveal the condition (temperature and wate r content) of the MEA. Siegel et al. suggested using a tw o dimensional model that includes the

PAGE 40

27 transport of the water, gaseous species, protons, energy, and dissolved water in the membrane. The goal of the model is to describe the operating regimes dominated by mass transport limitations resulting from the transport and formation of water. Siegel stresses that in order to model a membrane, the physical properties of the M EA (porosity of the gas diffusion and catalyst layers, ionic conductivity of the membrane, mass transfer coefficients, reaction surface area, and structural properties) must be accurately measured. Once the model’s results were fitted to actual test data the performance limitations can then be isolated and quant ified [48]. An extension of this concept was explored by references [49 and 50] where a three dimensional analysis of the transport and electrochemical reactions were completed. Most thr ee dimensional models incorporate some type of computational fluid dynamic (C FD) code, thus allowing for parallel computing making high resolution steps pr actical. Most CFD models account for the convective and diffusive transport allowing for predictions of species concentrations. Since the entire membr ane structure can be investigated locally, the heat load provided by the electr ochemical reaction can be determined accurately. The model developed by Sive rtsen et al. solves for the ionic and electric potentials in the electrodes and the membrane, as well as the local activation over-potential in order to predict local current densities. Sivertsen and Djilali [50] concluded that the changi ng relative influence of ohmic losses and activation over-potential losses drastically alters the current distribution within the model, thus confirming that the local activation potential needs to be considered during modeling efforts. Since the m odel utilizes three dimensional governing equations, the mass transport limitation of di ffering flow fields can be evaluated as well [45]. Um and Wang [ 49] found that at high curre nt densities (greater than .8 A/cm2), the interdiginated flow field (similar to that of figur e 8) allowed local current densities to increase by 40%.

PAGE 41

28 Figure 8: Interdiginated flow patterns By utilizing one of these three tec hniques of approximating a cell’s operation, authors [51-53] each focused on a different aspect of the membrane in order to elucidate the processes occurring in that specific regi on. For example, Mishra et al. [51] concentrated on det ermining the optimal design and operating conditions based on systematic parametr ic avenues. The design parameters considered for the optimiz ation include the cell’s pot ential, power density, maximum temperature rise, and minimum hydrat ion. The results of the this study conclude that operating temperature, t he anode’s relative humidity and pressure are critical parameter s that must be adjus ted to obtain optimum operation [46]. Kulikovsky [52] sought to determine expressions for the membrane’s resistance and voltage loss due to the incomp lete humidification of the cell. The author determined that both of these parameters depend on the ratio of the mass transport coefficients of the water in the membrane and the GDL backing layer. The simulation accounts for this ratio when determining the effect of the level of hydration upon the membrane resistance and voltage loss. It was determined that if this ratio was le ss than one, then there exists an optimal current density which minimizes the membr ane’s resistance [52]. This same concept can be exten ded to describe the dependence of the cathode’s potential on current density. Hs uen [53] validates the one dimensional model approach by accurately determining a cell’s performance based on the oxygen depletion within the cathode. Hsuen’s investigati ons prove that all one,

PAGE 42

29 two and three dimensional models, can accu rately estimate the operation of the cell as long as the complexity of the in vestigation is paired with the appropriate model [53]. In order to simulate the effects of sub-zero start and operation, Sundaresan and Moore state t hat models must have the abi lity to map internal heating methods and the effe ct that the endplate the rmal mass has upon the end cells [12]. Although current models do account for the internal and external heating methods, the lumped parameter assumption does not adequately describe the localized effects of these heating strategies. A major evaluation criterion of a heating schem e should include observation of the localized stack temperatures and water content. In addition, the effect that the thermal mass of the endplate has upon the stack’s temper ature distribution cannot be fully described without a layered model. Sundaresan proposed the use of a layered one dimensional model to quickly and accurately describe a cell’s sub-zero operation characteristics. Furthermo re, Sundarsen’s laye red one-dimensional model is of interest since the goal of the model is to evaluate systems for certification of the 2010 D epartment of Energy’s perfo rmance criterion. The model has been designed using Matlab/Simulink software by conducting energy and mass balance equations for each layer of the model. Each layer encompasses all of the pertinent cell processes including: sensible energy flows for the coolant, anode, and cathode gases, water generation and phase changes (vapor, liquid, and ice) stack heat loss to environment, and internal heat generation re sulting from electrochem ical reactions and ohmic resistances. The primary result of thes e considerations is a midpoint temperature of the individual layer as a function of time. The pertinent regions that should be included in the sub-zero model include: the bipolar plates, gas diffusion layers, catalyst support, membrane, and end plate assembly.

PAGE 43

30 Preliminary results from the layered model simulating differing storage and startup procedures have shown an increas ed accuracy and ability to predict the effectiveness of a mitigation strategy. For example, during cold storage applications under 12 hours in length at an ambient temperatur e of -20C, the model, described in [12], predicted that it is energetically favorable to prevent ice formation with resistance heater s (~ .05 W per cell inter nal heat input and 10 W input at the endplates) instead of trying to thaw out the fr ozen membrane. In order to correctly model the cold start up situation each cell is assumed to have the same ice mass as the coldest cell in the stack (providing the worst case scenario). Both models clearly show t hat current mitigation strategies do not provide a viable solution to satisfy the Department of En ergy’s 2010 goals.

PAGE 44

31 Chapter 3: Experimental f acility and testing procedures The test bed was designed to provi de the user with control of the parameters deemed necessary to obtain sati sfactory experimental results. The following functions were determined to be e ssential for the operation fuel cell test station: control/monitoring of the ambient temperature, fu el cell temperature, fuel line humidity, contamination, mass fl ow, and pressure. An automated data acquisition system, a de-ionized water s upply, drain for water output, and a power supply to drive all the dat a acquisition and control elements. These parameters include thermocouple re adings of the fuel cell’s anode, cathode, hydrogen fuel line, air fuel line, ambient air temper ature, cryogenic freezer temperature and the temperature of the respec tive gas inside of each humidification bubbler. Other sensors incorporated into this test bed include exhaust gas pressure sensors fo r both anode and cathode lines, mass flow controllers for both intake lines, humidity sensors for the intake lines, and a fuel cell current and voltage detector. In order to stabilize and control such variables as fuel cell temperature, hydrogen/air stream humidity and te mperature active heating elements were installed. These elements are automatic ally controlled by one or multiple feedback devices described above.

PAGE 45

32 Test bed configuration phase: build 1 Purpose The goal of the first build of the fuel cell test bed was to design and implement a testing center that could re cord important baseli ne data that would be compared to the performance data in or der to determine the extent of cell degradation. Overview This phase of the project is analo gous to the prototyping phase of any engineering design. The goal of this bu ild was to explore how well the Labview expansion data acquisition cards could c ontrol and acquire the data required to produce the baseline and experimental resu lts. The build 1 test bed design included all of the basic fuel cell system components noted previously. Ideally, the implementation of build 2 would only include the addition of a cryogenic climate chamber used to house the f uel cell while environmental stressing occurs. Of course, nothing is perfect and unfortunately other modifications were needed to optimize the cell’s performance while improving the over-all system error. Figures 8, 9, and 10 show build 1’s control box and test bed layout.

PAGE 46

33 Figure 9: Fuel cell test bed build 1 Figure 10: Data acquisition build 1

PAGE 47

34 A schematic of the test bed lay out has been included in Figure 11. Figure 11: Physical layout of build 1 Data acquisition The data acquisition system for the f uel cell was developed by Patricia Alexander in the fall of 2005 at the Univ ersity of Florida. The data acquisition system has been designed using the National Instrument’s Lab view program along with I/O Tech’s Daq Book 200, a 16 bi t parallel port data acquisition device. The DaqBook has been fitted with two expansion cards I/O Tech’s DBK/15 general voltage card and DBK/19 a thermoco uple acquisition card. This system was designed to acquire all of the required fuel cell related data as well as to

PAGE 48

35 control the hydrogen and air line temperatures, mass flow rates, humidity, fuel cell temperature, and exhaus t line back pressure. This allows the user to simultaneously observe, record, control and manipulate the experimental data in order to obtain the most accurate results. A complete list of all data acquisition devices has been placed in appendix B. Figu re 12 below is a schematic of the data acquisition system: Figure 12: Test bed data acquisition schematic

PAGE 49

36 Test bed configuration phase: build 2 Purpose The approach of build 2 was to inco rporate a climate chamber into the current test bed configuration and impl ement the required design upgrades as determined necessary. Refinements such as new line heaters, insulation, and gas path routing were integrated into t he test bed before cold startup/ operative testing commenced. Overview The second build of the test bed was focused on the design and integration of an environm ent chamber into build 1’s layout. This was accomplished by surveying the current te chnologies available to satisfy the -40 degrees Celsius temperature requiremen t, and choosing the most practical system. The two cooling methods explored were mechanical (2-stage compressor system) or direct gas expans ion option. First, the mechanical avenue will be briefly discussed in order to provide a better understanding of the final decision. A compound refrigerati on system has many advantages such as minimal long duration operating costs, pr ecision temperature control and direct factory implementation. Wh ile these points are valid fo r consideration, it was found that a cryogenic evapor ation system was the much cheaper option. There was an approximate $3,500 cost differ ential between the two systems with only a few drawbacks such as less precise temper ature control and a longer fabrication time. The cryogenic gas expansion system was chosen because it had a lower initial setup cost and the design was simp le enough to fabricate in the lab.

PAGE 50

37 Cryogenic expansion climate chamber Many design considerations were considered before deciding on the current configuration of t he cryogenic liquid ex pansion climate c hamber such as external and internal heat loads from the fuel cell and the surrounding environment, size considerations (the fuel cell’s dimensions are 5 cm x 5 cm x 2cm), chamber door leakage, porting leaka ge (due to ported holes in the side of the chamber to pass electrical wiring and fuel intake/exhaust lines in to the chamber), and heat exchanger design. In order to minimize the cooling lo ad the chamber was designed to be as small and thermally insulated as possible. This goal was accomplished by a generous donation from A&W Re frigeration which included all of the building and insulation materials needed to complete the chamber’s infrastructure. The chamber was constructed of 4 inch thick walk-in freezer panels with a thermal Rvalue of approximately 12-14. These panels were cut in order to provide 1 cu-ft of volume within the chamber. Each panel edge was cut at a 45 degree angle to minimize air infiltration and to ensure even application of the expanding thermal foam. The exterior panel joints we re stabilized with extruded aluminum siding and sealed using Hilti CF 116-14 #314722 ex panding foam. This same foam was also used to seal the ported intake/e xhaust holes. Next the heat exchanger was constructed of inch diameter copper tubing formed into a shape similar to that of figure 13.

PAGE 51

38 Figure 13: Climate chamber heat exchanger Liquid nitrogen was chosen as the ex pandable refrigerant because it was the cheapest cryogenic gas t hat would fulfill the c hamber’s low temperature requirement. A Jefferson Valves cryogenic solenoid, catalog number YC1390BT2UCT, was chosen to control t he flow of liquid nitrogen as determined by the PID temperature cont roller. Due to the high ve locity of the evaporating gas, a needle valve was installed on the ex haust side of the heat exchanger to maximize the amount of time available for heat transfer to occur. This controller relies on the feedback from a thermocouple inside the chamber to determine the liquid nitrogen flow rate. Figure 14 illus trates the basic construction of the chamber.

PAGE 52

39 Figure 14: C.A.D. representation of the climate chamber

PAGE 53

40 Figure 15: Completed build 2 test bed

PAGE 54

41 Design upgrades In order for the build 2 goals to be accomplished, other modifications to the test bed were requir ed as outlined below: The hydrogen and air line heaters and insu lation needed replacement as they had become unshielded and a fire hazard. The fuel line thermocouples had to be relocated in order to accurately determine the relative humidity in the fuel line. The humidity sensor’s ori entation was adjusted so t hat water would no longer collect on the probe causing it to short circuit. After the calibration testing of the build 2 design was completed, three process controllers were identified as being unsuited for their designed task and needed to be replaced. The issues encountered with the CN800 and the CNi32 controllers were due to the lack of ability to control the Proportional, Integral, and Damping (P.I.D.) parameters. These controllers were replaced with Omega model CN9000A which provi ded greater P.I.D. control and adjustment. Although these controllers functioned properly they still were not the perfect solution because the controllers’ manufacturers di d not provide the same adjustments for both output channels. This limitation disa bled the derivative polling ratio which automatically adjusts the pr oportional band during large se t point deviations, thus minimizing overshoot. This overshoot was minimized manually on the climate chamber by closing the back pressure valve on the heat exchanger to further restrict or even stop the flow until the chamber’s internal temperature could equalize. This overshoot does not pose as much of a problem for the line controllers since it only affects the calcul ated humidity values and not the actual line humidity. The impact of the fuel line temperature (i.e. long start-up time) can be minimized by not allowing any gas to flow through the lines until they have reached the set point temperature.

PAGE 55

42 Testing procedures Baseline tests The baseline test represents the cont rol data to which the future cold temperature investigations wi ll be compared. Before t he fuel cell is exposed to the sub zero temperatures, initial power curves are generated from the recorded data at standard temperatur e and pressure and at 30 psig fuel line pressure. These power or “polarization” curves will be used to determi ne the amount of cell degradation experienced during sub zero stor age. The loss of efficiency will be quantified by the differenc e between the experimental and control polarization curves. For the baseline test each membra ne was allowed ample time to hydrate and reach full power output before the bas eline data will be acquired. The data was recorded as the cell powers down from this maximum level at ten second intervals. In order to accurately repres ent the polarization cu rve, the entire power band of the fuel cell was explored. This was accomplished by starting with an open circuit and varying the load from 10 to .01 ohms. This process was repeated a minimum of three times in or der to insure proper MEA break-in and hydration.

PAGE 56

43 Cold storage investigations In order for the degradation effects to be investigated the test parameters must first be outlined. The cold storage experiments will be evaluating the performance loss caused by internal ice fo rmation resulting from low temperature storage. The following procedure was follo wed for each test in order to ensure repeatability and test integrity. First, the cell was “broken in” according to the manufacturer’s specifications and a base line test was completed as outlined above. Next, one of the three variable parameters was chosen to be explored and a goal was defined. For example, to determine the effects that the hydration state prior to freezing has upon the perfo rmance, the MEA was subjected to a dehydration period in which t he fuel gases were humidi fied to the experimental value and allowed to flow through the cell at 250ml/min for ten mi nutes. After the dehydration period, the ce ll was properly shut down, and the climate chamber was properly adjusted for the correct subz ero temperature. The cell was allowed enough time to reach equilibrium with the climate chamber before the freeze/thaw time was recorded. The co ld storage tests are unique in that they focus on the shutdown regimes and their effect on the cell’s operat ion. The cell’s water content during shutdown is very important because of the effect that residual water has upon MEA degradation, so this parameter was investigated by varying the dehumidification leve l during the shutdown procedure.

PAGE 57

44 The actual time of the co ld storage is of particular interest since the water in the fuel cell reacts differently dependi ng on its location within the cell. Longer duration freeze times may be required to soli dify the water in these regions of the cell. Finally, the start-up time/proc edures were evaluated to determine the proper way to recover the cell from the extreme temperatures as quickly as possible. Methodology To isolate the performance degradation caused by internal ice formation from the numerous other caus es of performance loss (i.e impurity contamination, low water content, insufficient fuel flow) a simple yet focused procedure must be developed. The performance loss of t he cell was extrapolated from the polarization curves before and after each fr eeze/thaw cycle. To characterize the cold storage degradation, multip le M.E.A.’s of similar th ickness, platinum content and fabrication technique were investigated. In order to determine the effect that membrane hydration has upon the cold storage degradation both the GDL and Nafion membrane pore size were determined via a B.E.T. por osity scanner. Prior to exposing the membrane to a subzero environment, the ph ysical structure and pore size was quantified using a scanning Hitatchi S800 electron mi croscope, energy dispersive x-ray spectroscopy, and a B.E.T. porosity sc anner. Further information regarding these instruments can be located in appendix B. This allows for quantification of the relative pore size as well as visual in spection of the pore structure. Storage effects were investigated with the me mbrane at 90%, 80%, 70%, 60%, and 50% relative humidity at a temperature of 60 degrees Celsius. Once the baseline power curves were obtained the test st ation was shutdown according to the operating instructions. Next the climat e chamber was adjusted to the correct

PAGE 58

45 experimental set point (0C -10C, -20C, -40C) and allowed ample time to equilibrate. The freezer was allowed to operate until the inter nal temperature of the fuel cell had stabilized at the set point temperatur e for one hour. The freezer was then turned off and the cell was allowed to return to the ambient temperature by the process of natural convection. At this point the memb rane’s power output is quantified as discussed previously, and then the membrane is ready to begin the next freeze/thaw test at the next colder temperature. It should be noted that the 60% and 70% relative humidity cold storage tests were each conducted with new membranes, however, due to the lack of degradation the same membrane was used for the 70%, 80%, and 90% runs. In order to better quantify the amount of degradation caused by the membrane’s water content, the amount of time of the freeze/thaw cycles had to be increased in order to cause a meas urable degradation. It was determined that a twenty four hour freez e/thaw duration would be better suited to cause the expected membrane degradation. This inve stigation was conducted in a similar fashion as outlined above, but instead of varying the temperat ure of freeze/thaw cycle only the humidity altered. The coldes t temperature previously investigated (40 C) will be utilized for a ll NM4 membrane tests. It should be noted that due to the sample size restrictions of the S.E.M. and BET scanner the physical quantificat ion of the meas ured degradation was conducted at the membrane’s end of lif e and not after each experiment. Thus, any physical degradation identified by these techniques is the result of an entire set of freeze/thaw cycles and can not be attributed to any particular set of variable combinations.

PAGE 59

46 Investigated membrane electrode assembly The M.E.A. used to quantify the fr eeze/thaw degradation was purchased from www.fuelcell.com. This model of the M.E.A. chosen is FC25-MEA, which has an area of 25 cm^2 and a platinum content of 1 mg/cm^2 or (20% wt Pt/C). All of Electrochem’s membranes have been fabricated with proprietary manufacturing processes that use up to 30% Teflon in the gas diffusion and catalyst layers. This M.E.A. also utilizes a hydrophobic carbon paper upon which the catalyst and proprietary coatings ar e applied and subsequently hot pressed in to a Nafion membrane [54].

PAGE 60

47 Chapter 4: Results Cold storage investigation The goal of this research was to obtain a general idea of how resilient a P.E.M. fuel cell membrane is at vary ing climate conditions. This task was accomplished by storing the cell at di fferent temperatures and varying the humidity as outlined in the methodology sect ion. The results show that a P.E.M. fuel cell can be frozen and thawed without catastrophic destruction to the membrane. In fact the cell showed very little degr adation during certain storage conditions. Polarization Curves for One Hour Duration Freeze/Thaw Cycles0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 02004006008001000120014001600 Current Density (mA/cm^2)Voltage (V) Cho et al. (Base Line) Cho (after -10C freeze) NM3 (Base Line) NM3 after -10C freeze Figure 16: Polarization curve comparison

PAGE 61

48 The results presented in figure 16 compare the NM3 one hour freeze/thaw cycles to similar results observed by Cho et al. [22]. Similarities between the polarization curves can be seen even t hough the experimental procedures and membranes differ. Some differences s hould be noted as well; for example, Cho et al. [22] chose to utilize a one hour freeze thaw duration from which he observed significant power losse s after the cell was thawed. The NM3 results for the freeze/thaw duration of one hour show a greater resilience to the ice degradation. It should also be observed that mode of degradation experienced by each cell is very similar. According the polarization graph, both membranes had increased ohmic resistance and flooding probability, while the activation energies stayed fairly constant. It can be seen in figure 17 that the cell did not degrade significantly as the membrane was exposed to a one hour freez e/thaw cycle. This membrane was dehydrated for ten minutes with 70% rela tive humidity gas before freezing. Figures 17 through 35 illustrate the El ectrochem membrane’s operational map after being exposed to several freeze thaw cycle at various temperatures and degrees of dehydration.

PAGE 62

49 Polarization Curve0 0.2 0.4 0.6 0.8 1 1.2 02004006008001000 Curent Density (mA/cm^2)Voltage (v) 0 C freeze -10 C freeze -20 C freeze -40 C freeze Baseline Figure 17: One hour freeze/thaw cycle investigation polarization graph 60% relative humidity

PAGE 63

50 Power vs. Current0 2 4 6 8 10 12 14 16 18 02004006008001000 Current Density (mA/cm^2)Power (W) 0 C Freeze -10 C freeze -20 C freeze -40 C freeze Baseline Figure 18: One hour freeze/thaw cycle investigation – power v. current graph 60% relative humidity

PAGE 64

51 Voltage vs Power0 0.2 0.4 0.6 0.8 1 1.2 05101520 Power (W)Voltage (V) NM2 (0 C freeze) NM2 (-10 C freeze) NM2 (-20 C freeze) NM2 (-40 C freeze) Baseline Figure 19: One hour freeze/thaw cycle investigation – voltage v. power graph 60% relative humidity Figure 19 illustrates that there is very little degradation of this particular P.E.M. fuel cell membrane after it exper ienced several freeze/thaw cycles for a duration of one hour. The fuel cell was placed in the subzero climate chamber and allowed to equilibrate to the set poi nt temperature for one hour before the duration cycle time was started. If the 60% dehumidified te st run is compared with the 70% case presented in figure 20, it can be postu lated that the amount of dehydration has a negligible effect on the performance fo r this storage duration.

PAGE 65

52 Polarization Curve 70% RH0 0.2 0.4 0.6 0.8 1 1.2 02004006008001000 Current Density (mA/cm^2)Voltage (V) Base Line 0 C Freeze -10 C Freeze -20 Freeze -40 C Freeze Figure 20: One hour freeze/thaw cycle investigation – polarization graph 70% relative humidity

PAGE 66

53 Power vs. Current 70% RH0 2 4 6 8 10 12 14 16 02004006008001000 Current Density (mA/cm^2)Power (W) Base Line 0 C Freeze -10 C Freeze -20 C Freeze -40 C Freeze Figure 21: One hour freeze/thaw cycle investigation – power v. current graph 70% relative humidity

PAGE 67

54 Voltage vs. Power 70% RH0 0.2 0.4 0.6 0.8 1 1.2 0246810121416 Power (W)Voltage (V) Base Line 0 C Freeze -10 C Freeze -20 C Freeze -40 C Freeze Figure 22: One hour freeze/thaw cycle investigation – voltage v. power graph 70% relative humidity The insignificance of the dehydrati on can be confirmed by the following cold storage data for the 80% relative hum idity case. The conclusion from these experiments is that the water content and temperature of t he subzero storage had little effect on the performance of the cell with a one hour freeze cycle duration.

PAGE 68

55 Polarization Curve 80% RH0 0.2 0.4 0.6 0.8 1 1.2 02004006008001000 Current Density (mA/cm^2)Voltage (V) Base Line 0 C Freeze -10 C Freeze -20 C Freeze -40 C Freeze Figure 23: One hour freeze/thaw cycle investigation – polarization graph 80% relative humidity

PAGE 69

56 Power vs. Current 80% RH0 2 4 6 8 10 12 14 16 02004006008001000 Current Density (mA/cm^2)Power (W) Base Line 0 C Freeze -10 C Freeze -20 C Freeze -40 C Freeze Figure 24: One hour freeze/thaw cycle investigation – power v. current graph 80% relative humidity

PAGE 70

57 Voltage vs. Power 80% RH0 0.2 0.4 0.6 0.8 1 1.2 0246810121416 Power (W)Voltage (V) Base Line 0 C Freeze -10 C Freeze -20 C Freeze -40 C Freeze Figure 25: One hour freeze/thaw cycle investigation – voltage v. power graph 80% relative humidity Due to the membrane’s resilience to freeze/thaw degradation, the duration was extended to intensify the degradation in order to better indentify the physical consequences. This investigation involv es freezing a fully humidified membrane (80+ %) to -40 degrees Celsius for varyi ng amounts of time. The results of these runs have been illustrated in figures 26, 27, and 28.

PAGE 71

58 Polarization Curve -40 C Freeze0 0.2 0.4 0.6 0.8 1 1.2 02004006008001000 Current Density (mA/cm^2)Voltage (V) Base Line 6 hour freeze 12 hour freeze 18 hour freeze 24 hour freeze Figure 26: Freeze/thaw cycle duration investigation – polarization graph

PAGE 72

59 Power vs. Current0 2 4 6 8 10 12 14 16 02004006008001000 Current Density (mA/cm^2)Voltage (V) Base Line 6 hour freeze 18 hour freeze 24 hour freeze Figure 27: Freeze/thaw cycle duration investigation – power v. current graph

PAGE 73

60 V oltage vs. Power 0 0.2 0.4 0.6 0.8 1 1.2 0246810121416 Power (W)Voltage (V) Base Line 6 hour freeze 18 hour freeze 24 hour freeze Figure 28: Freeze/thaw cycle duration investigation – voltage v. power graph The graphs above demonstrate that al though the curve of the power map is similar, the maximum power output has been reduced. This can be best illustrated by figure 29 whic h plots the same data as figure 26, but modified slightly to better magnify the performance loss.

PAGE 74

61 Voltage vs. Power0.62 0.63 0.64 0.65 0.66 0.67 0.68 0.69 0.7 0.71 0.72 500550600650700750800850900 Current Density (mA/cm^2)Voltage (V) Base Line 6 hour freeze 18 hour freezer 24 hour freeze Figure 29: Freeze/thaw cycle duration investigation – voltage v. power (high load) It can now be seen from figure 29 abov e that the cell’s power output is affected as each freeze/thaw cycle is c onducted for longer durations of time. Since this particular membrane had pr eviously experienced twelve, one hour, freeze/thaw cycles (70%, 80% and 90% relative humidity tests) before the beginning of this experimen t, and very little degradati on was previously recorded therefore it can be assumed t hat the duration of the freeze thaw cycle is a critical parameter for the facilitat ion of power degradation. To determine the effect that the humidity has upon the membrane’s degradation a twenty four hour freeze/thaw cycle will be utilized in order to properly explore the low temperature operational r ange of this membrane.

PAGE 75

62 The results from the twent y four hour freeze/thaw cycles were similar to the previous freeze/thaw investigations ; in that there wa s an initial degradation from the base line along with minimal performance lo ss from the subsequent cycles. This is illustrated in figures 30, 31, and 32. Polarization Curve0 0.2 0.4 0.6 0.8 1 1.2 02004006008001000 Current Density (mA/cm^2)Voltage (v) Base Line 50% RH 60% RH 70% RH 80% RH Figure 30: Twenty four hour freeze/thaw cycle – polarization graph Figure 30 shows that in contrast with the previous NM3 membrane, the NM4 membrane has produced noticeable lo sses for the entire operational range of the cell. This is sign ificant because the NM3 memb rane only produced limited losses near full power operation point. One plausible explanation is the separation of any of the f our critical transport areas (i.e. between the GDL and MPL layers, and between the MPL and Nafion layers.) Since the membrane is symmetric about its center, both the cat hode and anode have similar interfaces. An increase in the ohmic resistance w ould explain why the entire power curve was affected. The delamination of these components would decrease the

PAGE 76

63 membrane’s ion conductivity over the entire operational range, thus restricting the power output as seen in figure 30. The degradation effects can be confi rmed by the followi ng polarization curve illustration. Figure 31: Polarization graph tutorial Figure 31 describes how use the polarizat ion curve to help identify three types of possible degradation namely acti vation, ohmic and concentration losses. Each type is identified on one specific regi on of the graph. By utilizing figure 31, the first mode of degradation could be a ttributed to an increase in cell’s resistance which occurs because of reduced ionic conductivity through electrolyte and electrode. Physically t he ohmic resistance could have increased because of the destruction of interconnecting ionic channels, or the separation of the GDL, MPL, and Nafion layers as mentioned above [55]. This ohmic resistance loss is quantified by the Area Specific Resistance (ASR) which has units of 2* cm ohm and is a function of membrane’s design, material selection,

PAGE 77

64 manufacturing technique, and operating co nditions. This parameter can be determined using the slope of the linear portion of the polarization curve as shown in figure 31 or more accurately by impedance spectroscopy. If the ASR of the baseline polari zation curve is compared to the subsequent curves a slight increase in the slope can be seen, but the difference between the slopes is too small to quant ify as shown in figure 32 below. NM4 ASR Polarization Curvesy = -0.0003x + 0.8804 y = -0.0003x + 0.8565 0.6 0.7 0.8 0.9 0200400600800 Current Density (mA/cm^2)Voltage (v) baseline 80% RH Linear (baseline) Linear (80% RH) NM Figure 32: Post twenty four hour freeze/thaw cycle area specific resistance After the initial freeze thaw degradat ion from the baseline power curve the membrane continued to follow the same voltage/current relationship for each subsequent cycle. It cannot be assumed that the cell did not experience some type of degradation during these cycles.

PAGE 78

65 Power vs Current0 2 4 6 8 10 12 14 16 02004006008001000 Current Density (mA/cm^2)Power (W) Base Line 50% RH 60% RH 70% RH 80% RH Figure 33: NM4 power v. current In fact, figure 33 illustrates that the cell does experience the same type of power degradation as seen in the NM3 memb rane. It can be s een that the 80% humidity run had a 5% decrease in current and power output when compared to the previous 70% test run. This ty pe of degradation can be related to the mass transport losses which occur at high current densities due to the dilution of the fuel by the products produced at the ca thode. This increased dilution could be caused by GDL, MPL and Nafion pore enl argement which would reduce the layers’ hydrophobic properties and allow wa ter vapor to flood the electrode. The potential difference which is produced by the change in concentration has been formulated by United States ’ Department of Energy [ 55]. Two forms of this equation have been presented below. The first form is the potential difference in terms of the species concentrations, and t he second is related to the current and limiting current.

PAGE 79

66 B S concC C nF RT E ln Where E represents the potential differ ence produced by the concentration change Conc represents the losses caused by the loss constituent concentration R represents the total cell re sistance including electronic, ionic, and contact sources T represents t he Temperature of the Cell n represents t he number of electrons participating in the reaction F represents Faraday’s c onstant: 96,487 coulombs/g-mole electron BC represents the bul k species concentration Cs represents the surface concentration by using a Fick’s law relation the equation becomes ) 1 (l conci i Ln nF RT i represents the rate of mass transport to and electrode surface Li represents t he limiting current of the maximu m rate of which the reactant can be supplied to the electrode By comparing the maximum current produced after the 80% relative humidity freeze to the base line val ue; the change in the potential was determined to be .029 V. This value c an be compared to the actual potential difference calculated by subtracting t he voltage at maximum power of the 80% relative humidity case from the cont rol membrane’s voltage at maximum power. This difference was calculated to be 666 .6337= .0323 V. The experimental values compared well with the theoret ical formula well with an error of approximately 10%.

PAGE 80

67 V oltage vs Power 0 0.2 0.4 0.6 0.8 1 1.2 0246810121416 Power (W)Voltage (v) Base Line 50% RH 60% RH 70% RH 80% RH Figure 34: Twenty four hour freeze/thaw cycle voltage v. power graph It should be noted that similar degradation losses were not experienced between the 50%-60%, and 60%-70% test runs. This suggests that the dehumidification of the membrane had a mi tigating effect on the cell’s power loss, but just not as pronounced as r eported by Hou [21]. The membrane’s fabrication method and proprietary hy drophobic treatments are probably responsible for this performance.

PAGE 81

68 Polarization Curve (Full Load Region)0.6 0.62 0.64 0.66 0.68 0.7 0.72 500550600650700750800850900 Current Density (mA/cm^2)Voltage (v) Base Line 50% RH 60% RH 70% RH 80% RH Figure 35: Twenty four hour freeze/thaw cycle – polarization graph (full load region)

PAGE 82

69 B.E.T. investigations To identify the affect that the ice formation has upon the pore size, volume, and membrane surface area a B.E. T. porosity scanner was utilized. A complete description of scientific principl es employed with this type of analysis has been included in appendix C. The por osity scanner determines the surface area of both meso-porous and micro-por ous structures by measuring the nitrogen physisorption of the material. Figure 36 compares the presented results to that of Cho et al. [22] This graph illustrates the pore volume plott ed against the pore diameter as calculated by the Horvath – Kututhar method [58] for a particular membrane before and after the experimental freeze/thaw cycl es. Both investigations observed enlargement of the pore volume after the thermal cycles. It can be seen that the Electrochem membrane has a smaller initia l pore size, but a larger pore volume when compared with Cho et al.’s membrane.

PAGE 83

70 Figure 36: Pore size distribution The surface area, pore diameter and volume were characterized for an unfrozen membrane (OS1), the 60% de-humidified membrane (NM2), the membrane used for the 70%, 80%, and prelim inary 24 hour investigations (NM3), and a reference membrane which was frozen in standard freezer at 0 degrees Celsius for approximately for 14 days (REF1). The results from these investigations have been outlined below in figures 37 and 38.

PAGE 84

71 y = 270.09x0.8191R2 = 0.9979 y = 167.58x + 7.8715 R2 = 0.9999 y = 37.074x R2 = 0.9838 y = 30.594x R2 = 0.99590 10 20 30 40 50 60 70 80 90 100 0.000.050.100.150.200.250.30 Relative Pressure, P/P01/W[((P0/P)-1)]BET Surface Area = 10.92 m2/gOS1BET Surface Area = 19.85 m2/gNM2BET Surface Area = 102.6 m2/gREF1 NM3BET Surface Area = 119.2 m2/g Figure 37: Surface area: adsorbed nitrogen v. relative pressure

PAGE 85

72 0.00E+00 2.00E-03 4.00E-03 6.00E-03 8.00E-03 1.00E-02 1.20E-02 1.40E-02 0102030405060Pore Volume (cc/ /g)Pore Diameter, OS1 NM2 REF1 NM3 Figure 38: Pore volume v. pore diameter

PAGE 86

73 Figure 37 illustrates that as the dura tion of subzero exposure increases the amount of nitrogen that is physisorbed decreases, thus indicating an increase of membrane surface area. It was extrapolated from the figure 37 that the surface area increased by an order of magnitude from the OS1 to the NM3 tests. It should be noted that an incr ease of the surface area does not imply that an increase of the active surface area (regions where the catalyst, hydrogen and oxygen are present) was measured. In fact, the exact opposite is true since an increase of the surface area would indica te an increase of the pore size or volume, therefore decreasing the active surface area. These results agree with the performance data acquired from the sa mples, in that, if the sample had shown larger performance losses, then it also experienced large increases in pore volume and surface area. A similar re lationship is true for the control and NM2 samples; smaller performance loss – smaller pore volume and surface area increases. Therefore the performance lo ss must have a direct correlation with the membrane’s pore volume an d surface area increases. Figure 38 shows the pore volume in relation to the pore diameter as calculated by the DA method [58]. These re sults show that the pore size did not increase by any appreciable amount; ther efore the majority of the physical destruction must have occurred within the pore structures. The destruction can be quantified by analyzing the pore volume vari ations of different membranes. It can be seen that the pore volume incr eased by a factor of 10 between OS1 and NM3 cases. This does present a slight contradict ion with the results obtained from the HK method, but the literature [58] says that the DA method may not describe all samples properly. Specifically it noted t hat if the sample had larger amounts of active carbon may map erroneous result s. It should be understood that HK method provides an accurate analysis for this particular application.

PAGE 87

74 S.E.M. investigations A Hitachi Scanning Electron Microscope was utilized to correlate the recorded power loss to certain structural anomalies such as la yer delamination or separation. Due to the sample size limit ations, the optical in vestigation of the membrane was conducted after the me mbrane’s performance data had been collected. Each sample had similar preparation procedures which included: Cut a 1 x 22cm rectangle from the parent sample using a new industrial razor blade. Place the rectangular sample into a c limate chamber for 1 hour at 90C to remove as much water as possible, before degrading the Nafion layer. Sputter approximately 15 nm of gold on to the surface using a Hummer™ sputter coater. Ground the sample to the sample holder using a piece of conductive copper tape. The membranes that were investig ated include the control membrane OS1, NM3, and NM4.

PAGE 88

75 First, the cross sectional view wa s analyzed for any apparent separation or tearing of the individual layers. Figure 39 illustrates a typical polymer electrolyte membrane. The image has been annotated to indicate the approximate thickness of each of the layers of assembly. Gas Diffusion Layer ~ 120 m Micro Porous Layer ~ 100 m Proton Exchange Layer ~ 50 m Figure 39: Proton exchange membrane cross-section

PAGE 89

76 When figure 39 was investigated at a hi gher magnification, it was apparent that some amount of delamin ation occurred between the la yers of the M.E.A. The results of this invest igation can be summarized by the meso-porous section below. Meso-porous investigations Figure 40 below shows the interface between the micro-porous (catalyst) and proton exchange (Nafion) layers of an unfrozen membrane. At 1100x magnification there is no apparent gap or tearing of the interface. Figure 40: Control (OS1) proton exchange – M.P.L. interface x1100

PAGE 90

77 Figure 41: NM3 proton exchange – M.P.L. interface x2500 Figure 41 depicts the same inte rface in the NM3 membrane which experienced approximately 20 freeze/thaw cycles. This image shows a more pronounced division between the two layers.

PAGE 91

78 Figure 42: Control (OS1) proton exchange – M.P.L. interface x10,000 Figure 42 shows the control me mbrane’s interface at 10,000x magnification. A relatively smooth trans ition is still seen between the layers even at this magnification.

PAGE 92

79 Figure 43: NM3 proton exchange – M.P.L. interface x5,000 In contrast, figure 43 shows a distinct channel dividing the two layers at half the magnification. This channel would be enough to increase the ohmic proton exchange resistance of the membrane. By increasing the magnification yet furt her the approximate size of the fore mentioned channel can be determined in bo th the control and experimental samples.

PAGE 93

80 Figure 44 below shows the control memb rane at 70,000x magnification. At this magnification level the approx imate channel gap of the interface is measured with the Edax Genesis software. The approximate size of the channel is .283 microns with an uncertainty of +/15nm. Figure 44: Control (OS1) proton exchange – M.P.L. interface x70,000 Figure 45 shows the captured image of the NM3 membrane after being physically strained by internal ice formation. The effects of this strain are evident by the few stay strands of the polymer that still cl ing to the proton exchange layer. It can be seen from the picture t hat the approximate size of the channel is 1.35 microns, almost an order of magnitude larger than the control sample. The associated error with this m easurement would be +/67nm.

PAGE 94

81 Figure 45: Proton exchange – M.P.L. interface x35,000 After the micro-porous layers were investigated, the effects of the ice formation on the G.D.L. structure quantified. Gas diffusion layer investigations The gas diffusion layer has been identif ied as a possible cause of the post freeze cathodic flooding. One possibl e explanation of this flooding is the enlargement of the macro and micro matrix of the structure.

PAGE 95

82 Figure 46 illustrates the G.D.L. la yer’s macro pore matrix which is comprised of larger macro pores connec ted with a fine micro-porous weave as shown in figure 47. Figure 46: Control (OS1) gas diffusion layer x100

PAGE 96

83 Figure 47: Control (OS1) gas diffusion layer x2000

PAGE 97

84 Figure 48: Control (OS1) gas diffusion layer x12,000 It can be seen from figures 49 and 50, t hat there is littl e degradation to the structure of this layer. The pore si ze and shape of the G.D.L layer remains unaffected by the ice as it crystallizes.

PAGE 98

85 Figure 49: Comparison of NM3 and control (OS1) G.D.L. layers (macro sized) Figure 50: Comparison of NM3 and control (OS1) G.D.L. layers (micro sized)

PAGE 99

86 Chapter 5: Conclusions One hour freeze/thaw cycle investigations Based on the results presented in the preceding section, it was determined that a one hour freeze/thaw cycle was not long enough to show the anticipated degradation in this particular me mbrane. Therefore, no conclusion can be made about the effects that t he cell’s water content has upon the membrane. Superficially this experiment seems lik e a failure, but it did provide some valuable insights which helped to pr operly quantify the causes of the degradation. For example, it cannot be assumed that there was not any degradation; in fact the degr adation observed was actually due to the dilution of fuel gases at the cathode. This power loss may be caused by delamination of additives designed to maintain the delic ate water balance needed to operate at maximum power.

PAGE 100

87 Extended duration investigations These experiments were conducted to explore the power loss of the membrane as it is exposed to longer s ubzero climate durations. From these results it was clear that another investigat ion similar to the one hour investigation, but longer in duration was needed. Although the membrane used in these investigations showed more power loss t han previously obtained the full extent could not be understood because of the effects of the previous fifteen freeze/thaw cycles. Twenty-four hour freeze/thaw duration investigations Based on the results from the ext ended duration testing, a new membrane was broken in and utilized to explore t he effects of the membranes hydration level on the cell’s degradation after a twent y four hour freeze/thaw cycle time. The only difference between the one hour and twenty four hour investigations is that the twenty four hour te sts were all conducted at a temperature of -40 Celsius to isolate the hydration effects. The membrane was found to degrade not only near the maximum power region, but al ong the whole power curve as well. According the Fuel Cell Handbook [55] there are multiple modes of degradation occurring simultaneously i.e. delami nation of the membranes, and the enlargement/ or destruction of the meso -porous structures. From this investigation it is evident that the ti me of the subzero st orage is a critical parameter in determining how much degradation will be experienced by the membrane.

PAGE 101

88 B.E.T. investigations These tests were conducted to analyze the micro and meso-porous structures for apparent destruction and/or enlargement. After comparing the physisorbtion of nitrogen between several samp les, it was clear that the internal ice formation affected the membrane’s de licate pore structures. The B.E.T. determined that the samples whic h had experienced large numbers of freeze/thaw cycles, or had experienced long subzero climate exposure had the average pore volume and active surface area increased by a factor of 10. Being in agreement with the performance results, a relationship can be established between the increase of the membr ane’s pore volume and the observed performance losses. S.E.M. – optical investigation Delamination By analyzing the experimental membr anes with the electron microscope, possible causes of the performance loss we re observed. First, separation of the membrane electrode assembly layers wa s captured on the NM 3 membrane. A clear distinction can be m ade between the control and experimental sample as seen figures 40 and 41. Again this delamination is further apparent after comparing figures 42 and 43. The line tool has identified a clear physical tearing or separation of these particular layers wh ich increased the channel by a factor of 4.3.

PAGE 102

89 G.D.L. micro-structure Another interesting finding was that mi cro-porous weave that is used to hold the larger carbon chains to the matrix was unaffected by the repeated freeze/thaw cycles. This can be conf irmed after analyzing figures 49 and 50. This is notable because it proves that not all of the porous structures of the membrane are negatively affected by the fr eeze/thaw cycles. The most plausible explanation for this lack of degradation pertains to the application of the hydrophobic coatings to the Gas Diffusion Layer.

PAGE 103

90 References 1. Department of Energy (Editor), U.S., Hydrogen, Fuel Cells & Infrastructure Technologies Program Washington, D.C. 2003. 2. Chalk, S.G. and J.F. Miller, Key challenges and recent progress in batteries, fuel cells, and hydroge n storage for clean energy systems. Journal of Power Sources, 2006. 159(1): p. 73-80. 3. Larminie, J., Fuel Cell Systems explained 2nd ed. West Sussex: John Wiley & Sons Ltd, 2003. 4. Oszcipok, M., et al., Portable proton exchange membrane fuel-cell systems for outdoor applications. Journal of Power Sources, 2006. 157(2): p. 666-673. 5. Yan, Q., et al., Effect of sub-freezing tem peratures on a PEM fuel cell performance, startup and fuel cell components. Journal of Power Sources, 2006. 160(2): p. 1242-1250. 6. Scheiba, F., et al., Electron microscopy techniques for the analysis of the polymer electrolyte distribution in proton exchange membrane fuel cells. Journal of Power Sources, 2008. 177(2): p. 273-280. 7. Heitner-Wirguin, C., Recent advances in perfluorinated ionomer membranes: structure, properties and applications. Journal of Membrane Science, 1996. 120(1): p. 1-33.

PAGE 104

91 8. Giddey, S., F.T. Ciacch i, and S.P.S. Badwal, Design, assembly and operation of polymer electrolyte me mbrane fuel cell stacks to 1 kWe capacity. Journal of Power Sources, 2004. 125(2): p. 155-165. 9. Hou, J., et al., Analysis of PEMFC freeze degradation at -20 [degree sign]C after gas purging. Journal of Power Source s, 2006. 162(1): p. 513520. 10. Hottinen, T., O. Himanen, and P. Lund, Performance of planar freebreathing PEMFC at temperatures below freezing. Journal of Power Sources, 2006. 154(1): p. 86-94. 11. Ishikawa, Y., et al., Behavior of water below the freezing point in PEFCs. Journal of Power Sources, 2007. 163(2): p. 708-712. 12. Sundaresan, M. and R.M. Moore, Polymer electrolyte fuel cell stack thermal model to evalua te sub-freezing startup. Journal of Power Sources, 2005. 145(2): p. 534-545. 13. Hou, J., et al., Investigation of resided water effects on PEM fuel cell after cold start. International Journal of Hydrogen Energy, 2007. 32(17): p. 4503-4509. 14. Shichiri, T. and Y. Araki, Nucleation mechanism of ice crystals under electrical effect. Journal of Crystal Grow th, 1986. 78(3): p. 502-508. 15. Cappadonia, M., J.W. Er ning, and U. Stimming, Proton conduction of Nafion(R) 117 membrane between 140 K and room temperature. Journal of Electroanalytical Chemistr y, 1994. 376(1-2): p. 189-193.

PAGE 105

92 16. Saito, M., et al., Mechanisms of proton trans port in alcohol-penetrated perfluorosulfonated ionomer me mbranes for fuel cells. Solid State Ionics, 2007. 178(7-10): p. 539-545. 17. Datta, B.K., G. Velayutham, and A.P. Goud, Fuel cell power source for a cold region. Journal of Power Source s, 2002. 106(1-2): p. 370-376. 18. Watanabe, M., M. To mikawa, and S. Motoo, Experimental analysis of the reaction layer structure in a gas diffusion electrode. Journal of Electroanalytical Chemistr y, 1985. 195(1): p. 81-93. 19. Guo, Q. and Z. Qi, Effect of freeze-thaw cycl es on the properties and performance of membrane-electrode assemblies. Journal of Power Sources, 2006. 160(2): p. 1269-1274. 20. Kim, S. and M.M. Mench, Physical degradation of membrane electrode assemblies undergoing freeze/thaw cyc ling: Micro-structure effects. Journal of Power Sources, 2007. 174(1): p. 206-220. 21. Hou, J., et al., Electrochemical impedance investigation of proton exchange membrane fuel cells exper ienced subzero temperature. Journal of Power Sources, 2007. 171(2): p. 610-616. 22. E. Cho, J.K., H. Ha, S. Hong, Fuel Cell Applications. Electrochemical Society, 2003: p. A1667-A1670. 23. K. Weisbrod, J.H., J. Tafoya, R. Borup, M. Inbody, 2000 Fuel Cell Seminar Preview. Fuel Cells Bulletin, 2000. 3(25): p. 3.

PAGE 106

93 24. Zhang, S., et al., Effects of Freeze/Thaw Cycles and Gas Purging Method on Polymer Electrolyte Membrane Fuel Cells. Chinese Journal of Chemical Engineering, 2006. 14(6): p. 802-805. 25. Mathias, M.F., Roth, J., Fleming, J.,Lehnert,W., Vielstich,W., Gastegier,H., Lamm, A., Handbook of Fuel Cells: F undamentals, Technology and Applications Vol. 3. 2003: Wiley. 26. Wood, D.L., Xie, J., Pacheco, S.D., Davey, J.R., Borup, R.L., Garzon, F. Atanassov, P. Durability Issues of t he PEMFC GDL and MEA Under Steady-State and Drive-Cycle Operating Conditions in Fuel Cell Seminar 2004. San Antonio, TX. 27. Frisk, J., Hicks, M., Radoslav, A.T., Boand, W.M., Schmoeckel, A.K., Kurkowski, M.J. How 3M Developed a New GDL Construction for Improved Oxidative Stability in Fuel Cell Seminar 2004. San Antonio, TX. 28. Kangasniemi, K.H., Condi t, D.A., Jarvi, T.D., Characterization of vulcan electro -chemically oxidized under si mulated PEM fuel cell conditions. Journal of Electroanalytical Chemistry, 2004. 151. 29. Williams, M.V., H.R. Kunz, and J.M. Fenton, Operation of Nafion(R)-based PEM fuel cells with no external hu midification: influence of operating conditions and gas diffusion layers. Journal of Power Sources, 2004. 135(1-2): p. 122-134. 30. Lee, C. and W. Merida, Gas diffusion layer durability under steady-state and freezing conditions. Journal of Power Sour ces, 2007. 164(1): p. 141153.

PAGE 107

94 31. Reiser, C.A., Battery boosted rapid star tup of frozen fuel cell 2004, UTC fuel cells LLC. 32. Wheeler, T.F.F.D.J., Start up of a frozen fuel cell 2000, Internation Fuel Cell Corporation: United States of America. 33. Vengatesan, S., et al., Operation of a proton-e xchange membrane fuel cell under non-humidified conditions using thin cast Nafion membranes with different gas-diffusion media. Journal of Power Sources, 2006. 156(2): p. 294-299. 34. Palan, V., J.W.S. Shepard, and K.A. Williams, Removal of excess product water in a PEM fuel cell stack by vibrational and ac oustical methods. Journal of Power Source s, 2006. 161(2): p. 1116-1125. 35. Wang, H., et al., Effects of reverse voltage and subzero startup on the membrane electrode asse mbly of a PEMFC. Journal of Power Sources, 2007. 165(1): p. 287-292. 36. Olsommer, R.J.A.W. T.U.L.A.B.A.P.G.B.C., Method and apparatus for preventing water in a fuel ce ll from freezing during storage 2004, UTC fuel cells LLC. 37. Sun, S., et al., Catalytic hydrogen/oxygen reaction assisted the proton exchange membrane fuel cell (PEMFC) startup at subzero temperature. Journal of Power Sources, 2008. 177(1): p. 137-141. 38. Hickner, M.A., et al., Alternative Polymer Systems for Proton Exchange Membranes (PEMs). Chem. Rev., 2004. 104(10): p. 4587-4612.

PAGE 108

95 39. Martwiset, S., et al., Intrinsically conducting polymers and copolymers containing triazole moieties. Solid State Ionics, 2007. 178(23-24): p. 13981403. 40. Vengatesan, S., et al., Operation of a proton e xchange membrane fuel cell under non-humidified conditions usi ng a membrane-electrode assemblies with composite membrane and electrode. Journal of Power Sources, 2007. 167(2): p. 325-329. 41. Di Vona, M.L., et al., Hybrid materials for polymer electrolyte membrane fuel cells: Water uptake, mechanical and transport properties. Journal of Membrane Science, 2007. 304(1-2): p. 76-81. 42. Sgreccia, E., et al., Mechanical properties of hybrid proton conducting polymer blends based on sulf onated polyetheretherketones. Journal of Power Sources. In Press, Corrected Proof: p. 81. 43. Licoccia, S., et al., SPPSU-based hybrid proton conducting polymeric electrolytes for intermediate temperature PEMFCs. Journal of Power Sources, 2007. 167(1): p. 79-83. 44. Alberti, G., R. Na rducci, and M. Sganappa, Effects of hydrothermal/thermal treatments on the water-uptake of Nafion membranes and relations with changes of conformation, counter-elastic force and tensile modulus of the matrix. Journal of Power Sources. In Press, Corrected Proof. 45. Kreuer, K.D., On the development of proton conducting polymer membranes for hydrogen and methanol fuel cells. Journal of Membrane Science, 2001. 185(1): p. 29-39.

PAGE 109

96 46. Ho Jung, U., et al., Improvement of water management in air-breathing and air-blowing PEMFC at low temper ature using hydrophilic silica nanoparticles. International Journal of Hy drogen Energy, 2007. 32(17): p. 44594465. 47. Nakajima, H., T. K onomi, and T. Kitahara, Direct water balance analysis on a polymer electrolyte fuel cell ( PEFC): Effects of hydrophobic treatment and micro-porous layer addition to the gas diffusion layer of a PEFC on its performance during a simulated start-up operation. Journal of Power Sources, 2007. 171(2): p. 457-463. 48. Siegel, N.P., et al., A two-dimensional computational mo del of a PEMFC with liquid water transport. Journal of Power Sources, 2004. 128(2): p. 173-184. 49. Um, S. and C.Y. Wang, Three-dimensional analysis of transport and electrochemical reactions in pol ymer electrolyte fuel cells. Journal of Power Sources, 2004. 125(1): p. 40-51. 50. Sivertsen, B.R. and N. Djilali, CFD-based modelling of proton exchange membrane fuel cells. Journal of Power Sources, 2005. 141(1): p. 65-78. 51. Mishra, V., F. Yang, and R. Pitchumani, Analysis and design of PEM fuel cells. Journal of Power Sources, 2005. 141(1): p. 47-64. 52. Kulikovsky, A.A., The effect of cathodic water on performance of a polymer electrolyte fuel cell. Electrochimica Acta, 2004. 49(28): p. 51875196.

PAGE 110

97 53. Hsuen, H.-K., Performance equations for cathodes in polymer electrolyte fuel cells with non-uniform wate r flooding in gas diffusers. Journal of Power Sources, 2004. 137(2): p. 183-195. 54. Electrochem, I. Membrane Electrode Assembly 2006 [cited; Available from: http://www.fuelcell.com/index.asp?P ageAction=VIEWCATS&Category=52. 55. U.S.D.o.E., I., Fuel Cell Handbook, Editor EG&G Technical Services. 2004: Morgantown. 56. Hugh W. Coleman, W.G.S., Experimentation and Uncertainty Analysis for Engineers 2nd ed. 1999, New York: Wiley Interscience 57. FESEM types of signals New Mexico Tech. 58. SEM technology ETH Zrich website. [cited; Available from: http://www.quantachrome.co m/gassorption/index.html

PAGE 111

98 Appendices

PAGE 112

99 Appendix A: Calibrations To ensure the accuracy and repeatability of the experimental results, all of the instrumentation was tested according to manufacturer specifications. If the measurements were determined err oneous then a calibration sequence was completed. In general, each type of input measurement is calibrated in a similar way. This rule applies to all of th e thermocouples except for the feedback temperature for the environm ent chamber which will be discussed separately. The T-type thermocouples were calibrated with reference to a known temperature such as ice or boiling wa ter at atmospheric pressure. Each thermocouple was submersed in both of these references and adjusted appropriately. A summary of the dat a acquisition system calibrations and approximate degree of accuracy are presented in the table below:

PAGE 113

100 Appendix A (Continued) Table 1: Instrument calibrations Calibrations Instrument Manufacturer/Model Tested Parameter Additional Uncertainty Associated Controller (+/Error) T-type Thermocouple Omega Hydrogen Line OmegaCN9000a 1C T-type Thermocouple Omega Oxygen Line OmegaCN9000a 1C T-type Thermocouple Omega Anode 1C T-type Thermocouple Omega Cathode Omega-CNi32 1C T-type Thermocouple Omega Hydrogen Bubbler OmegaCN76000 1C T-type Thermocouple Omega Oxygen Bubbler OmegaCN76000 1C T-type Thermocouple Omega Freezer OmegaCN9000a 1C Mass Flow Controller Cole Parmer/32708-xx Hydrogen 1.50% Mass Flow Controller Cole Parmer/32708-xx Oxygen 1.50% Pressure Transducer Cole Parmer/C206 Hydrogen Line Back Pressure 0.13% Pressure Transducer Cole Parmer/C206 Hydrogen Line Back Pressure 0.13% Humidity/Temperature Vaisala/HMT330 Hydrogen Line Humidity RH 5/Temp. 2C/Press. 20hpa Humidity/Temperature Vaisala/HMT330 Oxygen Line Humidity RH 5/Temp. 2C/Press. 20hpa Current Transducer AAC(American Aerospace Controllers)/S770-30 Fuel Cell Current 1% Voltage Transmitter CR Magnetics/CR5320-1 Fuel Cell Voltage 0.50%

PAGE 114

101 Appendix B: Uncertainty analysis In order to determining the accuracy of the measurements a general uncertainty analysis in the form of equations one and two below: Equation one: ) ,..., (2 1 jX X X r r Equation two jX j X X rU X r U X r U X r U2 2 2 2 2 2 2 1 2* ) ( .... ) ( ) (2 1 [56] The uncertainty values of the measured voltage, current density, and power were determined to be as follows: Table 2: Uncertainty Parameter Associated Uncertainty Acquisition Range Uncertainty Range Voltage 0.50% 0 – 1 Volts 0 .005 v Current (Amps) 1% 0 – 30 Amps 0 .3 Amps Power (Watts) 0 – 20 Watts 0 .3 Watts S.E.M. Uncertainty The Hitatchi scanning electron microscopes generally have a measurement accuracy of +/5%. In some cases measurement errors can be as high as +/15% if the insulative sample is not prepared proper ly. To ensure the highest degree of accuracy obtainable, t he insulative sample was coated with 15nm of gold, and secured to the sample holder with an electronically conducting copper tape to minimize the charging effects. 2 2* 000025 09 I V Up

PAGE 115

102 Appendix C: Scanning electr on microscope (S.E.M.) The Scanning Electron Microscope ( SEM) is a type of microscope capable of producing high resolution images of a sample surface using electrons rather than light to form an image. Electron micr oscopy takes advantage of the wave nature of rapidly moving electrons. Wher e visible light has wavelengths from 4,000 to 7,000 Angstroms, electrons accelera ted to 10,000 KeV have wavelengths of 0.12 Angstrom s. Optical microscopes have their resolution limited by the diffraction of light to about 1000 diameters magnification. The Hitachi S800 scanning electron microscope in t he present study is limited to magnifications of around 3,000,000 diameters [57]. The SEM uses the secondary electrons when a focused electron beam is incident on the specimen to form t he image. The secondary electron signal provides information about the surface of a specimen. Since secondary electrons do not diffuse much inside the specimen, they are most suitable for observing fine structure of the specimen surface. figure 51 show the si gnals generated in an electron beam and specimen interaction.

PAGE 116

103 Appendix C (Continued) Figure 51: Sample electron scattering The basic diagram of the operation of the Hitachi S800 SEM is shown in figure 52. Electrons from a filament in an electr on gun are beamed at the specimen inside a vacuum chamber. That beam is collimated by electromagnetic condenser lenses, focused by an objective lens and then swept across the specimen at high speed. The secondary electrons are detected by a scintillation material that produces flashes of light from the electrons. The light flashes are then detected and amplified by a photomultiplier tube.

PAGE 117

104 Appendix C (Continued) Figure 52: S.E.M. layout Energy dispersive x-ray spectroscopy (EDS) Another important signal that can be analyzed by the Hitachi S800 SEM (when the electron beamspecimen interact ion occurs) is the x-ray emission. EDS identifies the elemental composition of material s imaged in a Scanning Electron Microscope (SEM) for all element s with an atomic num ber greater than boron (B). Most elements are detected at concentrations on the order of 0.1%, excluding hydrogen. When t he electron beam of the SEM hits the sample surface, it generates x-ray fluorescence fr om the atoms in its path. The energy of each x-ray photon is characteristic of the element which produced it. The EDS microanalysis system collects the x-rays, sorts and plots them by energy, and automatically identif ies and labels the elements res ponsible for the peaks in this energy distribution. The liquid nitrogen-co oled detector is used to capture and map the x-ray counts continuously [58].

PAGE 118

105 Appendix C (Continued) Burnauer, Emmett and Teller (B.E.T.) su rface area and pore size distribution measurements Catalysts and photo catalysts are often characterized by their interaction with gases. At low temperat ures, non-reactive gases (n itrogen, argon, krypton, etc.) are physisorbed by the surface. Th rough gas physisorption the total surface area of the sample can be calculated by the BET method [59]. The Autosorb-1C from Quantachrome Instru ments has been employed to determine the surface area and pore size distribution of the electrode membranes. A known amount of sample (~100 mg) was placed in a gla ss tube and the sample was out-gassed at 150 C for 3 hours. Multi point BET method using nitrogen as the adsorbate gas, the isotherms has been measured at 77 K. The BET surface area and pore size measurement principles are giv en in the following figure 53 [59]. The tendency of all solid surfaces to attract surrounding gas molecules gives rise to a process called gas sorp tion. Monitoring the gas sorption process provides a wealth of useful information abo ut the characteristics of solids such as surface area and pore size. Before performing a surface area analysis or pore size measurement, solid surfaces must be freed from contam inants such as water and oils. Surface cleaning (degassing) is most often carried out by placing a sample of the solid in a glass cell and heating it under a vacuum or a flow of dr y, inert gas. Once clean, the sample is brought to a constant temperature by means of an external bath, typically a Dewar fla sk containing a cryogen like liquid nitrogen. Then, small amounts of a gas (the absorbat e) are admitted in steps into the evacuated sample chamber

PAGE 119

106 Appendix C (Continued) Absorbate molecules quickly find their way either bounce off or stick to the surface. Gas molecules that stick to the surface are said to be adsorbed. The strength with which adsorbed molecules in teract with the surface determines if the adsorption process is to be considered physical (weak) or chemical (strong) in nature. Figure 53: Nitrogen adsorption process Figure 53 illustrates that the surface area is calculated from the monolayer amount, often using the B.E. T. method, and pore size is calculated from pore filling pressures.


xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 002001089
003 fts
005 20090428134228.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 090428s2008 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0002565
035
(OCoLC)319537141
040
FHM
c FHM
049
FHMM
090
TJ145 (Online)
1 100
Fedock, John Andrew.
0 245
Low temperature polymer electrolyte fuel cell performance degradation
h [electronic resource] /
by John Andrew Fedock.
260
[Tampa, Fla] :
b University of South Florida,
2008.
500
Title from PDF of title page.
Document formatted into pages; contains 106 pages.
502
Thesis (M.S.M.E.)--University of South Florida, 2008.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
520
ABSTRACT: The goal of this research was to quantify the degradation experienced by a polymer electrolyte fuel cell after storage at subzero temperatures ranging from 0 to -40¨C. The performance loss was determined by comparing the polarization and other applicable power curves before and after the subzero storage cycle. The causes of this performance degradation were investigated by the use of Scanning Electron Microscope, Energy Dispersive x-ray Spectroscopy, and porosity scanning technologies. It was found that there are two distinct types of degradation experienced by the membrane. The first type was identified as a variance of the actual voltage current relationship of the cell. The membrane experienced a 2 15% power reduction depending on the load applied to the cell. This mode of degradation only pertained to the initial freeze/thaw cycle and was not observed after any number of subsequent cycles.The cause of this type of degradation has been hypothesized to be related to the delamination of the proton exchange, gas diffusion, and micro porous layers. The second type of degradation was only observed during the subsequent cycles, and mainly affected the high power regions of the operating range. A 5% reduction in current density and power output was observed as a result of further freeze/thaw cycles. Mass transport limitations may have been caused by the destruction of the meso-porous gas diffusion and micro-porous layers. The pore size, volume, and membrane surface area were quantified using a B.E.T. porosity scanner. The results showed that the pore diameter of the catalyst and proton exchange layer did not increase significantly. The porosity scanner did indicate that a pore volume increased by a factor of ten and was confirmed by the surface area measurements of the membrane. The S.E.M. investigations allowed visual inspection of the membrane's structural integrity.Physical separation of the catalyst and gas diffusion layers was observed in the experimental sample, while a more homogeneous assembly was seen in the control sample.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
590
Advisor: D. Yogi Goswami, Ph.D.
653
Subzero
Alternative energy
Catalytic reactor
Cryogenic
Ice formations
690
Dissertations, Academic
z USF
x Mechanical Engineering
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.2565