Issuu on Google+

Analysis of Transport Phenomena in Optical Fiber Manufacturing

Ranga Pitchumani Advanced Materials and Technologies Laboratory Department of Mechanical Engineering University of Connecticut Storrs, Connecticut http://www.engr.uconn.edu/amtl

Presented at Corning Incorporated • August 6, 2008


Outline

 Introduction to Advanced Materials and Technologies Laboratory  Optical Fiber Drawing  Numerical Modeling  Analysis of dopant concentration and refractive index profile evolution during manufacturing  Simulation and optimization under uncertainty  Analysis of coating thickness variations  A sampling of other recent research activities in the Laboratory

Advanced Materials and Technologies Laboratory • University of Connecticut


Advanced Materials and Technologies Laboratory  MISSION: Conduct research towards improving the fundamental description and understanding of complex physical phenomena governing materials processing and design, and emerging technologies. The fundamental description and understanding are applied towards practical development, design, optimization and control.  LAB PERSONNEL:  6 Graduate Students, 1-2 Postdocs  2 Undergraduate Students  2-3 High School Students each summer  RESEARCH AREAS:  Advanced Materials Processing  Microsystems and Microfabrication  Fuel Cells/Energy  Design and Manufacturing  Core Sciences $6M total funding Advanced Materials and Technologies Laboratory • University of Connecticut


Representative Recent Projects  Polymer and Composite Materials Processing (NSF, ONR, AFOSR)  Processing of thermosetting and thermoplastic-matrix composites  Fundamental transport property determination: Conductivity; Permeability  Investigations on microscale and nanoscale phenomena  Process optimization and control using physics-based models  Microsystems (Sandia National Lab, NSF)  Fabrication of high-aspect ratio microstructures (HARMs)  Reliability of microdevices  Fuel Cells (Army)  Novel cell designs–functionally graded electrodes and catalyst layers  Passive fuel cells and mFuel Cells  Cell modeling and performance optimization under uncertainty  Measurement and prediction of contact resistance  Information Technology in Materials Processing (NSF-ITR)  Advanced computing techniques for rapid simulation and optimization of materials processing under uncertainty–Application to Optical Fiber Drawing  Life Prediction of Aircraft Engine Components Using Neural Networks (P&W)  New Material Designs for Contact Heads and Ablative Walls in Electrical Circuit Breakers (GE)

Advanced Materials and Technologies Laboratory • University of Connecticut


Optical Fiber Drawing

nr(r)

coating

 Optical fibers are used to transmit light signals  The manufacturing process consists of: fiber drawing, cooling and coating steps.  Preform is SiO2 doped with oxides of Ge, P or B, with an initial concentration profile.  The index of refraction profile is a function of the dopant concentration which evolves during the drawing and cooling steps; at the end of the cooling step, the index of refraction profile is finalized.  The cooled fiber is coated with a polymer.

Advanced Materials and Technologies Laboratory • University of Connecticut


Optical Fiber Drawing Model Conservation equations are formulated in the glass and surrounding fluid domains; It is assumed that the flow and temperature fields are not significantly affected by the dopant (taken to be GeO2).

Continuity



Momentum

 Energy

Species (Dopant Diffusion) 

u 1 (rv)  0 z r r u 2  1 (ruv) p   u  1   u v      2 m  rm   z r r z z  z  r r  r r  ( uv) 1 (rv 2 ) p   u v  2   v  2mv     m   rm  z r r r z  r z  r r  r  r 2

( C p uT) 1 rC p vT   T  1   T    k  rk  z r r z  z  r r  r    u 2 u 2 v 2  u v 2   m2               r  r  z   z r   

 E  (uc) 1 (rvc)   c  1   c    D  rD ; D  D0 exp  z r r z  z  r r  r   kB T 

+ Boundary and Interface Conditions 

Advanced Materials and Technologies Laboratory • University of Connecticut


Optical Fiber Drawing Model The preform/fiber domain can be represented in one of two ways: 1.

A fixed necking profile given by a parametrized function (such as the logarithmic function of an 8th order polynomial, as in Lee and Jaluria, 1996)

2.

Profile that is determined as part of the solution: In approach, the necking profile is parametrized as a hyperbolic tangent function

r(z)  c1 tanh c 2 z c 5  c 3   c 4

in which, the constants are determined as solution to an optimization problem seeking to maintain uniform tension (obtained from the viscous v  force gradient in fiber drawing simulation) along the axis  FT  3mA  



z 

Simulation Outputs: 

Temperature, Velocity, and Concentration, profiles in the preform fiber, and fluid domains

Refractive index is calculated from the dopant concentration using the Lorentz-Lorentz equation (Huang, Sarkar, Schutz, 1978): 1 2 n ; 1 

n   n

2 r,i

2 r,i

1c i

 2i

ci



i

Advanced Materials and Technologies Laboratory • University of Connecticut


Necking Profiles for Different Parameters

zL = 15 cm

Advanced Materials and Technologies Laboratory • University of Connecticut


Validation and Parametric Studies Y. Yan and R. Pitchumani, Int. J. Heat and Mass Transfer (2006)

Numerical simulations show good agreement to results in literature The validated numerical simulations are used to investigate the effects of non-dimensional parameters representing the flow (i.e. the Reynolds numbers), the geometry (i.e. furnace radius), and the mass transport (i.e. Peclet number of the dopant), on the drawing process by observing the surface temperature, dopant concentration, and refractive index distributions. Advanced Materials and Technologies Laboratory • University of Connecticut


Effects of Reynolds Number Y. Yan and R. Pitchumani, Int. J. Heat and Mass Transfer (2006)

Re f 



u f rf

 f ,m

; 

T  T ; T0  T

z r ;  zL r(z)



As Ref increases, the fiber residence time in the furnace and cooling zones decreases, reducing the rate of temperature rise and decay in the respective zone. Increasing Ref decreases dopant diffusion and increases the radial refractive index variation.

Advanced Materials and Technologies Laboratory • University of Connecticut


Effects of Peclet Number of Dopant Y. Yan and R. Pitchumani, Int. J. Heat and Mass Transfer (2006)

The surface temperature is independent of Pe, due to the model assumption.

u f rf T  T ;  ; Dm T0  T z r  ;  zL r(z) Pe 

As Pe increases, centerline dopant concentration decays less along the axis (due to poor diffusion), and the refractive index varies considerably across the radius. The results suggest that refractive index profile may be tailored to a degree through processing conditions (in addition to material choices)



Advanced Materials and Technologies Laboratory • University of Connecticut


Outline

 Introduction to Advanced Materials and Technologies Laboratory  Optical Fiber Drawing  Numerical Modeling  Analysis of dopant concentration and refractive index profile evolution during manufacturing  Simulation and optimization under uncertainty A. Mawardi and R. Pitchumani, IEEE J. Lightwave Technology (2008) C. Acquah, et al., Ind. Eng. Chem. Res. (2006)

 Analysis of coating thickness variations  A sampling of other recent research activities in the Laboratory

Advanced Materials and Technologies Laboratory • University of Connecticut


Stochastic Analysis: Uncertain Parameters  Uncertain Parameters  Radiation Parameter: – Spectral, hemispherical emissivity of the silica preform, (a). The absorption coefficient a is given as a function of the wavelength . The magnitude of a is considered to be uncertain.  Concentration Parameters: – Thermophysical dopant properties (temperature dependent): Diffusion coefficients: D0, E  Rheological Parameters: – Viscosity of the glass melt varies with temperature following an Arrhenius relationship; the preexponential factor (m0) and the activation energy (Em)are considered uncertain.  Temperature Parameters: – Furnace temperature, Tw – Cooling gas temperature, Td  Parameters with uncertainties are represented as probability distributions with mean m, and standard deviation .  Define: coefficient of variance /m  quantifies the severity of uncertainty in the parameters  represents the ―width‖ of the distribution Advanced Materials and Technologies Laboratory • University of Connecticut


Quality Parameters Considered Drawing induced defects [Hanafusa, et al. (1985)] Point defects associated with the interstitial Ge atoms (Ge E’ centers) occurs in GeO2-doped silica due to quenching from high temperatures: associated with oxygen vacancies in neighboring the Ge atoms. The concentration of the thermally induced defects, nd, is expressed as:  E A  n d (T)  n p exp f    kT V 

Draw Tension [Paek and Runk (1978)] Combination of viscous force, forces due to surface tension, inertia, and shear force exerted by external fluid.  v FT (r)  Fm  F  FI  Fe  Fg  3mR 2 z

Residual Stress [Paek and Kurkjian (1975)]: Combination of mechanical and thermal stresses. Affects the optical characteristic of the fiber F  z (r)   m   t (r)    A2

T r 

E(r,T)

 1  (r,T) c   (r,T)dT

TC

Index of Refraction (IR) Correction [Hibino, et al. (1989)] Residual stress affects refractive index through photoelastic effects. A correctionto the refractive index is given by:

nr  Ca  r  Cb (   z )  Cb  z r

Advanced Materials and Technologies Laboratory • University of Connecticut


Sampling-Based Stochastic Analysis Tw

Td

m0

Em

D0

ED a

Uncertain Inputs Distributions

Output Distributions nd Sampling (LHS)

zz,max Deterministic Inputs u1, u2,…uM

Nonlinear Optical Fiber Drawing Simulation

FT,max

Based on the stochastic convergence analysis, a sample size of 500 was chosen for each stochastic simulation Advanced Materials and Technologies Laboratory • University of Connecticut


  max

n r r

r





n r,0 r



n r r  n r,0 r 100% n r,0 r


Quality Variability Measure  Typically, standard deviation is used as a measure of ―variation‖ of a distribution  However, standard deviation is sensitive to outliers.  In this study, a measure of variance based on percentiles values is used to quantify distribution variation. 25% 50%

75%

Interquartile Range (IQR)

Advanced Materials and Technologies Laboratory • University of Connecticut


Output Distributions

Advanced Materials and Technologies Laboratory • University of Connecticut


Output Variations

The IQRs of all output parameters pronouncedly increases with the COV of Tw. The IQRs of max. residual stress and max. refractive index variation also significantly increase with the COV of ED, while the IQR of max. tension increases with COV of Em. Advanced Materials and Technologies Laboratory • University of Connecticut


Design Windows

As input COV increases, the output variations (IQR) increases. As the wall temperature increases, the IQRs of the defects and max. refractive index variation increase, while the IQRs of the residual stress and maximum tension decrease. Advanced Materials and Technologies Laboratory • University of Connecticut


Remarks

 The wall temperature is the most important of the uncertain parameters, and uncertainty in the wall temperature affects significantly the variations in all output parameters, and especially the Ge(3) defects  Uncertainty in the diffusion affects the variation in residual stress and the index of refraction profile

 Uncertainty in the viscosity most significantly influences the variability in maximum tension  Low draw temperature is desired to minimize the variations in Ge(3) defects and index of refraction profile, while high temperature minimizes the variability in residual stress and draw tension

Advanced Materials and Technologies Laboratory • University of Connecticut


Optimization Under Uncertainty

Problem Statement:  Design problem: to maximize drawing velocity while satisfying constraints on product quality requirements in the forms of: Ge(3) defects, residual stress, drawing tension, and a desired refractive index profile, given uncertainty in parameters affecting the fiber drawing process

Advanced Materials and Technologies Laboratory • University of Connecticut


Sampling-Based Optimization Under Uncertainty

Sampler Input Parameters

Deterministic Model

Stochastic Model

Output Variabilities

Optimum Design

Optimizer

Confidence Level (pc)

 The optimization problem:

Minimize : E Pc ( f (di )) di

pdf

pc

(Probabilistic Objective Function)

f

fcrit 1

Subject to: E Pc (g j )  g j,crit

(Probabilistic Constraints)

pc

cdf f

fcrit



 pc is the confidence level, which allows different level of risk management.  Optimization may be solved using any deterministic non-gradient based optimization method.

Advanced Materials and Technologies Laboratory • University of Connecticut


Two-Stage Optimization Problem (TSOP) C. Acquah, et al., Ind. Eng. Chem. Res. (2006) 

Using the optical fiber drawing numerical model as the basis, a deterministic TwoStage Optimization Problem (TSOP) and a Split and Bound (SB) algorithm is employed to solve the optimization under uncertainty of a fiber drawing process.

Two-Stage Optimization Problem: the two stages represent the initial design stage followed by the operational stage.

In the design stage, design variables, dk, are selected and are fixed as the process runs. During the process operation, another set of variables, zi, referred to as control variables, are varied to compensate for the effect of uncertainties.

In TSOP uncertainty, the objective function is evaluated as an expected (mean) value, whereas the constraints are treated deterministically for each realization of  within the domain of the uncertain parameter. The optimization problem can be expressed as: Maximize E ( V )

Tw ,Td ,ra ,c i ,z L , V

Subject to:

g1  n d (dk ,z i , j )  n d ,crit  0

 j  Tw ,Td ,a, m0 E m,,D0, E  dk  Tw ,Td ,ra ,c 0 ,zL,,u f 

g2  FT (dk ,z i , j )  Fcrit  0 g3   (dk ,z i , j )   crit  0 g4  max (dk ,z i , j )  max,crit  0

z i  ucg ,uig 

g5  ave (dk ,z i , j )  ave,crit  0 Advanced Materials and Technologies Laboratory • University of Connecticut


Outline

 Introduction to Advanced Materials and Technologies Laboratory  Optical Fiber Drawing  Numerical Modeling  Analysis of dopant concentration and refractive index profile evolution during manufacturing  Simulation and optimization under uncertainty  Analysis of coating thickness variations Q. Jiang, F. Yang, and R. Pitchumani, IEEE J. Lightwave Technology (2005)

 A sampling of other recent research activities in the Laboratory

Advanced Materials and Technologies Laboratory • University of Connecticut


The Coating Process Coating thickness and its variation govern the reliability of the fiber The average coating thickness can be developed analytically based on analysis of fluid mechanics of the coating polymer in the applicator Coating thickness variation has been studied mostly through experiments, with little supporting analysis The present goal is to be able to predict the thickness variations as a function of the coating process and material parameters. Advanced Materials and Technologies Laboratory • University of Connecticut


Coating Process Modeling: Problem Formulation Assumptions:  Incompressible Newtonian fluid  Axisymmetric laminar flow  No-slip on the fiber surface

Governing Equations: u u w   0 t r z    1  ur    2 u  u u  p  u    u  w     m    2 r z  r  t  r  r r  z   1   w   2 w  w w  p  w   g  u w     m r  2  t  r  z  z r  r  r     z  

Boundary Conditions: (1) no-slip conditions at the fiber surface, (2) vanishing shear stress at the free surface, (3) stress balance at the free surface, and (4) kinematic condition Stream function: Velocity components are expressed in terms of a stream function, 

Nondimensionalization:

u

1  (radial ) r z

w 1

1  (axial) r r

 3  3 p  pa  h0   p  ;   ;   2  ;    4  W 02 W 0 h02  4 m g 

z 

z h0

;r 

R r  W 0 t mW 0 W 0 h0 h  ;t  ;Ca0  ;Re 0  ;a  f ;h   h0 h0  m h0 h0

Advanced Materials and Technologies Laboratory • University of Connecticut




Solution Approach: Perturbation Analysis  *  0*  1*  O( 2 ) p*  p*0  p1*  O( 2 ) Substituting the power series expansions into the nondimensional form of the governing equations and associated conditions, and collecting terms  of the first two powers of the perturbation parameter— the wave number, — yields two differential equations for the two unknowns * and p*. Substituting these into the kinematic condition and defining a nondimensional coating thickness: (t*,z*) = h*(t*,z*) – 1 yields a partial differential equation for , whose solution is of the form: d ZW  i 0 cosZ  dr t  ;    V f 

  Ai exp 

Z  z 

Vf  t W0

dr : the dimensionless linear wave speed di : dim ensionless growth rate of the amplitude

Amplitude of the thickness variation at the entrance to the cure region (Zf) = coating thickness in the final  product d Z W  i f 0 A f  Ai exp  V    f 

Ai is still unknown

Advanced Materials and Technologies Laboratory • University of Connecticut


 



Initial Perturbation Amplitude Correlation Ai = f(Re, S, pin/pa, Rd/Rf. L/Rf) Re 

V f R f : Re ynolds Number m

Rd R f : Radius Ratio

L R f : Aspect Ratio

pin pa : Inlet Pr essure Ratio

OhCa  S

2

Fr

gm 4

 3



Ai  C1 Rd R f  L R f   pin pa  S n 4 exp S n 5Re n 6  f1 exp f 2 Re n 6  n1

n2

n3

Using experimental data from Panoliaskos, et al., (1985), Wagatsuma, et al., (1986) and Kobayashi, et al., (1991), the coefficients and exponents are determined empirically to get

Ai  0.079Rd R f 

1.5

L R f 

0.5

pin

pa 

3.85

S 0.1023 exp S 0.964 Re 3 

Advanced Materials and Technologies Laboratory • University of Connecticut


Effects of Process and Geometry on Final Amplitude

Advanced Materials and Technologies Laboratory • University of Connecticut


Design of Process Parameters & Applicator Geometry

 The maximum Reynolds number decreases with increasing S or decreasing inlet pressure ratio  The maximum Reynolds number increases with increasing Rd/Rf or L/Rf

Advanced Materials and Technologies Laboratory • University of Connecticut


Remarks

Linear Perturbation Analysis was applied to the study of coating thickness fluctuation in optical fiber manufacturing. Effects of S, Re, pin/pa, Rd/Rf and L/Rf on the coating thickness variation were presented. Process parameters have a more pronounced influence on the fluctuation than the geometric parameters. Af increases with increase in Re or S, and with decrease in pin/pa, Rd/Rf or L/Rf. A methodology for identifying design and process parameters based on maximum allowable thickness fluctuation was presented.

Advanced Materials and Technologies Laboratory • University of Connecticut


Outline

 Introduction to Advanced Materials and Technologies Laboratory  Optical Fiber Drawing  Numerical Modeling  Analysis of dopant concentration and refractive index profile evolution during manufacturing  Simulation and optimization under uncertainty  Analysis of coating thickness variations  A sampling of other recent research activities in the Laboratory

Advanced Materials and Technologies Laboratory • University of Connecticut


Composites Processing


Flow Control in Liquid Composite Molding The Process

The Problems Vacuum Line Vent

Resin Line Inlet

0.00

0-0.

The Approach


Processing-Interphase-Property Relationships AFOSR MEANS Program Tf tramp

Tj

[(CT)T ]  T  (kT T ) t y y



[ CT ]  T   (k )C EO HR (1v f ) t y y t



30 t'=10.80 t'=54.00

25

t'=0.11

0

20

t'=10.80

2.0

t'=54.00

1.8

40 60 Molecular Layer

80

100

1.2

5.0

Interphase

1.0 4.5

t'=147.18

1.6 1.4

di

0.5

Glass Fiber

2

Modulus, E [GPa]

pph PACM20

t'=147.18

20

y* = 0.0

t'=0.11

2.2

35

y/L 0.0

5.5

2.4

y* = 0.0



6.0

E , [GPa]

40

y

Matrix

di 0

10

20 30 Molecular Layer

40

4.0

0

50 100 150 Cure Cycle Ramp Time, t'

200

r

50

5.8 5.6

y/L 0.0

5.4

0.5

5.2 5.0

1.0

4.8 4.6 4.4 -0.1

0 0.1 0.2 0.3 0.4 0.5 Cure Cycle Hold Temperature,  f

0.6


Thermoplastic Composites Processing Stage 1: PREPREGGING

Fiber Prepregging Matrix Prepreg

CONSOLIDATION

Advanced Materials and Technologies Laboratory • University of Connecticut


Thermoplastic Tape Laying/Tow Placement

 Heat Transfer

(Temperature History)

 Compaction

(Void Growth/Reduction)

 Intimate Contact  Polymer Healing  Polymer Degradation  Polymer Crystallization

(Strength Development)

(Mechanical Properties)

Advanced Materials and Technologies Laboratory • University of Connecticut


Placement Head

Compaction Force Main Heater

Preheater

Advanced Materials and Technologies Laboratory • University of Connecticut


Process Simulations and Processing Runs Polymer Crystallization

Polymer Degradation

Statistical DOE was carried out on the fabrication of 8� dia. rings using AS4/PEEK prepregs. Products evaluated in regard to interlaminar shear strength (short beam shear test), void fraction, and microstructural quality

Advanced Materials and Technologies Laboratory • University of Connecticut


Process Design Considerations

Objective

Maximize Interlaminar Bond Strength Constraints  Void Content (at Roller 2 exit)

v vmax  Dimensional Change (Degree of Compaction) 

   hi  h f Dc    hi  

   Dc,max  

 Polymer Degradation (of mass [and properties] due to prolonged  exposure of polymer to high temperatures)

  max

Advanced Materials and Technologies Laboratory • University of Connecticut


Parametric Effects 1

0.6

Degree of Bonding, D b

T 1 = 700 oC; T 2 = 800 oC

Void Fraction at Roller Exit [%]

N

2 4 6 8

0.8

0.6

10 25

0.4

0.2

0 0

10

20

30

40

50

60

70

0.5 0.4 0.3 0.2 0.1

T 1 = 700 oC; T 2 = 800 oC

0 0

80

N 2 4 6 8 10 25

10

Line Speed, V [mm/s]

40

50

60

70

80

0.01 N

N

0.15

T1 = 700oC ; T = 800oC 2

25 10 8

0.13

Degree of Degradation, ď Ą

Degree of Compaction, D c

30

Line Speed, V [mm/s]

0.17

0.11 6

0.09 0.07

4 2

0.05 0.03 0.01

20

0

10

20

30

40

50

60

Line Speed, V [mm/s]

70

80

T1 = 700oC; T2 = 800oC

25 0.008 10 0.006 8 0.004 6 0.002

0 10

4 2 15

20

Line Speed, V [mm/s]

25


Processing Windows T 1 = 600 oC; T 2 = 700 oC

Line Speed, V [mm/s]

100

V op tuc based on Max. Strength V m in based on  Š 0.001 V based on v Š 0.2% m ax f V m in based on Dc Š 0.1

80

60 Optimum Line Speed B

40 Processing Window 20

D C A

0

1

5

9

13

17

21

25

Number of Layers T 1 = 600 oC; T 2 = 800 oC

100

T 1 = 600 oC; T 2 = 900 oC

100

V uc based on Max. Strength op t V m in based on  Š 0.001 V based on v Š 0.2% m ax f V based on D Š 0.1

based on Max. Strength V based on  Š 0.001 m in V based on D Š 0.1 m in c V based on v Š 0.2%

80

m ax

60

f

Optimum Line Speed B

40

Processing Window D

20

C

Line Speed, V [mm/s]

Line Speed, V [mm/s]

V uc op t

80

m in

c

Optimum Line Speed

60 B

Processing Window

40

D C

20 A

A

0

1

5

9

13

17

Number of Layers

21

25

0

1

5

9

13

17

Number of Layers

21

25


Product Quality

Advanced Materials and Technologies Laboratory • University of Connecticut


Fabrication of High Aspect Ratio Microstructures (HARMs)


Microcasting of Ceramic and Metallic Microparts Al2O3 Microgear

Fabrication of Ceramic and Metallic Microparts Slurry

Microchannel

Mold Top View

Micropart

Capillary-driven Mold Filling Microcast Slurry, (arrows indicate slurry flow) Cured and Planarized

Stainless Steel Microgear with nominally 8 mm teeth

Binder Removal/Sintering Demolded Composite Preform (Shrinkage)

Advanced Materials and Technologies Laboratory • University of Connecticut


Fuel Cells


Fuel Cell Performance and Related Issues  Measured in terms of a Cell voltage – Current density variation, referred to as a

Cell Voltage, V

polarization curve.

Decreasing Ecell

Average Current Density, I

 Local current density could be high and its spatial variation is a factor influencing membrane reliability; it is desirable to minimize the spatial variation.

 A second issue pertains to the system complexity associated with balance of plant; it is desirable to reduce system complexity.

Advanced Materials and Technologies Laboratory • University of Connecticut


Planar PEM Micro Fuel Cell (µPEMFC)  Traditional fuel cell designs are based on a ―sandwich‖ construction of stacking constituent layers; Micro and miniature fuel cell designs have also focused primarily on the layered design, while shrinking the size  Proposed concept is that of combining microfabrication techniques with fuel cell technologies to achieve compact, modular, and scalable designs for high power density  A unique feature of the mPEMFC design is that the cathode and the anode channels are patterned in a planer configuration. The design also eliminates gas diffusion layers; Design lends itself to fabrication of fuel cell arrays with associated circuitry on a single wafer

Advanced Materials and Technologies Laboratory • University of Connecticut


Fabrication of mPEMFC

1. Solution casting 40~50 mm Nafion® 500 mm Si Substrate

2. Hot pressing of microchannels Si Micro Die

Nafion® Si Substrate

3. Current collector deposition Au

Au mask

Top view of the cell

Substrate 4. Catalyst deposition Pt/C Catalyst mask Nafion

Substrate 5. Sealing with PDMS

PDMS H2

H+

O2

Substrate

Pattern Fidelity


mPEMFC Testing Cell dimensions: 25mm x 25mm x ~5mm Width of microchannels: 500mm       0.9

Single cell prototype Patterned Nafion® micro channels No gas diffusion layers H2 and air operation Concept also applicable to DMFC Active area ~0.33 mm2

80

0.8

Comparison with Peer Designs

70

0.7 60 0.6

V (V)

2

0.5

P (mW/cm )

50

40 0.4

V (V) P (mW/cm2)

30 0.3 20 0.2

10

0.1

0 0.00

42.67

85.33

128.00

218.66

0 253.33

I (m A/cm2 )

Advanced Materials and Technologies Laboratory • University of Connecticut


mPEMFC Arrays and Interdigitated Stacks

Parallel stack of 7 mPEMFCs

Series-parallel stack of 21 mPEMFCs

Advanced Materials and Technologies Laboratory • University of Connecticut


Acknowledgments  Funding  National Science Foundation  Office of Naval Research  Army Research Office  Air Force Office of Scientific Research

 Sandia National Laboratories  Students and Collaborators  C. Acquah, Q. Jiang, Dr. A. Mawardi, Dr. F. Yang, Dr. R.J. Johnson, B. Elolampi, Dr. A. Morales (Sandia)

Thank you! Advanced Materials and Technologies Laboratory • University of Connecticut


Functionally Graded Material Designs

Advanced Materials and Technologies Laboratory • University of Connecticut


Air Breathing Fuel Cells/Prototype Hybrid Fuel Cell/Battery System

Stack Voltage, EStack [V]

15.0

35 30 25

10.0

20 15 5.0

10

0.0 0

1

Voltage (V) Power (W)

5

3

0 4

2

Power, P [W]

Design specifications:  25 W avg. power  Power delivery at 12 VDC (regulated)  Hydrogen fuel, and air-cooled  Provide 1800 Wh of energy over 72 hrs

T = 25 oC, RHa = 0 %, H2 Pressure=5 psi

Current, I [A]

Fuel Cell

10.00

Power (W) 25

12 VDC

8.00

20

Voltage (V)

6.00

15

4.00 10

2.00 5.0

Long Term Test Five units (10 fuel cells), room temperature, flow rate = 0.75lpm 0.00

0.0

0

10

20

30

40

50

Time (Hrs)

Advanced Materials and Technologies Laboratory • University of Connecticut

Power (W)

Voltage (V)

DC/DC Converter

30


Corning Talk