저 시-비 리- 경 지 2.0 한민

는 아래 조건 르는 경 에 한하여 게

l 저 물 복제, 포, 전송, 전시, 공연 송할 수 습니다.

다 과 같 조건 라야 합니다:

l 하는, 저 물 나 포 경 , 저 물에 적 된 허락조건 명확하게 나타내어야 합니다.

l 저 터 허가를 면 러한 조건들 적 되지 않습니다.

저 에 른 리는 내 에 하여 향 지 않습니다.

것 허락규약(Legal Code) 해하 쉽게 약한 것 니다.

Disclaimer

저 시. 하는 원저 를 시하여야 합니다.

비 리. 하는 저 물 리 목적 할 수 없습니다.

경 지. 하는 저 물 개 , 형 또는 가공할 수 없습니다.

공학박사 학위논문

자가 가습 연료전지의 시동 특성 및

성능 향상에 관한 연구

Studies on the start-up characteristics and

performance improvement of self-humidified proton

exchange membrane fuel cells

2015년 2월

서울대학교 대학원

기계항공공학부

공 임 모

i

Abstract

Studies on the start-up characteristics and

performance improvement of self-humidified

proton exchange membrane fuel cells

Im Mo Kong

Department of Mechanical and Aerospace Engineering

The Graduate School

Seoul National University

In this study, the start-up characteristics of self-humidified PEMFCs

under dry conditions were investigated to provide useful information and

better understanding on self-humidified PEMFCs. In addition, the effects of

structure design and hydrophobicity of GDBL were investigated to improve

the start-up and operating performance of self-humidified PEMFCs. In

PEMFCs, the proton conductivity of the membrane determines the cell

performance and it depends on the hydration level of the membrane. In

conventional system, external humidifiers are used to hydrate the membrane.

ii

However, an external humidification system introduces disadvantages such as

increased complexity of the system, parasitic power loss, increased weight,

large volume, and high manufacturing cost. Therefore, operating PEMFC

without external humidifier is a tremendous issue in this field. Although there

have been some researches on self-humidified fuel cell, the start-up

characteristic under dry condition is not clear yet. In addition, although

relatively low performance of self-humidified fuel cell has been improved, it

should be improved further.

In this study, we investigated the characteristics of dry start-up process of

self-humidified PEMFCs. Specifically, we evaluate the hydrogen crossover

rate across the membrane, the influence of the direct reaction of hydrogen and

oxygen producing water on the dry start-up process. The effect of starting

temperature was also evaluated with different flow arrangement. As a result, It

was found that start-up performance with counter flow is effective than co-

flow and the available operating temperature increases with counter flow.

However, the dry start-up was failed at a high temperature of 45ºC. In order to

solve this problem, an advanced dry start-up process was applied and the

result was successful. The results showed that the WSP played an important

role during the dry start-up and the initial cell performance was remarkably

improved. Moreover, WSP made dry start-up possible at high cell

iii

temperatures, without the need for a long time to cool down the fuel cell after

the previous shut-down.

In order to improve the start-up and operating performance of self-

humidified PEMFCs, the effect of structure design and hydrophobicity of

GDBL on water management of GDL was investigated with an analytic model

based on the capillary pressure–saturation relationship with the Leverett J-

function. In this model, structure design and hydrophobicity of GDBLs were

represented by the measurable parameters, porosity and contact angle,

respectively. With this analytic model, the liquid water saturation distribution

and the amount of water remaining in GDL were evaluated as a function of

porosity, contact angle, and thickness ratio of each GDBL. As a result, it was

concluded that porosity gradient in negative direction and hydrophobicity

gradient in positive direction is effective on water retention. Based on the

analytic results, start-up and steady-state performance of fuel cell was

measured with the modified GDL containing double GDBL with different

porosity in negative direction (GDBL with lower porosity near the flow

channel). The result showed that shorter duration of WSP was required for

successful start-up and high performance with modified GDL. In addition, the

effect of stacking was also evaluated with the GDL containing double GDBL

with the same porosity. The result showed that the structural design of the

iv

GDBL had a major effect on its water retention capacity, whereas stacking has

negligible effect on the water retention capacity. In additions, the self-

humidification effect and the performance of the fuel cell containing the

structurally modified GDBL were found to be significantly improved over a

wide range of operating conditions.

Lastly, functionalized GDL is optimized for water retention and/or water

removal. For this, gas diffusion coefficient is considered. Although there were

many difficulties in manufacturing multi-layer GDBL in present time, with

the advanced manufacturing techniques, it is expected that manufacturing

functionalized multi-layer GDBL is possible in near future. Then,

optimization of functionalized GDL could provide important information to

manufacturer.

Keywords: Proton Exchange Membrane Fuel Cell, Start-up, Water

Management, Gas Diffusion Layer, Porosity, Hydrophobicity

Identification Number: 2009-20650

v

Contents

Abstract .....................................................................................................i

Contents .................................................................................................... v

List of Figures .............................................................................................. viii

List of Tables ................................................................................................ xiii

Nomenclature ................................................................................................ xiv

Chapter 1. Introduction .................................................................................. 1

1.1 Background of the study ...................................................................... 1

1.2 Literature survey .................................................................................. 5

1.3 Objectives and scopes ........................................................................ 10

Chapter 2. Advanced process for dry start-up ............................................ 12

2.1 Introduction ........................................................................................ 12

2.2 Start-up with dry gases ....................................................................... 14

2.2.1 Experimental apparatus and test procedure ............................. 12

2.2.2 Transient behavior of cell performance ................................... 22

2.2.3 Analysis on long term behavior of cell temperature ................ 23

2.3 Water balance at OCV condition ........................................................ 27

2.3.1 OCV variation during purge process ....................................... 27

2.3.2 MEA weight measurement ...................................................... 28

2.3.3 Hydrogen concentration measurement .................................... 33

vi

2.4 Start-up with an water storage process ............................................... 40

2.4.1 Water storage process .............................................................. 40

2.4.2 The effect of water storage process ......................................... 41

2.4.3 Analysis on long term behavior with WSP .............................. 46

2.5 Durability test ..................................................................................... 49

2.6 Summary ............................................................................................ 52

Chapter 3. Numerical analysis on water retention capacity of multi-layer

GDL ......................................................................................... 54

3.1 Introduction ........................................................................................ 54

3.2 Multi-layer GDL model ...................................................................... 55

3.2.1 Multi-layer GDL ...................................................................... 55

3.2.2 Analytic model ........................................................................ 58

3.3 Single GDBL with/without MPL ....................................................... 64

3.4 Double GDBL with MPL ................................................................... 69

3.4.1 The effect of porosity gradient ................................................ 69

3.4.2 The effect of hydrophobicity gradient ..................................... 75

3.5 Triple GDBL with MPL ..................................................................... 80

3.6 Summary ............................................................................................ 84

Chapter 4. Performance improvement of self-humidified fuel cell with

multi-layer GDL ......................................................................... 85

4.1 Introduction ........................................................................................ 85

vii

4.2 Dry start-up with modified GDL ........................................................ 86

4.2.1 Preparation of the GDL ........................................................... 86

4.2.2 The results of dry start-up with multi-layer GDL .................... 89

4.3 The effect of stacking ......................................................................... 95

4.4 Self-humidification effect of modified GDL .................................... 107

4.4.1 Preparation of the GDL ......................................................... 107

4.4.2 Experimental procedure ......................................................... 107

4.4.3 Experimental result ................................................................. 112

4.5 Summary .......................................................................................... 124

Chapter 5. Optimization of functionalized GDL ...................................... 125

5.1 Introduction ...................................................................................... 125

5.2 Optimization for water retention ...................................................... 126

5.3 Optimization for water removal ....................................................... 135

5.4 Summary .......................................................................................... 146

Chapter 6. Concluding remarks ................................................................. 147

References ............................................................................................... 150

Abstract (in Korean) ................................................................................... 168

viii

List of Figures

Figure 1.1 Membrane conductivity as a function of water content .............. 4

Figure 2.1 Two types of self-humidified PEMFCs ..................................... 13

Figure 2.2 Schematic of the experimental setup ......................................... 15

Figure 2.3 The front panel of LabView program for automatic measurement

................................................................................................... 16

Figure 2.4 Schematic design of the flow field machined on the bipolar

plates ......................................................................................... 18

Figure 2.5 Dry start-up procedure with/without WSP ................................ 20

Figure 2.6 The result of dry start-up without water storage process .......... 25

Figure 2.7 Cell temperature change after start-up at 25°C ......................... 26

Figure 2.8 OCV change during purge process with dry gases immediately

after the external load was disconnected ................................... 30

Figure 2.9 Measured weight of the MEA ................................................... 32

Figure 2.10 Schematic procedure of hydrogen concentration measurement 37

Figure 2.11 Water production rate due to hydrogen crossover and its direct

reaction with oxygen at the cathode catalyst ............................. 38

Figure 2.12 Calculated RH of exit air .......................................................... 39

Figure 2.13 Water storage rate during WSP with the assumption that all the

oxygen in the supplied air is completely utilized and the exit air

is fully saturated ........................................................................ 43

Figure 2.14 The result of dry start-up with WSP at 25°C ............................. 44

Figure 2.15 The result of dry start-up with WSP at 45°C. ............................ 45

ix

Figure 2.16 Transient behaviors of dew point, cell temperature, and cell

performance during WSP .......................................................... 48

Figure 2.17 Polarization curves of the fuel cell measured before and after the

durability test ............................................................................. 50

Figure 2.18 Transient behavior of cell voltage during the durability test ..... 51

Figure 3.1 Schematic design of multi-layer GDL ...................................... 57

Figure 3.2 Liquid water saturation profile across the single GDBL (350 μm)

without MPL ............................................................................. 66

Figure 3.3 Liquid water saturation profile across the MPL (50 μm) coated

uniform GDBL (300 μm) .......................................................... 67

Figure 3.4 The effect of MPL on the water removal in GDL ..................... 68

Figure 3.5 The effect of double GDBL with porosity gradient in positive

direction on the liquid water saturation and water retention

capacity of GDL ........................................................................ 72

Figure 3.6 The effect of double GDBL with porosity gradient in negative

direction on the liquid water saturation and water retention

capacity of GDL ........................................................................ 73

Figure 3.7 Effective diffusion factor profile across the GDL with the same

average porosity of 8.5 for GDBL ............................................ 74

Figure 3.8 The effect of double GDBL with hydrophobicity gradient in

positive direction on the liquid water saturation and water

retention capacity of GDL ......................................................... 77

Figure 3.9 The effect of double GDBL with hydrophobicity gradient in

negative direction on the saturation and water retention capacity

x

of GDL ...................................................................................... 78

Figure 3.10 Effective diffusion factor profile across the GDL with the same

average contact angle of 130º for GDBL .................................. 79

Figure 3.11 The amount of water remaining in GDL containing triple GDBL

with linear arrangement and the consequential average gas

diffusion factor of GDL ............................................................ 82

Figure 3.12 The amount of water remaining in GDL containing triple GDBL

with parabolic arrangement and the consequential average gas

diffusion factor of GDL ............................................................ 83

Figure 4.1 Schematic of the conventional GDL and functionalized ML-

GDL for water retention ............................................................ 87

Figure 4.2 The result of dry start-up with functionalized ML-GDL at 45°C

................................................................................................... 92

Figure 4.3 The improvement of steady-state performance with

functionalized ML-GDL under various temperatures at 6.67

A/cm2 (SR 2.0 and 1.5 for air and hydrogen, respectively) ...... 93

Figure 4.4 The result of impedance measurement ...................................... 94

Figure 4.5 Cross-sectional morphology of GDLs prepared to verify the

effect of stacking ..................................................................... 101

Figure 4.6 Schematic of apparatus for measuring the resistance of GDL and

the results ................................................................................ 102

Figure 4.7 Schematic of apparatus for measuring the vapor permeation rate

through the GDL and the results ............................................. 103

Figure 4.8 Comparison of the cell performances at 65°C with fully

xi

humidified hydrogen, SR 2.0 and 1.5 for air and hydrogen,

respectively. ............................................................................. 105

Figure 4.9 The effect of RH and SR of supplied air on the performance of

fuel cells at constant current density of 0.778 A/cm2 .............. 106

Figure 4.10 Schematic of multi-layer GDLs with double GDBL prepared to

enhance water retention capacity of GDL ............................... 109

Figure 4.11 The effect of hydrogen humidity and operating temperature on

the performance of fuel cells with dry air (SR 2.0 and 1.5 for air

and hydrogen, respectively) ..................................................... 110

Figure 4.12 The effect of flow rate of supplied dry air on the performance of

the fuel cell at 65°C with fully humidified hydrogen .............. 117

Figure 4.13 The effect of relative humidity of supplied air on the

performance of the fuel cell at 65°C with fully humidified

hydrogen (SR 2.0 and 1.5 for air and hydrogen, respectively) 118

Figure 4.14 Dynamic responses of fuel cells with the GDL........................ 119

Figure 4.15 Cell performances with multi-layer GDL for water retention with

porosity gradient and hydrophobicity gradient ....................... 122

Figure 4.16 Effective porosity of GDL after PTFE treatment .................... 123

Figure 5.1 Maximum water retention and average gas diffusion factor with

uniform porosity and various contact angle arrangements ...... 127

Figure 5.2 Water retention and flooding with a contact angle arrangement of

120° and 140° for GDBL-1 and GDBL-2, respectively .......... 129

Figure 5.3 The optimized water retention with a contact angle arrangement

of 120° and 140° for GDBL-1 and GDBL-2, respectively ..... 130

xii

Figure 5.4 The optimized water retention with a contact angle arrangement

of 140° and 140° for GDBL-1 and GDBL-2, respectively. ..... 131

Figure 5.5 The optimized water retention with contact angle arrangement of

120° and 120° for GDBL-1 and GDBL-2, respectively .......... 132

Figure 5.6 Minimum water retention at each porosity arrangements with

optimized thickness of GDBL-1 with/without hydrophobicity

gradient ................................................................................... 138

Figure 5.7 Minimum water retention and thickness of GDBL-1 for this,

with the fixed porosity of 0.9 for GDBL-2 ............................. 139

Figure 5.8 Minimum water retention thickness of GDBL-1 for this, when

the porosities of GDBLs are the same ..................................... 140

Figure 5.9 Comparison of single GDBL and optimized double GDBL ... 144

Figure 5.10 Effective available range of arrangement of porosity and contact

angle for each GDBL to improve the water removal ability ... 145

xiii

List of Tables

Table 2.1 Operating conditions of fuel cell for dry start-up test ............... 21

Table 2.2 Experimental conditions for measuring MEA weight ............... 31

Table 2.3 Experimental conditions for hydrogen concentration

measurement ............................................................................. 36

Table 3.1 Values of variables for multi-layer GDL model ....................... 63

Table 4.1 Properties of GDBLs employed in dry start-up test .................. 88

Table 4.2 Properties of GDLs prepared to verify the effect of stacking . 100

Table 4.3 Operating conditions of fuel cell to verify the effect of stacking

................................................................................................. 104

Table 4.4 Operating conditions of fuel cell to verify the self-humidification

effect of GDL ........................................................................... 111

Table 5.1 Optimal arrangement of double GDBL for water retention with a

constraint, β 0.3 and the same MPL ............................ 133

Table 5.2 Optimal arrangement of double GDBL for water retention with a

constraint, β 0.5 and the same MPL ............................ 134

Table 5.3 Optimal arrangement of double GDBL for water removal ..... 141

Table 5.4 GDBL properties for a comparison of uniform single GDBLs

and ‘Opt #1’ ............................................................................ 142

Table 5.5 GDBL properties for a comparison of uniform single GDBLs

and ‘Opt #2’ ............................................................................ 143

xiv

Nomenclature

F Faraday constant (C mol-1)

M Molecular weight (g mol-1)

i Current density (A cm-2)

A Active area (cm2)

J LeverettJ-function

S Liquid water saturation

P Capillary pressure (Pa)

K Permeability (m2)

k Relative permeability

k Kozeny constant

ε Porosity

d Fiber diameter (m)

V Volume (m3)

q Flux (kg s-1 m-1)

x GDL thickness directional distance (μm)

δ Thickness(μm)

σ Surface tension (N m-1)

ρ Density (kg m-3)

μ Dynamic viscosity (Pa s)

ν Kinematic viscosity (m2 s-1)

θ Contact angle (°)

xv

α Effective diffusion coefficient of water through membrane

Q Liquid water remaining in GDL (water retention capacity) (mg)

C Integral constant

D Gas diffusivity

β Gas diffusion factor

P Cell power (W)

I Cell current (A)

V Cell voltage (V)

T Temperature (°C)

U Overall heat transfer coefficient (W cm-2 K-1)

Subscript

nw Non-wetting phase

w Wetting phase

l Liquid phase

g Gas phase

water

avg average

amb ambient

cell fuel cell

1

Chapter 1. Introduction

1.1 Background of the study

Energy crisis and environmental degradation due to the use of fossil fuels

are among the most important issues in recent years [1]. The combination of

overabundance of vehicles and industrial machinery powered by fossil fuels

and the declining crude oil supplies and political instability in the regions of

the world with large oil reserves had led to an energy crisis [2]. Additionally,

the widespread use of fossil fuels is considered to be the largest source of

anthropogenic emissions of carbon dioxide within the current energy

infrastructure, and is largely blamed for global warming and climate change

[3]. As the energy crisis becomes more acute, alternative energy sources are

becoming increasingly attractive A viable alternative fuel must be technically

feasible, economically competitive, environmentally acceptable, and readily

available [4]. Hydrogen is the most abundant element in the universe and can

be produced from a wide variety of primary energy sources using various

technologies [5]. A proton exchange membrane fuel cell (PEMFC) system is

one of such technologies that can use hydrogen to produce energy. This

system produces electricity by means of electrochemical reaction between

2

hydrogen and oxygen. The by-product of reaction is only water, which is

environmentally friendly material. PEMFCs have relatively low operating

temperatures, short start-up times, and quick response. Therefore, PEMFCs

are expected to be used widely in portable, stationary, and transportation

systems in the near future [6].

In PEMFCs, the proton conductivity of the membrane is a crucial factor

that determines the cell performance. As shown in Fig 1.1, the proton

conductivity depends on the hydration level of the membrane [7]. In most of

conventional systems, external humidifiers are used to hydrate the membrane

by humidifying incoming air. However, using an external humidification

system introduces disadvantages such as increased complexity of the system,

parasitic power loss, increased weight, large volume, and high manufacturing

cost. Therefore, operating the PEMFCs with dry gases would be advantageous,

since it would dispense with the need for an additional humidification system.

Moreover, the amount of water produced during the operation of the PEMFCs

is sufficient to humidify the supplied gas under proper operating condition. In

addition, the presence of excess water causes flooding. Moreover, it is

difficult to install external humidifiers in portable applications. Therefore, it is

important to study self-humidified PEMFCs.

There have been a lot of researches on self-humidified fuel cells. Many

3

researchers have attempted to operate PEMFCs with dry gases and have

evaluated the effects of operating parameters [8-14]. In addition, the in-plane

distribution of water on the membrane has been investigated under dry

conditions [15, 16]. As a result of these efforts, it has been demonstrated that

the fuel cells can be stably operated using dry gases without external

installation.

However, most of the previous studies were only focused on the steady-

state performance without considering the start-up of self-humidified fuel cell.

Although Yu and Ziegler [11] studied transient behavior of cell performance

with dry air and hydrogen, the fuel cell was conditioned with humidified gas

before the measurement. Then, the gas was switched to dry gases for the

designed ramp sweep measurement. Despite the importance of start-up, there

have been few or no discussions on the start-up characteristics of self-

humidified fuel cell under dry conditions.

In this study, the start-up characteristics of self-humidified PEMFCs

under dry conditions were investigated to provide useful information and

better understanding on self-humidified PEMFCs. In addition, the effects of

structure design and hydrophobicity of GDBL were investigated to improve

the start-up and operating performance of self-humidified PEMFCs

4

Fig. 1.1 Membrane conductivity as a function of water content

0.00

0.03

0.06

0.09

0.12

0.15

0.18

0 5 10 15 20 25

Co

nd

uct

ivit

y, σ

(S/c

m)

Water content, λ

T = 25

T = 45

T = 65

5

1.2 Literature survey

Since now, many researchers have tried to operate PEMFCs without

external humidifiers. Some researchers have proposed alternative

humidification systems. Choi et al. [17, 18] studied dead-end operations with

pulsation flow and Kim and Kim [19] proposed an exhaust air recirculation

system for PEMFCs as a kind of internal humidification system. However, the

issue of system complexity was not solved in these studies. On the other hand,

some researchers have demonstrated that the performance of PEMFC can be

improved without external installations by adding hydrophilic particles into

the membrane. Uchida et al. [20] and other researchers [21-22] developed

self-humidifying polymer electrolyte membranes by adding hydrophilic

nanoparticles to Nafion. These membranes do enhance the self-humidification

effect, but they block the passing of protons and reduce the proton

conductivity of the membrane [23]. Other researchers [24-28] added

hydrophilic particles to the catalyst layer. However, these methods also have

limitations due to electrical insulation.

In self-humidified PEMFCs, external humidifiers are not used and

ambient air is supplied to the fuel cell without additional humidification.

Since ambient air is not always sufficiently humidified, many researchers

employed dry air condition for self-humidified PEMFCs [8-14]. In fact, in the

6

winter, the outdoor temperature goes down below zero and it makes dry air

condition. Moreover, the water remaining inside the fuel cell is purged during

the shut-down process to prevent problems such as degradation, freezing, and

so on [29, 30]. This makes the membrane dehydrated. Hence, in some cases,

the start-up of fuel cell has to be conducted with dehydrated membrane using

dry air and hydrogen. This is termed as dry start-up in this study. Despite the

importance of start-up under dry condition, there have been few or no

discussions on the start-up characteristics of self-humidified fuel cell.

In PEMFCs, the gas diffusion layer (GDL) plays on important role on

water management in addition to gas diffusion, electron transport and heat

transfer [31]. Conventional GDL is composed of gas diffusion backing layer

(GDBL, substrate layer) and micro-porous layer (MPL). The MPL effectively

removes the product water from catalyst so that the reactant gases can reach

the reactions sites through the relatively dry pores [32]. The other functions of

MPL are reducing contact resistance and protecting the catalyst layer from

physical damage caused by penetration of substrates [31]. Hence, MPL is

needed for both water retention and water removal. For this reason, in order to

enhance the water retention capacity and self-humidification effect of GDLs,

some researchers coated double hydrophobic/hydrophilic MPL on GDBL and

observed performance improvements versus under certain conditions [33-40].

7

However, since MPL has micro-scale pores and low gas permeability,

additional MPL causes mass transport problems. In addition, as the current

density increases, hydrophilic MPL easily causes a flooding problem by

blocking the pores even though the fuel cell is operated under low humidity

conditions [41].

On the other hand, GDBL has a relatively high water retention capacity

due to its higher porosity and thickness versus MPL. Many researchers

developed flow model in porous media [42-46] and investigated the effect of

pore size on cell performance [47, 48]. Based on the results of these studies,

Chu et al. [49] suggested GDL with porosity gradient and investigated the

effect of average porosity on the oxygen transport. Roshandel et al. [50]

evaluates the effects on cell performance of porosity distribution variation

resulting from the compression pressure corresponding to assembly process.

Zhan et al. [51, 52] analyzed GDL with porosity gradient for water removal.

The effect of porosity changes on gas diffusion and liquid saturation due to

water remaining in GDL has been calculated based on a simplified one-

dimensional model. Chen et al. [53] developed a two-phase flow model based

on the multiphase mixture concept to investigate the transport characteristics

in the cathode GDL of a PEMFC with a porosity gradient. Moreover, Huang

et al. [54] predicts the enhancement of the water transport for linear porosity

8

gradient in the cathode GDL with a three-dimensional, two-phase, non-

isothermal model. With these efforts, it was demonstrated that a GDL with a

porosity gradient is more favorable for liquid water removal from the catalyst

layer to the gas channel. However, their researches were only focused on

water removal. In addition, there have been few or no discussions in the

literature on the effect of hydrophobicity gradient in GDBL except an

experimental result by Kumar et al. [55]. Hydrophobicity related to contact

angle [31] and the degree of hydrophobicity has been evaluated by measuring

the contact angle [15-17, 38, 39]. A hydrophobic treatment is conducted by

dipping a GDBL into the hydrophobic agent or spraying, brushing it on the

GDBL. Then, hydrophobic agent permeates through the GDBL and it makes

hydrophobicity gradient in GDBL. Hence, some degree of hydrophobicity

gradient is an inevitable phenomenon, whether it is intended or not. Moreover,

Kumar et al. [55] intentionally developed GDBL with cross sectional PTFE

gradient applying multiple PTFE loading stages. Their result provides the

possibility and needs of GDBL with PTFE gradient.

There have been remarkable advance in manufacturing GDBL and

hydrophobic treatment. Gao et al. [56] developed a new carbon paper made up

of carbon nanotube (CNT) and other materials. Gerteisen et al. [57, 58] made

an attempt to drill holes into the carbon paper with laser perforation.

9

Fushinobu et al. [59] and Zhang et al. [60] made micro-fabricated metal GDL

using micromachining technology. In addition, a lot of metal substrates were

developed with stainless steel mesh [61], titanium mesh [62-67], nickel mesh

[68], nickel-chromium alloy foam [69], and titanium substrate [70, 71].

Hydrophobic treatment technique was also advanced with a lot of

hydrophobic agent or methods such as polytetrafluoroethylene (PTFE) [72-

76], fluorinated ethylene propylene (FEP) [77-79], polyvinylidene fluoride

(PVDF) [80] perfluoropolyether (PFPE) [79, 81], perfluoroalcoxy (PFA) [79],

CF4 plasma treatment [82].

With this advanced techniques, it is expected the result of this study will

provides an inspiration on the start-up strategy of self-humidified fuel cell and

functionalized GDL with multi-layer GDL.

10

1.3 Objectives and scopes

There have been a lot of studies on the performance of self-humidified

PEM fuel cell. However, most of them only focused on the stable operation

and performance improvement under defined conditions. Moreover, some of

them clearly mentioned that humidified gas was used for the start-up of self-

humidified fuel cell. Despite the importance of start-up, there have been few

or no discussions on the start-up characteristics of self-humidified fuel cell

under dry conditions.

The objective of the present study is to improve the performance of self-

humidified PEMFCs in a respect of start-up and stable operation. For this, the

start-up characteristics under dry conditions were investigated. In addition, the

effect of structure design and hydrophobicity of GDBL was evaluated.

In chapter two, in order to provide useful information and better

understanding on dry start-up process, the effect of starting temperature was

evaluated with different flow arrangement. In addition, the amount of water

produced by hydrogen crossover was measured using a mass spectrometer.

Moreover, in order to improve the available starting cell temperature, the

effect of water storage process was evaluated.

In chapter three, a functionalized multi-layer GDL was suggested to

improve the performance of fuel cell. The effects of structure design and

11

hydrophobicity were investigated numerically. In addition to porosity, the

effects of contact angle and thickness ratio of GDBLs were analyzed based on

the capillary pressure–saturation relationship with the Leverett J-function.

In chapter four, the effect of multi-layer GDL with double GBBL was

investigated experimentally. The structure of a GDBL was modified to

enhance the self-humidification effect in PEMFCs and dry start-up

performance was evaluated with modified GDL. In order to evaluate the effect

of stacking, some characteristics of the GDLs such as contact angle, resistance,

and vapor permeation rate were measured as well as electrochemical

performance. In addition, self-humidification effect of modified GDL was

evaluated.

In chapter five, functionalized multi-layer GDL was optimized for water

retention and water removal based on the analytic model. Although there were

many difficulties in manufacturing multi-layer GDBL in present time, with

the advanced manufacturing techniques, it is expected that manufacturing

functionalized multi-layer GDBL is possible in near future. Then, the result of

this study could provide important information to manufacturer.

In chapter six, conclusions are given along with the brief summarization

of the results.

12

Chapter 2. Advanced process for dry start-up

2.1 Introduction

In this chapter, the start-up characteristics of self-humidified PEMFCs

under dry conditions were investigated to provide useful information and

better understanding on self-humidified PEMFCs. As shown in Fig. 2.1, self-

humidified PEM fuel cells can be classified into air-flow type and air-

breathing type. In air-flow type fuel cells, the air is supplied by forced

convection using blower or fan without external humidification and the cell

performance could be improved with proper operating condition [8]. The

operating temperature rises remarkably in air-flow type fuel cells due to its

high power. On the other hand, in air-breathing type fuel cells, the cathode

side is exposed to the atmosphere and the air supply is entirely depends on

natural convection. Consequently, air-breathing type fuel cells show

relatively low power due to its high mass transfer resistance [83] and the

operating temperature is slightly higher than ambient air condition. For this

reason, the start-up under dry condition is not an important issue in the air-

breathing type fuel cells. Therefore, in present study, an air-flow type fuel

was employed to investigate the dry start-up characteristics.

Fi

(b

ig. 2.1 Two t

1

(a) Air- flow

b) Air-breath

types of self-

3

w type

hing type

-humidified PPEMFCs

14

2.2 Start-up with dry gases

2.2.1 Experimental apparatus and test procedure

Fig. 2.2 shows a schematic diagram of the experimental setup used in

this study. For the electrochemical measurements, an electric loader (PLZ

664WA, Kikusui Electronics, Japan) was installed in the system. The flow

rates of the reactants were adjusted using mass flow controllers (HI-TEC

MFC, Bronkhorst, Netherlands), which were calibrated before the

experiments. Bubble-type humidifiers were installed as an external

humidification system for the activation process. The relative humidity (RH)

of the reactants was fixed by adjusting the temperature of the humidifier and

the operation temperature of the fuel cell. For the dry start-up test, the air

and hydrogen flows directly to the fuel cell without passing through the

humidifier by controlling two valves. In addition, temperatures and pressures

were also measured with thermocouples (T-type) and pressure transmitters

(PA-21SR, Keller, Switzerland), respectively. For electrochemical

measurements, an electric loader (PLZ 664WA, Kikusui Electronics, Japan)

was installed in the system. All the experiments are automatically controlled

by Labview program as shown in Fig. 2.3.

FFig. 2.2 Sche

1

ematic of the

5

e experimenttal setup

Figg. 2.3 The froont panel of

1

LabView pro

6

ogram for auutomatic mea

asurement

17

A single cell, composed of electrically insulated end plates, graphite

bipolar plates with flow channels, Teflon-type gaskets, commercial gas

diffusion layers (GDLs; 35BC, SGL Carbon Weisbaden, Germany), and a

membrane electrode assembly (MEA) was prepared. The active area of the

cell was 9 cm2. The MEA (CNL Energy, Korea) was composed of Nafion-

211 and a catalyst layer (0.4 mgPt/cm2) on both the anode and the cathode

sides.

Fig. 2.4 shows a schematic design of the flow fields in the anode and

the cathode bipolar plates. For the experiments, identical single serpentine

flow fields with channel depths, channel widths, and land-widths of 0.9 mm,

1 mm, and 1 mm, respectively, were machined on both the anode and the

cathode bipolar plates. Using these flow fields, two different combinations

of flow arrangements were applied in the experiments. In the first case, ports

1, 2, 3, and 4 were used as the anode inlet, anode outlet, cathode inlet, and

cathode outlet, respectively. This combination of flow arrangements was

denoted as the co-flow arrangement, because both the reactants flowed from

the top to the bottom of the flow field. The second type of flow arrangement

was denoted as the counter-flow arrangement, where the flow arrangement

on the anode side was in the reverse direction compared to the co-flow

arrangement (i.e., port 1 was the anode outlet and port 2 was the anode inlet),

Figg. 2.4 Schemmatic design o

1

of the flow fi

8

ield machineed on the bipo

olar plates

19

whereas the flow arrangement on the cathode side remained the same as the

co-flow arrangement. These two flow arrangement combinations have been

used in a previous study [8] employing series-parallel flow fields.

The single cell was activated for more than 10 h after each assembly

process. The activation process was conducted under atmospheric pressure,

using air and hydrogen at a RH of 100% at the operating temperature of

65°C in the constant voltage mode. Then, the fuel cell was cooled down to

25°C and the residual water was eliminated for dry start-up test.

Fig. 2.5 shows the dry start-up procedures used in this experiment. The

operating conditions are listed in Table 2.1. The dry start-up procedure

consisted of three steps. The first step was a purge process. During the shut-

down process in an actual PEMFC system, the remaining water inside the

fuel cell is purged without an external load. In the present study, the purge

process was conducted using dry air and hydrogen for more than 1 h. During

this process, the gas flow rates were maintained at constant values of 299.4

mL/min 94.1 mL/min for dry air and hydrogen, respectively, in the open

circuit voltage (OCV) mode. These flow rates were equivalent to a

stoichiometric ratio (SR) of 2.0 and 1.5 for air and hydrogen, at a current

density of 1.0 A/cm2. During idling process, no reactants were supplied to

the fuel cell for 1 min simulating shut-down.

20

Fig. 2.5 Dry start-up procedure with/without WSP

OCV (Open circuit voltage) Load (0.4 V)

① Purge

② Idling

③ Power generation

WSP

60 min ~ 1 min 1 min 1 ~ 5 min

21

Table 2.1 Operating conditions of fuel cell for dry start-up test

Operating parameters Unit Values

Temperature °C 25, 45

Relative Humidity % 0

Supplied gas (cathode/anode) - Air/H2

Flow

rates

Purge process mL/min 299/94

Idling process mL/min 0/0

Power generation

process mL/min 299/94

Water storage

process (WSP) mL/min 20/10

22

The final step of the dry start-up procedure was the power generation

process. This process simulated the restart-up of the PEMFC system after

shut-down and was composed of two sub-steps. In the first sub-step, the

OCV mode was maintained for 1 min, in order to allow sufficient time for

the diffusion of the supply gases. Following this, the external load was

connected.

In most control devices, a time lag between the input signal and the

response is inevitable. The mass flow controller used in this experiment also

had a conspicuous time lag, when the flow rate was changed from 0 mL/min

to 20 mL/min for air and 10 mL/min for hydrogen. The measured time lag

was about 20~30 s. However, it occurred during the operation of the fuel cell

in the OCV mode and the effect of the time lag would be negligible.

2.2.2 Transient behavior of cell performance

Fig. 2.6(a) shows the results of a dry start-up at 25°C. From the result, it

was found that both the flow arrangements and starting temperatures were

very important parameters in determining the success of the dry start-up. As

shown in Fig. 2.6(a), at 25°C, dry start-up was successfully accomplished

with both flow arrangements. However, more than 3 min was required in the

23

case of co-flow before normal operation occurred and the initial performance

was very poor compared to that achieved with counter-flow. On the other

hand, as shown in Fig. 2.6(b), dry-start up completely failed at 45°C with

both flow arrangements. Based on these results, it is deduced that the

hydration level of the membrane is not sufficient to allow dry start-up at high

temperatures. In addition, the fuel cell has to be cooled down adequately to

start-up successfully with dry gases after the shut-down process. It needs a

lot of time. Also, considering the temperature rise of fuel cell during

operation, it might be a barrier for the commercial use of self-humidified

fuel cells.

2.2.3 Analysis on long term behavior of cell temperature

Fig. 2.7 shows the long term behavior of cell temperature as a function

of time at 0.4 V, with gas flow rates of 299 mL/min and 94 mL/min for air

and hydrogen, respectively. Assuming that the water product was in the

vapor form, the heat release rate of the PEMFCs ( ) is given as

(1.25 )cell cell cellQ I V (2.1)

(1.25 )cellcell cell

cell

PQ V

V (2.2)

( )cellcell cell amb cell

dTQ UA T T V

dt (2.3)

24

where Icell and Vcell represent the current (A) and the voltage (V) of the

cell, respectively.

Since the voltage was maintained at a constant value, the rate of heat

production was proportional to the current. A portion of the heat produced

was released into the atmosphere and supplied gases, whereas the rest of the

heat raised the temperature of the fuel cell. Thus, the increment of cell

temperature means the increment of cell performance and the steady-state

cell temperature is the optimal temperature for cell performance. As shown

in Fig. 2.7, the temperature of the fuel cell increased up to 48°C and 33°C

with counter-flow and co-flow arrangements, respectively. Based on the

result, it is concluded that counter-flow is advantageous in terms of

enhanced water retention capability compared to co-flow.

25

(a) Start-up at 25°C

(b) Start-up at 45°C

Fig. 2.6 The result of dry start-up without water storage process

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0 200 400 600 800 1000

Cu

rren

t d

ensi

ty (

A/c

m2)

Time (s)

Co-flow (25)

Counter-flow (25)

Load on

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0 200 400 600 800 1000

Cu

rren

t d

ensi

ty (

A/c

m2 )

Time (s)

Co-flow (45)

Counter-flow (45)

Failed in dry start-up

Load on

26

Fig. 2.7 Cell temperature change after start-up at 25°C

0

10

20

30

40

50

60

0 3000 6000 9000 12000

Tem

per

atu

re (

oC

)

Time (s)

Counter-flow arrangement

Co-flow arrangement

27

2.3 Water balance at OCV condition

2.3.1 OCV variation during purge process

It is well known that gas crossover reduces the OCV of a fuel cell [84-

88]. Fig. 2.8 shows the OCV change during purge process, immediately after

the external load was disconnected. During the experiment, the gas flow rate

and the cell temperature were maintained at constant values. As shown in Fig.

2.8, the OCV increased initially and then dropped, eventually converging to

a certain value after a sufficient amount of time. This result is reasonable,

considering the water behavior and the hydration level of the membrane

during the purge process. Initially, excess water was eliminated causing the

membrane to be dehydrated. The membrane flexibility, which has a

significant effect on the gas crossover rate, depends on the hydration level of

the membrane [87]. Therefore, it was concluded that the OCV increased

because of the decrease in the gas crossover rate, which was caused by the

decrease in the membrane flexibility. The membrane thickness also depends

on the hydration level of the membrane [88] and the membrane shrinks

under dry conditions. The gas crossover rate generally increases in thin

membranes [84, 87, 89] and therefore, the decrease in OCV observed is

28

reasonable. However, the OCV did not decrease indefinitely and converged

to 1.01 V for a fresh membrane. It should be noted that this value can vary

depending on the degree of membrane degradation. It was concluded that the

gas crossover rate and hydration level of the membrane were maintained at

constant values, after a sufficient amount of time corresponding to the

convergence of the OCV. The hydrogen crossover measurements were

conducted after the convergence of the OCV.

2.3.2 MEA weight measurement

Fig. 2.9 shows the weight of the membrane electrode assembly (MEA),

which was measured using an electronic balance. The corresponding

experimental conditions are listed in Table 2.2. Before measuring the weight,

the fuel cell was treated by three different methods. The methods included

maintaining the cell at OCV with a supply of dry Air/H2, maintaining the cell

at OCV with a supply of dry N2/H2, heating the cell up to 105°C, and supply

of dry Air/H2 again after heat treatment. The conductivity of the membrane

(σ, S/cm) at the operating temperature of T K has been given by Springer

et al. [90] and is shown in Eq. (2.4).

30

1 1( ) exp 1268

303 TT

(2.4)

29

In the above equation, σ refers to the membrane conductivity at

30°C as a function of the water content λ and is given by Eq. (2).

30 ( ) 0.005139 0.00326 (2.5)

The hydration level of the membrane is directly related to its water

content and the dry start-up will fail if the membrane is completely

dehydrated. However, as shown in Fig. 2.9, the membrane was not

completely dehydrated. Further, it was found that the weight of membrane

increased when air and hydrogen were supplied, implying that the oxygen in

the air and hydrogen crossed over reacted under OCV conditions, resulting

in the production of water. Moreover, the weight of the membrane recovered

when dry Air/H2 supplied again after heat treatment. Although the cathode

and the anode were separated by the membrane, gas crossover across the

membrane enabled the reaction between air and hydrogen [84-88]. In other

words, the hydration level of the membrane could be recovered by the

hydrogen crossover across the membrane and its direct reaction with oxygen

at the cathode side. This would enable a dry start-up.

30

Fig. 2.8 OCV change during purge process with dry gases immediately

after the external load was disconnected

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1.1

1.2

0 600 1200 1800 2400 3000

Op

en c

ircu

it v

olt

age

(V)

Time (s)

31

Table 2.2 Experimental conditions for measuring MEA weight

Test No. method

① Air/H2 20/10 mL/min for 10 h at 25°C

② N2/H2 20/10 mL/min for 10 h at 25°C

③ Heat treatment 105°C for 1 h

④ Air/H2 After heat treatment, 20/10 mL/min for 50 h at 25°C

32

Fig. 2.9 Measured weight of the MEA

23.0

23.5

24.0

24.5

25.0

1 2 3 4

ME

A w

eig

ht

(mg

/cm

2 )

Test condition

Air / H2 N2 / H2Heat

treatmentAir / H2

After Heat treatment

33

2.3.3 Hydrogen concentration measurement

The amount of hydrogen crossover was measured using a mass

spectrometer. The corresponding operating conditions are listed in Table 2.3.

As stated previously, the flow rates of 299 mL/min and 94 mL/min for air

and hydrogen, respectively, are equivalent to a SR of 2.0 and 1.5 for air and

hydrogen at current density of 1.0 A/cm2. Fig. 2.10 shows the procedure of

hydrogen concentration measurement.

The total hydrogen crossover rate across the membrane was obtained by

measuring the hydrogen concentration at the cathode outlet when nitrogen

was supplied on the cathode side and hydrogen was supplied on the anode

side. The remaining hydrogen crossover rate after its direct reaction with

oxygen was obtained by measuring the hydrogen concentration at the

cathode outlet when air was supplied on the cathode side and hydrogen was

supplied on the anode side. Finally, the actual hydrogen concentration in

fresh air was measured. Then, the hydrogen consumption rate c , mL/min)

and the water production rate m ,mg/min) are obtained by calculating

given equations.

2 2 2

2

/ /

6

( )

10N H Air H Air

H CA

x x xc V

(2.6)

2 2 2H O H H O

Pm c M

RT (2.7)

34

where V is flow rate of gas (mL/min) supplied on the cathode side, x

is hydrogen concentration (ppm) measured at the cathode outlet, P is

operating pressure (atm), and M and R represent molecular weight

(g/mol) of water and ideal gas constant (atm·L/mol·K), respectively.

In this study, the experiments were repeatedly conducted with different

flow arrangements. The water production rates were then obtained using the

hydrogen concentration measurements and the above equations. Fig. 2.11

shows the water production rate due to the hydrogen crossover and its direct

reaction with oxygen at the cathode catalyst under OCV conditions. The

operating temperature was found to have a significant effect on the water

production rate, whereas the influences of flow arrangements and flow rates

were negligible. An increase in the flow rates of the reactants did not cause a

significant increase in the extent of direct reaction or the amount of water

produced. Therefore, the water production rate could be approximated to an

empirical formula in the Arrhenius-form and can be expressed as shown in

Eq. (2.8).

2expH Om

T

(2.8)

In this study, the values of α and β were determined to be 16.24 and

1767.68 by a regression analysis on the results of the experiments. Then,

with the assumption that all product water is leaving the cell in the vapor

35

phase at the cathode side, the RH of exit air could be calculated. Fig. 2.12

shows the result. The RH of exit air is lowered as the cell temperature rises.

The water production rate calculated from the empirical formula provides

useful information for simulations and it provides insights into the reasons

behind the failure of dry start-up with high gas flow rates and high

temperature.

36

Table 2.3 Experimental conditions for hydrogen concentration

measurement

Operating parameters Unit Values

Temperature °C 25, 45, 65

Relative Humidity % 0

Supplied gas (cathode/anode) - Air/H2, N2/H2, air/air

Flow rate (cathode/anode) mL/min 299/94, 299/299

37

Fig. 2.10 Schematic procedure of hydrogen concentration measurement

Hydrogen concentration

Mea

sure

d c

on

dit

ion

③ Consumed amount

① Total amount that crossed over from anode side

② Unreacted amount

N2 / H2

Air / H2

Air

38

Fig. 2.11 Water production rate due to hydrogen crossover and its direct

reaction with oxygen at the cathode catalyst

0.00

0.04

0.08

0.12

0.16

15 25 35 45 55 65 75

Wat

er p

rod

uct

ion

rat

e (m

g/m

in)

Temperature (oC)

Co-flow (CA 299 / AN 94)

Co-flow (CA 299 / AN 299)

Counter-flow (CA 299 / AN 94)

Counter-flow (CA 299 / AN 299)

Flow arrangement (flow rate, mL/min)

39

Fig. 2.12 Calculated RH of exit air

0

2

4

6

8

10

25 35 45 55 65

RH

of

exit

air

(%

)

Temperature (oC)

20

50

100

200

300

Flow rate of supplied air (mL/min)

40

2.4 Start-up with an water-storage process

2.4.1 Water storage process

In order to enhance the dry start-up performance, an advanced start-up

process, defined as water storage process (WSP), was applied in the power

generation process. The water produced during the operation of the fuel cell

is expected to be accumulated and absorbed by the membrane during the

WSP. As a result, the conductivity of the membrane could increase, which

enables high cell performance. In this study, during the WSP, the OCV mode

was maintained for 1 min, and then an external load was connected from 1

min to 5 min, with gas flow rates of 20 mL/min and 10 mL/min for air and

hydrogen, respectively. These flow rates were equivalent to a SR of 1.0 and

1.2 for air and hydrogen at a current of 1.2 A. With these flow rates, the

current of 1.2 A was obtained at 0.4 V in the experiments irrespective of the

flow arrangements and the operating temperatures. This implies that the

oxygen in the supplied air was almost completely utilized during the fuel cell

reactions. Therefore, these flow rates were selected for the experiments. As

shown in Fig. 2.13, the amount of water produced by fuel cell reaction

during the WSP is 100 times more than that of crossover. However, it should

41

be noted that the flow rates must be selected carefully on a case-by-case

basis for each fuel cell.

2.4.2 The effect of water storage process

In this study, the effect of WSP was evaluated as a possible solution for

the poor start-up and initial performance observed in the self-humidified fuel

cells. As mentioned previously, during the WSP, the water is absorbed into

the membrane and the conductivity of the membrane increases. Fig. 2.14

shows the effect of the WSP on the dry start-up performance at 25°C. It was

found that the initial cell performance was significantly improved after WSP

was applied to the start-up process. As the duration of the WSP is increased,

the membrane can absorb more water. However, there is a limit to the

amount of water that can be absorbed by the membrane, implying that an

excessive WSP time may be ineffective. This was verified by the results of

the experiments. As the duration of the WSP was increased, the cell

performance was improved but the increment was decreased. In addition, the

cell performance with counter-flow arrangement was almost the same with

the co-flow arrangement. Therefore, it is concluded that the flow

arrangement has little effect on the cell performance at 25°C, if the fuel cell

42

goes through the WSP during the dry start-up. However, the flow

arrangement has a great effect on the cell performance at 45°C. Fig. 2.15

shows the result of start-up test with WSP at 45°C. The dry start-up was

successful upon applying the WSP for more than 2 min at a cell temperature

of 45°C. With 1 min of WSP, the fuel cell started to operate, but shut-down

immediately when high flow rates were applied. It was found that 1 min of

WSP was too short to increase the membrane conductivity sufficiently. In

addition, the fuel cell operated stably at the cell temperature of 45°C, only

with the counter-flow arrangement. As shown in Fig. 2.15(a), while the fuel

cell with the co-flow arrangement showed increased cell performance

initially, the performance dropped rapidly over time, which is attributed to

the gradual removal of the water retained during the WSP. It is obvious that

the WSP played an important role in the dry start-up and the cell

performance was improved, irrespective of the flow arrangement or the

operating temperature.

43

Fig. 2.13 Water storage rate during WSP with the assumption that all the

oxygen in the supplied air is completely utilized and the exit air

is fully saturated

0

2

4

6

8

10

25 35 45 55 65

Wa

ter

sto

rag

e ra

te (

mg

/min

)

Temperature (°C)

1.0 A (17 mL/min)

1.2 A (20 mL/min)

1.5 A (25 mL/min)

44

(a) Co-flow

(b) Counter-flow

Fig. 2.14 The result of dry start-up with WSP at 25°C

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0 200 400 600 800 1000

Cu

rren

t d

ensi

ty (

A/c

m2)

Time (s)

12345

WSP duration (min)

Load on

AN

CA

25oC

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0 200 400 600 800 1000

Cu

rren

t d

ensi

ty (

A/c

m2 )

Time (s)

12345

WSP duration (min)

Load on

AN

CA

25oC

45

(a) Co-flow

(b) Counter-flow

Fig. 2.15 The result of dry start-up with WSP at 45°C

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0 200 400 600 800 1000

Cu

rren

t d

ensi

ty (

A/c

m2 )

Time (s)

12345

WSP duration (min)

Load on

AN

CA

45oC

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0 200 400 600 800 1000

Cu

rren

t d

ensi

ty (

A/c

m2 )

Time (s)

12345

WSP duration (min)

Load on

AN

CA

45oC

46

2.4.3 Analysis on long term behavior with WSP

In order to evaluate the water management capability of the fuel cells

with different flow arrangements, dew point analysis was performed by

conducting the WSP for more than 3 h on the cathode side. As shown in Fig.

2.16, the results obtained with co-flow and counter-flow arrangements show

similar trends and the results can be divided into three stages. In stages 1 and

2, the measured dew points were similar to the cell temperature, implying

that the cathode outlet gas was fully saturated with liquid water and any

excess water could be absorbed by the membrane. However, a fluctuation in

the cell performance was found in stage 2. Based on the result, it was

concluded that the membrane is fully hydrated after stage 1 and the excess

liquid water possibly accumulated in the flow channel and diffused to the

anode side during stage 2, starting to cause a flooding problem. In stage 3,

the measured dew points were significantly higher than the cell temperature.

Therefore, in this stage, the excess water was discharged with the cathode

outlet gas and the cell performance fluctuated periodically. Since the dew

point hygrometer was installed at a certain distance from the fuel cell,

fluctuations were not detected from the dew point measurements. The

counter-flow arrangement showed a shorter duration of stage 1 indicating

that the membrane was hydrated more quickly in this case compared to the

47

case of the co-flow arrangement, due to the back diffusion of the water from

the cathode outlet into the anode inlet and the diffusion of the water from the

anode outlet to the cathode inlet due to the electro-osmotic drag force. The

water diffusion also explains the longer duration of stage 2 and shorter

flooding period during stage 3, in the cell with the counter-flow arrangement.

It is concluded that the counter-flow arrangement retained more water

compared to the co-flow arrangement under the same operating conditions.

48

(a) Co-flow

(b) Counter-flow

Fig. 2.16 Transient behaviors of dew point, cell temperature, and cell

performance during WSP

0.0

0.1

0.2

0.3

0.4

0.5

20

0

20

40

60

80

0 3000 6000 9000 12000

Po

we

r d

ens

ity

(W/c

m2 )

Tem

per

atu

re (

oC

)

Time (s)

Power density

Dew point temperature

Cell temperature

Stage 1 Stage 3Stage 2

AN

CA

-

0.0

0.1

0.2

0.3

0.4

0.5

20

0

20

40

60

80

0 3000 6000 9000 12000

Po

we

r d

ens

ity

(W/c

m2 )

Tem

per

atu

re (

oC

)

Time (s)

Power density

Cell temperature

Dew point temperature

Stage 1 Stage 3Stage 2

AN

CA

-

49

2.5 Durability test

It is well known that load change, start-up, idling and high power

causes the performance degradation of PEMFCs [91]. There's a lot of

concern about the degradation result in repeated start-up and dry/wet

environment [92, 93]. In self-humidified PEMFCs, the membrane and GDL

are wetted during start-up and dried during purge process. Therefore, in this

study, a durability test was conducted to evaluate the effect of dry-start up on

performance degradation of PEMFCs. For this, the start-up/purge process

was repeated automatically more than 2000 times by LabView program.

Fig. 2.17 shows the polarization curves of the fuel cell measured before

and after the durability test. The performance measurement was conducted at

a temperature of 25ºC. The SRs of air and hydrogen was fixed at 1.5 and 2.0.

Although there's a lot of concern about the degradation, any noticeable

performance degradation was not found. In addition, as shown in Fig. 2.18,

the same transient behavior of cell voltage was repeated during durability

test except for the first cycle. Since the first cycle was started with extremely

dehydrated membrane, the difference in transient behavior of cell voltage

during first cycle was a reasonable result.

50

Fig. 2.17 Polarization curves of the fuel cell measured before and after the

durability test

0.2

0.4

0.6

0.8

1.0

1.2

0.0 0.2 0.4 0.6 0.8 1.0 1.2

Vo

ltag

e (V

)

Current density (A/cm2)

Co-flow / before

Co-flow / after

Counter-flow / before

Counter-flow / after

51

Fig. 2.18 Transient behavior of cell voltage during the durability test

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0 5400

Vo

ltag

e (V

)

# Cycle (1 Cycle = 600s)

# 1 # 500 # 1000 # 1500 # 2000

52

2.6 Summary

In this study, the possibility of dry start-up and the effect of various

parameters on dry start-up were investigated. The amount of water produced

by hydrogen crossover was evaluated under OCV conditions. In addition, the

rate of water production by the direct reaction of hydrogen and oxygen at the

cathode catalyst was determined. It was determined that hydrogen crossover

had positive effects on the hydration level of the membrane and the

operating temperature was a major parameter affecting the water production

rate. Even under dry gas conditions, the membrane was not completely dry,

due to the hydrogen crossover and the direct reaction of the reactants

yielding water, which made dry start-up possible at a low temperature of

25°C. However, dry start-up was unsuccessful at a higher cell temperature of

45°C, in this study, which might be a barrier for the commercial use of self-

humidified fuel cells.

The problem with dry start-up at high temperatures was overcome with

the WSP, which enabled the successful operation of the fuel cell at the cell

temperature of 45°C. In addition, the initial cell performance was also

improved significantly with WSP. It was found that while the initial cell

performance depended on the duration of the WSP, the duration higher than

a certain optimum value was ineffective. Therefore, the duration of the WSP

53

has to be decided carefully. In addition, the reactant flow arrangement was

also an important parameter in achieving a stable startup and normal

operation. With the WSP process, the fuel cell showed a stable start-up and a

normal operating performance even at 45°C with the counter-flow

arrangement. However, with the co-flow arrangement, while the fuel cell

started to operate initially, its performance dropped rapidly at 45°C. The dew

point analysis showed that the water retention capacity of the fuel cell was

enhanced by the counter-flow arrangement.

The results showed that the WSP played an important role during the

dry start-up and enhanced the initial cell performance. Moreover, WSP made

dry start-up possible at high cell temperatures, without the need for a long

time to cool down the fuel cell after the previous shut-down. The results of

this work provide useful information for improving the simulations and the

operating strategies of self-humidified PEMFCs. This work can contribute to

the commercialization of portable fuel cells. Further, the results are also

applicable to automotive and residential fuel cells.

54

Chapter 3. Numerical analysis on water retention

capacity of multi-layer GDL

3.1 Introduction

In most of self-humidified fuel cells, the limiting factor of cell

performance is low proton conductivity result from dehydration of the

membrane. In PEMFCs, the gas diffusion layer (GDL) plays on important

role on water management in addition to gas diffusion, electron transport and

heat transfer. The dry gases supplied to the fuel cell can be humidified

passing through the wetted GDL. In terms of mass transport, optimized

structure design and hydrophobicity of GDL resulted in an effective water

management and improved gas diffusion kinetics. Since the dry gases

supplied to the fuel cell can be humidified passing through the wetted GDL,

start-up and operating performance of self-humidified fuel cell can be

improved by enhancing the water retention capacity of GDL.

Therefore, in this chapter, In addition, the effect of structure design and

hydrophobicity of GDBL was investigated numerically based on the

capillary pressure–saturation relationship with the Leverett J-function.

55

3.2 Multi-layer GDL model

3.2.1 Multi-layer GDL

Considering the manufacture difficulties, a GDL with linear gradient in

porosity and/or hydrophobicity is more about hope than reality. More

practical design is multi-layer GDL with certain change in porosity and

hydrophobicity. Fig. 3.1 shows the schematic design of the multi-layer GDL.

In this study, an analytic model was obtained based on the capillary

pressure–saturation relationship with the Leverett J-function. In multi-layer

GDL model, each uniform layer has high degree of freedom in property

arrangements of fiber diameter, porosity, contact angle, and thickness. Also,

the number of layers can be expanded to infinitely.

For convenience, MPL coated double GDBL was analyzed in this study.

Then, if the porosity and contact angle of GDBLs were identical, the multi-

layer GDL was equal to conventional GDL with single MPL and single

GDBL. In addition, the gradient GDBL was equivalent to arrangement of

different porosity, contact angle for each layer. With this multi-layer GDL

model, the saturation distribution and the amount of water remaining in GDL

were evaluated as a function of porosity, contact angle, and thickness ratio of

56

each GDBL.

It should be noted that positive direction of certain property was defined

as GDBL with lower property near the MPL and higher property near flow

channel and negative direction of certain property was defined as GDBL

with higher property near the MPL and lower property near flow channel. In

addition, as the amount of water remaining in GDL was increased, water

retention capacity of GDL was increased while water removal ability of

GDL was decreased.

57

Fig. 3.1 Schematic design of multi-layer GDL

MPL

GDBL-1

GDBL-2

εGDBL-1θGDBL-1

εGDBL-2θGDBL-2

δGDBL-2

δGDBL-1

δMPL εMPL ,θMPL

GDBL-n︙

58

3.2.2 Analytic model

The capillary pressure is defined as the difference in pressure across the

interface between the non-wetting phase and the wetting phase. For the GDL,

gas and liquid phase correspond to non-wetting phase and the wetting phase,

respectively.

c nw w l gP P P P P (3.1)

In porous media, the capillary pressure is expressed as [48-54]

0.5

cos( )

( / )c

cP J SK

(3.2)

The Leverett J-function, J S is a dimensionless function of liquid

water saturation S V V⁄ and given as

2 3

2 3

1.417(1 ) 2.120(1 ) 1.263(1 ) ( 90 )( )

1.417 2.120 1.263 ( 90 )c

c

S S SJ S

S S S

(3.3)

In hydrophobic GDL (θ 90° liquid phase pressure is higher than

gas phase pressure (P > 0) so that the product water is transported from

catalyst layer to channel through the GDL.

In recent years, various attempts have been made to relate the

permeability to other more readily measurable properties and the most

broadly known is the Kozeny-Carman relation [98]. Tomadakis and

Robertson [98] also proposed a more comprehensive relation to predict the

59

anisotropic permeability. Tamayol and Bahrami [99, 100] related the

permeability of fibrous media to the microstructure geometrical parameters.

In addition, they modified their result to accommodate the effects of

mechanical compression and PTFE contents. Among them, Kozeny-Carman

equation was employed in this study because of its simplicity and wide use

[51, 52]. For random non-overlapping fiber structures [98], the Kozeny-

Carman equation becomes

3 2f

216 (1 )K

dK

k

(3.4)

The water flux (m ) produced by the electrochemical reaction at the

cathode is expressed as

2 2 2H O H O

im M

F (3.5)

Then, the relation of water flux and capillary pressure is given as [51-54]

2

1 22

l rlH O c

l

KkiM p

F

(3.6)

In this relation, it is assumed that all the product water exists in liquid

phase and the phase change of water inside the GDL is not considered. For

liquid water, the relative permeability is known as [51, 52, 101]

3rlk S (3.7)

Then, with the additional assumption that the water flux driven by

60

electro-osmotic drag is balanced by back diffusion [101], Eq. (3.6) is

simplified to Eq. (4.7)

2

3

2l

H O cl

KSiM p

F

(3.8)

Substitute above values Eqs. (3.2), (3.3), (3.4) into Eq. (3.8), Eq. (3.9)

is obtained as the one-dimensional steady-state model based on a capillary

pressure–saturation relationship

2

3 4 5

2f

(1.417 4.240 3.789 )

4(1 )

2 cosK

H Oc

dSS S S

dx

kiM

F d

(3.9)

In multi-layer model, the porosities are uniform at each layer and the

right hand side of Eq. (3.9) could be assumed as a constant depending on

each layer. This provides outstanding convenience for calculation compared

to gradient model.

Then, integrating Eq. (3.9), an analytic solution is obtained.

2

4 5 6

2f

1.417 4.240 3.789( )

4 5 6

4(1 )

2 cosK

H Oc

S S S

kiM x C

F d

(3.10)

The value of integral constant, C is obtained from the boundary

condition of each layer.

The liquid water saturation defined as the ratio of the pore volume

61

occupied by water to the entire pore volume. Assuming constant density of 1

g/cm3 for liquid water, the amount of liquid water remaining in GDL under

steady conditions is calculated from Eq. (3.11).

00.1A ( ) ( )

GDL

lQ x S x dx

(3.11)

With a fixed liquid water flux flows through GDL, Q should be as

small as possible for water removal [51-52]. On the other hand, Q should

be as large as possible for water retention.

The remaining water in GDL occupies pore volume so that gas

diffusion is reduced. The gas diffusion flux through porous media is

described by Fick’s Law and given as

effij ij

Ciq D

x

(3.12)

where the effective diffusivity, effijD , is given as

1.5

( , ) refeffij ij ref ref

ref

P TD D T P

P T

(3.13)

1.5 1m

o S (3.14)

where is effective diffusion factor and m is a constant. In this

study, m = 2 was assumed. [52]. Then, the average diffusion factor can be

calculated using Eq. (3.15)

62

0

1avg dx

(3.15)

The liquid water saturation distribution in each layer is governed by Eq.

(3.10). However, at the interface of layers, the liquid water saturation has a

sudden change because of change in some material or structure properties.

Hence, liquid water saturation at the interface is determined from capillary

pressure equilibrium [101],

0.5

1 110.5

1 1

cos( )

( / )

cos( )

( / )

MPL

MPL

MPL MPLMPL x

MPL MPL

GDBL GDBLGDBL x

GDBL GDBL

J SK

J SK

(3.16)

1

1

1 110.5

1 1

2 220.5

2 2

cos( )

( / )

cos( )

( / )

MPL GDBL

MPL GDBL

GDBL GDBLGDBL x

GDBL GDBL

GDBL GDBLGDBL x

GDBL GDBL

J SK

J SK

(3.17)

Also, in this study, S = 0.01 is assumed at the interface of GDL and gas

channel to determine the integral constant because it depends on the channel

condition and a constant value is assigned, in general [51, 52, 101]. The

values of variables used in the calculations are listed in Table 3.1.

63

Table 3.1 Values of variables for multi-layer GDL model

Parameters Values

Thickness of GDL (MPL/GDBL) 350 (50/300) μm

Kozeny constant 6 [51,52]

Fiber diameter of GDBL 9 μm [100]

Fiber diameter of MPL 1 μm [51,52]

Porosity of MPL 0.5 [101]

Saturation at channel 0.01 [51,52]

Faraday constant 96485 C mol-1

Molecular weight of water 18 g mol-1

Surface tension 0.0725 N m-1 [101]

Kinematic viscosity 3.5 10-7 m2 s-1 [101]

Density of water 1 g cm-3

Current density 1 A cm-2

Active area 1 cm2

64

3.3 Single GDBL with/without MPL

Fig 3.2 shows the liquid water saturation profile across the single

GDBL (350 μm) with various porosity and contact angle. For this, identical

properties were used for GDBL-1, GDBL-2 and MPL. In addition, Fig 3.3

shows the liquid water saturation profile across the MPL (50 μm) coated

single GDBL (300 μm). In this case, identical properties were used for

GDBL-1 and GDBL-2, while the properties of MPL were changed to the

values listed in Table 3.1. The same contact angle of 140˚ and the same

porosity of 0.8 were used for the GDBL as default values. The liquid water

saturation in GDBL is decreased as the porosity and contact angle were

increased (high contact angle means high hydrophobic). On the other hand,

MPL makes a sudden change in local saturation near the interface of MPL

and GDBL-1 (at δ=50 μm). The local saturation must satisfy capillary

pressure equilibrium at the interface. The small porosity and fiber diameter

of MPL makes a large capillary pressure gradient through the MPL and it is

effective for water removal. Fig. 3.4 shows the amount of water remaining

(water retention) in GDL as a function of porosity and contact angle of

single GDBL at a current density of 1 A/cm2. It was found that the amount of

water remaining in GDL is decreased inserting MPL. Zhan et al. [40, 41]

found that liquid water volume remaining in the GDL is decreased and the

65

water flux through the GDBL is increased inserting MPL between catalyst

layer and GDBL. They also concluded that thin MPL with high porosity

effectively removes water in GDL. In addition to their result, it was found

that the liquid water saturation in MPL is almost unchanged with the same

properties of MPL even though the properties of GDBL are changed. The

reason is that the local saturation in MPL near the interface was almost

unchanged because the MPL has extremely lower porosity than GDBL.

66

(a) Porosity (with contact angle of 140˚)

(b) Contact angle (with porosity of 0.8)

Fig. 3.2 Liquid water saturation profile across the single GDBL (350 μm)

without MPL

0.00

0.05

0.10

0.15

0.20

0.25

0 50 100 150 200 250 300 350

S

x (μm)

ε = 0.6ε = 0.7ε = 0.8ε = 0.9

0.00

0.05

0.10

0.15

0.20

0.25

0 50 100 150 200 250 300 350

S

x (μm)

θ = 120°θ = 130°θ = 140°θ = 150°

67

(a) Porosity (with contact angle of 140˚)

(b) Contact angle (with porosity of 0.8)

Fig. 3.3 Liquid water saturation profile across the MPL (50 μm) coated

uniform GDBL (300 μm)

0.00

0.05

0.10

0.15

0.20

0.25

0 50 100 150 200 250 300 350

S

x (μm)

ε = 0.6ε = 0.7ε = 0.8ε = 0.9

0.00

0.05

0.10

0.15

0.20

0.25

0 50 100 150 200 250 300 350

S

x (μm)

θ = 120°θ = 130°θ = 140°θ = 150°

Fig. 3.4 The effect

6

of MPL on t

8

the water remmoval in GD

L

69

3.4 Double GDBL with MPL

3.4.1 The effect of porosity gradient

Porosity, which is highly related to the structure design of GDL, is an

important parameter in terms of water management. Figs. 3.5 and 3.6 show

the effect of porosity gradient on liquid water saturation and the amount of

water remaining in GDL. In order to investigate the effect of gradient

direction on water management, positive gradient of 0.8/0.9 and negative

gradient of 0.9/0.8 was introduced to GDBL-1/GDBL-2. The contact angle

of GDBL was fixed to 140˚. In addition, the thickness of GDBL-1 was

adjusted from 0 μm (the same with single GDBL of GDBL-2) to 300 μm

(the same with single GDBL of GDBL-1) to evaluated the effect of thickness

ratio of GDBLs.

Fig. 3.5 shows the effect of porosity gradient in positive direction

(0.8/0.9 for GDBL-1 and GDBL-2, respectively) on the liquid water

saturation and the amount of liquid water remaining in each layer. The

thickness of GDBL-1 was changed from 0 μm (0% of total GDBL thickness)

to 300 μm (100% of total GDBL thickness) with the same contact angle of

140˚. When the thickness of GDBL-1 was 0 μm, The GDL is the same to

70

MPL coated single GDBL with uniform porosity of 0.8. As shown in Fig 3.5,

the porosity gradient in positive direction helps water removal with proper

range of thickness ratio. In this case, the thickness of 120 μm (20%) for

GDBL-1 was the best choice for water removal.

Fig. 3.6 shows the effect of porosity gradient in negative direction

(0.9/0.8 for GDBL-1 and GDBL-2, respectively) on the liquid water

saturation distribution and the amount of liquid water remaining in each

layer. It was found that the porosity gradient in negative direction hindered

water removal. If effective water retention is important (i.e. self-humidified

fuel cell), however, it is expected that porosity gradient in negative direction

is useful on performance improvement. Total amount of water remaining in

GDL was increased as the thickness of GDBL-1 was increased up to 190 μm

and then it started to decrease. The porosity gradient in negative direction

helps water retention with proper range of thickness ratio. In this case, with

the thickness of 190 μm (20%) for GDBL-1, the amount of water remaining

in GDL was maximized. Since the supplied dry gas can be humidified by the

produced water passing through the GDL, it is expected that the performance

of self-humidified fuel cell will be improved. However, since the water

remaining in the GDL occupies the pore volume, gas diffusion should be

considered.

71

Fig. 3.7 shows effective diffusion factor profile across the GDL

calculated by Eq. (3.14). Since the thicknesses of GDBL-1 and GDBL-2

were fixed to 150 μm, the average porosities of GDBL were the same to 0.85

in all GDLs. Compared with the result of uniform single GDBL, effective

diffusion factor of GDL is increased with porosity gradient in positive

direction. It means that the amount of water remaining in GDL is reduced.

Thus, it is concluded that porosity gradient of double GDBL in positive

direction is effective on water removal. On the other hand, effective

diffusion factor of GDL is decreased with porosity gradient in negative

direction. It means that the amount of water remaining in GDL is increased.

Thus, it is concluded that porosity gradient of double GDBL in negative

direction is effective on water retention.

72

(a) Liquid water saturation profile

(b) The amount of water remaining in each layer

Fig. 3.5 The effect of double GDBL with porosity gradient in positive

direction on the liquid water saturation and water retention

capacity of GDL

0.0

0.1

0.2

0.3

0 50 100 150 200 250 300 350

S

x (μm)

(0, 300) (60, 240)

(120, 180) (180, 120)

(240, 60) (300, 0)

(δGDBL-1, δGDBL-2), μm

εGDBL-1=0.8, εGDBL-2 =0.9

0.0

1.0

2.0

3.0

4.0

5.0

0 50 100 150 200 250 300

Q (

mg

)

δGDBL-1 (μm)

MPLGDBL-1GDBL-2Total

εGDBL-1=0.8, εGDBL-2 =0.9

73

(a) Liquid water saturation profile

(b) The amount of water remaining in each layer

Fig. 3.6 The effect of double GDBL with porosity gradient in negative

direction on the liquid water saturation and water retention

capacity of GDL

0.0

0.2

0.4

0.6

0.8

0 50 100 150 200 250 300 350

S

x (μm)

(0, 300) (60, 240)

(120, 180) (180, 120)

(240, 60) (300, 0)

(δGDBL-1, δGDBL-2), μm

εGDBL-1=0.9, εGDBL-2 =0.8

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

0 50 100 150 200 250 300

Q (

mg

)

δGDBL-1 (μm)

MPLGDBL-1GDBL-2Total

εGDBL-1=0.9, εGDBL-2 =0.8

74

Fig. 3.7 Effective diffusion factor profile across the GDL with the same

average porosity of 8.5 for GDBL

0.0

0.2

0.4

0.6

0.8

1.0

0 50 100 150 200 250 300 350

Eff

ecti

ve d

iffu

sio

n f

acto

r, β

x (μm)

0.85/0.850.80/0.900.90/0.80

εGDBL-1/εGDBL-2

GDBL-1 GDBL-2MPL

75

3.4.2 The effect of hydrophobicity gradient

Hydrophobicity, which is represented by contact angle, is also

important parameter in terms of water management. Fig. 3.8 shows the effect

of hydrophobicity gradient in positive direction on the liquid water

saturation and the amount of water remaining in GDL. Fig. 3.9 shows the

effect of hydrophobicity gradient in negative direction. For convenience,

contact angle arrangements of 120°/140° and 140°/120° were introduced to

GDBL-1/GDBL-2 for positive gradient and negative gradient, respectively.

The porosity of GDBL was fixed to 0.8. The thickness of GDBL-1 was also

adjusted from 0 μm (single GDBL with GDBL-2) to 300 μm (single GDBL

with GDBL-1).

In the case of hydrophobicity, the amount of water remaining in GDL is

increased by hydrophobicity gradient in negative direction. Then, the amount

of water remaining in GDL is decreased by hydrophobicity gradient in

positive direction. In terms of capillary pressure, this result was reasonable.

If the saturation is assumed as a constant, in hydrophobic media, the

capillary pressure decreases with increase of porosity and decrease of

contact angle. Hence, it was concluded that the arrangement of GDBL-1

with relatively high capillary pressure and GDBL-2 with relatively low

capillary pressure (porosity gradient in positive direction or hydrophobicity

76

gradient in negative direction) are effective on water removal. On the other

hand, the porosity gradient in negative direction and hydrophobicity gradient

in positive direction are effective on water retention.

Fig. 3.10 shows effective diffusion factor profile across the GDL

calculated by Eq. (3.14). Since the thicknesses of GDBL-1 and GDBL-2

were fixed to 150 μm, the average porosities of GDBL were the same to 130°

in all GDLs. Compared with the result of uniform single GDBL, effective

diffusion factor of GDL is increased with hydrophobicity gradient in

negative direction. Thus, it is concluded that hydrophobicity gradient of

double GDBL in negative direction is effective on water removal. On the

other hand, effective diffusion factor of GDL is decreased with

hydrophobicity gradient in positive direction. Thus, it is concluded that

hydrophobicity gradient of double GDBL in positive direction is effective on

water retention.

77

(a) Saturation profile

(b) The amount of water remaining in each layer

Fig. 3.8 The effect of double GDBL with hydrophobicity gradient in

positive direction on the liquid water saturation and water

retention capacity of GDL

0.0

0.1

0.2

0.3

0 50 100 150 200 250 300 350

S

x (μm)

(0, 300) (60, 240)(120, 180) (180, 120)(240, 60) (300, 0)

(δGDBL-1, δGDBL-2), μm

θGDBL-1=120°, θGDBL-2 =140°

0.0

1.0

2.0

3.0

4.0

5.0

0 50 100 150 200 250 300

Q (

mg

)

δGDBL-1 (μm)

MPLGDBL-1GDBL-2Total

θGDBL-1=120°, θGDBL-2 =140°

78

(a) Saturation profile

(b) The amount of water remaining in each layer

Fig. 3.9 The effect of double GDBL with hydrophobicity gradient in

negative direction on the saturation and water retention capacity

of GDL

0.0

0.1

0.2

0.3

0 50 100 150 200 250 300 350

S

x (μm)

(0, 300) (60, 240)(120, 180) (180, 120)(240, 60) (300, 0)

(δGDBL-1, δGDBL-2), μm

θGDBL-1=140°, θGDBL-2 =120°

0.0

1.0

2.0

3.0

4.0

5.0

0 50 100 150 200 250 300

Q (

mg

)

δGDBL-1 (μm)

MPLGDBL-1GDBL-2Total

θGDBL-1=140°, θGDBL-2 =120°

79

Fig. 3.10 Effective diffusion factor profile across the GDL with the same

average contact angle of 130º for GDBL

0.0

0.2

0.4

0.6

0.8

1.0

0 50 100 150 200 250 300 350

Eff

ecti

ve d

iffu

sio

n f

acto

r, β

x (μm)

130°/130°120°/140°140°/120°

ϴGDBL-1/ϴGDBL-2

GDBL-1 GDBL-2MPL

80

3.5 Triple GDBL with MPL

In order to provide further information about the effect of multi-layer

GDL, the studies were extended to triple GDBL. For convenience, the

thicknesses of GDBL-1, GDBL-2 and GDBL-3 were fixed to the same value

of 100 μm.

Fig. 3.11 shows the effect of porosity gradient in triple GDBL. Porosity

arrangements of 0.90/0.85/0.80, 0.85/0.85/0.85, 0.80/0.85/0.90 were

introduced to GDL-3LL1, GDL-3LU, and GDL-3LL2, respectively. Thus,

porosity gradient of GDBL was negative direction in the case of GDL-3LL1

and positive direction in the case of GDL-3LL2. The average porosities of

GDBL were the same to 0.85 in all GDLs. Compared to the result of GDL-

3UP with uniform GDBL, the liquid water remaining in GDL was reduced

and average gas diffusion factor was increased by porosity gradient in

positive direction, whereas the liquid water remaining in GDL was enhanced

and average gas diffusion factor was decreased by porosity gradient in

negative direction. This tendency was the same with the result of double

GDBL.

Fig .3.12 shows the effect of parabolic arrangement. Porosity

arrangements of 0.85/0.80/0.85, 0.85/0.85/0.85, 0.85/0.90/0.85 were

introduced to GDL-3LP1, GDL-3LU, and GDL-3LP2, respectively. Thus,

81

the porosity of the middle GDBL, GDBL-2, was relatively low in the case of

GDL-3LP1 and relatively high in the case of GDL-3LP2. Compared to the

result of GDL-3LU, the liquid water remaining in GDL was enhanced and

average gas diffusion factor was decreased in both of parabolic arrangements.

With parabolic arrangement of porosity, the water retention effect and the

water removal effect appeared at the same time in a GDL. However, the

water retention is stronger than water removal. It was the reason for the

result of parabolic arrangement.

82

Fig. 3.11 The amount of water remaining in GDL containing triple GDBL

with linear arrangement and the consequential average gas

diffusion factor of GDL

0.0

0.2

0.4

0.6

0.8

0

2

4

6

8

10

GDL-3LL1 GDL-3LU GDL-3LL2

βav

g

Q (

mg

)

GDL with triple GDBL

βavg

MPLGDBL-1GDBL-2

GDBL-3

Qlayer

0.50

0.90

0.85

0.80

0.50

0.850.85

0.85

0.500.800.850.90

83

Fig. 3.12 The amount of water remaining in GDL containing triple GDBL

with parabolic arrangement and the consequential average gas

diffusion factor of GDL

0.0

0.2

0.4

0.6

0.8

0

2

4

6

8

10

GDL-3LP1 GDL-3LU GDL-3LP2

βav

g

Q (

mg

)

GDL with triple GDBL

βavg

MPLGDBL-1GDBL-2

GDBL-3

Qlayer

0.50

0.850.85

0.85

0.50

0.85

0.80

0.85

0.500.85

0.90

0.85

84

3.6 Summary

The effect of structure design and hydrophobicity of GDBL on water

management of GDL was investigated to improve the start-up and operating

performance of self-humidified PEMFCs. For this, an analytic model was

obtained based on the capillary pressure–saturation relationship with the

Leverett J-function. In this model, structure design and hydrophobicity of

GDBLs were represented by the measurable parameters, porosity and

contact angle, respectively.

With this analytic model, the liquid water saturation distribution and the

amount of water remaining in GDL were evaluated as a function of porosity,

contact angle, and thickness ratio of each GDBL. As a result, it was

concluded that porosity gradient in positive direction and hydrophobicity

gradient in negative direction are effective on water removal. On the other

hand, porosity gradient in negative direction and hydrophobicity gradient in

positive direction is effective on water retention. Although effective gas

diffusion factor was also considered, further work would be needed to

optimize the structural design and manufacturing of GDBLs.

85

Chapter 4. Performance improvement of self-

humidified fuel cell with multi-layer GDL

4.1 Introduction

In this chapter, the effect of multi-layer GDL with double GBBL was

investigated experimentally. The structure of a GDBL was modified to

enhance the self-humidification effect in PEMFCs. Although it is expected

that manufacturing functionalized multi-layer GDBL is possible in near

future, there were many difficulties in manufacturing multi-layer GDBL in

present time. Hence, in this study, the modified GDLs were prepared by

stacking two GDLs for convenience. Therefore, in order to evaluate the

effect of stacking, some characteristics of the GDLs such as contact angle,

resistance, and vapor permeation rate were measured as well as

electrochemical performance. In addition, self-humidification effect of

modified GDL was evaluated.

86

4.2 Dry start-up with modified GDL

4.2.1 Preparation of the GDL

Fig. 4.1 shows a schematic illustration of the GDL design employed in

this study. In Chap. 2, GDL-A was used to measure the dry start-up

performance of fuel cell. GDL-A and GDL-A′ had conventional designs and

consisted of a single MPL and a single GDBL (GDBL-A and GDBL-A′ for

GDL-A and GDL-A′, respectively), whereas the modified GDL, GDL-A´C,

contained a single MPL and double GDBL. GDL-A´C was prepared by

stacking GDL-A´ and GDBL-C. As listed in Table 1.1, in order to improve

the water retention capacity of GDL-A′C, GDBL-C had a lower porosity

than GDBL-A′. In all the cases, carbon paper-type GDBLs were used, to

exclude the possibility of anisotropic effects [93]. The thicknesses of the

GDLs were measured using a thickness gauge (Mitutoyo Co., Japan). The

measured thicknesses of GDBL-C were 120±5 µm, whereas the total

thicknesses of GDL-A´C (355±3 µm) were approximately equal to GDL-A

(348±4 µm). The thickness of the MPL can be calculated by subtracting the

total thickness of the GDBL from the total thickness of the GDL.

87

(a) GDL-A (conventional GDL)

(b) GDL-A′C (functionalized ML-GDL)

Fig. 4.1 Schematic of the conventional GDL and functionalized ML-GDL

for water retention

MPL

GDBL-A

MPL

GDBL-C

GDBL-A’

εlow

εhigh

88

Table 4.1 Properties of GDBLs employed in dry start-up test

Property Unit GDL-A GDL-A′C

GDBL-A GDBL-A′ GDBL-C

Type - Carbon

paper

Carbon

paper

Carbon

paper

PTFE wt% 5 5 5

Porosity - 0.9 0.88 0.8

Areal weight g/m2 54 40 44

Thickness 305±6 190±4 120±5

Total thickness 348±4 (GDL-A) 355±3 (GDL-A′C)

89

4.2.2 The results of dry start-up with multi-layer GDL

With GDL-A, the dry start-up was successful applying the WSP for

more than 2 min at a cell temperature of 45°C. With 1 min of WSP, the fuel

cell started to operate during WSP, but shut-down immediately when high

flow rates were supplied. It was found that 1 min of WSP was too short to

increase the membrane conductivity sufficiently. In consideration of

practical use of self-humidified fuel cell, the duration of WSP should be

reduced.

As shown in Fig 4.2, during WSP, the cell performance was recovered

more quickly with GDL-A′C than GDL-A. It means that the membrane

could be hydrated more quickly with GDL-A′C. It may be the reason for the

successful dry start-up only with 1 min of WSP. Moreover, stable

performance was observed when high flow rates were supplied after 1 min

of WSP regardless of flow arrangement. With GDL-A, while the fuel cell

with the co-flow arrangement showed increased cell performance initially,

the performance dropped rapidly over time, which is attributed to the gradual

removal of the water retained during the WSP.

Fig. 4.3 shows the steady-state performance of fuel cell under various

temperatures. This experiment was conducted by maintaining continuous

90

operation at a constant current of 6 A with gas flow rates of 200 mL/min and

63 mL/min for air and hydrogen, respectively. These flow rates are

equivalent to a SR of 2.0 and 1.5 for air and hydrogen at the current of 6 A.

The operating temperatures were controlled by an external heating system

and were measured when the cell was in a steady state for more than 1 h. It

was found that while the performance of the cell with the co-flow

arrangement dropped at temperatures above 35°C, the performance of the

cell with the counter-flow arrangement was improved up to 50°C, beyond

which it started to drop.

It was found that steady-state performance at each temperature was

improved applying GDL-A′C. In addition, stable operation was possible at

higher temperatures. It can be concluded that self-humidification effect of

GDL was enhanced due to modified structure of GDBL. Fig. 4.4(a) shows

the result of electrochemical impedance spectroscopy (EIS). During the

measurement, the current density was fixed at 1 A so that the same amount

of water was produced by fuel cell reaction. The water remaining in the GDL

occupies the pore volume and mass transfer resistance increased. Since the

diameter of semicircle of EIS reflects the size of mass transfer resistance, Fig.

4.4(a) provides the evidence of enhanced water retention capacity of GDL

with modified structure. In addition, the result of high frequency resistance

91

(HFR) shown in Fig. 4.4(b) provides information about the proton

conductivity of the membrane. Then, it can be concluded that the self-

humidification effect of GDL was remarkably improved with GDL-A′C.

However, it should be noted that excessive water retention capacity of GDL

causes mass transfer problem, so called flooding.

92

(a) Co-flow arrangement

(b) Counter-flow arrangement

Fig. 4.2 The result of dry start-up with functionalized ML-GDL at 45°C

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0 200 400 600 800 1000

Cu

rren

t d

ensi

ty (

A/c

m2)

Time (s)

GDL-A'C (1)

GDL-A (1)

GDL-A (2)

(WSP duration, min)

Load on0.4 V

AN

CA

45°C

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0 200 400 600 800 1000

Cu

rren

t d

ensi

ty (

A/c

m2)

Time (s)

GDL-A'C (1)

GDL-A (1)

GDL-A (2)

Load on

(WSP duration, min)

0.4 V

45°C

AN

CA

93

(a) Co-flow arrangement

(b) Counter-flow arrangement

Fig. 4.3 The improvement of steady-state performance with functionalized

ML-GDL under various temperatures at 6.67 A/cm2 (SR 2.0 and

1.5 for air and hydrogen, respectively)

0.0

0.2

0.4

0.6

0.8

30 40 50 60 70 80

Vo

ltag

e (V

)

Temperature (oC)

GDL-A'CGDL-A

6 A

AN

CA

0.0

0.2

0.4

0.6

0.8

30 40 50 60 70 80

Vo

ltag

e (V

)

Temperature (oC)

GDL-A'CGDL-A

6 A

AN

CA

94

(a) EIS (Electrochemical Impedance Spectroscopy)

(b) HFR (High Frequency Resistance) at 45°C

Fig. 4.4 The result of impedance measurement

0

20

40

60

80

0 20 40 60 80 100

Rea

ctan

ce (

mΩ

)

Resistance (mΩ)

GDL-A (25°C)GDL-A (45°C)GDL-A'C (25°C)GDL-A'C (45°C)

0

20

40

60

80

100

120

140

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

HF

R (

mΩ

·cm

2)

Current density (A/cm2)

GDL-A

GDL-A'C

95

4.3 The effect of stacking

Although it is expected that manufacturing functionalized multi-layer

GDBL is possible in near future, there were many difficulties in

manufacturing multi-layer GDBL in present time. Hence, in this study, the

modified GDLs were prepared by stacking two GDLs for convenience.

The interfacial gap formed by the stacking of GDLs has been reported

to have little effect on the relative permeability of water [102]. In order to

understand the effect of stacking the GDLs, GDL-A´B was prepared by

stacking GDL-A´ and GDBL-B, whereas GDL-A´C was prepared by

stacking GDL-A´ and GDBL-C. As listed in Table 4.2, GDBL-A′ and

GDBL-B had the same porosity. Although the porosity of GDBL-A was

somewhat higher than that of GDBL-A´, the actual difference in the

porosities may be negligible owing to the manufacturing error and

compression effects. Therefore, comparing the performances of GDL-A´B

and GDL-A would provide a means of evaluating the effect of stacking the

GDLs.

The GDL samples were analyzed using a digital microscope (Dino-Lite

AM-413T5, AnMo Electronics Co., Taiwan) at a magnification of ×500. The

compression ratio of the GDL samples were adjusted to 20%, which is the

96

same value employed in the electrochemical performance tests. Fig. 4.5

shows the cross-sectional digital micrographs of the GDLs. Macro-scale

spaces were found at the interface of the GDBLs in the case of GDL-A´B

and GDL-A´C. Although stacking is believed to have little effect on the

relative permeability of water, it may still increase the contact resistance and

water accumulation at the GDBL interface. Therefore, the electrical

resistance, contact angle, and vapor permeation rate of the GDLs were

measured to evaluate the effects of stacking and the modified structural

design.

Fig. 4.6 shows the results of the resistance measurements. The

resistances of circular GDL samples (2.48 mm diameter) were measured

using gold-coated plates and a DC milli-Ohm meter (GOM-802, GW

INSTEK, Japan). Under low pressure conditions below 0.1 MPa, the

resistances of the modified GDLs (GDL-A´B and GDL-A´C) were

significantly higher than that of GDL-A. At a pressure of 0.02 MPa, the

measured resistances of GDL-A, GDL A´B, and GDL-A´C were greater than

140, 300, and 230 mΩ cm2, respectively. The high values of resistances of

GDL-A´B and GDL-A´C were attributed to the stacking of two GDLs,

which caused additional contact resistances at the interface of the GDLs. In

addition, the resistance of GDL-A´C was lower than that of GDL-A´B,

97

because the areal weight of GDBL-C was higher than that of GDBL-B.

However, the resistances of the GDLs decreased rapidly as the pressure

increased. This relationship between the resistance of the GDLs and the

compression pressure was consistent with previous results reported in the

literature [103]. In general, during fuel cell assembly, the GDL is

compressed to the thickness of the gasket, in order to reduce contact

resistance and prevent the leakage of reactant gases. In this study, the GDLs

were compressed at pressures above 1 MPa, which resulted in a decrease in

the resistances below 15 mΩ•cm2 and the gap of resistances were negligible.

Consequently, although the resistances of GDL-A´B and GDL-A´C were

slightly higher than that of GDL-A, it was reasonable to conclude that the

increase in the contact resistance had negligible effect on the fuel cell

performance.

The degree of hydrophobicity can be evaluated by measuring the

contact angle [15-17, 38, 39]. The average contact angles of the GDBLs used

in this study were distributed around 149.5°. In porous media such as GDLs,

the capillary pressure is a function of cos(θ), where θ is the contact angle

[32-38]. Therefore, the differences in the hydrophobicity among the various

GDBLs may have negligible effect on the cell performance.

Self-humidification could be improved by enhancing the water

98

retention capacity in the GDLs [35-40]. Fig. 4.7(a) shows the apparatus

proposed in previous studies for evaluating the vapor permeation rates

through the GDL [33-35]. The vapors under and inside the GDLs are swept

by the upper gas stream. Thus, a higher vapor permeation rate implies a

lower water retention capacity of the GDLs. The tests were conducted using

9 cm2 pieces of the GDL samples. Pure water (300 mL) at 65°C was held in

a container below the GDL samples, whereas dry air was supplied above the

GDL samples at a flow rate of 0.5 L/min. The vapors passing through the

GDL were condensed and weighed using an electronic balance (HM-300,

AND, Japan). As shown in Fig. 4.7(b), the vapor permeation rate was

significantly reduced with GDL-A´C, whereas GDL-A´B showed a similar

vapor permeation rate as GDL-A.

In order to investigate the effect of stacking GDBL more clearly,

electrochemical performance of fuel cells were measured under the

conditions listed in Table 4.3. As shown in Fig 4.8, GDL-A´B showed a

similar cell performance compared to GDL-A, whereas GDL-A´C showed

significantly improved cell performance. Fig. 4.9 shows the effects of RH

and SR on the performance of fuel cells containing GDLs at a current

density of 0.778 A/cm2. As mentioned above, compared to GDL-A and

GDL-A´B, the cell containing GDL-A´C performed better at an RH of 0%,

99

whereas it exhibited a poorer performance at 100% RH. In addition, the cell

performance could be readily recovered by increasing the flow rate of air. At

an RH of 100%, the performance of the cell containing GDL-A´C was

almost the same as that of the cell containing GDL-A at an air SR of 3.5.

Based on the results, it may be concluded that the structural design of

the GDBL has a major effect on the water retention capacity of GDLs,

whereas stacking GDBLs has negligible effect on the water retention

capacity. Owing to its simplicity, stacking can be applied readily to improve

the performance of self-humidified PEMFCs without additional changes to

the system.

100

Table 4.2 Properties of GDLs prepared to verify the effect of stacking

Property Unit GDL-A GDL-A′B GDL-A′C

Type (GDBL) - Carbon

paper

Carbon

paper

Carbon

paper

PTFE (GDBL) % 5 5/5 5/5

Porosity (GDBL) - 0.9 0.88/0.88 0.88/0.8

Areal weight

(GDBL) g/m2 54 40/28 40/44

Thickness (GDBL) 305±6 311±4 309±4

Total thickness 348±4 357±5 355±3

Fig. 4.5 Cross-s

of stack

sectional mo

king

10

(a) GDL

(b) GDL-

(c) GDL

rphology of

01

L-A

-A′B

-A′C

GDLs prepa

ared to verify

y the effect

Fig

(a) S

(

. 4.6 Schem

the resu

0

50

100

150

200

250

0.0

Res

ista

nce

(mΩ

·cm

2 )

Schematic of

(b) The resul

atic of appar

ults

0.2

10

the apparatu

lt of the resis

ratus for me

0.4 0

Pressu

02

us for resistan

stance measu

easuring the r

0.6 0.8

ure (MPa)

nce measurem

urement

resistance of

1.0

GDL-A

GDL-A'B

GDL-A'C

ment

f GDL and

1.2

(

Fig.

(a) Schemati

(b) Th

. 4.7 Schem

through

300

350

400

450

500

550

600

Vap

or

per

mea

tio

n r

ate

(g/h

·m2 )

c of apparatu

e result of th

atic of appa

h the GDL a

GDL-A

10

us for the vap

he vapor perm

ratus for me

nd the result

GD

G

03

por permeati

meation rate

easuring the

ts

DL-A'B

GDL

ion rate meas

measuremen

vapor perm

GDL-A'C

surement

nt

eation rate

C

104

Table 4.3 Operating conditions of fuel cell to verify the effect of

stacking

Condition Unit Values

Operation temperature 65

Relative Humidity % Cathode (Air): 0, 25

Anode (H2): 100

Stoichiometric ratio - Cathode (Air) : 2.0

Anode (H2) : 1.5

105

(a) RHair 0%

(b) RHair 25%

Fig. 4.8 Comparison of the cell performances at 65°C with fully

humidified hydrogen, SR 2.0 and 1.5 for air and hydrogen,

respectively.

0.2

0.4

0.6

0.8

1.0

1.2

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Vo

ltag

e (V

)

Current density (A/cm2)

GDL-A

GDL-A'B

GDL-A'C

0.2

0.4

0.6

0.8

1.0

1.2

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Vo

ltag

e (V

)

Current density (A/cm2)

GDL-A

GDL-A'B

GDL-A'C

106

Fig. 4.9 The effect of RH and SR of supplied air on the performance of

fuel cells at constant current density of 0.778 A/cm2

0.3

0.4

0.5

0.6

0.7

0.8

1.5 2.0 2.5 3.0 3.5

Vo

ltag

e (V

)

SRair

GDL-A

GDL-A′B

GDL-A′C

RHair 0%

RHair 50%

RHair 100%

107

4.4 Self-humidification effect of modified GDL

4.4.1 Preparation of the GDL

As a functionalized GDL on water retention, GDL-A´D was prepared

by stacking GDL-A´ and GDBL-D in addition to GDL-A´C. Fig. 4.10

schematic of multi-layer GDLs with double GDBL prepared to enhance

water retention capacity of GDL. GDBL-D was similar to GDBL-C except

the PTFE contents. For the GDBL-D, 20 wt% of PTFE was treated whereas

5 wt% of PTFE was treated for other GDBLs. The measured contact angle of

GDBL-D was about 153° and somewhat higher than that of other GDBLs.

Since contact angle represents hydrophobicity, GDL-A´D was MPL coated

double GDBL with hydrophobicity gradient in positive direction.

4.4.2 Experimental procedure

The same apparatus explained in chap. 2 was also used in this study.

After the activation process was completed, a verification test was performed

to ensure successful assembly of the fuel cell. The fuel cell performance was

then characterized by measuring current-voltage polarization curves under

various experimental conditions. In order to evaluate the effect of flow rate,

108

the cell performances were measured with various SR of air, namely 1.5, 2.0,

2.5, 3.0, and 3.5. The SR of hydrogen was fixed at 1.5 in all the cases.

Whenever the RH of air was adjusted to a different value, the fuel cell was

reactivated and the performance measurements were repeated.

Fig. 4.1 shows the cell performance with GDL-A. The experiment was

conducted with dry hydrogen at 25°C, fully humidified hydrogen at 25°C

and fully humidified hydrogen at 65°C. In all cases, dry air was supplied.

When fully humidified hydrogen is supplied, the cell performance with co-

flow arrangement improved remarkably. The cell performances were almost

the same at 25°C. Furthermore, when the cell temperature increased up to

65°C, the cell performance with co-flow arrangement is better than that with

counter-flow arrangement. Thus, co-flow arrangement is better than counter-

flow arrangement if hydrogen recirculation is considered. It is because the

effect of produced water is concentrated to the membrane at cathode outlet in

the case of co-flow arrangement. On the other hand, in the case of counter-

flow arrangement, the produced water is distributed and the overall

conductivity is lowered at high temperature. Therefore, co-flow arrangement

was employed for the operating conditions listed in Table 4.4.

109

(a) GDL-A′C

(b) GDL-A′D

Fig. 4.10 Schematic of multi-layer GDLs with double GDBL prepared to

enhance water retention capacity of GDL

MPL

GDBL-C

GDBL-A’

εlow

εhigh

MPL

GDBL-D

GDBL-A’ θlow

θhigh

110

Fig. 4.11 The effect of hydrogen humidity and operating temperature on

the performance of fuel cells with dry air (SR 2.0 and 1.5 for air

and hydrogen, respectively)

0.2

0.4

0.6

0.8

1.0

1.2

0.0 0.2 0.4 0.6 0.8 1.0 1.2

Vo

ltag

e (V

)

Current density (A/cm2)

25 / 0% / Co-flow25 / 100% / Co-flow65 / 100% / Co-flow25 / 0% / Counter-flow25 / 100% / Counter-flow65 / 100% / Counter-flow

T / RHair / flow arrangement

111

Table 4.4 Operating conditions of fuel cell to verify the self-

humidification effect of GDL

Condition Unit Values

Operation temperature 65

Supplied gas (cathode/anode) - Air/H2

Relative Humidity % Cathode : 0, 25, 50, 75, 100

Anode : 100

Stoichiometric ratio - Cathode : 1.5, 2.0, 2.5, 3.0, 3.5

Anode : 1.5

112

4.4.3 Experimental result

Fig. 4.12 shows the polarization curves of the fuel cells containing

GDLs measured at various SRs of air. The RH of the supplied air was set to

0%, whereas that of the supplied hydrogen was set to 100%. The SR of

hydrogen was set to 1.5. As shown in Fig. 4.12(a), the performance of the

fuel cell containing GDL-A slightly improved as the flow rate of air was

increased, up to an air SR of 2.0, beyond which the fuel cell performance

dropped significantly. It may be noted that a SR of 1.0 implies that just the

right amount of gas was used for the fuel cell reaction. If air is supplied at

the exact SR of 1.0, concentration loss occurs due to oxygen starvation at the

outlet of the channel. To prevent concentration loss, air is generally

oversupplied to the fuel cell [19]. In general, the oversupply of air has

conflicting effects on the cell performance under low humidity conditions.

As the flow rate of air increases, the oxygen concentration increases and it

positively affects the cell performance. On the other hand, the membrane is

dehydrated by an excess supply of dry air, resulting in low proton

conductivity of the membrane, which negatively affects the cell performance.

It is worth noting that excessive water can be removed by the oversupply of

air under high humidity conditions. In the present study, the structure of

GDBL was modified to enhance the water retention capacity of the GDL,

113

which was expected to highly humidify the supplied gas and improve the

hydration level of the membrane. Although it is difficult to directly correlate

the humidification effect to the amount of water retained in the GDL, it is

clear that the supplied dry gas can be humidified by the produced water as it

passes through the GDL [104]. As intended, the performance of the cell

containing GDL-A´C was significantly better than that of the cell containing

GDL-A, including at high current densities under all SR conditions, except

for the SR of 1.5. The concentration of oxygen in dry air is only 21%.

Moreover, the water retained in the GDL occupies the pores, as a result of

which the mass transfer resistance increases. Therefore, at an air SR of 1.5,

the performance of the cell containing GDL-A´C dropped rapidly,

particularly at high current densities.

Fig. 4.13 shows the polarization curves of the fuel cells containing

GDLs under various RH conditions of supplied air with SR of 2.0 and 1.5

for air and hydrogen, respectively. The RH of supplied hydrogen was set to

100%. As the humidity of the supplied air was increased, a higher proportion

of the water produced in the cell condensed and membrane dehydration was

eliminated. At low current densities, the amount of water produced was

relatively small and the porosity of the GDBL was large enough to prevent

flooding. Consequently, the structure of GDBL had little effect on the cell

114

performance at low current densities. Thus, for all the GDLs, the

performance of the fuel cells improved at low current densities as the RH of

the supplied air was increased. However, in the case of GDL-A´C,

concentration losses were observed at high current densities as the RH of the

supplied air was increased. At an RH of 100%, the supplied air is fully

hydrated and all the water produced exists in the liquid state. Consequently,

membrane dehydration disappears and flooding occurs. In the case of GDL-

A´C, the outer GDBL had smaller pores than the inner GDBL, posing

difficulties in discharging the water produced in the cell. This implies that

although GDL-A´C is resilient against dehydration, it is more vulnerable to

flooding than GDL-A.

Fig. 4.14 shows the dynamic responses of the fuel cell with prepared

GDLs. It is important to study the correlation between the transient response

of the PEMFCs and the characteristics of the GDLs [105]. Moreover, since

the GDBL structure was modified to enhance the water retention capacity,

unstable performance was expected with changes in the external load, owing

to transient water flooding. Therefore, the dynamic responses of the cells

were analyzed using the results obtained during the activation process,

because the dynamic responses of current with sudden changes in voltage

were measured during the activation process. As shown in Fig. 4.14, current

115

overshoot was observed in the case of GDL-A´C, particularly under high RH

conditions. When an external load was applied (OCV → 0.65 V), the fuel

cell started producing electricity and water. In addition, the cell performance

increased as the RH increased at 0.65 V. Therefore, it can be concluded that

the membrane was not sufficiently hydrated under this condition, which also

resulted in performance improvement with GDL-A´C at 0.65 V. When the

voltage was decreased from 0.65 V to 0.5 V, a current overshoot was

observed in the case of the cell containing GDL-A´C, particularly under high

RH conditions. However, in the cases of GDL-A, the fuel cell performance

improved as the RH increased, without overshoot at 0.5 V. Therefore, it may

be concluded that the membrane was not hydrated sufficiently. The amount

of water produced by the fuel cell reaction is directly related to the current.

Therefore, as the current increased, the pores in GDL were gradually

occupied by the produced water. In addition, under high RH conditions,

increased amounts of the produced water existed in the liquid phase without

evaporation and the effective gas permeability of the GDL decreased [102].

Therefore, mass transfer resistance increased and the cell performance

decreased, which explains the current overshoot in the case of GDL-A´C.

However, under low humidity conditions, the membrane was dehydrated and

most of the produced water was reused to humidify the supplied air.

116

Therefore, the current overshoots disappeared gradually as the RH of the

supplied air decreased. In addition, in the case of GDL-A´C, the cell

performance was higher than that with other GDLs at an RH of 0%, owing to

the enhanced water retention capacity. As shown in Fig. 4.14(b), the highest

overshoot was observed at the RH of 50%. At RH values above 50%, the

excess water in the GDL-A´C caused flooding, limiting the cell performance,

owing to increased mass transfer resistance. The performance of the fuel cell

containing GDL A´C at an RH of 0% was equivalent to that of the cell

containing GDL-A at an RH of 25%. In addition, the fuel cells containing

GDL-A´C exhibited improved performance for a wide range of external

loads up to an RH of 50%. The amount of vapors in the air having an RH of

100% at 25°C (saturated vapor at standard conditions) is equivalent to that in

the air having an RH of 12.66% at 65°C. Based on the results, it can be

concluded that self-humidification was improved significantly in fuel cells

using GDL-A´C.

117

(a) GDL-A

(c) GDL-A′C

Fig. 4.12 The effect of flow rate of supplied dry air on the performance of

the fuel cell at 65°C with fully humidified hydrogen

0.2

0.4

0.6

0.8

1.0

1.2

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Vo

ltag

e (

V)

Current density (A/cm2)

1.5

2.0

2.5

3.0

3.5

SRair

0.2

0.4

0.6

0.8

1.0

1.2

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Vo

ltag

e (

V)

Current density (A/cm2)

1.5

2.0

2.5

3.0

3.5

SRair

118

(a) GDL-A

(c) GDL-A′C

Fig. 4.13 The effect of relative humidity of supplied air on the performance

of the fuel cell at 65°C with fully humidified hydrogen (SR 2.0

and 1.5 for air and hydrogen, respectively)

0.2

0.4

0.6

0.8

1.0

1.2

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Vo

lta

ge

(V)

Current density (A/cm2)

0%

25%

50%

75%

100%

RHair

0.2

0.4

0.6

0.8

1.0

1.2

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Vo

ltag

e (V

)

Current density (A/cm2)

0%

25%

50%

75%

100%

RHair

Fig. 4.14 Dynamic

11

(a) GDL

(b) GDL-

responses o

19

L-A

-A′C

f fuel cells wwith the GDL

Ls

120

In the same manner, the performance test was conducted for GDL-A´D.

Although the amount of hydrophobic agent can be directly related to contact

angle, it is obvious that the hydrophobicity of GDL increases as the amount

of treated agent is increased. Thus, it was expected that this GDL is

functionalized on water retention. Fig. 4.6 shows the polarization curve of

fuel cells at 65°C with fully humidified hydrogen of SR 2.0 and 1.5,

respectively. As shown in the Fig. 4.18(b), the cell performance was

significantly improved applying GDL-A´D. Although contact angle of

GDBL-D is slightly higher than others, the improvement is remarkable.

Considering the decrease in porosity by PTFE treatment, this result was

reasonable. If the PTFE is added on the carbon paper GDL, the pore volume

is randomly filled by the PTFE. The effective porosity can be expressed

approximately as a function of PTFE content ω [106].

(1 )0.9

1o

PTFE o

(4.18)

where o is the original porosity before PTFE treatment.

Fig. 4.7 shows the effective porosity of GDL after PTFE treatment. If

the original porosity is 0.9, the effective porosity of GDBL with 20 wt%

PTFE is 0.878 while the effective porosity of GDBL with 5 wt% PTFE is

0.895. This could makes porosity gradient in negative direction and the

121

water retention would be enhanced. It was concluded that the self-

humidification effect of GDL and the performance of fuel cell were

significantly improved by applying functionalized GDBL.

122

(a) GDL-A′C

(b) GDL-A′D

Fig. 4.15 Cell performances with multi-layer GDLs for water retention

with porosity gradient and hydrophobicity gradient

0.2

0.4

0.6

0.8

1.0

1.2

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Vo

ltag

e (V

)

Current density (A/cm2)

GDL-A /

GDL-A'C /

GDL-A /

GDL-A'C /

RHair 0%

RHair 0%

RHair 100%

RHair 100%

0.2

0.4

0.6

0.8

1.0

1.2

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Vo

ltag

e (

V)

Current density (A/cm2)

GDL-A /

GDL-A'D /

GDL-A /

GDL-A'D /

RHair 0%

RHair 0%

RHair 100%

RHair 100%

123

Fig. 4.16 Effective porosity of GDL after PTFE treatment

0.5

0.6

0.7

0.8

0.9

1.0

0 5 10 15 20 25 30

Po

rosi

ty, ε P

TF

E

PTFE contents, ω (%)

0.9

0.8

0.7

εo = 0.9εo = 0.8εo = 0.7

124

4.5 Summary

Based on the analytic results, start-up and steady-state performance of

fuel cell was measured with the modified GDL containing double GDBL

with different porosity in negative direction (GDBL with lower porosity near

the flow channel). The result showed that shorter duration of WSP was

required for successful start-up and high performance with modified GDL.

In addition, since the modified GDL was prepared by stacking GDL, the

effect of stacking was also evaluated with the GDL containing double GDBL

with the same porosity. The result showed that the structural design of the

GDBL had a major effect on its water retention capacity, whereas stacking

has negligible effect on the water retention capacity. In additions, the self-

humidification effect and the performance of the fuel cell containing the

structurally modified GDBL were found to be significantly improved over a

wide range of operating conditions. We believe that this work provides new

insights into improving the performance of self-humidifying fuel cells by

structural design of the GDBL, rather than by modifying the MPL.

125

Chapter 5. Optimization of functionalized GDL

5.1 Introduction

In previous chapters, functionalized multi-layer GDL was suggested

and the effects of porosity and hydrophobicity (represented by contact angle)

gradient in GDL were investigated for the purpose of enhanced water

management. Porosity gradient in positive direction and hydrophobicity

gradient in negative gradient helps water removal. On the other hand,

porosity gradient in negative direction and contact angle gradient in positive

gradient helps water retention. The results are verified by experiments. The

cell performance was improved significantly by applying the functionalized

multi-layer GDL. As a result, it was concluded that functionalized multi-

layer GDL is effective on water management. However, further work is

needed to optimize the multi-layer GDL.

126

5.2 Optimization for water retention

In order to improve self-humidify effect, the water retention capacity of

GDL should be enhanced. In other words, the amount of liquid water

remaining in GDL should be as large as possible for effective water retention.

However, the remaining water in GDL occupies porous volume so that gas

diffusion is reduced. Hence, the gas diffusivity is also considered and there

are a lot of optimum arrangements for water retention.

The total thickness of GDBL was fixed to 300 µm. Then, the water

retention in GDL, Q is a function of porosity of GDBL-1, porosity of

GDBL-2, contact angle of GDBL-1, contact angle of GDBL-2, and thickness

of GDBL-1.

1 2 1 2 1( , , , , )GDBL GDBL GDBL GDBL GDBLQ f (5.1)

Fig. 5.1(a) shows the effect of contact angle gradient on water retention

with the same porosity of GDBLs. The result shows that the maximum Q

was obtained where the contact angle arrangement is 120° and 140° for

GDBL-1 and GDBL-2, respectively. With this value, Q increases remarkably

while the change in effective diffusion factor, β is negligible. Then, it was

expected that if the porosity gradient is also introduced to GDBL, the water

retention capacity will enhanced.

Fig. 5.1 Maxim

uniform

mum water re

m porosity an

12

(a)

(b)

etention and

nd various co

27

d average ga

ontact angle

s diffusion f

arrangemen

factor with

ts

128

Fig. 5.2(a) shows the maximum water retention capacity of GDL with

optimized thickness of GDBL-1. The result shows that water retention

capacity of GDL increases as the higher porosity gradient is applied.

However, higher porosity gradient causes flooding as shown in Fig. 5.2(d).

Hence, in order to avoid flooding, a constraint about β was introduced. Fig.

5.3 shows the results. As the minimum required average diffusion factor,

β increases, available range of porosity gradient decreases. When β

is 0.3 and 0.5, the optimal values of thickness of GDBL-1, porosity

arrangement, and water retention were obtained. Then, this process was

repeated for the contact angle arrangement of (120°, 120°) and (140°, 140°)

and the results are shown in Fig. 5.4 and Fig. 5.5, respectively. When β

is 0.3, the optimal values of thickness of GDBL-1, porosity arrangement, and

water retention were listed in table 5.1. When β is 0.5, the optimal

values of thickness of GDBL-1, porosity arrangement, and water retention

were listed in table 5.2.

Fig

(a)

(c)

. 5.2 Water r

120° an

retention and

nd 140° for G

12

d flooding w

GDBL-1 and

29

(b)

(d)

with a contact

d GDBL-2, re

t angle arran

espectively

ngement of

Fig

(a)

(c)

(e)

. 5.3 The opt

120° an

0

0.3

0.5

timized wate

nd 140° for G

13

(

(d

(f

er retention w

GDBL-1 and

30

(b)

d)

f)

with a contac

d GDBL-2, re

0

0.3

0.5

ct angle arran

espectively

ngement of

Fig

(a)

(c)

(e)

. 5.4 The opt

140° an

0

0.3

0.5

timized wate

nd 140° for G

13

(

(d

(f

er retention w

GDBL-1 and

31

(b)

d)

f)

with a contac

d GDBL-2, re

0

0.3

0.5

ct angle arran

espectively

ngement of

Fig

(a)

(c)

(e)

. 5.5 The op

120° an

0

0.3

0.5

ptimized wat

nd 120° for G

13

(

(d

(f

er retention

GDBL-1 and

32

(b)

d)

f)

with contact

d GDBL-2, re

0

0.3

0.5

t angle arran

espectively

ngement of

133

Table 5.1 Optimal arrangement of double GDBL for water retention

with a constraint, β 0.3 and the same MPL

Contact angle (°)

(GDBL-1/GDBL-2) Properties GDBL-1 GDBL-2

120/140

Porosity 0.9 0.74

Thickness (µm) 122 178

Q (mg) 13.8646

140/140

Porosity 0.9 0.74

Thickness (µm) 113 187

Q (mg) 11.4425

120/120

Porosity 0.9 0.72

Thickness (µm) 107 193

Q (mg) 11.7332

134

Table 5.2 Optimal arrangement of double GDBL for water retention

with a constraint, β 0.5 and the same MPL

Contact angle (°)

(GDBL-1/GDBL-2) Properties GDBL-1 GDBL-2

120/140

Porosity 0.9 0.8

Thickness (µm) 293 7

Q (mg) 5.8347

140/140

Porosity 0.9 0.76

Thickness (µm) 288 12

Q (mg) 5.7583

120/120

Porosity 0.9 0.74

Thickness (µm) 296 4

Q (mg) 5.7475

135

5.3 Optimization for water removal

Fig. 5.6 shows the minimized water retention capacity of GDL

Q at each porosity and contact angle arrangements with optimized

thickness of GDBL-1. The same values were used for variables. Based on

the result, it was found that optimal point appears at the porosity of 0.9 for

GDBL-2 regardless of contact angle arrangement. Thus, it was expected that

the water removal is most effective when the porosity of GDBL-2 is 0.9

(maximum value in this experiment). In addition, when the porosity of

GDBL-1 is larger than that of GDBL-2 (negative gradient of porosity),

Q with the contact arrangement of 140°/130° and 140°/120° is the same

to the result of 140°/140° except near the range where the porosity of

GDBLs are the same. It is because negative gradient of porosity increases

water retention and decreases water removal so that the Q appears when

the thickness of GDBL-1 is 300 µm (single GDBL with GDBL-1)

Fig. 5.7(b) shows optimized thickness of GDBL-1 for minimizing the

amount of liquid water remaining, when the fixed porosity of 0.9 is applied

to GDBL-2. The result shows that Q could be reduced applying positive

gradient of porosity ( ε 0.9 and Q remarkably changed

between 0.8 and 0.9 of porosity, while ε 0.8,the change in Q

is negligible. The same Qmin in low porous layer increases the risk of

136

flooding. In this point of view, the low porosity of GDBL should be avoided.

The lowest Q appeared when the arrangements of porosity and contact

angle of GDBLs were 0.82, 140° and 0.9, 140° for GDBL-1 and GDBL-2,

respectively. Hence, this is the optimized porosity and contact angle

arrangement. For this, the optimized thicknesses were 120 and 180 μm for

GDBL-1 and GDBL-2, respectively. With this optimized arrangement, Q

was lowered about 9.2% compared to the lowest value of Q without

gradient (atε ε 0.9, θ θ 140° . The

result also shows that contact angle gradient is not necessary but higher

contact angle for both GDBLs more improves water removal.

Nevertheless, contact angle gradient is still important and necessary. Fig.

5.8 shows the minimum amount of liquid water remaining in GDL and

thickness of GDBL-1 for this, when the porosities of GDBLs are the same.

The results show that Qmin was significantly lowered as the higher contact

angle arrangement is applied. Considering the manufacture difficulty and

cost, GDL with contact angle gradient is more favorable than that with

porosity gradient. The lowest Q appeared with the contact angle

arrangement of 140° and 120° for GDBL-1 and GDBL-2, respectively,

regardless of porosity.

In addition, the optimized Q appeared where ε 0.9

137

(maximum value in this study) with thickness arrangement of 203 and 97 μm

for GDBL-1 and GDBL-2, respectively. With this optimized arrangement,

Q was lowered about 5.6% compared to the lowest value of Q

without gradient. The results are listed in Table 5.3.

Fig. 5.9 shows a comparison of MPL coated single GDBL and

optimized GDL. The properties of GDLs compared with ‘Opt #1’ are listed

in Table 5.4 and the properties of GDLs compared with ‘Opt #2’ are listed in

Table 5.5. Compared to the ‘case # 1-2’ (the porosity of this GDL is average

value of ‘Opt #1’ considering the thickness ratio), the amount of water

remaining in GDBL was reduces by 13.33% and average gas diffusion factor

increased 2.88%. In addition, Compared to the ‘case # 1-2’ (the contact angle

of this GDL is average value of ‘Opt #2’ considering the thickness ratio), the

amount of water remaining in GDBL was reduces by 7.79% and average gas

diffusion factor increased 1.5%. Based on the result, it is expected that the

cell performance will be improved remarkably with multi-layer GDL

functionalized on water removal. Fig. 5.10 shows effective available range

of porosity arrangement and contact angle arrangement for each GDBL to

improve the water removal ability. For this, optimal thickness ratio was

applied for each arrangements and the result was compared to the MPL

coated single GDBL with uniform porosity (0.9) and contact angle (140°).

Fig. 5.6 Minim

optimi

gradie

(a) Wi

(b) With

mum water r

ized thickne

ent

13

ith contact an

hout contact

retention at

ess of GDB

38

ngle gradien

angle gradie

each poros

BL-1 with/w

nt

ent

ity arrangem

without hydr

ments with

ophobicity

139

(a)

(b) Optimized

Fig. 5.7 Minimum water retention and thickness of GDBL-1 for this, with

the fixed porosity of 0.9 for GDBL-2

2.2

2.3

2.4

2.5

2.6

2.7

2.8

0.5 0.6 0.7 0.8 0.9

Qm

in(m

g)

εGDBL-1

(120°, 120°) (130°, 130°)

(140°, 140°) (140°, 130°)

(140°, 120°) (130°, 120°)

εGDBL-2=0.9

(θGDBL-1, θGDBL-2)

0

50

100

150

200

250

300

0.5 0.6 0.7 0.8 0.9

δG

DB

L-1

(μm

)

εGDBL-1

(120°, 120°) (130°, 130°)

(140°, 140°) (140°, 130°)

(140°, 120°) (130°, 120°)

(θGDBL-1, θGDBL-2)

εGDBL-2=0.9

140

(a)

(b) Optimized

Fig. 5.8 Minimum water retention thickness of GDBL-1 for this, when the

porosities of GDBLs are the same

2.2

2.4

2.6

2.8

3.0

3.2

3.4

0.5 0.6 0.7 0.8 0.9

Qm

in(m

g)

εGDBL

(120°, 120°) (130°, 130°)(140°, 140°) (140°, 130°)(140°, 120°) (130°, 120°)

εGDBL-1=εGDBL-2

(θGDBL-1, θGDBL-2)

0

50

100

150

200

250

300

0.5 0.6 0.7 0.8 0.9

δG

DB

L-1

(μm

)

εGDBL

(140°, 130°)

(140°, 120°)

(130°, 120°)

(θGDBL-1, θGDBL-2)

εGDBL-1=εGDBL-2

141

Table 5.3 Optimal arrangement of double GDBL for water removal

Optimal

arrangement Properties GDBL-1 GDBL-2

Opt # 1

Porosity 0.82 0.9

Contact angle (°) 140 140

Thickness (µm) 120 120

Opt # 2

Properties 0.9 0.9

Porosity 140 120

Contact angle (°) 203 97

142

Table 5.4 GDBL properties for a comparison of uniform single GDBLs

and ‘Opt #1’

Case

Porosity

GDBL-1

(120 μm)

GDBL-2

(180 μm)

Average

value

Ref 0.90 0.90 0.90

# 1-1 0.82 0.82 0.82

# 1-2 0.87 0.87 0.87

Opt #1 0.90 0.82 0.87

143

Table 5.5 GDBL properties for a comparison of uniform single GDBLs

and ‘Opt #2’

Case

Contact angle (°)

GDBL-1

(203 μm)

GDBL-2

(97 μm)

Average

value

Ref 140 140 140

# 2-1 120 120 120

# 2-2 133.53 133.53 133.53

Opt #2 140 120 133.53

Fig. 5.9 Com

(a)

(b)

mparison of

14

Opt #1 (see

Opt #2 (see

single GDBL

44

Table 5.4)

Table 5.5)

L and optimiized double G

GDBL

Fig. 5.10 Effect

angle

(a)

(b) Co

tive available

for each GD

14

Porosity arr

ontact angle

e range of ar

DBL to impro

45

rangement

arrangement

rrangement o

ove the water

t

of porosity a

r removal ab

and contact

ility

146

5.4 Summary

In the previous chapters, the effects of porosity and hydrophobicity

gradient in GDBL were investigated numerically and experimentally. As a

result, it is expected that the performance of self-humidified fuel cell could

be improved by functionalized multi-layer GDL. In this chapter,

functionalized multi-layer GDL was optimized for water retention and water

removal based on the analytic model obtained in Chap.3. For this, the

arrangement of porosity, contact angle, thickness ratio was considered.

Although there were many difficulties in manufacturing multi-layer

GDBL in present time, with the advanced manufacturing techniques, it is

expected that manufacturing functionalized multi-layer GDBL is possible in

near future. Then, the result of this study could provide important

information to manufacturer.

147

Chapter 6. Concluding remarks

In a proton exchange membrane fuel cell (PEMFC), adequate hydration

of the membrane is crucial for achieving sufficient proton conductivity and

high performance. Conventionally, membrane hydration is achieved using an

external humidifier. However, their use increases the complexity, weight,

volume, and parasitic power loss in fuel cell systems. For this reason, self-

humidified PEMFCs are being studied.

In this study, we investigated the characteristics of dry start-up process,

which is a fuel cell start-up using dry reactant gases after any remaining

water from the previous operation of the cell has been completely purged.

Specifically, we evaluate the hydrogen crossover rate across the membrane,

the influence of the direct reaction of hydrogen and oxygen producing water

on the dry start-up process. The effect of starting temperature was also

evaluated with different flow arrangement. As a result, It was found that

start-up performance with counter flow is effective than co-flow and the

available operating temperature increases with counter flow. However, the

dry start-up was failed at 45°C. In order to solve this problem, an advanced

dry start-up process was applied and the result was successful. The results

148

showed that the WSP played an important role during the dry start-up and

the initial cell performance was remarkably improved. Moreover, WSP made

dry start-up possible at high cell temperatures, without the need for a long

time to cool down the fuel cell after the previous shut-down.

In consideration of practical use of self-humidified fuel cell, the

duration of WSP should be reduced and relatively low operating

performance of self-humidified fuel cell is another problem to be solved.

Therefore, the effect of structure design and hydrophobicity of GDBL on

water management of GDL was investigated to improve the start-up and

operating performance of self-humidified PEMFCs. For this, an analytic

model was obtained based on the capillary pressure–saturation relationship

with the Leverett J-function. In this model, structure design and

hydrophobicity of GDBLs were represented by the measurable parameters,

porosity and contact angle, respectively. With this analytic model, the liquid

water saturation distribution and the amount of water remaining in GDL

were evaluated as a function of porosity, contact angle, and thickness ratio of

each GDBL. As a result, it was concluded that porosity gradient in negative

direction and hydrophobicity gradient in positive direction is effective on

water retention.

Based on the analytic results, start-up and steady-state performance of

149

fuel cell was measured with the modified GDL containing double GDBL

with different porosity in negative direction (GDBL with lower porosity near

the flow channel). The result showed that shorter duration of WSP was

required for successful start-up and high performance with modified GDL.

In addition, the effect of stacking was also evaluated with the GDL

containing double GDBL with the same porosity. The result showed that the

structural design of the GDBL had a major effect on its water retention

capacity, whereas stacking has negligible effect on the water retention

capacity. In additions, the self-humidification effect and the performance of

the fuel cell containing the structurally modified GDBL were found to be

significantly improved over a wide range of operating conditions.

Lastly, functionalized GDL was optimized for water retention and/or

water removal. For this, gas diffusion coefficient was considered. Although

there were many difficulties in manufacturing multi-layer GDBL in present

time, with the advanced manufacturing techniques, it is expected that

manufacturing functionalized multi-layer GDBL is possible in near future

and this work provides important information to manufacturer. We believe

that this work provides new insights into improving the performance of self-

humidifying fuel cells by structural design of the GDBL, rather than by

modifying the MPL.

150

References

[1] Barnwal BK, Sharma MP. Prospects of biodiesel production from

vegetable oils in India. Renewable and Sustainable Energy Reviews

2005;9(4):363-378.

[2] Campen A, Mondal K, Wiltowski T. Separation of hydrogen from syngas

using a regenerative system. Int’l J. Hydrogen Energy 2008;33(1):332-

339.

[3] Zanganeh KE, Shafeen A. A novel process integration, optimization and

design approach for large-scale implementation of oxy-fired coal power

plants with CO2 capture. Int’l J. Greenhouse Gas Control 2007;1 (1):47-

54.

[4] Meher LC, Sagar DV, Naik SN. Technical aspects of biodiesel

production by transesterification - a review. Renewable and Sustainable

Energy Reviews 2006;10(3):248-268.

[5] Balat M. Potential importance of hydrogen as a future solution to

environmental and transportation problems. Int’l J. Hydrogen Energy

2008; 33(15):4013-4029.

151

[6] Wang Y, Chen KS, Mishler J, Cho SC, Adroher XC. A review of

polymer electrolyte membrane fuel cells: Technology, applications, and

needs on fundamental research. Applied Energy 2011;88(4):981-1007.

[7] Pourcelly G, Oikonomou A, Gavach C, Hurwitz HD. Influence of the

water content on the kinetics of counter-ion transport in

perfluorosulphonic membranes. J. Electroanal Chem. Interfacial

Electrochem. 1990;287(1):43-59.

[8] Buchi FN, Srinivasan S. Operating proton exchange membrane fuel cells

without external humidification of the reactant gases.J. Electrochemical

Society 1997;144(8):2767-2772.

[9] Berg P, Promislow K, Pierre J St., Stumper J, Wetton B. Water

management in PEM fuel cells. J. Electrochemical Society

2004;151(3):A341-A353.

[10] Yan WM, Yang CH, Soong CY, Chen F, Mei SC. Experimental studies

on optimal operating conditions for different flow field designs of PEM

fuel cells. J. Power Sources 2006;160(1):284-292.

[11] Yu H, Ziegler C. Transient behavior of a proton exchange membrane

fuel cell under dry operation. J. Electrochemical Society

2006;153(3):A570-A575.

152

[12] Zhang J, Tang Y, Song C, Cheng X, Zhang J, Wang H. PEM fuel cells

operated at 0% relative humidity in the temperature range of 23–120°C.

Electrochimica Acta 2007;52(15):5095-5101.

[13] Guvelioglu GH, Stenger HG. Flow rate and humidification effects on a

PEM fuel cell performance and operation. J. Power Sources

2007;163(2):882-891.

[14] Riascos LAM, Pereira DD. Limit operating temperature in polymer

electrolyte membrane fuel cells. J. Electrochemical Society

2009;156(9):B1051-B1058.

[15] Feindel KW, Bergens SH, Wasylishen RE. Insights into the distribution

of water in a self-humidifying H2/O2 proton-exchange membrane fuel

cell using 1H NMR microscopy. J. American Chemical Society

2006;128(43):14192-14199.

[16] Kim TJ, Kim JR, Sim CM, Lee SW, Kaviany M, Son SY, Kim MH.

Experimental approaches for distribution and behavior of water in

PEMFC under flow direction and differential pressure using neutron

imaging technique. Nuclear Instruments and Methods in Physics

Research Section A 2009;600(1):325-327.

[17] Choi JW, Hwang YS, Cha SW, Kim MS. Experimental study on

enhancing the fuel efficiency of an anodic dead-end mode polymer

153

electrolyte membrane fuel cell by oscillating the hydrogen. Int’l J.

Hydrogen Energy 2010;35(22):12469-12479.

[18] Choi JW, Hwang YS, Seo JH, Lee DH, Cha SW, Kim MS. An

experimental study on the purge characteristics of the cathodic dead-end

mode PEMFC for the submarine or aerospace applications and

performance improvement with the pulsation effects. Int’l J. Hydrogen

Energy 2010;35(8):3698-3711.

[19] Kim BJ, Kim MS. Studies on the cathode humidification by exhaust gas

recirculation for PEM fuel cell. Int’l J. Hydrogen Energy

2012;37(5):4290-4299.

[20] Uchida H, Ueno Y, Hagihara H, Watanabe M. Self-humidifying

electrolyte membranes for fuel cells preparation of highly dispersed

TiO2 particles in nafion 112. J. Electrochemical Society

2003;150(1):A57-A62.

[21] Hagihara H, Uchida H, Watanabe M. Preparation of highly dispersed

SiO2 and Pt particles in nafion®112 for self-humidifying electrolyte

membranes in fuel cells. Electrochimica Acta 2006;51(19):3979-3985.

[22] Yang HN, Lee DC, Park SH, Kim WJ. Preparation of Nafion/various

Pt-containing SiO2 composite membranes sulfonated via different

154

sources of sulfonic group and their application in self-humidifying

PEMFC. J. Membrane Science 2013; 443:210-218.

[23] Ke CC, Li XJ, Shen Q, Qu SG, Shao QG, Yi BL, Investigation on

sulfuric acid sulfonation of in-situ sol–gel derived Nafion/SiO2

composite membrane, Int’l J. Hydrogen Energy 2011;36(5):3606-3613.

[24] Jung UH, Park KT, Park EH, Kim SH. Improvement of low-humidity

performance of PEMFC by addition of hydrophilic SiO2 particles to

catalyst layer. J. Power Sources 2006;159(1):529-532.

[25] Chao WK, Lee CM, Tsai DC, Chou CC, Hsueh KL, Shieu FS.

Improvement of the proton exchange membrane fuel cell (PEMFC)

performance at low-humidity conditions by adding hygroscopic γ-Al2O3

particles into the catalyst layer. J. Power Sources 2008;185(1);136-142.

[26] Su H, Xu L, Zhu H, Wu Y, Yang L, Liao S, Song H, Liang Z, Birss V.

Self-humidification of a PEM fuel cell using a novel Pt/SiO2/C anode

catalyst. Int’l J. Hydrogen Energy 2010; 35(15):7874-7880.

[27] Choi IS, Lee KG, Ahn SH, Kim DH, Kwon OJ, Kim JJ. Sonochemical

synthesis of Pt-deposited SiO2 nanocomposite and its catalytic

application for polymer electrolyte membrane fuel cell under low-

humidity conditions. Catalysis Communications 2012;21:86-90.

155

[28] Inoue N, Uchida M, Watanabe M, Uchida H. SiO2-containing catalyst

layers for PEFCs operating under low humidity Electrochemistry

Communications. Electrochemistry Communications 2012;16(1):100-

102.

[29] Ha TH, Cho JH, Park JM, Min KD, Kim HS, Lee ES, Jyoung JY.

Experimental study of the effect of dissolution on the gas diffusion layer

in polymer electrolyte membrane fuel cells. Int’l J. Hydrogen Energy

2011:36(19):12427-12435.

[30] Kim SI, Baik KD, Kim BJ, Lee NW, Kim MS. Experimental study on

mitigating the cathode flooding at low temperature by adding hydrogen

to the cathode reactant gas in PEM fuel cell. Int’l J. Hydrogen Energy

2013;38(3):1544-1552.

[31] Cindrella L, Kannan AM, Lin JF, Saminathan K, Ho Y, Lin CW, Wertz

J. Gas diffusion layer for proton exchange membrane fuel cells—A

review. J. Power Sources 2009;194(1):146-160.

[32] Weber AZ, Newman J, Effects of microporous layers in polymer

electrolyte fuel cells. J. Electrochemical Society 2005;152(4): A677–

A688.

156

[33] Antolini E, Passos RR, Ticianelli EA. Effects of the cathode gas

diffusion layer characteristics on the performance of polymer electrolyte

fuel cells. J. Applied Electrochemistry 2002;32(4):383-388.

[34] Chen J, Matsuura T, Hori M. Novel gas diffusion layer with water

management function for PEMFC. J. Power Sources 2004;131(1-

2):155-161.

[35] Yoshikawa Y, Matsuura T, Kato M, Hori M. Design of low-

humidification PEMFC by using cell simulator and its power generation

verification test. J. Power Sources 2006;158(1):143-147.

[36] Chun JH, Park KT, Jo DH, Lee JY, Kim SG, Park SH, Lee ES, Jyoung

JY, Kim SH. Development of a novel hydrophobic/hydrophilic double

micro porous layer for use in a cathode gas diffusion layer in PEMFC.

Int’l J. Hydrogen Energy 2011;36(14):8422-8428.

[37] Choun MH, Chung SH, Jeon HR, Uhm SH, Lee JY. Atomic-layer-

deposited TiO2 on cathode gas diffusion layer for low humidity

operation in hydrogen fuel cells. Electrochemistry Communications

2012;24:108-111.

[38] Kitahara T, Nakajima H, Mori K. Hydrophilic and hydrophobic double

microporous layer coated gas diffusion layer for enhancing performance

157

of polymer electrolyte fuel cells under no-humidification at the cathode.

J. Power Sources 2012;199(19):29-36.

[39] Kitahara T, Nakajima H, Morishita M. Water vapor exchange system

using a hydrophilic microporous layer coated gas diffusion layer to

enhance performance of polymer electrolyte fuel cells without cathode

humidification. J. Power Sources 2012;214:100-106.

[40] Kitahara T, Nakajima H, Inamoto M, Morishita M. Novel hydrophilic

and hydrophobic double microporous layer coated gas diffusion layer to

enhance performance of polymer electrolyte fuel cells under both low

and high humidity. J. Power Sources 2013;234:129-138.

[41] Ito K, Ashikaga K, Masuda H, Oshima T, Kakimoto Y, Sasaki K.

Estimation of foding in PEMFC gas diffusion layer by differential

pressure measurement. J. Power Sources 2008;175(2):732-738.

[42] Adler PM, Jacquin CG, Quiblier JA. Flow in simulated porous media.

Int’l J. Multiphase Flow 1990;16(4):691–712.

[43] Ioannidis MA, Chatzis I. Network modelling of pore structure and

transport properties of porous media. Chem. Eng. Sci. 1993;48(5):951–

972.

[44] Wang CY, Cheng P.Multiphase flow and heat transfer in porous media.

Adv. Heat Transfer 1997;30:183-196.

158

[45] Wang CY. A fixed-grid numerical algorithm for two-phase flow and

heat transfer in porous media. Numerical Heat Transfer B, Fundam.

1997;31(1):85-105.

[46] Blunt MJ. Flow in porous media — pore-network models and

multiphase flow. Current Opinion in Colloid & Interface Science

2001;6(3):197-207.

[47] Lee HK, Park JH, Kim DY, Lee TH. A study on the characteristics of

the diffusion layer thickness and porosity of the PEMFC. J. Power

Sources 2004;131(1-2):200-206.

[48] Larbi B, Alimi W, Chouikh R, Guizani A. Effect of porosity and

pressure on the PEM fuel cell performance. Int’l J. Hydrogen Energy

2013;38(20)8542-8549

[49] Chu HS, Ha CY, Chen FL. Effects of porosity change of gas diffuser on

performance of proton exchange membrane fuel cell. J. Power Sources

2003;123(1):1-9.

[50] Roshandel R, Farhanieh B, Saievar-Iranizad E. The effects of porosity

distribution variation on PEM fuel cell performance. Renewable Energy

2005;30(10):1557-1572.

159

[51] Zhan Z, Xiao J, Li D, Pan M, Yuan R. Effects of porosity distribution

variation on the liquid water flux through gas diffusion layers of PEM

fuel cells. J. Power Sources 2006;160(2):1041-1048.

[52] Zhan Z, Xiao J, Zhang Y, Pan M, Yuan R. Gas diffusion through

differently structured gas diffusion layers of PEM fuel cells. Int’l J.

Hydrogen Energy 2007;32(17):4443-4451.

[53] Chen F, Chang M, Hsieh P. Two-phase transport in the cathode gas

diffusion layer of PEM fuel cell with a gradient in porosity. Int’l J.

Hydrogen Energy 2008;33(10)2525-2529

[54] Huang YX, Cheng CH, Wang XD, Jang JY. Effects of porosity gradient

in gas diffusion layers on performance of proton exchange membrane

fuel cells. Energy 2010;35(12):4786-4794.

[55] Kumar JFR, Radhakrishnan V, Haridoss P. Enhanced mechanical and

electrochemical durability of multistage PTFE treated gas diffusion

layers for proton exchange membrane fuel cells. Int’J J. Hydrogen

Energy 2012;37(14):10830-10835.

[56] Gao Y, Sun GQ, Wang SL, Zhu S. Carbon nanotubes based gas

diffusion layers in direct methanol fuel cells. Energy 2010;35(3):1455-

1459.

160

[57] Gerteisen D, Heilmann T, Ziegler C. Enhancing liquid water transport

by laser perforation of a GDL in a PEM fuel cell. J. Power Sources

2008;177(2):348-354.

[58] Gerteisen D, Sadeler C. Stability and performance improvement of a

polymer electrolyte membrane fuel cell stack by laser perforation of gas

diffusion layers. J. Power Sources 2010;195(16):5252-5257.

[59] Fushinobu K, Takahashi D, Okazaki K. Micromachined metallic thin

films for the gas diffusion layer of PEFCs. J. Power Sources

2006;158(2):1240-1245.

[60] Zhang FY, Advani SG, Prasad AK. Performance of a metallic gas

diffusion layer for PEM fuel cells. J. Power Sources 2008;176(1):293-

298.

[61] Oedegaard A, Hebling C, Schmitz A, Moller-Holst S, Tunold R.

Influence of diffusion layer properties on low temperature DMFC. J.

Power Sources 2004;127(1-2):187-196.

[62] Yu EH, Scott K. Direct methanol alkaline fuel cell with catalysed metal

mesh anodes. Electrochemistry Communications 2004;6(4):361-365.

[63] Yang CC, Chiu SJ, Lin CT. Electrochemical performance of an air-

breathing direct methanol fuel cell using poly(vinyl

161

alcohol)/hydroxyapatite composite polymer membrane. J. Power

Sources 2008;177(1):40-49.

[64] Shao ZG, Zhu FY, Lin WF, Christensen PA, Zhang HM. PtRu/ Ti

anodes with varying Pt:Ru ratio prepared by electrodeposition for the

direct methanol fuel cell. Physical Chemistry Chemical Physics

2006;8:2720-2726.

[65] Shao ZG, Zhu FY, Lin WF, Christensen PA, Zhang HM, Yi BL.

Preparation and characterization of new anodes based on Ti mesh for

direct methanol fuel cells. J. Electrochemical Society

2006;153(8):A1575-A1583.

[66] Lim C, Scott K, Allen RG, Roy S. Direct methanol fuel cells using

thermally catalysed Ti mesh. J. Applied Electrochemistry

2004;34(9):929-933.

[67] Shao ZG, Lin WF, Zhu FY, Christensen PA, Zhang HM, Yi BL. A

tubular direct methanol fuel cell with Ti mesh anode. J. Power Sources

2006;160(2):1003-1008.

[68] Liu P, Yin GP, Lai QZ. Gold-plated Ni mesh as the gas diffusion

medium for air-breathing direct methanol fuel cell. Int’l J. Energy

Research 2009;33(1):1-7.

162

[69] Chen R, Zhao TS. A novel electrode architecture for passive direct

methanol fuel cells. Electrochemistry Communications 2007;9(4):718-

724.

[70] Ioroi T, Oku T, Yasuda K, Kumagai N, Miyazaki Y. Influence of PTFE

coating on gas diffusion backing for unitized regenerative polymer

electrolyte fuel cells. J. Power Sources 2003;124(2):385-389.

[71] Wittstadt U, Wagner E, Jungmann T. Membrane electrode assemblies

for unitised regenerative polymer electrolyte fuel cells. J. Power Sources

2005;145(2):555-562.

[72] Bevers D, Rogers R, VonBradke M. Examination of theinfluence of

PTFE coating on the properties of carbon paperin polymer electrolyte

fuel cells. J. Power Sources 1996;63(2):193-201.

[73] Park GG, Sohn YJ, Yang TH, Yoon YG, Lee WY, Kim CS. Effect of

PTFE contents in the gas diffusion media on the performanceof PEMFC.

J. Power Sources 2004;131(1-2):182-187.

[74] Lin GY, Nguyen TV. Effect of thickness and hydrophobic polymer

content of the gas diffusion layer on electrode flooding level in a

PEMFC. J. Electrochemical Society 2005;152 (10):A1942–A1948.

163

[75] Prasanna M, Ha HY, Cho EA, Hong SA, Oh IH. Influence ofcathode

gas diffusion media on the performance of the PEMFCs. J. Power

Sources 2004;131(1-2):147-154.

[76] Park S, Popov BN. Effect of cathode GDL characteristics on mass

transport in PEM fuel cells. Fuel 2009;88(11):2068-2073.

[77] Staiti P, Poltarzewski Z, Alderucci V, Maggio G, Giordano N,Fasulo A.

Influence of electrodic properties on water management in a solid

polymer electrolyte fuel-cell. J. Applied Electrochem. 1992;22(7):663-

667.

[78] Lim C, Wang CY. Effects of hydrophobic polymer content in GDL on

power performance of a PEM fuel cell. Electrochimica Acta

2004;49(24):4149-4156.

[79] Latorrata S, Stampino PG, Cristiani C, Dotelli G. Novel

superhydrophobic microporous layers for enhanced performance and

efficient water management in PEM fuel cells. Int’l J. Hydrogen Energy

2014;39(10):5350-5357.

[80] Cabasso I, Yuan Y, Xu X. Gas diffusion electrodes based on

poly(vinylidene fluoride) carbon blends US Patent No. 5783325.1998.

[81] Stampino PG, Latorrata S, Molina D, Turri S, Levi M, Dotelli G.

Investigation of hydrophobic treatments with perfluoropolyether

164

derivatives of gas diffusion layers by electrochemical impedance

spectroscopy in PEM-FC. Solid State Ionics 2012;216:100-104.

[82] Pai YH, Ke JH, Huang HF, Lee CM, Jyh-Myng Z, Shieu FS. CF4

plasma treatment for preparing gas diffusion layers in membrane

electrode assemblies. J. Power Sources 2006;161(1):275-281.

[83] Zhang Y, Pitchumani R. Numerical studies on an air-breathing proton

exchange membrane (PEM) fuel cell. Int’l J. Heat and Mass Transfer

2007;50(23-24): 4698-4712.

[84] Baik KD, Hong BK, Kim MS. Effects of operating parameters on

hydrogen crossover rate through Nafion® membranes in polymer

electrolyte membrane fuel cells. Renewable Energy 2013;57:234-239.

[85] Baik KD, Hong BK, Kim MS. Novel technique for measuring oxygen

crossover through the membrane in polymer electrolyte membrane fuel

cells. Int’l J. Hydrogen Energy 2013;38(21):8927-8933.

[86] Baik KD, Kong IM, Hong BK, Kim SH, Kim MS. Local measurements

of hydrogen crossover rate in polymer electrolyte membrane fuel cells.

Applied Energy 2013;101:560-566.

[87] Inaba M, Kinumoto T, Kiriake M, Umebayashi R, Tasaka A,Ogumi Z.

Gas crossover and membrane degradation in polymer electrolyte fuel

cells. Electrochimica Acta 2006;51(26): 5746-5753.

165

[88] Kocha SS, Yang JD, Yi JS. Characterization of gas crossover and its

implications in PEM fuel cells. AIChE Journal 2006; 52( 5):1916-1925.

[89] Zhang J, Tang Y, Song C, Zhang J, Wang H. PEM fuel cell open circuit

voltage(OCV) in the temperature range of 23°C to 120°C, J. Power

Sources 2006; 163(1):532-537.

[90] Springer TE, Zawodzinski TA, GottesfeldS. Polymer electrolyte fuel-

cell model, J. Electrochemical Society 1991;138(8):2334-2342.

[91] Pei P, Chang Q, Tang T. A quick evaluating method for automotive fuel

cell lifetime. Int’l J. Hydrogen Energy 2008;33(14):3829–36.

[92] Kang J, Kim J. Membrane electrode assembly degradation by dry/wet

gas on a PEM fuel cell. Int’l J. Hydrogen Energy 2010;35(23):13125-

13130.

[93] Seo JH, Baik KD, Kim DK, Kim S, Choi JW, Kim M, Song HH, Kim

MS. Effects of anisotropic bending stiffness of gas diffusion layer on the

MEA degradation of polymer electrolyte membrane fuel cells by

wet/dry gas. Int’l J. Hydrogen Energy 2013;38(36): 16245-16252.

[94] Ous T, Arcoumanis C. The formation of water droplets in an air-

breathing PEMFC. Int’l J. Hydrogen Energy 2009;34(8):3476-3487.

166

[95] Manoj P, Parthasarathy V. A passive method of water management for

an air-breathing proton exchange membrane fuel cell. Energy

2013;51:457-461.

[96] Wang ZH, Wang CY, Chen KS.Two-phase flow and transport in the air

cathode of proton exchange membrane fuel cells. J. Power Sources

2001;94(1):40–50.

[97] Ihonen J, Mikkola M, Lindbergh G. Flooding of gas diffusionbacking in

PEFCs: physical and electrochemical characterization, J.

Electrochemical Society 2004;151(8):A1152–A1161.

[98] Tomadakis MM, Robertson TJ. Viscous permeability of random fiber

structures: comparison of electrical and diffusional estimates with

experimental and analytical results, J. Comp. Materials 2005;39(2):163-

188.

[99] Tamayol A, Bahrami M. Transverse permeability of fibrous porous

media, Physical Review 2011;83(4):046314-1 - 046314-9.

[100] Tamayol A, McGregor F, Bahrami M. Single phase through-plane

permeability of carbon paper gas diffusion layers. J. Power Sources

2012;204:94-99.

167

[101] Shi W, Kurihara E, Oshima N. Effect of capillary pressure on liquid

water removal in the cathode gas diffusion layer of a polymer

electrolyte fuel cell. J. Power Sources 2008;182(1):112-118.

[102] Hussaini IS, Wang CY. Measurement of relative permeability of fuel

cell diffusion media. J. Power Sources 2010; 195(12): 3830-3840.

[103] Mason TJ, Millichamp J, Neville TP, El-kharouf A, Pollet BG, Brett

Daniel JL. Effect of clamping pressure on ohmic resistance and

compression of gas diffusion layers for polymer electrolyte fuel cells. J.

Power Sources 2012;219:52-59.

[104] Lee Y, Kim B, Kim Y. Effects of self-humidification on the dynamic

behavior of polymer electrolyte fuel cells. Int’l J. Hydrogen Energy

2009;34(4):1999-2007.

[105] Cho J, Park J, Oh H, Min K, Lee E, Jyoung JY. Analysis of the

transient response and durability characteristics of a proton exchange

membrane fuel cell with different micro-porous layer penetration

thicknesses. Applied Energy 2013;111:300-309.

[106] Hao L, Cheng P. Lattice Boltzmann simulations of anisotropic

permeabilities in carbon paper gas diffusion layers. J. Power Sources

2009;186(1):104-114.

168

국 문 초 록

고분자 전해질막 연료전지에서 고분자막의 전도도는 연료전지의

성능을 결정짓는 중요한 요소이며, 고분자막의 수화도에 의해

결정된다. 고분자막의 수화도를 유지하기 위하여, 외부 가습기가

주로 사용되고 있지만, 시스템이 복잡해지고, 동력 손실이

증가하며, 제작 비용이 추가되는 등의 단점이 있다. 따라서 외부

가습기 없이 고분자 전해질막 연료전지를 운전하는 시스템에 대한

관심이 커져가고 있다. 지금까지 자가 가습 연료전지에 관한

다양한 연구가 수행되어 왔지만, 건조한 상태에서의 시동에 관한

연구는 거의 이루어지지 않았다. 또한 자가 가습 연료전지의

성능이 기존보다 향상되었지만 아직까지 많이 낮기 때문에 성능

향상을 위한 연구가 필요하다. 본 연구에서는 자가 가습

연료전지의 성능 향상을 위한 연구를 수행하였다. 이를 위하여,

먼저 건시동에 관한 특성을 분석하였다. 수소 크로스오버에 의한

물 생성이 고분자막의 수화도에 미치는 영향을 평가하였으며, 공급

기체의 유동 배열에 따른 시동성을 평가하였다. 그 결과, 대항류에

의해 건시동 성능 및 운전 가능한 온도가 향상됨을 확인하였다.

또한 고온에서의 건시동 문제를 해결하기 위하여 물 저장 과정

(water storage process)을 포함한 시동 프로세스를

제안하였다. 이를 적용한 결과, 고온 재시동에 성공하였으며, 초기

성능이 크게 향상되는 것을 확인하였다. 상대적으로 낮은 자가

가습 연료전지의 성능은 지속적으로 관심을 받아온 문제이다. 또한

본 연구에서는 능동적인 물관리를 통한 자가 가습 연료전지의

성능 향상을 위하여, 하나의 미세기공층 (single MPL)과 이중의

기재층 (double GDBL)로 이루어진 기능성 다중층 기체확산층

(Multi-layer GDL)을 제안하였다. 해석적인 분석을 통하여

기재층의 다공성, 소수성, 두께 배열에 따라 향상된 물

보존(water retention) 효과 또는 물 배출 (water removal)

169

효과를 확인하였으며, 이를 바탕으로 실험을 수행하였다. 물 보존

효과가 높은 이중 기재층을 적용한 결과, 고온에서 연료전지의

성공적인 시동과 안정적인 운전을 위해 요구되는 물 저장 과정

적용 시간을 단축시킬 수 있었다. 또한 이중 기재층 구조에 의한

자가 가습 효과의 향상을 확인하였다. 마지막으로, 기체확산층의

분석 모델을 이용하여 최적화를 수행하였다. 본 연구의 결과로

고분자 전해질막을 이용한 자가 가습 연료전지의 안정적인 시동과

운전이 가능해질 것으로 기대된다. 특히 고온에서 건시동에

성공하였으며, 이에 따라 자가 가습 연료전지의 상용화에 크게

기여할 수 있을 것으로 기대된다. 또한 기능성 다중층

기체확산층에 대한 분석 기법과 결과는 다양한 적용이 가능하기

때문에 활용도가 높을 것으로 기대된다.

주요어: 고분자 전해질막 연료전지, 시동, 물 관리, 기체확산층,

기공도, 소수성

학 번: 2009-20650