In-situ XAFS studies of fuel cell catalysts · Develop methods for characterization of catalysts in...

October 20, 2006 - NSLS

In-situ XAFS studies of fuel cell catalysts

Carlo U. Segre

Center for Synchrotron Radiation Research & Instrumentation andDepartment of Biological, Chemical & Physical Sciences,

Illinois Institute of Technology

Workshop on XAFS studies of nanoparticles and chemical transformationsOctober 19-21, 2006 - NSLS


Acknowledgements

Graduate studentsStanislav Stoupin - IIT PhysicsEun-Hyuk Chung - IIT ChEHarry Rodriguez – UPRRobert Richard Diaz – UPRRamesh Viswananthan - Intel

CollaboratorsEugene Smotkin - NEUSoma Chattopadhyay - IIT

FundingArmy Research Office – long term support of fuel cell program

MRCAT is supported by contributions from MRCAT member institutions.

The APS is funded by the U. S. Department of Energy, Office of Basic Energy Sciences under Contract number W-31-109-Eng-38.


Outline

• Brief introduction to fuel cells• Direct Methanol Fuel Cells

– Anode mechanisms → mixed metal catalysts• X-ray spectroscopy primer• EXAFS and XANES of operating liquid feed DMFC

– Experimental challenges– Selection of potential window for experiments– Analysis of XANES and EXAFS– Lattice parameter analysis of catalysts

• Conclusions


Types of Fuel Cells

Alkali Fuel Cell (AFC)@60oC

Phosphoric Acid Fuel Cell (PAFC)

Molten Carbonate Fuel Cell (MCFC)

@600ºC

Solid Oxide Fuel Cell (SOFC)

@1000ºC

Hydrogen/air Fuel Cell

(PEMFC) @60-100ºC

Liquids in a matrix Polymer electrolytes Ceramics and molten salts

Direct MethanolFuel Cell


Ano

de C

atal

yst

Cat

hode

Cat

alys

t

Nafion

Load

Voltmeter6e- 6e-

6H+

Methanolcrossover

CO2 + H2O+MeOH

Graphiteflow field

Carbon PaperDiffusion Layer

Carbon ClothGas Diffusion

Layer

Air

AirCompressor

ISCO Pump

Pure MeOH

Gear Pump

DI Water

H2O, N2, O2CO2 from crossover

DMFC Schematics

Anode : CH3OH + H2O → 6H+ + CO2 + 6e- Eºanode = 0.016VCathode : 1.5O2 + 6H+ + 6e- → 3H2O Eºcathode = 1.23VOverall reaction : CH3OH + 3/2O2 → CO2 + 2H2O Eºcell = 1.214V


Fuel Cell Voltage/Current Characteristics

0.5

1.0Region of Activation Polarization (Reaction Rate Loss)

Theoretical EMF or Ideal Voltage (OCV)

Region of Ohmic Polarization(Resistance Loss)

Region of Concentration Polarization(Gas Transport Loss)

Total LossOperation Voltage Curve

Current Density (mA/cm2)0

Cell

Volta

ge

Ref. Fuel cell Handbook (6th Ed.) by EG&G Technical Services, Inc


Major DMFC issues

• Electrocatalyst– Anode

- Methanol oxidation rate using Pt alloys is sluggish - CO tolerance of Pt alloys is poor

– Cathode- Oxygen reduction kinetics are sluggish (even with H2 fuel cells)- Cathode performance degraded by methanol crossover

• Membrane electrolyte– Methanol is permeable through membrane


Motivation for synchrotron studies

Develop methods for characterization of catalysts in fully operating fuel cells

• Nanoparticle structure during operation• Surface chemistry• Conditioning effects• Degradation mechanisms

Initial experiments focus on anode• Supported catalyst in hydrogen/air fuel cell (2001)• Unsupported Pt/Ru catalyst in DMFC


20 40 60 80 100 120

Pt50Ru50

2 Θ

Pt44Ru41Os10Ir5

Pt80Ru20

Inte

nsity

(a.

u.)

Pt65Ru25Os10

Pt80Ru10Os10

Pt80Os20

Pt

What do we know about the catalyst?

• Arc-melted alloys

• Nanoparticle catalysts

All catalysts are fcc


Lattice parameters of nanoparticle catalysts

• Total pattern fitting

• Internal Si standard

• Accurate lattice parameters

• Particle size broadening

a = 3.883 Å


fcc lattice spacings of catalysts and arc-melted alloys

FCC lattice parameter comparisons

Borohydridereduction

Arc melted


Pt/Ru Catalyst: Bifunctional Mechanism

Methanol AdsorptionPt + CH3OH → Pt-(CH3OH)ads

C-H bond ActivationPt-(CH3OH)ads → Pt-(CH3O)ads + H+ + e-

Pt-(CH3O)ads → Pt-(CH2O)ads + H+ + e-

Pt-(CH2O)ads → Pt-(CHO)ads + H+ + e-

Pt-(CHO)ads → Pt-(CO)ads + H+ + e-

Water AdsorptionRu + H2O → Ru-OH +H+ + e-

CO OxidationPt-(CO)ads + Ru-OH → Pt + Ru + CO2 + H+ + e-

OverallCH3OH + H2O → CO2 + 6H+ + 6e-


The X-ray absorption experiment

Double crystalmonochromator

PolychromaticX-Rays

MonochromaticX-Rays

Incident FluxMonitor

Transmitted FluxMonitor

Standard fuel cell

Io = Incident FluxI = Transmitted Fluxx = Sample Thicknessµ(E) = Absorption Coefficient at photon energy E


Nafion 117membrane

Graphiteblock

Gasket CurrentCollector

End Plate

X-Rays

Anode Catalyst Pt-Ru Cathode Catalyst Pd/C

ELAT GDL Flow Channel

Anode Gas Inlet

Cathode GasInlet

Cathode GasOutlet

Anode GasOutlet

R. Viswanathan et al., J. Phys. Chem. B 106, 3458 (2002).



Planning DMFC synchrotron experiments

• DMFC spectroscopy challenges– Density fluctuations

• CO2 bubbles at the anode• Flooding at the cathode

– Argonne safety requirements• Solutions

– 35oC cell temperature– Slight backpressure at the anode– 4% H2 balanced N2 at the cathode


Fuel cell

A C

4%H2

MFCPump Anode Cathode

Potentiostat- + PC

BPR

Fuelexhaust

Fuelreservoir

CathodeExhaust

BPR : Back Pressure Regulator MFC : Mass Flow Controller

DMFC x-ray setup


Experimental conditions

• DMFC– Anode: PtRu (1:1)– Cathode: Pd/C (30wt%)– Cell temperature: 35ºC– Fuel composition: (1) H2O, (2) 0.1M (3) 2M MeOH– Cathode: 4% H2 balanced N2– Potential vs DHS: 250mV, 300mV, 350mV, 400mV, 450mV

• In-situ x-ray absorption– XANES and EXAFS data: Separately taken at Ru K and Pt L3 edges. – Absorption edge jumps: ∆µx = 0.05 for Ru and ∆µx = 0.17 for Pt.– References: Pt foil, Ru metal, RuO2, RuO2-hydrate, as received PtRu– Monochromator: Double crystal Si (111) – Harmonic Rejection Mirror: Pt for Ru edge, Rh for Pt– Ion chamber detector gases: Incident beam; 80% He- 20% N2:

Transmission; pure N2.


What is the interesting potential range?

0 5 10 15 20 25 30 350.0

0.2

0.4

0.6

0.8 Full cell performance curve Anode polarization curve Cathode polarization curve

Pote

ntia

l (V)

Current density (mA/cm2)

35ºC, 0.1M MeOH, 1mL/min

450mV

250mV


-0.66

-0.64

-0.62

-0.6

-0.58

-0.56

21750 22250 22750 23250Energy, eV

Abs

orpt

ion

Experimental approach

• X-ray transmission experiments conducted in continuous scan mode, minimizing absorption transition effects.

• Enables > 85% use of data for averaging.

• Example of transitions in absorption due to density fluctuations (e.g. CO2)

• Note magnitude of Ru edge jump!


XANES fitting• Data were normalized and

aligned using Athena. • Least squares fitting of Ru edges

with Sixpack. • The standards for the least

squares fits were RuO2-hydrate and Ru powder.

11550 11560 11570 11580 11590 11600 116100.0

0.3

0.6

0.9

1.2

1.5

1.8

Nor

mal

ized

χ µ

(E)

Energy (eV)

250mV 300mV 350mV 400mV 450mV Pt Black Pt foil PtRu(66:34) as received catalyst

22080 22100 22120 22140 22160 22180 222000.0

0.2

0.4

0.6

0.8

1.0

1.2

250mV 300mV 350mV 400mV 450mV Ru metal PtRu(66:34) as

received catalyst Ru Oxide Ru oxide H2O

Norm

aliz

ed χ

µ (E

)

Energy (eV)22080 22100 22120 22140 22160 22180 22200

0.0

0.2

0.4

0.6

0.8

1.0

1.2

450mV (Potential vs DHE) Water 0.1M MeOH 2M MeOH

Nor

mal

ized

χ µ

(E)

Energy (eV)

Pt L3 edge

Ru K edge Ru K edge

As received


RuO2-hydrate fraction by XANES

200 250 300 350 400 450 500 550 6000.12

0.14

0.16

0.18

0.20 water 0.1M MeOH 2M MeOH

The

frac

tion

of R

u ox

ide

hydr

ate

Potential vs DHE (mV)


Pt foil XAFS Analysis

Remove background

Fit with simplemodel at multiplek-weightings

k-weight = 3 k-weight = 1

XANES region

EXAFS region


Pt EXAFS

• Potential dependent EXAFS at 0.1M. Pt EXAFS has excellent fit with a totally metallic environment. All data are nearly identical.

• FT range for k space is 2 Å to 13 Å.• Fit range for R space is 1.5 Å to 3 Å.

0 2 4 6 8 10 12

-0.3

-0.2

-0.1

0.0

0.1

0.2

0.3 250mV 300mV 350mV 400mV 450mV

k χ,

k

k, Å-1

2 4 6 8 10 12

-0.3

-0.2

-0.1

0.0

0.1

0.2

0.3 Exp. data at 350mV Fit with model

k χ,

k

k, Å-1

0 2 4 6

0

2

4

6

8 Exp. data at 350mV Fit with model

|χ(R

)|, Å

-4

R, Å


Ru EXAFS Fitting

• Addition of Ru-O neighbors improves the EXAFS fit.• The peak at about 1.3Å is ascribed to oxygen bound to Ru.• The asymmetric distribution of the Ru-O peak is consistent with disorder

0 2 4 6

0

1

2

3

4

5 Exp. data at 350mV Fit with Ru

|χ(R

)|, Å

-4

R, Å0 2 4 6

0

1

2

3

4

5 Exp. data at 350mV Fit with Ru and Ru-O

|χ(R

)|, Å

-4R, Å


Ru EXAFS

• Potential dependent EXAFS at 0.1M MeOH

• [MeOH] dependent EXAFS • Model fit at 350mV

0 2 4 6 8 10

-0.3

-0.2

-0.1

0.0

0.1

0.2

0.3 250mV 300mV 350mV 400mV 450mV

k χ,

k

k, Å-12 4 6 8 10

-0.3

-0.2

-0.1

0.0

0.1

0.2

0.3 Exp. data at 350mV Fit with model

k χ,

k

k, Å-1

2 4 6 8 10-0.2

-0.1

0.0

0.1

0.2

0.3 Water 0.1M MeOH 2M MeOH

kχ, k

k, Å-1


Metal cluster structural model

FCC structure, count first shell neighbors only from Pt and Ru edgesThis only “sees” atoms in the metallic cluster

Average Coordination #’sNRu

Pt = Pt around RuNRu

Ru = Ru around RuNPt

Ru = Ru around PtNPt

Pt = Pt around PtnPt = Pt coordinationnRu = Ru coordination

Fractional Coordination #’sY = Ru around PtX = Pt around Ru

Frenkel 1998, Shibata 2003


[Pt]nPtN

BPt-RuY =

[Ru]nRuN

BRu-PtX =

BPt-Ru = BRu-Pt


Metallic nanoparticle structure• First shell analysis

• Fit Pt and Ru EXAFS simultaneously at each potential. No potential dependence observed.

• Simultaneously fit Pt and Ru data at all potentials. Identical overall average coordination was observed.

• Use fractional coordination numbers, X (Pt around Ru) and Y (Ru around Pt) and total coordination number about each atom, n (Frenkel 1998, Shibata 2003)

• Bond lengths and Debye-Waller factors are consistent with literature values for C supported Pt-Ru catalyst (Russel 2001, Camara 2002)

0.27 ± 0.02Y0.54 ± 0.02X8.2 ± 0.2n

= 0.5=XY

[Pt][Ru]


Apply model to as-received catalyst

Pt not simply metallic but has oxygen near neighbors

Ru shows large increase in number of oxygen near neighbors


Catalyst structural changes

As received catalyst

• Ru oxidation ~58%• N = 5.6• [Ru]/[Pt] = 0.44• Pt-O bonds present• Ru-O bonds ~2.8 avg

In situ catalyst

• Ru oxidation ~15%• N = 8.2• [Ru]/[Pt] = 0.50• No Pt-O bonds• Ru-O bonds ~0.24 avg

S. Stoupin et al., J.Phys Chem B, 110, 9932 (2006).


[Ru]/[Pt] = 35/65 = 0.54


Possible Structural Model

FCC alloy phase with an amorphous Ru ghost phase

As received catalyst In situ catalyst

•Higher surface area may be critical in catalyst performance•Incorporated O eases CO oxidation


Conclusions

Metallic cluster of the catalyst nanoparticleComposition is about 2:1 Pt:Ru

Model fit suggests that the alloy is not totally randomized(i.e. X ≠ .65 & Y ≠ .35)

Pt is metallic within the potential window (250mV and 450mV) in water or aqueous methanol.

Ru–O bonds are not potential or [MeOH] dependent (Rolison)On the surface?

In a separate phase?

The potential transition point is not accompanied by ensemble changes at the surface.

Lots more to do!


Extra bonus material

Another example of the kind of experiment which is well-suited to an undulator beamline. Where the sample is damaged rapidly by the x-ray beam!


Dilute magnetic semiconductors

• Cations replaced by Mn, Co, Fe, etc. • Typical examples are: ZnO-Mn, CdS-Mn, ZnS-Mn, etc.• Host s-p band ⇔ Mn2+ d electron exchange interactions• Unusual magnetotransport and magnetooptical phenomena

Carrier induced ferromagnetism in InAs-Mn and GaAs-Mn

• DMS nanocrystals are unique systemssemiconductor confinement effectsmagnetic properties


Acknowledgements

Graduate studentsMehdi Ali – IIT PhysicsRanjani Viswanatha - IIS

CollaboratorsDipankar Das Sarma - IISSoma Chattopadhyay – MRCATTomohiro Shibata – MRCATMali Balasubramamian – XOR20Shelly Kelly – ANL BioSciences

FundingMRCAT is supported by contributions from MRCAT member institutions.

The APS is funded by the U. S. Department of Energy, Office of Basic Energy Sciences under Contract number W-31-109-Eng-38.


Sample preparation and characterization

• Wet chemical synthesis starting with Mn-acetate and Zn-acetate.

• Capping with polyvinylpyrollidone (PVP) results in smaller sized particles (5 nm or less) with uniform size distribution

• Bulk sample synthesized by annealing the powders at 1200°C for 12 hours in air. The size of the bulk particles is ~1.5 microns.

• Size calculated using Scherrer’s equation and verified by TEM.

• The percentage of Mn doping in the samples was estimated byEDAX and ICP-AES.

• Bandgap was measured by UV-VIS Absorption spectroscopy.

• Magnetic properties were measured by Electron Paramagnetic Resonance (EPR).


XRD , UV-VIS and EPR results

3 0 4 0 5 0 6 0 7 0

( i i ) B u l k x = 0 . 0 1

( v i i ) x = 0 . 0 2 3

( v i ) x = 0 . 0 1(1

02)

(110

)

(103

)

(200

)(1

12)

(201

)

(101

)(0

02)

(100

)

( i v ) x = 0 . 0

( v i i i ) x = 0 . 0 5

( i i i ) S i m u l a t e d f o r 4 . 7 n m d i a m e t e r

( v ) x = 0 . 0 0 5

2 θ ( d e g )

Inte

nsity

(arb

. uni

ts)

( i ) B u l k x = 0 . 0 3 0 0 3 2 0 3 4 0 3 6 0 3 8 0 4 0 0

( i i ) x = 0 . 0 0 5Inte

nsity

(a. u

.)

λ ( n m )

( v ) x = 0 . 0 5

( i v ) x = 0 . 0 2 3

( i i i ) x = 0 . 0 1

( i ) x = 0 . 0

B u l k B a n d g a p

C a p p e d Z n 1 - x M n x O : U V - a b s o r p t i o n

2900 3000 3100 3200 3300 3400 3500 3600

Magnetic Field (Gauss)

Inte

nsity

(arb

. uni

ts)

(ii) 4.7 nm

(i) Bulk

Signal [II]

Signal [I]

XRD shows formation of wurtzite nanocrystals

Increase in the bandgap compared to the bulk, some variation with Mn concentration

EPR spectra from the doped samples exhibit well resolved hyperfine splitting of isolated Mn2+ ions, suggesting that Mn-Mn interactions are rather weak.


Sample degradation of ZnO-Mn

Clear evidence of reduction of Mn with time exposed to x-raysObserved with bending magnet beam tooUndulator quick scans can help!


ZnO-Mn XANES

• Average valence state changes from bulk to nanoparticle sample

• Mn(II) dominates bulk sample

• Mn(III)-Mn(IV) dominates nanoparticle samples

6532 6536 6540 6544 65480.0

0.2

0.4

6530 6540 6550 6560 65700.0

0.4

0.8

1.2

1.6

2.0

x m

u

Energy (eV)

xmu

Energy (eV)

BULK NANO-Mn:1.0% NANO-Mn:5.0% MnO Mn3O4

Mn2O3

MnO2


Mn replaces Zn in the bulk sample but not in the nanoparticle

0 2 4 6 8 10 12-6-5-4-3-2-1012

Mn edge data:Bulk ZnO-Mn(1%) Zn edge data:Bulk ZnO-Mn(1%) Mn edge data:Nano ZnO-Mn(1%) Zn edge data:Nano ZnO-Mn(1%)

k(χ)

k (Å-1)

EXAFS of bulk and nanoparticle ZnO-Mn(1%)

Bulk Zn and Mn spectra are similar

Nanoparticle Zn and Mn spectra are different


ZnO structural model

Zn

O7

Zn1

Zn1

Zn1

Zn1

Zn4

Zn5Zn5

Zn2

Zn3

Zn3

Zn3

Zn2

Zn3

Zn5Zn3

Zn3

Zn4

Zn5O8

O1

O7

O5

O8

O8O9

O8

O7

O7

O8

O8

O2

O7O4

O3

O6

O9

O7

O10

O1

O6

4.574Zn5

3.816O7

3.802O6

3.212Zn2

4.961O10

4.962O9

4.576O8

4.572Zn4

3.801O5

3.256Zn3

3.221O4

3.214Zn1

1.991O3

1.9751O2

1.972O1

Bond Length (Å)

Path DegeneracyAtoms

ZnO has wurtzite structure:a = b = 3.250 Å; c = 5.207 Åα = β = 90°. γ = 120°.

Mn


Bulk samples fits

0 2 4 6 8 10 12 14-6-5-4-3-2-1012

k2 χ(k

)(Å-2)

Mn edge data:Bulk ZnO-Mn(1%) Zn edge data:Bulk ZnO-Mn(1%) Mn edge fit Zn edge fit

k (Å-1)

0 2 4 6 8 10-1.2

-0.8

-0.4

0.0

0.4

0.8

1.2

Bulk Zn Data and Fit

Bulk Mn Data and Fit

M

ag o

f FT

(kw

= 2

)

R (Å)

0 2 4 6 8 10

-3

-2

-1

0

1

Bulk Zn Data and Fit

Bulk Mn Data and Fit

Rea

l Par

t of F

T (k

w =

2)

R (Å)

Bulk Mn and Zn spectra sets were simultaneously fit in R-space

No question that Mn substitutes for Zn in bulk


Fit results for bulk ZnO-Mn

0.1±1.4

10±2

σ 2Zn3

(·103 Å2)

Bulk: Zn edgeZnO-Mn (1%)

-2.3 ±1.5

16±47±14.0±16.0(1)5.34(2)3.19(2)

Bulk: Mn edgeZnO-Mn (1%)

E0 (eV)σ 2

Zn2

(·103 Å2)σ 2

Zn1

(·103 Å2)σ2

05

(·103 Å2)σ2

01

(·103 Å2)c (Å)a (Å)SAMPLE

• Model simultaneously optimized to both data sets with common parameters

• The model can describe the features of both data sets simultaneously.

• The σ2 values and distances are the same for the Mn edge and the Zn edge spectra

• The same model was applied to the Zn and Mn spectra from the nano sample, but was not successful in reproducing both spectra.


Nanoparticle XAS

23 ± 5 %17 ± 4 %16 ± 5 %28 ± 6 %Mn2O3

57 ± 4 %64 ± 3 %60 ± 4 %57 ± 4 %Mn3O4

19 ± 2 %19 ± 2 %29 ± 3 %14 ± 3 %MnO2

2 ± 2 %0 %3 ± 2 %2 ± 2 %ZnO-Mn1% BULK

SampleZnO-Mn(5.0%)

SampleZnO-Mn(2.3%)

SampleZnO-Mn(1.0%)

SampleZnO-Mn(0.5%)Standards

0 2 4 6 8 1 0 1 2 1 4 1 6 1 8-5

-4

-3

-2

-1

0

1

N an o Z n O -M n (0 .5% )

N an o Z n O -M n (1% )

N an o Z n O -M n (2 .3% )

N an o Z n O -M n (5% )

L in e a r C o m b in a tio n F it O f M n D a ta

k2 χ(k

) (Å

-2)

k (Å -1)


Summary: DMS nanoparticles

• XANES results suggest that the valence state of the nanoparticle samples is very different than that of the bulk sample

• EXAFS results show that Mn atoms replace Zn atoms in the bulk ZnO-Mn.

• Nanoparticles (4.7 nm) of ZnO-Mn with Mn doping varying from 0.5% to 5%, appear to be in core-shell structure with Mn located on the surface

• The shell consists of various oxides of Mn.• Preliminary analysis on CdS-Mn indicates same core-shell

structure and radiation damage similar to ZnO-Mn

Date post:	18-Oct-2020
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times

In-situ XAFS studies of fuel cell catalysts · Develop methods for characterization of catalysts in...

Documents