2016-05-21

Institute of New Energy (INE)

() China University of Petroleum, Beijing (CUPB)

2016

NRC

1.

Ballard

2.

Toyota

3.

Nissan

2016

NRC

1.

Ballard

2.

Toyota

3.

Nissan

2016

PEM properties

Dself x109 m2/s =5 =10 =15 =22

H2 2.57 2.79 3.80 5.51

O2 0.15 0.40 0.41 0.71

Pore volume

=5 0.24 ml/g

=10 0.29 ml/g

=15 0.37 ml/g

=22 0.48 ml/g

Equilibrated Nafion membrane

Water channel

Gas diffusivity in PEM

2016

Modeling of Nafion membrane

2016

Microstructure of Nafion membrane

2016

Degradation reactions of PEM

Reaction constants

Simulation of PEM degradation reactions using Kinetic Monte Carlo

Degradation reactions of Nafion:

-CF2CF2-COOH + OH -> -CF2COOH + CO2 + HF (main chain unzipping)

-CF2CF2SO3H + OH-> -CF2COOH + CO2 + HF (side chain scission)

-CF2CF2SO3H + OH-> -CF2COOH + CO2 + HF (ether cleavage)

Working scheme of modeling

Experimental validation: Gravimetry, HPLC-MS, Ion

chromatography, FTIR & 19F NMR

Model PEM degradation in atomic details

2016

Kinetic Monte Carlo (KMC) method:

KMC algorithm is used to simulate degradation processes.

The algorithm follows:

1. Set the time t at t = 0

2. Construct a list of all possible events. Suppose that there are N events in total with the rate constants r1, , rN in units of events per unit of time. It is assumed that these events are independent.

3. Select one of the events k with a probability pk = rk/ri4. Execute this event and advance the time for the next event (k+1) by

tk+1 = tk+ |ln(f)|/ ri, where f is an uniformly distributed random number between 0 and 1.

5. Repeat this process, starting from step 2.

Important factors: Reaction definition & Reaction rates

Modeling of PEM chemical degradation

2016

Degradation of Nafion Main chain unzipping

Polymer Degradation and Stability 94 (2009) 14361447

Reaction #1

-CF2CF2-COOH + OH -> -CF2COOH + CO2 + HF

Reaction rate r1 = k1[-COOH][OH] (k1 = 1)

Reaction #2

-CF2CFO-COOH + OH -> -CF2COOH + CO2 + HF

Reaction rate r2 = k2[-COOH][OH] (k2 = 0.01)

Obtaining reaction rates:

1. DFT estimation of reaction enthalpy

2. Fitting relevant experimental data

Need to predefine

Carboxylation rate

Need to predefine

Carboxylation rate

2016

Degradation of Nafion Side chain scission

ECS Transactions, 16 (2) 235-255 (2008)

Reaction #3

-CF2CF2SO3H + OH-> -CF2COOH + CO2 + HF

Reaction rate r3 = k3[-SO3H][OH] (k3 = 0.01)

Mostly take place in Fentons test of gas phase (dry conditions)

Obtaining reaction rates:

1. DFT estimation of reaction enthalpy

2. Fitting relevant experimental data

Need to predefine -SO3-

deprotonation rate

2016

Degradation of Nafion ether cleavage

Cause backbone & side chain scission

Reaction #4

-CF2CF2SO3H + OH-> -CF2COOH + CO2 + HF

Reaction rate r4 = k4[-CFO-][OH] (k4 = 0.002)

Macromolecules 2007, 40, 8695-8707

2016

Snapshot of degraded Nafion

Main chain unzipping Side chain scission

2016

Simulation validation - Ion chromatography

Simulation results

Experimental results

Validate exhaust components by Ion chromatography

2016

Simulation validation - HPLC-MS

Validate side-chain products by

HPLC-MS

CF3-COOH SO3H-CF2-COOH

Nafion side chain

2016

Simulation validation - FTIR

Validate backbone composition by FTIR

2016

Simulation validation - 19F NMR

Validate Nafion composition by 19F NMR

2016

NRC

1.

Ballard

2.

Toyota

3.

Nissan

2016

Development of MDM and MEAPLS

MEA Model(global chemical

degradation model

from T5 using

Kinetic model from

T1)

Sorption Model

(nature of water

domains T1)

Hierarchical

Fiber Bundle

Model (micro-crack

initialization from

T1)

Finite element fracture

model

(a new fracture

mechanics model

combined with standard

FEM model and FEM

Fatigue model from T2)

Molecular simulation

(mechanical properties of fibers and fundamental

degradation mechanisms from

T3)

MDM

MEAPLS

FEM

(in situ stress

from T2)

critical crack sizecrack density

2016

Molecular simulation of mechanical degradation

Molecular simulation:

1. Model Nafion membrane (20-50 nm)

2. Extract mechanical property of Nafion fiber unit (2-5nm)- Mechanical degradation

3. Model chemical degradation process of Nafion membrane - Chemical degradation

4. Calculate mechanical property of Nafion fiber unit upon chemical degradation Coupling Chemical/mechanical degradation

5. Explore structural change of Nafion under chemical and mechanical degradation

Membrane

degradation

Mechanical stress

Chemical dissolution

RH cycling

Structural

analysis

+

+

Gas

permeation

2016

Analysis of PEM mechanical failure:

Crazes in polymers are regions of highly localized deformation, which require the presence of hydrostatic tension.

Crazes nucleate in regions with stress concentrations such as crack-tips, pinholes, and surface defects, is sufficiently large.

Two categories of craze growth : (1) craze tip advance, or crack formation, and (2) craze widening, which corresponds to fibril formation and deformation. As the fibrils deform and elongate in the latter case, crazes grow in width and eventually the fibrils break down, resulting in separation of the material behind the crack-tip, similar to crack propagation.

Craze growth can be considered as a precursor to crack propagation.

Fibril elongation leads to localized plastic deformations and slowing the crack propagation by bridging the crack plane. When the fibrils are strong enough to hold the crack, craze formation occurs

Modeling Fatigue Failure of PEM

2016

Calculating PEM mechanical properties

Mech Time-Depend Mater (2008) 12: 205220

The stress-strain behavior of cross-linked ionic polymeric networks was investigated using molecular dynamics simulations.

2016

Reference of micro-crack simulation in polymer membrane

Crazing of entangled polymer chains

Growth, microstructure, and failure of crazes in glassy polymers, J. Rottler and M. O. Robbins, Phys Rev E, 68, 011801 (2003).Jamming under tension in polymer crazes, J. Rottler and M. O. Robbins, Phys Rev Lett, 89, 195501 (2002).Cracks and crazes: On calculating the macroscopic fracture energy of glassy polymers from molecular simulations, J. Rottler, S. Barsky, M. O. Robbins, Phys Rev Lett, 89, 148304 (2002)

Tensile pull on adhesive polymer chains

Large-scale simulation of adhesion dynamics for end-grafted polymers, S. W. Sides, G. S. Grest, M. J. Stevens, Macromolecules, 35, 566-573 (2002)Effect of end-tethered polymers on surface adhesion of glassy polymers, S. W. Sides, G. S. Grest, M. J. Stevens, S. J. Plimpton, Journal of Polymer Science, Part B (Polymer Physics), 42, 199-208 (2004)

2016

Computation procedures

Parallel working scheme

System Ionomers Dimension cluster Required

cores

Running

time

Purpose

Small 45 10 nm lattice 32 5.5 h/ns Testing

Large 1600 30 nm parallel 240 21.9h/ns Formal

Simulation steps:

1. Randomize initial membrane

2. pre-stabilize membrane

3. Stabilize membrane

4. Anneal membrane

5. Equilibrate membrane

1 2

3

45 2016

Structure of small Nafion system

PAGE 24

Initial Nafion membrane with

=6

Dimension = 10 nm

Water domain

Nafion framework

2016

J. Polym. Res. 2006, 13, 379.

TEM Analysis of PEM

H+ sites are replaced by metal ions to

increase the image contrast.

Dark regions are microscopic water

cavities in Nafion membrane.

Gray regions are polymer backbone.

Hongwei Zhang et al., Int. J. Hyd. Energy, 37 (2012) 4657-4664

J.Memb. Sci. 356 (2010) 4451

2016

6 8 10 12 14 16 18

Dimension 10.10 10.28 10.46 10.63 10.79 10.95 11.10

Swelling - 1.8% 3.6% 5.2% 6.8% 8.4% 10%

=8

=16

Observation:

Reorganization of Nafion

framework caused by water

uptake

Effect of RH cycling

2016

Morphology change under RH cycling

=6 (initial) =12 (equilibrated) =6 (equilibrated)

Dimension [nm] =6 =12 =6

Cycle 1 10.07 10.63 10.17

Cycle 2 10.17 10.63 10.21

Cycle 3 10.21 10.62 10.25

2016

Simulation of stress-induced mechanical degradation

A structure-based modeling approach:

1. Resemble the working scheme of fiber bundle model

2. Crucial to develop breakable bond of ionomer bundles

3. Capable of describe the fracture process

(microvoid initialization craze elongation fibril breakage crack propagation)

4. TEM images of crack morphology of membrane samples 2016

Scheme of fracture process

Microvoidformation

Craze formation

Craze elongation

2016

Scheme of fracture process

Fibril recombination

Fibril breakage

2016

TEM images of Crack Morphologies

(50nm)

2016

Molecular simulation of chemical degradation

25%50%75%

Weight loss

2016

Molecular simulation of membrane strain

Begin(50% weight loss)

End (strain 200%)

2016

NRC

1.

Ballard

2.

Toyota

3.

Nissan

2016

Building unit of carbon black

The energetically the most stable finite two-dimensional structure

Hexagonal graphite sheets

Structural property

Core-shell structure

Shell unit d = 3.7 nm

Core unit d = 0.7 nm

Carbon density = 2 g/cm

Particle size = 20 nm

2016

Morphology of carbon black

Slice view Surface view

2016

Porosity of carbon black

Slice view of porous carbon (dmean=2 nm ) Slice view of porous carbon (dmean=4 nm )

2016

Porosity of carbon black

Reproduce of targeted PSD in carbon particles

2016

Example of carbon degradation

Simulation of degradation of carbon black primary particles (20 nm) in catalyst layer of fuel cell

50%

weight loss

Experimental validation

2016

Background

Synthesis:

1. Pyrolysis in NH3

Create micropores

Dope Nitrogen at carbon surface

2. Catalytic sites hosted only in micropores (

NRC

1.

Ballard

2.

Toyota

3.

Nissan

2016

Research strategy

Non-noble metal catalyst

Carbon support Nitrogen source Metal precursor

Mainly FeMolecules can be as simple as NH3

Nanoparticles with complex structure, composition, chemical and physical properties

The most important factor for the synthesis of non-noble metal catalysts (needs extensive study)

Desire to keep them standard and simple

2016

Carbon models

Graphitic

building

unit

d=2.7 nm d=3.2 nm d=3.7 nm

Graphitic

levelCarbon 44 (2006) 753761

2016

Formation of active site

Non-porous carbon

Active site formation in created micropore of carbon,

particularly between graphitic gaps, during pyrolysis

Need to be studied

2016

Oxidation of carbon

Original

carbon

Oxidized

carbon (50%

weight loss)

Carbon Vol. 36, No. 4, pp. 433 441, 1998

modelingexperiment

Formation of

micropores

2016

Porosity of oxidized carbon

Carbon Vol. 36, No. 4, 433-441, 1998

Growth of porosity in oxidized carbon

Applied materials & interfaces

VOL. 1, NO. 8, 16231639, 2009

Modeling

2016

Scale up active sites to

molecular level (nm)

Structure of active site

-3.901 eV

One possible structure of

Fe-N active sites from DFTPotential analysis in molecular simulation

2016

Modeling of active site formation

Initial state Equilibrated state

Observation:

1. Two types of active sites formed in carbon pores

2. Type I active sites is slightly less than Type II at Fe loading of 1 wt%

Fe loading 1 wt%

Graphene layers

Fe atoms (green bead)

Type I active site (-3.9 eV)

Type II active site (-1.9 eV)

2016

NRC

1.

Ballard

2.

Toyota

3.

Nissan

2016

Example of Pt degradation

Simulation of Carbon black supported Pt

catalyst in fuel cell

Pt degradation processes for

modeling

2016

Pt degradation: migration and coalescence

This process involves motion of Pt particles and coalescence where they meet on the carbon support

A sequence of STEM HAADF images showing the coalescence of Pt nanoparticles after (a) 45 s, (b) 115 s and (c) 165 s with the beam left on in between images.

This phenomenon is observed in simulation, but with a less pronounced effect

2016

Pt degradation: molecular snapshots

Pt on carbon black (Pt/C=1) at 300 K

Original Pt particles (1~3 nm)

Pt coalescence

2016

NRC

1.

Ballard

2.

Toyota

3.

Nissan

2016

Challenges of FCVs for commercializationCost reduction, High power density, Durability, Sub-zero startup

Catalyst Layers (CLs) : Key componentReduction of Pt as a result of achieving high Pt utilization

Motivation

Scale / mm10-3 10-2 10-1 100 101

FabricationMaterial Structure Properties

Performance(Pt Utilization)

CL design for high Pt utilization

Size complexity in CLs

10-4

2016

0.0001

0.001

0.01

0.1

0 20 40 60 80 100 120

Relative Humidity / %

Ion

om

er

Con

cuctivity /

Scm

-1

0.0001

0.001

0.01

0.1

0 20 40 60 80 100 120

Relative Humidity / %

Ion

om

er

Co

ncu

ctivity /

Scm

-1

Ref: Iden et al., J. Electrocem. Soc., 156, B1078 (2009)

Motivation

Graphitized Ketjen

Black (without Pt)

Graphitized Ketjen

Black (without Pt)

Graphitized Ketjen

Black (with Pt)

Ketjen Black

(without Pt)

Ketjen Black and Graphitized Ketjen Black have similar macro-scopic structure, but different micro-scopic structure and surface properties.

How carbon supports and Pt particles affect structures and transport properties???? ???

2016

Effect of Water Contents (Bare GC)A

tom

Num

ber

1000

040

0

50

0

400

0

F

S

O (water)

O(H3O

+)

Z Direction Distance (Perpendicular to Carbon Sheet) / nm

Graphite sheet

0 1 2 3 4-1

510

2215

l = 22 l = 5

Z

Y

X

Y

(A)

(B)

A) Water adsorption on the surface of Nafion , not in the vicinity of GC.

B) Nafion thickness is independent of water contents. (No swelling)

Water Back bone Side chain GC

Water content: l

2016

5.6 nm

1.8 nm

Morphology of Nafion and Water

l = 22 (High water content)Thickness in x-direction: 2 [nm]

Y

Z

Nafion is attached to the GC via back bones. Sulfonic acid groups are mainly oriented away from the GC. Surface of Nafion film is hydrophilic.

Water Hydronium ion Back boneSide chain GC

Sulfonic acid group

2016

GC w/ 6 Pt, 12 Oligomers

Water occupied free carbon surface with increasing water contents in the case with Pt clusters.

Nafion and water coverage was independent of water contents in the case without Pt clusters.

GC w/o Pt, 12 Oligomers

Nafion & Water Coverage on Carbon

2016

Water H3O+ Back bone Side chain PtGC w/ 6 Pt clusters, 12 Oligomers

(a)(b)

(c)(d)

(e)

(f)

(a) (b)

(c) (d) (e) (f)

Area covered with water or

hydronium ions

Nafion adsorption on Pt clusters Small water coverage

Morphology of Nafion on Pt

2016

Nafion and Water Coverage on Pt

Water H3O+ Back

bone Side chain Pt

22 15 10 5

Water contents l

High Nafion coverage. Active sites slightly increase with water contents.

2016

A) Primary adsorption on sulfonic acid groups and Pt surface

B) Secondary adsorption and capillary condensation on Nafion

(A)(B)

Water H3O+ Back bone Side chain Pt

Water Adsorption on GC with Pt

l = 22 (High water content) l = 5 (Low water content)

2016

Morphology of water Adsorbed on the surface of Nafion film Affected by carbon support properties Affected by Pt

Morphology of Nafion ionomer Thin film (different from bulk

membrane) No swelling No penetration into primary pores Affected by Pt

Carbon

Ionomer(back bone)Ionomer(Side chain)

Water Secondary pore

Primary pore

H+

Transport of H+

Water clusters adsorbed on the surface of Nafion film mainly contribute to H+

transport.

How Relevant to CLs?

2016

0.0001

0.001

0.01

0.1

0 20 40 60 80 100 120

Relative Humidity / %

Ion

om

er

Con

cuctivity /

Scm

-1

0.0001

0.001

0.01

0.1

0 20 40 60 80 100 120

Relative Humidity / %

Ion

om

er

Co

ncu

ctivity /

Scm

-1

Answer to the First Question

Graphitized Ketjen

Black (without Pt)

Graphitized Ketjen

Black (without Pt)

Graphitized Ketjen

Black (with Pt)

Ketjen Black

(without Pt)

Different carbon support

Water morphology

Proton conductivity

Presence of Pt

Ionomer morphology

Water morphology

Proton conductivity

2016

NRC

1.

Ballard

2.

Toyota

3.

Nissan

2016

CAD+FEM

2016

2016