Lalitha Stanford Presentation Feb2010[1]

30
“Bridging the gap between theory and experiment: which theoretical Leveraging Simulations approaches are best suited to solve real problems in nanotechnology and biology?” Leveraging Simulations to Gain Insights into Polymer Electrolyte Membrane F l C ll Fuel Cells Stanford University Dr. Lalitha Subramanian Sr. Director and Fellow 24 th February, 2010 Accelrys Inc.

Transcript of Lalitha Stanford Presentation Feb2010[1]

Page 1: Lalitha Stanford Presentation Feb2010[1]

“Bridging the gap between theory and experiment: which theoretical

Leveraging Simulations

g g g p y papproaches are best suited to solve real problems in nanotechnology and biology?”

Leveraging Simulations to Gain Insights into

Polymer Electrolyte Membrane F l C llFuel Cells

Stanford University

Dr. Lalitha SubramanianSr. Director and Fellow

24th February, 2010

Accelrys Inc.

Page 2: Lalitha Stanford Presentation Feb2010[1]

Outline

• PEM FC overview

• Rational PEM DesignRational PEM Design– Morphology of perfluorosulfonic acid

(i.e., Nafion®) membranes– Further PEM studies

• Proton transport mechanism• Chemical/mechanical durability• Alternate membrane materials

• Rational Electrocatalyst Design– High Throughput Screening

• Combines both experimental and• Combines both experimental and simulation/modeling insight

© 2008 Accelrys, Inc. 2

Page 3: Lalitha Stanford Presentation Feb2010[1]

Fuel Cell powered cars is a reality –all major manufacturers committed to FC vehicles and develop PEMFC stacks

• HONDA – FCX concept (on lease in the US i 2008)US since 2008)

• Toyota – FCHV (on lease from 2005/2006) and Fine-X concept

• General Motors – Chevrolet Equinox (planned lease from fall 2007)

• Peugeot-Citroen – GENERAC stack, London Taxi conceptLondon Taxi concept

• Nissan – X-Trail FCV• …

© 2008 Accelrys, Inc. 3

Page 4: Lalitha Stanford Presentation Feb2010[1]

Major FC related business/public initiatives

• USA DOE Hydrogen program (www.hydrogen.energy.gov)

• FreedomCAR USA• FreedomCAR, USA (www1.eere.energy.gov/vehiclesandfuels)

• California Fuel Cell Partnership (www.cafcp.org)

• Japan Hydrogen and Fuel Cell Demonstration Project (JHFC) (www.jhfc.jp)

• Clean Energy Partnership (Germany) (CEP) (www.cep-berlin.de)

• Icelandic New Energy (INE) (www.newenergy.is)

• EU Research Framework programsEU Research Framework programs

© 2008 Accelrys, Inc. 4

Page 5: Lalitha Stanford Presentation Feb2010[1]

Fuel Cell TechnologyPresents Challenges

Challenges

• Fuel/Hydrogen Storage

Presents Challenges

• Catalyst optimization

• Electrode reactions

• PEM optimization

• Data Flow Management

© 2008 Accelrys, Inc. 5

Page 6: Lalitha Stanford Presentation Feb2010[1]

PEM Fuel Cells Challenges

• Fuel/Hydrogen Storage− Material selection/optimizationMaterial selection/optimization

• Catalyst optimization− Selectivity, activity, stability− Varied feedstocks, promotersVaried feedstocks, promoters

• Electrode reactions− Hydrogen Evolution/Oxygen

Reduction ReactionReduction Reaction− Degradation (dissolution,

oxidation, poisoning)

• PEM optimization

d

yelectricitOHHO 222 22

PEM optimization− Microstructure, hydration− Proton transport properties− Mechanical/chemical durability

© 2008 Accelrys, Inc. 6

Anode:

Cathode:

eHH 442 2

OHeHO 22 244 • Data Flow Management

– Stack Assembly Engineering

Page 7: Lalitha Stanford Presentation Feb2010[1]

Rational Proton Exchange Membrane Design

© 2008 Accelrys, Inc. 7

Page 8: Lalitha Stanford Presentation Feb2010[1]

Alternate Energy Industry Requirements

• There is a pressing requirement to develop polymer electrolyte membranes (PEM)– conduct protons at low levels of

hydration, – do not degrade upon prolonged

operation at elevated temperature, – and offer selective ionic and

molecular transport. • To optimize the chemistry of membranes

f t t t i f d t lfor proton transport requires fundamental understanding of – proton transport,

Rational design of the next generation of– mechanical properties– chemical degradation

Rational design of the next generation of polymer membranes is needed

© 2008 Accelrys, Inc. 8

Page 9: Lalitha Stanford Presentation Feb2010[1]

Challenges in PEM Design

• Cost– The cost of fuel cell power systems must be reduced before they can be

competitive with conventional technologiescompetitive with conventional technologies

• Durability and Reliability– Match durability and reliability of current automotive engines [i.e., 5,000-

hour lifespan (150 000 miles)] and the ability to function over the fullhour lifespan (150,000 miles)] and the ability to function over the full range of vehicle operating conditions (40°C to 80°C). For stationary applications, more than 40,000 hours of reliable operation in a temperature at -35°C to 40°C will be required for market acceptancetemperature at -35 C to 40 C will be required for market acceptance

• System Size– The size and weight of current fuel cell systems must be further reduced

t t th k i i t f t bilto meet the packaging requirements for automobiles

• Air, Thermal, and Water Management

• Improved Heat Recovery Systems

© 2008 Accelrys, Inc. 9

• Improved Heat Recovery Systems

Page 10: Lalitha Stanford Presentation Feb2010[1]

Wish list for PEM Material

• A good performance at a temperature of 120 ºC without the need to pressurize, i.e(RH) ≤ 40%. At this temperature, about 50 ( ) pppm CO can be tolerated without air bleed

• Conductivity σ = 0.1 -1 cm-1

• Hydrogen oxygen gas permeability <• Hydrogen-oxygen gas permeability < 1x10−12 (mol cm)/(cm2 s kPa))

• Limited swelling in water

• Mechanical properties better than Nafion®

• A chemical stability similar or superior to Nafion, i.e., a durability of around 40,000 h (≤ 1 μV/h)

• A cost target of ≤ $10/kW at 500,000 stacks/y (for automotive application)

© 2008 Accelrys, Inc. 10

• …

Page 11: Lalitha Stanford Presentation Feb2010[1]

Membrane Reliability - A Multiscale Problem

• The task is challenging because

th i t f th b i l– the environment of the membrane is complex

– the pore network morphology is dynamic p p gy y

– and the membrane dynamics takes place on much longer scales compared to proton transferlonger scales compared to proton transfer

© 2008 Accelrys, Inc. 11

Page 12: Lalitha Stanford Presentation Feb2010[1]

Molecular & Mesoscale Simulation

• Two major areas of fundamental* investigation

– Characterizing the chemical features that affect performance– the chemical nature of protonation sites– local concentration of protons – and local level of hydration

– Characterizing the underlying polymer morphologyU d t di t di t ib ti l ti h d ti it– Understanding water distribution, percolation, hence conductivity

– Polymer structure-hydrated morphology relationships

* Excluding transport, CFD and FEM type models

© 2008 Accelrys, Inc. 12

Page 13: Lalitha Stanford Presentation Feb2010[1]

Morphology in hydrated perfluorosulfonic acid membranes

• Morphology of Nafion at the nanoscale?

• SAXS SANS:• SAXS, SANS: – Nanophase segregation into hydrophilic and hydrophobic domains, – Debate over the shape and structure of the ionic clusters: spherical, ellipsoid, or

lamellar?

• Observations of the surface morphology via TEM and AFM– Three-phase model consisting of spherical water clusters surrounded by sulfonic

acid interfaces.– Also observed the coalescence and growth of ionic clusters with an increasingAlso observed the coalescence and growth of ionic clusters with an increasing

water content using AFM.

• Use mesoscale modeling to compare and contrast with exp. Observations

Wescott, Qi, Subramanian and Capehart, J. Chem. Phys. 124, 134702 (2006) collaboration between Accelrys and General Motors

© 2008 Accelrys, Inc. 13

(2006) – collaboration between Accelrys and General Motors

Page 14: Lalitha Stanford Presentation Feb2010[1]

Models of hydrated perfluorosulfonic acid membranes

Gierke’s et al Cluster – Network Model Yeo and Eisenberg’s Model

Yeager and Steck’s Model Starkweather bilayer-> Litt model ->Haubold model

© 2008 Accelrys, Inc. 14

Page 15: Lalitha Stanford Presentation Feb2010[1]

Multiscale Approach

TIMEwater3.5 nm

Diffraction TEM, AFM

ConductivityModulus

Experiment

h

FINITEmin

ANALYTICAL MODELS

PERCOLATION THEORY

water

5 nm

s

s

MESODYN(field method)

ELEMENTANALYSIS

•The task is challenging because

ps

ns MOLECULARDYNAMICS

F= M A

(field method) Statistics

Flory-Huggins Modeling

–the environment of the membrane is complex

–the pore network morphology is dynamic

fsQUANTUM

MECHANICSH=E

ELECTRONS => ATOMS => BEADS => GRIDS => PARAMETERSAtomic potential

y ggParameters –and the membrane dynamics

takes place on much longer scales compared to proton transfer

© 2008 Accelrys, Inc. 15DISTANCE

1 A 1 nm 100nm micron mm

ELECTRONS > ATOMS > BEADS > GRIDS > PARAMETERS

Page 16: Lalitha Stanford Presentation Feb2010[1]

Increasing Length and Time Scales

Atomistic Mesoscopic

Length nm 100’s of nm (or more)

Units atoms Beads representing many atomsg y

Time ns as much as milli-seconds

Dynamics F=ma Diffusion, hydrodynamics

© 2008 Accelrys, Inc. 16

Page 17: Lalitha Stanford Presentation Feb2010[1]

Coarse graining strategy

CF2 CF2 CF CF2x y Hydrophobic fluorocarbon backbone

O CF2 CF O CF2 CF2 SO3

CF3

zM Nafion 117 (EW=1100)

y=1, z=1, x=7

Hydrophilic sidechain with Sulfonic groups

SF

W234 atoms ~ 3 beads

S: Side chain ~ 306Å3

F: 4 -[CF2-CF2]- monomers ~ 325Å3

W: 10 water molecules ~ 315Å3F, S and W Beads

© 2008 Accelrys, Inc. 17

W: 10 water molecules 315Å

Page 18: Lalitha Stanford Presentation Feb2010[1]

Mesoscale Parameterization

Interaction Flory-Huggins

mixing parameterbetween beads(Mesodyn Input)

: solubility parameters

for each bead(MD)

energybetween F, S

and W(Mesodyn)( y p )

RTV JIref

JI

2)(

( )

VEcohI /

( y )

/0 vIJIJ

=9 8 =15 7 =0 7FS=9.8, Fw=15.7, WS=0.7

F beadsat experimental

density of Nafion

S beadsat experimental

density of N fi 117

W beadsat experimental density of water

© 2008 Accelrys, Inc. 18

density of Nafion 117, 2.05g/cm3

F = 13.9 (MPa)½

Nafion 117, 2.05g/cm3

S = 21.1 (MPa)½

density of water 1g/cm3

W = 47.9(Mpa)½

Page 19: Lalitha Stanford Presentation Feb2010[1]

Three-phase Morphology

30nm

H2OYeager and Steck’s Model

SA

: water / sulfonic group

FC

Water clusters (~4nm) surrounded by sulfonic phase

Embedded in a hydrophobic PTFE matrix

with =8, or 20% water: water / sulfonic group

Embedded in a hydrophobic PTFE matrix Consistent with Yeager-Steck[1] three-phase model and Xue’s observation[2]

Order Parameter: (metric for degreeof phase separation)

V III drr

VP 22 )(1

• Water cluster/fluorocarbon degree of

© 2008 Accelrys, Inc. 19

[1] J. Electrochem. Soc. 128, 1880 (1981)[2] J. Membr. Sci. 45, 261 (1989)

• Water cluster/fluorocarbon degree ofPhase separation increases with increasing water volume fraction

Page 20: Lalitha Stanford Presentation Feb2010[1]

Three-phase Morphology at =8

W30nm

W F SWater clusters (~4nm) surrounded by sulfonic phase embedded in a hydrophobic PTFE matrixC i t t ith Y St k[1] thConsistent with Yeager-Steck[1] three

phase model and Xue’s observation[2]

[1] J. Membr. Sci. 45, 261 (1989)

© 2008 Accelrys, Inc. 20

[1] J. Membr. Sci. 45, 261 (1989)[2] J. Elctrochem. Soc. 128, 1880 (1981)

Page 21: Lalitha Stanford Presentation Feb2010[1]

Percolation of water domains/ Percolation for conductivity

%62 %208

• Simulated morphology consistent with the structural i f ti i f d f ll

Volume fraction of water

%6,2 %20,8information inferred from small-angle scattering• Simulated morphology at low gywater content produces spherical hydrophilic domains of reverse micelles - similar to

%11,4 %33,16of reverse micelles similar to model of Gierke• Simulated morphology at hi h t t thigher water content –domains deform into elliptical and barbell shapes – similar to

© 2008 Accelrys, Inc. 21

three-phase model of Yeager and Steck

Page 22: Lalitha Stanford Presentation Feb2010[1]

Compare with Experimental Diffraction Data

SANSSIMULATEDScattering Curves at Different Water Content

=8

ensi

ty

=2

Ionomer k

1

2

SANS

I2

0.01 0.10.01 0.1 Q (A-1)Q (A-1)

Inte peak

0.05 0.1 0.2 0.5

0.2

0.5I

=16

tens

ity

=4

1.5

2

3

I6

0.01 0.10.01 0.1Q (A-1)

Int

Q (A-1)0.05 0.1 0.2 0.5

1

© 2008 Accelrys, Inc. 22

Ionomer peak: associated with hydrophilic domains

2

,( ) ( ) exp( ) exp( )i j i j

i jI q S q F F iq r iq r

Page 23: Lalitha Stanford Presentation Feb2010[1]

Rational Electrocatalyst Design

© 2008 Accelrys, Inc. 23

Page 24: Lalitha Stanford Presentation Feb2010[1]

Overview

• Oxygen reduction reaction (ORR) is acritical performance limiting step inProton Exchange Membrane FuelCells (PEMFC).

• ORR is catalyzed by the cathodewhich must satisfy the followingy grequirements:– Fast ORR kinetics– Stability against oxidation and

contamination– Compatibility with other PEMFC

componentsL t i l d f t t– Low materials and manufacture cost

OHeHO 22 244

© 2008 Accelrys, Inc. 24

Page 25: Lalitha Stanford Presentation Feb2010[1]

Structure and Composition

Bulk and surface defects Alloying Clustering

Defect decoration Surface segregation Skin formation

© 2008 Accelrys, Inc. 25

Page 26: Lalitha Stanford Presentation Feb2010[1]

Cathode material optimization:

• Increasing number of skin alloys and core/shell nanoparticles are being

Key property:

Surface stoichiometry ≠ Nominal stoichiometryreported to have ORR activity superior to that of pure Pt:

PtNi PtCo PtY PtPd

Understanding surface phase diagram under relevant conditions is critical

• PtNi, PtCo, PtY, PtPd• PtAu• Pt double layers• …

• Durability of these systems is a main challenge:• Re-alloyingRe alloying• Leaching and dissolution• Coalescence • Detachment from support

© 2008 Accelrys, Inc. 26

Shuo Chen et al Am. Chem. Soc./ 2008, 130, 13818

Page 27: Lalitha Stanford Presentation Feb2010[1]

Adsorption and activation energies: ORR

EE

E0=E(O2+*)

ETS=E(O*-O*)

Reaction coordinate

E1=E(O2*)E2=2E(O*)

E1=E(O2*)

Reaction coordinate

Ediss=E2-E1 Eads1=E1-E0Ea=ETS-E1 Eads2=E2-E0E =E E

© 2008 Accelrys, Inc. 27

More reaction steps need to be added for electro-reduction

Eads1=E1-E0

Page 28: Lalitha Stanford Presentation Feb2010[1]

Summary

• Ab initio High Throughput Approach offers valuable insight into factors defining catalytic activity of materials.

I t tl it ll t dd i lt l th bl f f d• Importantly it allows to address simultaneously the problems of surface and chemical reactivity.

• Our approach streamlines calculations of descriptors such as d-band centre position, atomic fraction of solute atoms near the surface and electron workposition, atomic fraction of solute atoms near the surface and electron work function.

• This opens the possibility of in silico cathode material optimisation complimentary to the experiment.

© 2008 Accelrys, Inc. 28

Page 29: Lalitha Stanford Presentation Feb2010[1]

Acknowledgements

• Dr. James Wescott

• Dr. Patricia Gestoso-SoutoDr. Patricia Gestoso Souto

• Dr. Jacob Gavartin

© 2008 Accelrys, Inc. 29

Page 30: Lalitha Stanford Presentation Feb2010[1]

Thank You

For more information contact …

Dr. Lalitha Subramanian SSr. Director and FellowAccelrys, [email protected](858)799-5340(858)799 5340

© 2008 Accelrys, Inc. 30