Atomic Layer Deposition Of Platinum Catalyst In Hydrogen And Oxygen Proton Exchange Membrane Fuel Cell

國 立 清 華 大 學 碩士論文 以原子層沈積製備白金觸媒應用於 氫氧質子交換膜燃料電池 Atomic Layer Deposition of Platinum Catalyst in Hydrogen and Oxygen Proton Exchange Membrane Fuel Cell

系

所：材料科學工程學系

學

號：943508

研 究 生：劉玨 (Chueh Liu) 指導教授：彭宗平 (Tsong-Pyng Perng)

中華民國九十六年七月

1

誌謝 七百多天的碩士生涯，首先要感謝老師彭宗平的指導，諄諄教誨，點滴 在心，讓學生不僅在學術方面有所成長，做人處事方面亦獲益匪淺。其次 謝謝葉君棣老師與蕭健男副組長撥冗出席口試，給予學生許多建議與指 導，使此論文更加完整。 接著謝謝致傑學長在 SEM 的大力協助，以及實驗的討論與建議；志忠 學長對實驗的建議，對此論文完成亦有相當大的貢獻；孟諺學長帶領初入 實驗室的我，學習了一些儀器與基本觀念，為我的實驗生涯開啟了第一扇 窗；世宗學長與哲平學長，教導我燃料電池組裝與量測技術，使我有能力 比較觸媒好壞；忠義學長、鐸耀學長、炯文學長發起的出遊，為生活增添 了許多色彩；彥勳的辛勤採買，雅雯的藥品管理，亦為大家帶來了許多方 便。 感恩父母的辛勞，為孩子付出無怨無悔，使我在充滿關愛，毫無經濟壓 力的情況下順利完成學業，兒沒齒難忘。最後感謝對我以及實驗有幫助的 所有人，有了你們實驗才能平安順利完成！

2

摘要 隨著能源危機與污染問題日益嚴重，燃料電池成為極有潛力的替代能源 之一，但低觸媒活性與使用率，導致其成本過高；然而不同製程可改變觸 媒活性與使用率，故期盼原子層沈積能提高此兩種特性。 用原子層沈積白金在不同基板上，包含碳布、碳黑或碳管塗佈的碳布， 以作為質子交換膜燃料電池觸媒使用。本實驗以顆粒尺寸來探討白金原子 層沈積在矽基板上的自我限制行為，以及顆粒尺寸與循環數的比例關係。 藉由掃瞄電子顯微鏡影像證實了這兩個性質，也觀察到基板效應：白金不 能沈積在未經酸處理的碳布，也不能沈積在有較高鐵氟龍含量的碳黑或碳 管塗佈的碳布上；然而，因為表面官能基的改變，白金可以沈積在經過酸 處理的碳布，以及在較低鐵氟龍含量的碳黑或碳管塗佈的碳布上。 比較不同膜電極組的效能測試，對陽極與陰極相同的膜電極組，E-TEK 商用電極表現出最大的功率密度(W/cm2)，然而以原子層沈積白金一百循環 的碳布電極有最低的白金功率密度(gPt/kW: 每千瓦之白金克數)。在不同陽 極，但陰極都是 E-TEK 商用電極的膜電極組中，陽極是以原子層沈積白金 一百循環於碳管塗佈碳布的膜電極組展現了最好的白金功率密度，卻仍有 與 E-TEK 商用電極相似的功率密度。

3

Abstract With the increasing threats of energy crisis and pollutions, fuel cell has become one of the most prospective substitute energies, but low catalyst activity and utilization results in high cost; however different preparation methods can change catalyst activity and utilization, so it’s expected atomic layer deposition can increase these two properties. Atomic layer deposition (ALD) is used to deposit Pt as catalyst of proton exchange membrane fuel cell catalyst on different substrates, including carbon cloth, carbon black powders or carbon nanotubes (CNT) coated on carbon cloth. In this experiment, Particle size is used to discuss self-limiting behavior and proportionality between particle size and cycle numbers of ALD of Pt on Si substrate. Both of the two characteristics are verified by SEM images. Substrate effects are also observed: Pt can not deposit on carbon cloth without acid treatment, and can not deposit on carbon black powder or CNT coated carbon cloth with higher concentration of PTFE addition; however, because of the change of the surface functional groups, for carbon cloth with acid treatment and carbon black powder or CNT coated carbon cloth with lower concentration of PTFE addition, Pt can deposit on them. Performance tests of different membrane electrode assemblies (MEAs) are compared. For the MEAs of the same anode and cathode, as received E-TEK electrode shows the largest power density (W/cm2), while the electrode of carbon cloth with ALD of Pt for 100 cycles has the lowest Pt specific power density (gPt/kW: gram of Pt per kilowattage). For the MEAs of different anodes but cathodes are all as received E-TEK electrodes, anode of ALD of Pt on CNT coated carbon cloth for 100 cycles shows the best Pt specific power density, while it still has similar power density to E-TEK electrode.

4

Contents 誌謝 摘要 Abstract Chapter I.

II.

Introduction.................................................................................1 1-1

Why Do We Choose Fuel Cells?............................................1

1-2

Brief History of Fuel Cells........................................................1

1-3

Why Do We Choose Atomic Layer Deposition (ALD)?...........3

Literature Review......................................................................6 2-1

Basic Principles of Fuel Cell.....................................................6

2-1-1

Thermodynamic Aspects.....................................................6

2-1-2

Kinetic Aspects....................................................................8

2-1-3

Three-Phase Boundary.......................................................8

2-2

Proton Exchange Membrane Fuel Cell (PEMFC)...................10

2-2-1

Schematic of PEMFC........................................................10

2-2-2

Catalyst..............................................................................14

2-2-2-1 Materials: Pt, Pt alloy, Pt-Free...................................14 2-2-2-2 Particle Size Effect....................................................15 2-2-2-3 Carbon Catalyst Supports: Carbon Black, CNT........18 2-3

PEMFC Catalyst Deposition Methods....................................20

2-3-1

Wet Process.......................................................................20

2-3-1-1 Chemical Process.......................................................20 2-3-1-2 Electrochemical Process............................................23 2-3-2 Dry Process........................................................................25 2-3-3

Performance Comparison of Different Catalyst Deposition

5

Methods.............................................................................28 2-4

Atomic Layer Deposition (ALD)............................................28 2-4-1

ALD Characteristics......................................................28

2-4-2

ALD of Pt......................................................................32

III. Experimental................................................................... 38 3-1

Substrate Preparation...............................................................38

3-2

Deposition of Catalyst.............................................................38 3-2-1

ALD of Platinum...........................................................38

3-2-2

Sputtering......................................................................42

3-3

Preparation of MEA................................................................42

3-4

Single Cell Setup.....................................................................42

3-5

Fuel Cell Test Station and Performance Test Conditions........45

3-6

Characterization and Chemical Analysis of Pt Catalyst..........47

IV. Results and Discussion...........................................................48 4-1

ALD Characteristics................................................................48 4-1-1

Self-Limiting.................................................................48

4-1-2

Proportionality between Cycle Number and Particle Size................................................................................48

4-1-3

Substrate Effect.............................................................48

4-2

Pt Loading on Different Substrates.........................................56

4-3

Performance Tests of MEAs Made by ALD, Chemical and Sputtering M e t ho d s ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... .. 5 8

V.

Conclusions................................................................................62

References................................................................................. 63

6

Chapter I.

Introduction

1-1 Why Do We Choose Fuel Cells? Energy demand of human beings is increasing (Fig. 1-1) [1], but after the oil crisis, and with the getting worse environmental conditions, people have tried to make use of so called “clean energy” by different ways, including wind, solar, geothermal, ocean, biomass, and hydrogen energies [3] in order to lower the pollutions of environment and the dependence on oil. Hydrogen energy, a kind of chemical energy, can be converted directly into electrical energy by fuel cells. The reason why we choose hydrogen energy and fuel cells is because there are several advantages of using hydrogen and fuel cells, including: (i) electrical energy can be transported conveniently and converted into another kind of energy at practically 100% efficiency [4]; (ii) higher energy conversion efficiency, from 35% (practically) [5] to 90% (theoretically) [4] for chemical-electrical energy conversion, comparing to heat engine’s efficiency, from 25% (practically) to 50% (theoretically) for chemical-mechanical energy conversion [4]; (iii) lower or zero emission of particulate matters [6] (harmful to respiratory system), NOx (increasing O3 concentration in urban area), COx (causing greenhouse effect), SOx (causing acid rain, monuments corrosion and rain forest destruction); all those species are generated in operation of heat engine.

1-2 Brief History of Fuel Cells In the experiment of Grove’s “gaseous voltaic battery” in 1842 [5], with platinum electrodes at both anode and cathode, sulfuric acid electrolyte, and H2 and O2 as reactants, Grove found that current could be detected once there was a thin layer of sulfuric acid on the electrode. It was found that reaction depended on the “surface of action,” the area where electrolyte, gas reactants, and solid

7

Figure 1-1. Per capita energy consumption in Taiwan [1]. (LOE: liter oil equivalent = 9．106 joule [2].)

8

electrode meet. Grove was also able to electrolyze water by series connection of at least 26 such kind of cells (Fig. 1-2). He alleged “every chemical synthetic action may, by a proper disposition of the constituents, be made to produce a voltaic current.” From the concept of “surface of action,” Grove increased the reaction area by platinized platinum electrode (platinum particles deposited on platinum electrode). Lord Rayleigh (1882) used platinum gauze to get larger solid electrode area. Mond and Langer (1889) not only used platinum black to increase solid electrode area in contact with strips of conductor, but also confined the electrolyte in a porous earthenware matrix to prevent electrode from flooding (liquid water occupation of pores of the electrode, causing gas reactants not to reach active sites of catalysts). More recently, in order to prevent flooding or to increase ion-conductivity, Teflon® or Nafion® is mixed with catalysts, respectively. Since Grove’s experiment, many other types of electrolyte have been tested, such as molten KOH, solid oxide, molten alkali metal carbonate, alkaline aqueous solution, phosphoric acid, solid polymer---each has its own advantages and operating conditions (Table 1-1) [5]. Different fuels have also been tested: coal, CH4, CO, methanol. Direct methanol fuel cell uses methanol directly as a fuel, not reformed to H2 first.

1-3 Why Do We Choose Atomic Layer Deposition (ALD)? Some advantages of ALD make itself a good choice for Pt deposition: good uniformity, precise thin film thickness control, large-area and large-batch production [23, 25]. Precise film thickness control means precise Pt particle size control, so uniform particle size distribution can be obtained to achieve good activity. Good uniformity means good dispersion of catalyst, and this can increase catalyst utilization. So ALD is tried to prepare Pt catalyst. 9

Figure 1-2. Grove’s cells in series use H2 and O2 as reactants to electrolyze H2O [5].

10

Table 1-1

Comparisons between different fuel cells [5].

Fuel cell type

Electrolyte

Alkaline FC (AFC) Proton exchange membrane FC (PEMFC) Phosphoric acid FC (PAFC) Molten carbonate FC (MCFC)

KOH Solid Polymer (e.g.: Nafion®) Phosphoric acid

Solid oxide FC (SOFC)

Direct methanol FC (DMFC)

Lithium and potassium carbonate Solid oxide electrolyte (e.g.: yttria, zirconia) Solid Polymer (e.g.: Nafion®)

Charge Carrier OHH+

Cell Temp. (oC) 60-120 50-100

H+

~220

CO32-

~650

O2-

~1000

H+

~60

11

Fuel Pure H2 Pure H2, (tolerates CO2 ) Pure H2, (tolerates CO2,1% CO) H2,CO, CH4, other hydrocarbons (tolerates CO2) H2,CO, CH4, other hydrocarbons (tolerates CO2) Methanol

Electric Efficiency (%) 35-55 35-45

Ref.

40

[5]

>50

[5]

>50

[5]

20~25

[7, 8]

[5] [5]

Chapter II. Literature Review 2-1 Basic Principles of Fuel Cell 2-1-1

Thermodynamic Aspects

Gibbs free energy difference, ∆G, is the driving force of the electrical work, pushing electron from anode to cathode, and this constructs a potential difference between anode and cathode [4]: E =－∆G/nF, where n is the number of electron transferred per overall reaction, and F is the Faraday constant (96320 Coul/mole). For the reaction: H2 + 1/2 O2
H2O, E (25oC) = 1.229 V. Intrinsic maximum efficiency of fuel cell is: εi =

∆G T ∆S = 1− ; ∆H ∆H

most of the reaction has efficiency larger than 90%. If ∆S is positive, and note that ∆H is negative, εi may even be larger than unity. For the reaction: H2 + 1/2 O2
H2O, εi = 0.83 and 0.91 at 25oC and 150oC, respectively. ∆G will change with temperature and pressure, and E will also change. Thus to compare fuel cell performance, cell and reactant temperatures, and reactant pressures must be specified. The simple derivations are as follows: (i) Temperature:

dG = − SdT + VdP , ( (

∂G )P = −S , ∂T

∂∆G ) P = −∆S , ∂T

12

(

∂E ∆S )P = . ∂T nF

So we can know how E changes with temperature once ∆S is known. (ii) Pressure: For a reaction:

aA + bB = cC + dD ,

∆G = cµc + d µ d − aµ a − bµb , where µi = (

∂G )T , P ,n j ≠ ni is the partial molar Gibbs free energy of species i. ∂n i (

(

∂µ ∂µ ∂µ ∂µ ∂∆G )T = c ( c )T + d ( d )T − a ( a )T − b ( b )T , ∂P ∂P ∂P ∂P ∂P

∂µi ∂ ∂G ∂ ∂G ∂ )T = [ ( )T , P,n j ≠ ni ]T = [ ( )T ]T ,P ,n ≠n = [ V] =Vi j i ∂P ∂P ∂n i ∂n i ∂P ∂n i T ,P,n j ≠ni

where V i is the partial molar volume of species i. So (

∂∆G )T = cV c + dV d − aV a − bV b . ∂P

If all gas species behave ideally, total volume V = (na + nb + nc + nd )

[ (

RT RT = ∑ ni ( ). P P i

∂ RT V ]T ,P ,n ≠n = = V i , so j i ∂n i P

∂∆G RT RT )T = (c + d − a − b) = ∆n . ∂P P P

By integration, ∆GP2 = ∆GP1 + ∆nRT ln(

P2 ). P1

Once ∆GP1 is known, ∆GP2 can be calculated if both P1 and P2 are known.

13

2-1-2

Kinetic Aspects

Although the reversible cell potential, Erev, can be predicted from thermodynamics, the actual cell potential, Ecell, is always lower than Erev because of kinetic behavior of the electrode (Fig. 2-1). Taking proton exchange membrane fuel cell using H2 and O2 as example, their relationship is governed by the equation [34]:

Ecell = Erev ( PH

2

, PO2 ,T )

− ∆Eohmic − η ORR − ηtx ,

where ∆Eohmic is the ohmic loss due to resistance of electronic contact and proton conduction through membrane; ηORR is the oxygen reduction reaction (ORR) overpotential (voltage loss) caused by sluggish O2 reduction kinetics; ηtx is mass transport voltage loss resulted from poor O2 transport through gas diffusion layer. Among the three factors that lower Ecell, ηORR contributes the most, around 65% of the overall voltage drop at a current density 1.5 A/cm2 (Fig. 2-1) [34]. Since oxygen reduction reaction lowers the cell potential the most, there are some suggestions to solve the problem [35]: (i) increasing catalyst activity of ORR; (ii) increasing Pt utilization through better dispersion; (iii) improving catalyst layer structure. All of these can be adjusted by using different catalyst

deposition methods.

2-1-3

Three-Phase Boundary

As Grove mentioned, the cell reaction depends on the “surface of action,” i.e., the area where electrolyte, gas reactant and solid electrode meet. “Surface of action” is also called “three-phase boundary.” What the three phases are can be easily recognized by the anodic half-reaction of PEMFC: H2 (g)
2H+ + 2e－; H2 (g) is conducted through pores; proton (H+) is through conducting electrolyte 14

Figure 2-1. Performance of proton exchange membrane fuel cell using H2 and O2 as reactants (Eeq = Erev) [34].

15

(Nafion®), and e- through catalyst particle, carbon substrate and carbon paper support (Fig. 2-2). So the three phases are gas conducting media (pore), proton conducting media (proton conducting electrolyte, Nafion®) and electron conducting media (catalyst particles, carbon substrate, and carbon paper support). If one of the three phases is absent, reaction will not occur, and voltage and current will not be produced, either.

2-2 Proton Exchange Membrane Fuel Cell (PEMFC) 2-2-1

Schematic of PEMFC

As shown in Fig. 2-3 [32], a PEMFC consists of four parts: (i) proton exchange membrane (PEM), (ii) catalyst layer, (iii) gas diffusion layer (GDL), and (iv) bipolar plates (current collector and gas channel). (i), (ii) and (iii) are always combined together, called membrane electrode assembly (MEA). PEM is electronically resistive, forcing electrons to flow through outer circuit load. On the other land, PEM is proton conductive, enabling proton to transfer from anode to cathode. It can prevent H2 crossover, i.e., leaking of H2 gas from anode to cathode without hydrogen oxidation reaction [36]. The molecular structure of PEM is shown in Fig. 2-4. It has perfluorocarbon backbone bonded with sulfonic acid group, on which protons are free to migrate. The catalyst layer consists of catalyst particles and catalyst support, and is the location of reduction of O2 or oxidation of H2. It can be coated on PEM, called catalyst-coated membrane (CCM); or on GDL, called catalyst-coated substrate (CCS) [34]. Further information of catalysts will be discussed in a later section. The gas diffusion layer, usually carbon cloth or carbon paper, ensures effective diffusion of reactants to the catalyst layer, and conducts electrons to and from the catalyst layer, and helps water management. 16

Figure 2-2. Three-phase boundary: pores for gas diffusion; Nafion® for H+ conduction; catalyst particle, carbon substrate and carbon paper support for electron conduction [6].

17

Figure 2-3. Schematic of PEMFC [32].

18

(a)

(b) Figure 2-4. (a) Molecular structure of Nafion®; (b) Ion clustering in Nafion® [9].

19

The bipolar plate, usually made of graphite, can conduct current and heat, control gas flow in the channels, and remove water produced by reactions. It also provides mechanical stability of fuel cell stack and enables sealing by compression [5].

2-2-2

Catalyst

2-2-2-1 Materials: Pt, Pt alloy, Pt-Free For the reaction: H2 + 1/2 O2
H2O, anodic half-reaction is H2 (g)
2H+ + 2e . －

Although many metal catalysts show high activity for hydrogen oxidation, precious metals, such as Pt or Pd, have better stability than non-precious metals. Because of the high activity for hydrogen oxidation, a low loading of 0.05 mg/cm2 will be enough for the anodic side. While for the cathode half-reaction, － 4H+ +4e + O2
2H2O,

more Pt loading, usually larger than 0.2 mg/cm2, is needed because the catalytic activity of Pt for oxygen reduction is not good. One of the major costs of fuel cell comes from the precious metal catalysts [5]. When a reformer is used, carbon monoxide will be mixed with the hydrogen gas in the product. CO at a ppm level can poison Pt severely. Thus, some Pt alloys have been tried to substitute for Pt, such as Pt-Rh and Pt-Ru. Pt-Rh is found to be CO tolerant at temperatures between 108 to 163oC at 1.7% CO, and Pt-Ru will be equivalent to Pt at temperatures lower than 120oC. For PEMFC, Pt-Ru is more suitable because the operation temperature limit of PEM is lower than 100oC for H2O contained inside the membrane to conduct protons. There are some other Pt alloys tried for cathode catalysts, e.g., Pt-Ni, Pt-Co,

20

Pt-Cr, Pt-Mn, Pt-Fe, Pt-Ti, etc. Most of them show better activity and similar stability compared to pure Pt catalyst [26, 34]. However, leaching of the non-noble base metals (Ni, Co, Cr, Mn, etc.) is a problem. It may lower H+ conductivity of the membrane by displacing H+ sites with base metal cations leached out from the Pt alloy catalysts. Therefore, Pt alloy catalysts must be pre-leached before use in MEA. Catalysts without Pt are also suggested for ORR. They are transition metals, e.g., Fe, Co, etc., surrounded by organic macrocycles (phthalocyanines, porphyrins, etc.). Although these Pt-free catalysts are much cheaper than Pt-containing catalysts, their activity and stability are worse. If it is to compensate the lower activity by increasing working electrode area, then the cost of bipolar plate and GDL and PEM will also increase. On the other hand, if it is to increase catalyst layer thickness, then an upper limit of the thickness exists because the mass transport overpotential will increase severely after the critical thickness is reached. It’s suggested at least one tenth activity of the Pt-free catalyst be obtained for automotive applications [34].

2-2-2-2 Particle Size Effect A cubo-octahedral model is suggested to illustrate the atomic arrangement in the Pt particle. That contains eight (111) planes, six (100) planes, corner and edge atoms as shown in Fig. 2-5 (a). Mass-averaged distribution (MAD) is defined as the ratio of the number of “surface atoms of a specific plane or site” to the number of “total atoms contained in a Pt particle”, such as [N (100)/ N (t)] Surface-averaged distribution (SAD) is the ratio of the number of “surface atoms of a specific plane or site” to the number of “total surface atoms of Pt particle”, such as [N (e+c)/ N (s)], where “N (e+c)” means the “number of atoms on edge and corner sites.” SAD and MAD of cubo-octahedral model are plotted in Fig. 2-5 (b) and 21

(a)

22

(b)

(c) Figure 2-5. (a) Cubo-octahedral model of platinum particles. (b) SAD of cubo-octahedral model; (c) experimental results of specific activity and SAD (100) curve [33]. 23

Fig. 2-6 (a), respectively. It is found that the experimental results are very similar to the behavior of (100) and (111) planes of the cubo-octahedral model. Therefore, it’s suggested these two planes are responsible for oxygen reduction reaction for Pt particle. It has been found that Pt particle size will affect mass activity and specific activity [33]. Mass activity is Ampere per gram of Pt at a specific voltage (A/gPt), and specific activity is micro Ampere per centimeter square of Pt surface area at a specific voltage (µA/cm2Pt). For Pt particle size ranging from 0 to 15 nm, the specific activity decreases as particle size decreases, and the mass activity has a maximum at about 3.5 nm as shown in Fig. 2-5 (c) and 2-6 (b), respectively.

2-2-2-3 Carbon Catalyst Supports: Carbon Black, CNT The support for metal catalyst is needed because of many reasons [37]. It will increase surface to volume ratio of metal catalyst, the ability to get highly dispersed catalysts even at high Pt loading (e.g., 30 wt%), and stability of metal catalyst by preventing it from agglomeration. The limitation of the fuel cell operating conditions, i.e., both anode and cathode are very acidic and electron must be conducted to and from the reaction cites, makes carbon support one of the best choices. Carbon black is the most common support used in PEMFC. Its surface area per weight ranges from 20 to 1500 m2/g [5], depending on the source materials and manufacturing processes. Different types of carbon black have different surface areas, which affect the catalyst dispersion [40]. Vulcan XC-72 is one of the most used carbon black catalyst support. The crystal structure of carbon black is either amorphous or “para-crystallite,” i.e., the individual graphitic layer as the basic building block of carbon black particles. Carbon nanotube (CNT) has also been tried as catalyst support because of 24

(a)

(b) Figure 2-6. (a) MAD of cubo-octahedral model; (b) experimental results of mass activity and MAD (100) curve [33]. 25

its nanometer size, high accessible surface area, good electron conductivity and good stability [35, 38-39]. Before metal deposition, CNTs are treated in acid at elevated temperatures in order to modify the functional groups (OH, COOH, C=O) [35]. The original inert surface of CNTs is not easy for deposition of catalysts [38]. The following is an example of the reaction between the surface functional groups and metal ions: CNT－COOH + M+X
CNT－COO M+ + HX, －

－

where the proton on the carboxylic group exchanges with the metal ion to form HX.

2-3 PEMFC Catalyst Deposition Methods There are numerous methods to prepare catalyst for PEMFC. The deposition methods can be mainly divided into two categories: wet or dry process. “Wet” means the dispersive medium in the process is in the liquid phase, and “dry” is in the gas phase. Wet processes include chemical [10] and electrochemical processes [41-46]. Dry processes include sputtering [16-22, 48], dual ion-beam-assisted deposition [14, 15] and pulse laser deposition [47].

2-3-1

Wet Process

2-3-1-1 Chemical Process The chemical processes can be simplified as shown in Fig. 2-7. The metal precursors of ionic or molecular state will be converted by chemical reactions to form metal particles, and the size of which can be controlled by confining support (e.g., Vulcan XC-72) or protecting agent (e.g., surfactant molecules). Chemical processes include impregnation, colloidal, and microemulsion, as shown in Fig. 2-8. Impregnation method is to let the pores of support soak up metal ions,

26

Figure 2-7. Schematic of chemical synthesis of supported metal nanoparticles [10].

27

Figure 2-8. Size controllable chemical methods to produce Pt nanoparticles on carbon supports [10].

28

which will be reduced to metal particles by reducing agents such as hydrazine, borohydride, and hydrogen. The pore size and distribution thus affect the particle size and distribution, respectively. When a uniform pore size is not easily obtained in the support, the particle size will range from nanometer to micrometer scale. The colloidal method can protect the reduced metal particles from agglomeration by adding surfactants to provide steric hindrance or electrostatic repulsion. Before using these protected metal particles in fuel cells, removing the protective molecules is necessary to make sure the catalytic reactions can occur upon the surface. Unprotected colloids use ethyl glycol both as a solvent and protective molecules which is suitable for large-scale synthesis. The microemulsion method is to engulf the metal precursors inside a surfactant shell. The size of the shell can decide the particle size. Inside the shell, reduction of metal ion can occur by stirring the reducing agent containing solution where microemulsion droplets also exist. Since the surfactants used in the method are expensive and extra wash steps are needed, it may not be applied for large-scale production.

2-3-1-2 Electrochemical Process As shown in Fig.2-9, the apparatus for electrochemical deposition of Pt particles on the cathode consists of a platinum containing solution and a pulse generator (or DC power supply) connected to the anode (Pt gauze) and the cathode. The cathodess include Nafion® impregnated cathode (Fig. 2-9 (a)), PEM attached cathode (Fig. 2-9 (b)), and original cathode (no Nafion® or PEM added, Fig. 2-9 (c)). The cathode consists of hydrophobic porous carbon cloth impregnated with carbon particles and PTFE (polytetrafluoroethylene). In Fig. 2-9 (c), Pt is deposited on the electrically conductive surface since the electrons needed to reduce the Pt ions to Pt particles are provided by the 29

Pulse Generator

Anode

Cathode

(a)

Pulse Generator

Anode

Cathode

(b)

Pulse Generator

Anode

Cathode

(c)

Figure 2-9. Electrochemical deposition of Pt on the electrode: (a) Nafion® impregnated cathode: Pt particles on Nafion®/carbon interface [41, 42]; (b) PEM on cathode: Pt particles on PEM/carbon interface [43, 44]; (c) No PEM or Nafion® on cathode: Pt particles on carbon [45, 46].

30

pulse generator through the carbon support. In Figs. 2-9 (a) and (b), Pt particles are formed at the interface of Nafion® and carbon support because Pt ions are transported from the solution through Nafion® layer to the Nafion®-carbon interface, and thus this makes sure that the particles will deposit in the region where is proton and electron conductive. A pulse generator can control peak current density, and on/off time, while a DC power supply can only control current density, causing deficient ion concentration near the electrode. This can be alleviated during the off time of pulse generator.

2-3-2

Dry Process

A schematic of dry process is shown in Fig. 2-10, which is also called physical vapor deposition (PVD). Pt bulk is bombarded by the incident beam. In an e-beam evaporator, the beam consists of electrons. For sputtering, it is plasma, and for pulse laser, it is electromagnetic wave. Dry processes are suitable for low loading, and fewer catalyst preparation steps are needed compared to wet processes. For high loadings, uniform size distribution of catalyst particles can not be easily achieved because Pt particles can not penetrate into deep region of the carbon support. In dual ion-beam-assisted deposition as shown in Fig. 2-11, an e-beam evaporator is used to deposit Pt on the substrate, and ions extracted from the plasma are accelerated toward the substrate, causing substrate heating and a more uniform and denser Pt film. In sputtering, Fig. 2-12, different substrates (carbon cloth, PEM, or catalyzed carbon cloth) are tried. In some cases, multiple layer sputtering is also tried to increase the active Pt surface area and Pt loading by alternating application of “Nafion® carbon ink” (NCI: a mixture of Nafion® and XC-72) layers and Pt sputtered layers as shown in Fig. 2-13.

31

Substrate

Pt atom

Incident Beam

Pt bulk

Figure 2-10. Schematic of dry process [11-13].

4

3 1

2

5 Figure 2-11. Schematic of dual ion-beam-assisted deposition. (1: electron beam; 2: Pt atoms; 3: plasma; 4: carbon cloth; 5: Pt target) [14, 15].

32

3

1

2

4 Figure 2-12. Schematic of sputtering. (1: plasma; 2: Pt atoms; 3: carbon cloth [16], PEM [17-18], or catalyzed carbon cloth [19-21]; 4: Pt target).

1 2

3 Figure 2-13. Multiple Layer Sputtering. (1: PEM; 2: Pt sputtered; 3: Nafion®-carbon ink) [17, 22].

33

2-3-3

Performance Comparison of Different Catalyst Deposition

Methods In order to make fuel cells applicable to large-scale automotive application, the Pt specific power density (gPt/kW) must be lower than 0.2 and the MEA power density (W/cm2) must be larger than 0.8 [34]. The goal set for 2010 by the U.S. department of energy (DOE) for Pt specific power density is also 0.2, and for the MEA power density is 0.5. Moreover, in order to use fuel at high efficiency (> 55%), operation voltage of the cell must be larger than 0.65V [34]. Therefore, 0.65V is chosen to compare the performance of different catalyst preparation methods. Since not all performance test results of different catalyst preparation methods are good, only some results are shown in Table 2-1.

2-4 Atomic Layer Deposition (ALD) 2-4-1

ALD Characteristics

Atomic layer deposition is a thin film manufacturing process taking advantage of the self-limiting mechanism of the gas-solid reaction to achieve good uniformity, conformality and precise thin film thickness control by adjusting the cycle numbers [23, 25]. Source vapors are pulsed into the reactor alternatively and separated by purging gas (Fig. 2-14). One cycle is over after the four steps: (i) pulse of reactant A; (ii) purge of inert gas; (iii) pulse of reactant B; (iv) purge of inert gas. Whether a reaction will take place largely depends on the presence or absence of the surface reactive functional group As shown in Fig. 2-15 (a), TiCl4 molecules react with the hydroxyl group, while in Fig. 2-15 (b) intact or dissociative chemisorption occurs [25]. Therefore, the substrate effect, i.e., materials can be deposited on some substrates but not on others, will be observed.

34

Table 2-1

Performance of different types of catalyst deposition methods. Method

Operating Condition PO2(atm); Tcell(oC); Humidified degree

Pt Loading (mgPt/cm2) anode cathode

Power Density (W/cm2)

gPt/kW (0.65V)

Ref.

U.S. DOE’s goal at 2010

N;N;N

0.1

0.5

0.2

[52]

chemical

0.2; 80; fully (80oC)

0.4

0.4

0.71

1.13

[34]

chemical

0.2; 80; fully (80oC)

0.05

0.4

0.64

0.71

[51]

electro-deposition

1; 75; N

0.4

0.32

0.79

0.91

[45]

pulse laser

5; 80; fully (110oC)

0.017

0.4

0.41

1.0

[47]

dual ion-beam-assisted deposition

4.2; 80; N.

0.04

0.04

0.88

0.09

[14]

sputter

1; 21; slightly

0.005

0.005

0.08

0.12

[48]

N: Not mentioned; Pt loading of U.S. DOE’s goal is the summation of anode and cathode loading.

35

Figure 2-14. Schematic of one ALD reaction cycle [23].

36

(a)

(b) Figure 2-15. ALD of TiO2: (a) hydroxyl group terminated; (b) dehydroxylated surface after the water pulse [25].

37

It is often questioned whether the film is deposited through CVD or ALD mechanism. In general, CVD involves reactions both in gas phase and gas-solid interface, as shown in Fig. 2-16, whereas ALD reaction just take place in gas-solid interface. Thus self-limiting is not observed in the CVD process where growth per cycle will increase proportionally to precursor flux (Fig. 2-17). Even though alternative precursor pulses are applied and one monolayer of the material is deposited per cycle, a pseudo-ALD may occur in a CVD mechanism by controlling the precursor flux at a precise amount (point A in Fig.2-17). While in ALD no matter how much amount of precursor is used, the growth per cycle remains the same.

2-4-2

ALD of Pt

In Fig. 2-18, self-limiting is observed for ALD of Pt, either by varying Pt precursor pulse time at a constant air pulse time or varying air pulse time at a constant Pt precursor pulse time. In Fig. 2-19, the film thickness is proportional to the cycle numbers. In Fig. 2-20, Pt particles form instead of films. The suggested reactions for Pt precursor and air are as follows: Pt(s) + O2(g)
Pt-Ox, CH3C5H4Pt(CH3)3 + Pt-Ox
Pt(s) + CO2(g) + H2O(g) + other byproducts. In Table 2-2, both O2 and air are used for the reaction. It is found that when O2 is used the growth temperature can be lowered to 200oC than 250oC when air is used. The impurity content of Pt film when O2 is used is lower than that when air is used. The film deposited when O2 is used can pass the tape test, while when air is used it will not pass the tape test.

38

Figure 2-16. Schematic kinetics of CVD of Si [28]. Note that gas phase reactions take place and Si can deposit on surface by adsorption of Si vapor onto the substrate (R7).

39

Figure 2-17. Different relationship between GaAs growth thickness per cycle and trimethylgallium (TMG) flux for MOCVD and ALE (Atomic layer epitaxy: the older nomenclature for ALD) [29].

40

Figure 2-18. Self-limiting curve of ALD of Pt [24].

Figure 2-19. Film thickness is proportional to reaction cycle number [24].

41

Figure 2-20. SEM image of ALD of Pt [24].

42

Table 2-2

Comparison of ALD Pt using air or O2 as reactant B.

Precursor

Reactant B Reactant B

Deposition

pulse time

pulse time

Flow Rate

Temp. (oC)

(sec)

(sec)

(sccm)

MeCpPtMe3

Air

40

0.2~1.5

1~3

Substrate

Growth

Impurity

Tape

Rate

Content

Test

(Å/cycle)

(at%)

300

Al2O3

0.5

H: 0.5~1.0

Not

250

Borosilicate

~0

C: 0.5~1.0

passed

SiO2 MeCpPtMe3

Pure O2

0.7

1

20

[24]

O: 1.0~2.0

200

Al2O3

0.3

H: <0.5

300

Borosilicate

0.5

C: <0.5 O: <0.5

43

Ref.

Passed

[27]

Chapter III. Experimental 3-1 Substrate Preparation Carbon cloth A (E-TEK, B1A) was subjected to acid treatment. It was refluxed in 100 ml 65% HNO3 at 140oC for 6 hours, washed with water, and dried in oven at 50oC for 24 hr. Carbon cloth A with acid treatment was abbreviated as CC. Carbon cloth B (E-TEK, B-1/B/StdWP) with or without acid treatment was also used as a substrate to show the substrate effect. 0.02 gram of Vulcan XC-72 carbon black powder (Cabot Corp.) and 0.02 gram of 0.0006 wt% PTFE solution (diluted from 60wt% PTFE solution (Aldrich)) were mixed together to form an ink. It was ultrasonically treated for 30 min, pasted on an acid treated carbon cloth A upon a heating plate at 90oC. XC-72 loading was about 1.69 mg/cm2. This configuration is abbreviated as XC-72/CC. Carbon nanotubes after the same acid treatment as carbon cloth A was treated with the same procedure as Vulcan XC-72 carbon black powder to form CNT/CC. With 0.02 g CNTs, the CNT loading was about 2.3 mg/cm2. Another concentration of PTFE (0.06 wt%) was also tried to determine the effect of PTFE content on ALD of Pt on XC-72/CC and CNT/CC.

3-2 Deposition of Catalyst 3-2-1

ALD of Platinum

Schematic of ALD apparatus is shown in Fig. 3-1. O2 (99.999%) and N2 (99.9995%) bottles were supplied through needle valves to control the gas flow rate. The Pt precursor, MeCpPtMe3, was contained in a stainless bottle, the temperature of which was maintained at 45oC by a temperature controller (Microprocessor PID Temperature Controller, Hunjoen). The vapor pressure of MeCpPtMe3 at different temperature can be calculated from the equation [49]:

44

Rotary Pump

V1

V2

Vacuum Gauge

V3 Heating Plate

Temperature Controller

O2

N2 Pt precursor

Figure 3-1. Schematic of ALD apparatus.

45

ln Pvap(torr) = 15.3 －5.24．103/T(K), where Pvap is the vapor pressure of liquid MeCpPtMe3. At 45oC, vapor pressure of MeCpPtMe3 is 0.308 torr. Pneumatic valve 1 (V1) is responsible for O2 pulse, pneumatic valve 2 (V2) for N2 purge, pneumatic valve 3 (V3) for Pt pulse, and pneumatic valve 4 (V4) is for the rotary pump (SD 2015, Alcatel). The stop valve in front of Pt precursor stainless bottle is used to open or close the Pt bottle. The stop valve between V2 and V3 is for N2 to purge away remained Pt precursor inside the pipelines. The pipelines are heated to about 70oC to prevent condensation of Pt precursor on the wall of pipelines. The heating plate, with an area of 16 cm2 capable of heating to 350oC, is located inside the chamber. The samples can be loaded onto the plate by opening the upper lid of the chamber. Pressure of the chamber is measured by thermal couple gauge (275 Convenctron gauge, Helix) and capacitance gauge (MKS Instruments) Four steps for the ALD of Pt are shown in Table 3-1. Pt pulse is achieved by opening V3 and V4 to make rotary pump bring the Pt precursor into the chamber because carrier gas is not designed for the chamber, and diffusion rate of Pt precursor is slow. N2 purge is achieved by opening V2 and V4, and O2 pulse is controlled by V1. The opening time of the four pneumatic valves and total cycle number can be set by respective timers. For all the ALD processes, the O2 pulse time was set to 5 sec. The N2 flow rate is estimated to be 1980 atm．cc/min, with the pumping speed 18 m3/hr and steady state pressure 5 torr. The substrate temperature was maintained at 250oC. The cycle number, Pt pulse time, substrate type, and N2 purge time were varied according to different requirements. The N2 purge time was 20 sec when the Pt pulse time was less than 7 seconds, and it was extended to 60 sec for Pt pulse time larger than 7 sec, in order to purge away all of the remaining precursor.

46

Table 3-1

Four steps for the ALD reaction. V1

V2

V3

V4

1st

Pt pulse

X

X

O

O

2nd

N2 purge

X

O

X

O

3rd

O2 pulse

O

X

X

X

4th

N2 purge

X

O

X

O

O: open; X: closed

47

To test the self-limiting behavior, Pt was deposited on Si wafer with the Pt pulse time ranging from 1 to 15 sec for 20 cycles. For examination of the proportionality between the Pt particle size and cycle number, Pt was deposited on Si wafer for 200, 300, and 400 cycles at a Pt pulse time of 1 sec. In order to study the substrate effect, different substrates were tried. For carbon cloth B with or without acid treatment, the Pt pulse time was set to 1 sec for 200 cycles. For XC-72/CC and CNT/CC with different PTFE concentration, the Pt pulse time was 1 sec for 300 cycles. For catalyst deposition, Pt was deposited on different substrates: CC, XC-72/CC, and CNT/CC for Pt pulse time 1 sec and 100 cycles to form anode or cathode of the fuel cell.

3-2-2

Sputtering

An auto fine coater (JFC-1300, JEOL) was used to sputter Pt on CC. Electrodes were prepared by sputtering for 53 sec, the Pt loading of which was equal to 100 cycles of ALD of Pt (see Table 4-1).

3-3 Preparation of MEA The flow chart for preparation of MEA is shown in Fig. 3-2. Both anode and cathode were brushed with 5% Nafion® solution and dried at room temperature for 1 hour. The Nafion® 115 membrane were cleaned with H2O2 at 80oC for 1 hour to remove organic impurities, rinsed by deionized water, immersed in H2SO4 at 80oC for 1 hour, and then rinsed by deionized water again. The Nafion® 115 membrane were sandwiched with anode and cathode, hot pressed at 140oC and 600 psi for 90 sec.

3-4 Single Cell Setup A single cell setup consists of three parts as shown in Fig. 3-3. The MEA denoted as “C” with an active area of 4 cm2 was surrounded by two Teflon® 48

Figure 3-2. Flow chart of MEA preparation.

49

Figure 3-3. Conventional single cell setup [30].

50

gaskets on the margin. Two resin-impregnated graphite plates (Well Hand Co., denoted as “B”), with gas inlet and outlet holes and two-way-through serpentine patterns, provided distribution of reactant gases and current collector for MEA. Stainless steel end plates (denoted as “A”) were used to conduct current from MEA and to hold B and C together by locking screws through the eight holes on the plate as shown on the right side of Fig. 3-3. There are also holes for clubbed heater and thermocouple, as shown on the left side of Fig. 3-3.

3-5 Fuel Cell Test Station and Performance Test Conditions As shown in Fig. 3-4, the I-V curve of the MEA was measured by an electronic load (Agilent N3301A) interfaced through a PCI-GPIB (National Instruments) with NI-488.2 software and Labview program installed in a personal computer. The power supply (Cosel, UAW500S-5) was connected between the fuel cell and electronic load in order to measure the low voltage of the cell. The humidity of reactant gases was controlled by the temperature of the water bottles connected between the fuel cell and the gas bottles. The flow rate of H2 and O2 were controlled by the mass flow controller (Sierra 840-MP and Brooks 5850 E series) and a flow control box (Protec Instrument PC540). By holding at a specific voltage for 15 sec, the averaged current value was recorded. The voltage sweeping cycle started from 1.0 V to 0 V, then to 1.0 V again. At least 5 cycles were tested for one condition. All cells were activated for 1 to 2 hr to achieve a steady state of the operation condition. The cells were slightly humidified by means of the water bottles for H2 and O2 set at 33oC. The cell temperature was fixed at 60oC, and the flow rates of H2 and O2 were both at 50 sccm. Different electrodes were used in the performance test of the cell, including ALD of Pt 100 cycles on CC, XC-72/CC, CNT/CC, sputtering of Pt for 53 seconds on CC, and as received E-TEK electrode (Pt/C 20%, 0.5 mgPt/cm2). 51

Figure 3-4. Schematic of fuel cell test station [31].

52

The performance test was conducted by two ways. First, both anode and cathode of a MEA were the same. Second, the anodes were different, but the cathodes were all as received E-TEK electrode.

3-6 Characterization and Chemical Analysis of Pt Catalyst Inductively coupled plasma mass spectrometry (ICP-MS) was used to determine the amount of elements in the sample. The low detection limit for most elements (0.1 to 10 ppb), high accuracy and precision of ICP-MS makes it an important technique for elemental analysis [50]. A resolution of 0.1 ppb could be obtained for Pt quantitative measurement. Pt on carbon cloth was dissolved in aqua regia and HF solution with a total volume of 50 ml. The Pt loading was obtained by dividing the total amount of Pt by the carbon cloth surface area. A field emission scanning electron microscope (FESEM, JSM-6500F, JEOL) was used to observe the surface morphologies of different electrodes. Most of the magnification was set to 100,000 X.

53

Chapter IV.

Results and Discussion

4-1 ALD Characteristics 4-1-1

Self-Limiting

In order to use Pt as electro-catalyst in PEMFC, Pt particles are preferred than Pt films, which have a lower surface to volume ratio and lower active sites than particles. The Pt particle size shown in Fig. 4-1 was used to check the self-limiting behavior of ALD of Pt. With 20 cycles, Pt particle sizes were estimated to 6, 8, 8, 8 nm for 3, 7, 11, 15 sec of Pt pulse, respectively. Therefore it’s concluded that self-limiting was achieved.

4-1-2

Proportionality between Cycle Number and Particle Size

The Pt particle size varies with different cycle numbers as seen from Fig. 4-2. The Pt particle size increased from 10 nm to 20 nm and 30 nm when the cycle numbers increase from 200 to 300 and 400. Therefore, the proportionality between cycle number and particle size was achieved.

4-1-3

Substrate Effect

ALD reactions depend largely on surface functional groups, and Pt precursors react with oxygen on the surface. Originally, Pt could not deposit on carbon cloth B (CC-B) without acid treatment. After acid treatment with HNO3, however, Pt particles could be deposited well as seen in Fig. 4-3 (a) and (b). This may be contributed to the surface modification of functional groups (OH or COOH or C=O) to provide reactive sites for gas-solid reaction. The cycle number is 200, and the Pt pulse time is 1 sec for the above two substrates When XC-72 or CNT was covered with a higher concentration of PTFE, the fluorine atom of PTFE may not react with the Pt precursor. Thus no Pt could deposit on XC-72/CC and CNT/CC with 0.06 wt% PTFE as illustrated in Figs.

54

(a)

(b)

55

(c)

(d) Figure 4-1. SEM images of ALD of Pt on Si for 20 cycles at different Pt pulse time: (a) 3 sec, (b) 7 sec, (c) 11 sec, (d) 15 sec. 56

(a)

(b)

57

(c) Figure 4-2. SEM images of ALD of Pt on Si at Pt pulse time 1 sec for different cycles: (a) 200 cycles, (b) 300 cycles, and (c) 400 cycles

58

(a)

(b)

59

(c)

(d)

60

(e)

(f) Figure 4-3. ALD of Pt on different substrates: (a) CC-B without acid treatment, (b) CC-B with acid treatment, (c) XC-72/CC with 0.06 % PTFE, (d) XC-72/CC with 0.0006 % PTFE, (e) CNT/CC with 0.06 % PTFE, (f) CNT/CC with 0.0006 wt% PTFE.

61

4-3 (c) and (e). When the PTFE concentration was lowered to 0.0006wt%, however, Pt is able to deposit on XC-72/CC and CNT/CC as shown in Figs. 4-3 (d) and (f). For these four cases, the cycle number was 300, and Pt pulse time was 1 sec. Without addition of PTFE, which served as a binding agent between XC-72 power and carbon cloth, XC-72 powder would peel off from the carbon cloth easily. The concentration of 0.0006 wt% was determined by trials and errors, and might not be the best concentration for adhesion of XC-72 powder with CC, Pt deposition, and water proof of the substrate.

4-2 Pt Loading on Different Substrates The Pt loadings on different substrates are determined by ICP-MS as shown in Table 4-1. They are all in the microgram level which is not easily achieved by chemical methods (usually an order of magnitude larger). Based on the Pt loadings on CC for different cycles, it is observed that the Pt loading is lower when XC-72/CC is used and Pt loading is higher when CNT/CC is used. This may be explained by the more active sites in the acid treated CNTs than that in acid treated carbon cloth, or the larger surface area of the CNT coated carbon cloth than not coated carbon cloth (Figs. 4-3 (e) and (a)). For XC-72 coated carbon cloth, although the total surface area increases, the number of active sites must be lowered according to the ICP-MS results. By sputtering, the averaged Pt loading is 9.763．10－5 mgPt/(cm2．sec). Therefore 53 seconds of sputtering equals 0.005174 mgPt/cm2, equal to 100 cycles of ALD. It is also shown that Pt loading increases with cycle numbers. This is another indirect evidence of proportionality between particle size and cycle number, except that “300, CNT/CC” has a lower loading than “200, CNT/CC.” It may be resulted from bad distribution of CNT on CC during pasting by a

62

Table.4-1

Pt loadings by ALD on different types of substrate and by sputtering on CC. 2

Type

mgPt/cm

100, CC

0.005126

200, CC

0.008294

300, CC

0.013688

100, XC-72/CC

0.003362

200, XC-72/CC

0.004784

300, XC-72/CC

0.008795

100, CNT/CC

0.016881

200, CNT/CC

0.029917

300, CNT/CC

0.018988

Sputter 30, CC

0.00262

Sputter 60, CC

0.005819

Sputter 90, CC

0.009771

Note: “100, CC” means 100 cycles of ALD of Pt on CC; “Sputter 90, CC” means sputtering of Pt on CC for 90 sec.

63

dropper; but this still has a higher Pt loading than “300, CC.”

4-3 Performance Tests of MEAs Made by ALD, Chemical and Sputtering Methods Fig. 4-4 shows the performance test results of different MEAs. In Fig. 4-4 (a), both anode and cathode of the MEA are the same, and in Fig. 4-4 (b) the anodes are different but cathodes are all as received E-TEK electrodes. Electrodes discussed here are “100, CC,” “100, CNT/CC,” “as received E-TEK electrode,” “Sputter 53.” The “100, XC-72/CC” is not discussed here because flooding is very severe, and the current density is almost zero. In Fig 4-4 (a), “as received E-TEK electrode” has the best current density of the four electrodes in the whole range of voltage. In Fig. 4-4 (b), although the Pt loading of “100, CNT/CC” is lower than that of “E-TEK,” a similar performance is obtained. There are three possible reasons for this behavior. First, the lower loading of PTFE in the “100, CNT/CC” makes itself more conductive than “E-TEK,” or better humidification of the membrane of “100, CNT/CC” because the water bottle temperature is only 33oC (slightly humidified condition). The better humidification increases the conductivity of H+ in the membrane. Second, better catalyst distribution (Fig. 4-3 (f)) and the direct contact of catalyst with the PEM create more three-phase boundary than that of conventional chemical catalyst, which is the catalyst of E-TEK electrode. Third, gas diffusion layer of CNT coated CC may be lower than E-TEK electrode, resulting in less mass transfer voltage loss. The comparison of Pt specific power density (gPt/kW) at 0.65 V and power density (W/cm2) is shown in Table 4-2. For the case where both anode and cathode are the same in the MEA, “100, CC” has the best Pt specific power density, but its power density is just 3.6 mW/cm2, substantially lower than the

64

1.0 100, CC 100, CNT/CC E-TEK Sputter 53

Voltage (V)

0.8 0.6 0.4 0.2 0.0 0.0

0.2

0.4

0.6

0.8

1.0

2

Current Density (A/cm )

(a)

1.0 100, CC 100, CNT/CC E-TEK Sputter 53

Voltage (V)

0.8 0.6 0.4 0.2 0.0 0.0

0.2

0.4

0.6

0.8

1.0

2

Current Density (A/cm )

(b) Figure 4-4. Performance of the MEA at cell temperature 60oC: (a) both anode and cathode of the MEA are the same; (b) anodes are different; cathodes are all “as received E-TEK electrode.” (As received E-TEK electrode is abbreviated as “E-TEK”)

65

Table 4-2

Comparison of performance of different MEAs. Electrode

A

C

W/cm2 (0.65 V)

mgPt/cm2 A C

gPt/kW (A + C)

100, CC

100, CC

0.0036

0.005

0.005

2.78

100, CNT/CC

100, CNT/CC

0.0057

0.017

0.017

5.96

E-TEK

E-TEK

0.132

0.5

0.5

7.58

Sputter 53

Sputter 53

0.0025

0.005

0.005

4

100, CC

E-TEK

0.018

0.005

0.005

28.06

100, CNT/CC

E-TEK

0.129

0.017

0.5

4.01

E-TEK

E-TEK

0.132

0.5

0.5

7.58

Sputter 53

E-TEK

0.031

0.005

0.005

16.39

A: anode; C: cathode

66

goal of 0.5 W/cm2. For the case where anodes are different but cathodes are all “E-TEK,” “100, CNT/CC” has the lowest Pt specific power density of 4.01 which is better than 7.58 of “E-TEK” with similar power densities (0.129 for ‘100, CNT/CC” and 0.132 for “E-TEK”). It is supposed that there is a critical catalyst loading ratio of anode to cathode for which the Pt specific power density reaches a minimum. Any value larger or lower than the critical value will increase the Pt specific power density. For the reaction, 2H2 + O2
2H2O, to occur, two hydrogen molecules at anode must be oxidized, and one oxygen molecule at cathode is reduced. However, the catalytic activities of Pt for the two reactions are different. Hydrogen oxidation is easy, while oxygen reduction is rather sluggish [51]. Thus the cathode must have a larger Pt loading than the anode. If the Pt loading at the cathode remains constant, it will have a constant number of active sites for oxygen reduction reaction, and a constant number of protons are needed for the cathodic half reaction (4H+ +4e + O2
2H2O). This －

constant number of protons can be produced by a specific Pt loading at the anode. Any Pt loading larger than the critical value will be wasted. The same argument holds for the waste of cathodic Pt loading if the anodic Pt loading is constant and can just produce a certain amount of protons. That is why the Pt specific power density of “100, CC” is low when both anode and cathode are “100, CC.” For the case where anode is “100, CC” but cathode is “E-TEK,” the cathodic loading is wasted, so the Pt specific power density increases from 2.78 to 28.06.

67

Chapter V.

Conclusions

1. Characteristics of ALD---self-limiting, proportionality between particle size and cycle numbers, and substrate effect---are verified by SEM images. About self-limiting, constant particle size (about 8 nm) are observed for 20 cycles at Pt pulse time larger than 7 seconds. About proportionality between particle size and cycle numbers, particle size of 10 nm are observed for 200 cycles; 20nm, 300 cycles; 30nm, 400 cycles. About substrate effect, Pt can not be deposited on CC without acid treatment, or on high concentration PTFE bounded carbon support; Pt can be deposited on CC after acid treatment and carbon support with very low concentration of PTFE. 2. Microgram level of Pt loading is achieved from ALD. Pt loading of ALD on CC is increased by addition of CNT on CC, but is decreased by addition of XC-72 on CC. Pt loading increases with cycle numbers. 3. When both anode and cathode are the same for the MEA, “100, CC” shows the lowest Pt specific power density; when anodes are different and cathodes are all “E-TEK,” “100, CNT/CC” shows the lowest Pt specific power density. This may be resulted from: (i) the lower PTFE concentration of CNT/CC; (ii) better catalyst distribution and utilization; (iii) thinner gas diffusion layer. 4. Critical catalyst loading ratio of anode to cathode for which the Pt specific power density reaches a minimum explains the lower Pt specific power density of the MEA of the same anode and cathode with “100, CC” than that of MEA of anode being “100, CC” but cathode being “E-TEK.”

68

References [1]

中華民國經濟部能源局網頁.

[2]

詹益亮, 洪桂彬, ”科學發展,” 413, 18 (2007).

[3]

胡成春, “清潔新能源─21 世紀的能源,” 科學技術文獻出版社 (1995).

[4]

J. O’M Bockris, S. Srinivasan, “Fuel Cells: Their Electrochemistry,” McGraw-Hill (1969).

[5]

G. Hoogers, ”Fuel Cell Technology Handbook,” CRC Press, London and New York (2003).

[6]

L. Carrette, K. A. Friedrich, U. Stimming, Chem. Phys. Chem., 1, 162 (2000).

[7]

J. H. Wee, J. Power Sources, 161, 1 (2006).

[8]

N. W. Deluca, J. Polym. Sci., Part B: Polym. Phys., 44, 2201 (2006).

[9]

K. Kordesch, G.. Simader, “Fuel Cells and Their Applications,” VCH Publishers, Inc., New York, NY (1996).

[10] K.Y. Chan, J. Ding, J. Ren, S. Cheng ,K.Y. Tsang, J . Mater. Chem., 14, 505 (2004). [11] http://0rz.tw/b52QX. [12] http://0rz.tw/302KG. [13] J. Singh, D.E. Wolfe, J. Mater. Sci., 40, 1 (2005). [14] A. F. Gullá, M. S. Saha, R. J. Allen, S. Mukerjee, Electrochem. Solid-State Lett., 8, A504 (2005). [15] A.F. Gullá, M. S. Saha, R. J. Allen, S. Mukerjee, J. Electrochem. Soc., 153, A366 (2006). [16] S. Hirano, J. Kim, S. Srinivasan, Electrochim. Acta, 42, 1587 (1997). [17] S. Y. Cha, W. M. Lee, J. Electrochem. Soc., 146, 4055 (1999). [18] R. O’Hayre, S. Lee, S. Cha, F.B. Prinz, J. Power Sources, 109, 483 (2002).

69

[19] E. A. Ticianelli, C. R. Derouin, S. Srinivasan, J. Electroanal. Chem., 251,275 (1988). [20] S. Mukerjee, S. Srinivasan, A. J. Appleby, Electrochim Acta 38, 1661 (1993). [21] C. A. Cavalca, J. H. Arps, M. Murthy, US Pat. No. 6,300,000 (2001). [22] A. T. Haug, R. E. White, J. W. Weidner, W. Huang, S. Shi, T. Stoner, N. Rana, J. Electrochem. Soc., 149 (3), A280 (2002). [23] R. L. Puurunen, J. Appl. Phys., 97, 121301 (2005). [24] T. Aaltonen, M. Ritala, T. Sajavaara, J. Keinonen, M. Leskela, Chem. Mater., 15, 1924 (2003). [25] M. Ritala, M. Leskela, “Handbook of Thin Film Materials,” Academic Press, San Diego, Vol. 1, 103 (2002). [26] D. Thompsett, “Handbook of Fuel Cells---Fundamentals, Technology and Applications,” Volume 3, John Wiley & Sons (2003). [27] T. Aaltonen, M. Ritala, Y. L. Tung, Y. Chi, K. Arstila, K. Meinander, M. Leskelä, J. Mater. Res., 19, 3353 (2004). [28] A. K. Praturi, “Proceedings of the sixth international conference on chemical vapor deposition, 1977,” The Electrochemical Society, p.20. [29] T. Suntola, “Atomic layer Epitaxy,” Blackie, Glasgow and London (1990). [30] 郭政肇, “小型質子交換膜燃料電池白金觸媒研究,” 清華大學材料所 論文 (2001). [31] 蔡天翔, “質子交換膜燃料電池白金觸媒—高分散白金奈米粉體於不 同碳支撐材之製備與電化學特性,” 清華大學材料所論文 (2006). [32] S. Litster, G. McLean, J. Power Sources, 130, 61, (2004). [33] K. Kinoshita, “Electrochemical oxygen technology,” Wiley, New York (1992).

70

[34] H.A. Gasteiger, S.S. Kocha, B. Sompalli, F.T. Wagner, Appl. Catal., B, 56, 9 (2005). [35] K. Lee, J. Zhang, H. Wang, D.P. Wilkinson, J. Appl. Electrochem., 36, 507 (2006). [36] EG&G Technical Services, Inc., “Fuel Cell Handbook (7th Edition),” U.S. Department of Energy (2004). [37] T. Tada, “Handbook of Fuel Cells---Fundamentals, Technology and Applications,” Volume 3, John Wiley & Sons (2003). [38] Z. Liu, X. Lin, J.Y. Lee, W. Zhang, M. Han, L. M. Gan, Langmuir, 18, 4054 (2002). [39] M. M. Shaijumon, S Ramaprabhua, N. Rajalakshmi, Appl. Phys. Lett., 88, 253105 (2006). [40] A.L. Dicks, J. Power Sources, 156, 128 (2006). [41] N. R. K. Vilambi Reddy, E. B. Anderson, E. J. Taylor, US Pat. No. 5,084,144 (1992). [42] E. J. Taylor, E. B. Anderson, N. R. K. Vilambi, J. Electrochem. Soc., 139, L45 (1992). [43] M. W. Verbrugge, J. Electrochem. Soc., 141, 46 (1994). [44] G.D. Stäb, P. Urban, US Pat. No. 6,258,239 (2001). [45] H. Kim, N. P. Subramanian, B. N. Popov, J. Power Sources, 138, 14 (2004). [46] K. H. Choi, H. S. Kim, T. H. Lee, J. Power Sources, 75, 230 (1998). [47] N. Cunningham, E. Irissou, M. Lefe`vre, M.-C. Denis, D. Guay, J.-P. Dodelet, Electrochem. Solid-State Lett., 6, A125 (2003). [48] D. Gruber, N. Ponath, J. Muller, F. Lindstaedt, J. Power Sources, 150, 67 (2005). [49] Z. Xue, H. Thridandam, H. D. Kaesz, R. F. Hicks, Chem. Mater., 4, 162 (1992). 71

[50] D. A. Skoog, F. J. Holler, T. A. Nieman, “Principle of Instrumental Analysis,” pp. 255-271. [51] H. A. Gasteiger, J. E. Panels, S. G. Yan, J. Power Sources, 127, 162 (2004). [52] P. Atanassova, Y. Sun, D. Kountz, J. Schwartz, L. Wang, “VII.C.5 Development of High-Performance, Low-Pt Cathodes Containing New Catalysts and Layer Structure,” DOE Hydrogen Program, FY 2005 Progress Report.

72