Methanol Steam Reforming in a Pd-Ag Membrane Reactor:

Experiments & Modeling

Sameer H. Israni

(Advisor : Prof. Michael P. Harold)

Acknowledgement: Support by NSF CTS-0521977

Palladium based membranes

Dense Palladium & Palladium-Alloys have very high selectivity for H2

Non-porous Gas Mixture Pd Membrane Pure H2

Porous Substrate

α-Al2O3

Conventional or Top LayerMembrane

ELP

α-Al2O3

Pd nuclei

α-Al2O3

Pd nucleiPd or Pd-Alloy

Sensitization & Activation

α-Al2O3

Pd nuclei

-Al2O3

α-Al2O3

-Al2O3

α-Al2O3

Pd nuclei

- Al2O3

NanoporeMembrane

Sol-Gel Slip Casting

Sensitization & Activation

Sol-Gel Slip Casting

ELP

Pd

Pd / Pd-Alloy Membrane SynthesisConventional & Nanopore Membranes

B.R.K. Nair et al., J. Membrane Sci., 290 (2007) 182

• Current commercial membranes are ~ 50 - 100 m thick Too costly

• Thinner Pd membranes required (5-20 m)– High flux and low cost– More defects

Porous Substrate

-Al2O3 hollow fibers

Characterization of Membranes

Membrane characterization

• SEM – defects, grain size

• EDS / XPS – bulk composition, composition profile along depth of membrane

• XRD – bulk composition, alloy formation

Membrane Reactors for H2 Generation

Advantages – Process intensification– Overcome equilibrium limitations

or kinetic inhibitions

Disadvantages – Cost– Durability (leaks)– Productivity vs. Utilization issue

H2 generation from various fuels (CH4, ethanol, methanol, NH3 etc.)– Stationary H2 generation

– On-vehicle H2 generation

Single Fiber Packed Bed Membrane Reactor Multi Fiber Packed Bed Membrane Reactor

Objectives

1. Membrane flux under reaction conditions• Effect of reactant & product species

2. Methanol steam reforming: experiments & 2-D modeling

• Packed Bed Reactor (PBR)

• Single-fiber Packed Bed Membrane Reactor (PBMR)

3. 3-D model to simulate larger scale multi-fiber membrane reactor

Part 1

Membrane H2 Flux

In

Reaction Conditions

Pd-Ag (23 wt % Ag) Nanopore membrane synthesized for methanol steam reforming study

Total Pd-Ag thickness – 3.7 to 4.0 m

-Al2O3

(50 nm pore size)

-Al2O3

(5 nm pore size, 50% porosity)

Pd-Ag2.5 m

1.2 m

Hollow Fibre Membrane - α-Al2O3 (3.7 mm OD)

Supplier - Media & Process Tech Inc.© , PA, USA

Permeation Characteristics 3.4 micron Pd-(23 wt. %)Ag membrane

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0 50 100 150 200 250 300 350 400 450

[Pret0.5 - Pperm

0.5)] (Pa0.5)

H2

flu

x (m

ol/

m2 /s)

225 C

250 C

300 C

Pd-Ag Membrane Pure Gas Permeation Results

•Infinite H2/He selectivity

•No degradation in performance after exposure to reforming conditions – 12 days

)( 5.0,

5.0, 222 permHretHH PPQJ

Inhibition of H2 Flux in Reaction Conditions

Presence of CH3OH, H2O, CO2, CO reduces the flux of H2

1. Decrease in H2 partial pressure

2. Concentration polarization

3. Surface coverage – reduction in surface area

H2

Other Gas

Conc.

)( 5.0,

5.0, 222 permHretHH PPQJ

)()1( 5.0,

5.0, 222 permHretHH PPQJ

the fraction of membrane surface sites covered by

species other than hydrogen

Experimental Setup

2 “

H2 + Impurity

Retentate to GC

3/8” OD

Pd77Ag23 Membrane

Permeate H2 to GC (@ 1 atm)

Feed (varying concentrations)

1. Single Impurities

• H2 + CO

• H2 + CO2

• H2 + H2O

• H2 + CH3OH

2. Mixture of impurities

• H2 + CO + CO2 + H2O + CH3OH

Temperatures

• 225 oC

• 250 oC

• 300 oC

Retentate pressures

• 3 – 5 bars

Measured

• Decrease in H2 flux through membrane (compared to pure H2 feed)

Model Details

• Momentum balance (Navier-Stokes – weakly compressible)

• Mass balance (Stefan-Maxwell)

• Membrane poisoning (Langmuir adsorption)

i ii

T

MxT

p

UUUpUU

U

)].(32))(([).(

0).(

0)))((.(

p

pxxDU ijj

iijii

)()1( 5.0,

5.0, 222 permHretHH PPQJ

2-D, isothermal model

is the fraction of membrane surface sites covered by species other than hydrogen

JH2 - H2 flux thru the membrane Q – permeability of membrane P i – partial pressure of species i

Ki – adsorption coefficients of species I KI, KII – intermediate reaction constants -fraction of surface covered by non-H2 species

)(

)()()()()()(1

1)1(

2

33

2

33

3322222

2

21

21

31

HH

OHCHOHCHIII

HH

OHCHOHCHIOHCHOHCHCOCOCOCOOHOHHH

HH

PK

PKKK

PK

PKKPKPKPKPKPK

PK

Assumptions

• Steps (5), (10) & (12) are rate determining steps

• All other steps are in equilibrium

• Adsorption energies independent of surface coverage

• Multi-site adsorption model (Martinez & Basmadjian, Chem. Eng. Sci., 7

(1996) 1043)

1) H2 + 2S 2H-S

2) H2O + S H2O-S

3) CO + 3S CO-S3

4) CO2 + 2S CO2-S2

5) CO2-S2 + 2S CO-S3 + O-S

6) O-S + 2H-S H2O-S + 2S

7) CH3OH + 2S CH3OH-S2

8) CH3OH-S2 CH3O-S + H-S

9) CH3O-S + S CH2O-S + H-S

10) CH2O-S + S CHO-S + H-S

11) CHO-S +3S CO-S3 + H-S

12) 2CH3OH-S2 CH3O-S + CH3-S + H2O-S + S

13) CH3-S + H-S CH4-S + S

14) CH4-S CH4 + S

Surface mechanisms based on experimental results and previous studies reported in literature

S – surface site on Pd-Ag membrane

Proposed Surface Mechanism

Results – Experimental vs. Simulation

• Main cause of decrease in H2 flux is surface coverage / poisoning

– CO causes the largest decrease

• As temperature ↑ H2 flux ↑

• Drop in flux is reversible

Single impurity studies300 C, 5 bars

Results – Experimental vs. Simulation

• As temperature ↑ H2 flux ↑– except for CH3OH; maximum H2 flux at ~ 250 oC

• During CO2 experiments CO & H2O detected at outlet

• During CH3OH experiments CO detected at outlet (H2O also detected for certain conditions)

225 C, 3 bars

Single impurity studies

250 C, 3 bars

250 C, 5 bars

Results – Parameters Estimated

)exp(RT

HKK o

K0 H (kJ/mol)

KH2 3.33e-10 -58.46

KH2O 1.54e-10 -49.12

KCO2 3.67e-15 -106.2

KCO 6.38e-11 -88.42

KCH3OH 1.69e-16 -123.3

KI 3.77e+38 418.9

KII 4.44e+29 328.8

Estimated binding energies correspond well with literature

reported values

)(

)()()()()()(1

1)1(

2

33

2

33

3322222

2

21

21

31

HH

OHCHOHCHIII

HH

OHCHOHCHIOHCHOHCHCOCOCOCOOHOHHH

HH

PK

PKKK

PK

PKKPKPKPKPKPK

PK

Results – Gas mixtures

No major interactions between species

Temperature

(oC)

Pressure(bars)

CH3OH(mole

%)

CO(mole

%)

CO2

(mole %)

H2O(mole

%)

Experimental % drop in H2

flux

Simulated % drop in

H2 flux

225 3 20 3 10 20 89 ± 3 92.2

250 3 20 3 10 20 84 ± 3 90.2

250 5 20 3 10 20 76 ± 3 83.7

300 5 20 3 10 20 77 ± 3 84.4

225 3 5 10 25 5 85 ± 3 87.4

250 3 5 10 25 5 80 ± 3 86.4

250 5 5 10 25 5 78 ± 3 81.3

300 5 5 10 25 5 75 ± 3 79.8

Part 2

Methanol Steam Reforming in

PBR & Single-Fiber PBMR

Experimental Details – PBR & PBMR

• CH3OH : H2O 1:1 molar basis

• Catalyst BASF V1765

(CuO/ZnO/Al2O3 modified with ZrO2)

(avg. particle diameter 0.45 mm)

• Temperatures 225, 250, 300 oC

• Retentate pressures 3, 5 bars

Performance Metrics

1. CH3OH Conversion

2. Permeate H2 Purity

3. H2 Utilization %

4. H2 Productivity

flowrateOHCHInlet

flowrateHPermeate

3

2

3

32

ms

mol

volumereactor

flowratemolarHpermeate

4 “

CH3OH + H2O(1:1 Molar basis)

CuO/ZnO/Al2O3

Catalyst Bed

Retentate to GC

3/8” OD

Pd78Ag22 Membrane

Permeate to GC (@ 1 atm)

• Mass balance : Stefan Maxwell equation

– Incorporating membrane permeability inhibition factor (1-)

– Radial mass diffusivities : D.J. Gunn, Chem. Eng. Sci., 42(1987) 363

Model Details

2-D, non-isothermal, pseudo-homogeneous

y

yijji

ijii Rp

pxxDU )))((.(

• Heat balance : Convection + conduction

– Effective catalyst bed thermal conductivity + Wall to bed heat transfer

coefficient : Specchia et al, Chem. Eng. Commun., 4 (1980) 361

y

yypeff RHTUCTk .).(

• Momentum balance : Darcy-Brinkman equation

)( 2UPk

U

Model Details

))(1())()(1(

)1()(5.02

5.0*)1(

5.022

*)1(

5.022

*)1(

5.023

*)1(3

112323

25.023

*)1(3

HHHOHOHHCOHCOOHOHCHOCH

assOHOHCHSRCOHHOHCHOCHSRSR pKppKppKppK

CCppKeqppppKkR

))(1())()(1(

)1()(5.02

5.0*)2(

5.022

*)2(

5.023

*)2(3

2232

25.023

*)2(3

HHHOHOHHOHCHOCH

assOHCHMDCOHHOHCHOCHMDMD pKppKppK

CCpKeqppppKkR

))()(1(

)1()(5.022

*)1(

5.022

*)1(

5.023

*)1(3

21222

5.022

*)1(

HOHOHHCOHCOOHOHCHOCH

sOHCOWGSCOHHOHCOOHwWGS ppKppKppK

CppKeqpppppKkR

• Reaction Kinetics : Peppley et al, Appl. Cat. A: Gen., 179 (1999) 31

i) Methanol steam reforming

CH3OH + H2O 3H2 + CO2 dH= +49.4 kJ/mol

ii) Methanol Decomposition

CH3OH 2H2 + CO dH= +90.5 kJ/mol

iii) Water-gas shift

CO + H2O 3H2 + CO2 dH= -41.1 kJ/mol

• Reaction Kinetics : Peppley et al, Appl. Cat. A: Gen., 179 (1999) 31

i) Methanol steam reforming

CH3OH + H2O 3H2 + CO2 dH= +49.4 kJ/mol

Cs,i – concentration of active catalyst sites

Kinetic Parameters

obtained from PBR data

Membrane permeability

obtained from pure gas permeation results

Membrane Inhibition Factor

Obtained from experiments & modeling

PBMR Model

PBR ResultsExit Compositions

250 C, 5 bars

CH3OH+

H2O

CuO/ZnO/Al2O3

Catalyst

PBR ResultsTemperature profile at center of PBR

250 C, 5 bars

Kinetic Parameters

obtained from PBR data

Membrane permeability

obtained from pure gas permeation results

Membrane Inhibition Factor

Obtained from experiments & modeling

PBMR Model

PBMR ResultsExit Retentate Compositions

250 C, 5 bars

PBMR Results

)()2( 23

22 flowrateOHInletflowrateOHCHInlet

flowrateHPermeatenUtilizatioH

250 C, 3 bars

250 C, 5 bars

300 C, 5 bars Temp ↑, Utilization ↑

Press ↑, Utilization ↑

W/Fao ↑, Utilization ↑

H2 Utilization in PBMR

PBMR Results

32

ms

mol

volumereactor

flowratemolarHpermeatetyproductivi

250 C, 3 bars

250 C, 5 bars

300 C, 5 barsTemp ↑, Productivity ↑

Press ↑, Productivity ↑

Optimum W/Fao

Permeate H2 Productivity of PBMR

PBR vs. PBMRMethanol Conversion

PBR vs. PBMR undiluted catalyst3/8" OD

20

40

60

80

100

0 5 10 15 20 25

W/Fao (g. cat. hr/ gmoles)

Met

han

ol C

on

vers

ion

%

250 C, 3 bars

250 C, 5 bars

300 C, 5 bars

Solid lines – PBR

Dashed lines - PBMR

~ 5 - 15 % increase

Methanol Conversion

)()1( 5.0,

5.0, 222 permHretHH PPQJ

Fraction of membrane surface sites covered by species other than hydrogen300 C, 5 bars

PBMR – Simulation Results

Rate limiting step H2 flux through membrane Surface Poisoning

= 0.5) implies that 50 % of membrane surface

poisoned

300 C, 5 bars

PBMR – Simulation Results

Permeate H2 ProductivityOvercoming surface poisoning effects

Permeate Purity

• 100 % pure H2 obtained

• GC TCD detection limit ~10 ppm

• If separation factor is not infinite

Concentration Polarization also

severely affects purity

H2

Other Gas

Conc.

Part 3

3-D model to Simulate Large-Scale Multi-Fiber

PBMR

Large scale multi-fiber PBMR model

PBR & single-fiber PBMR model

Model for membrane inhibition factor

Multi-Fiber PBMR

z

y

x

x

y

H2 mole fraction

x

y

z

H2 mole fraction

Multi-Fiber PBMR Simulation Results

Comsol © software used for simulations

Reactor ID – 5”

Length – 6 m

Membrane OD -3.7mm

No. of fibers - 85

Inlet velocity – 1 m/s

Inlet Temp – 250 C

Wall Temp – 265 C

Temperature (K)

Conclusions

1. Membrane flux in reaction conditions

2. Experiments & 2-D model for single-fiber PBMR– Rate limiting step Membrane H2 flux

– Surface poisoning is the main cause of low membrane flux

3. 3-D model developed for large scale multi-fiber PBMRs– For understanding effects of design & operating parameters on productivity

)(*)1( 5.0,2

5.0,22 permHretHH PPQJ

))(

1)(()()()()()(1

)(1)1(

25.022

3315.0

333

13

1

22225.0

22

5.022

KPK

PKKPKPKPKPKPK

PK

HHOHCHOHCHOHCHOHCHCOCOCOCOOHOHHH

HH