Thin-Film Manufacturing & Product Operation
ModelingDAVIS HEMENWAY
DIRECT-ENGINEERING AND COLORADO STATE UNIVERSITY
Outline
Manufacturing Overview Processing Hardware
Modeling Approach Design Improvements and Results
Continuing Work
Product Prototype Operation Design Evaluation and Results Conclusion
Manufacturing Overview
Thin-film CdTe Photovoltaic (PV) device
TCO front contact
CdS window layer
CdTe absorber layer
Metallic back contact
Manufacturing Overview
In-line deposition process
Multiple processing stations
3 x 3” PV cell created in under one hour
Processing Hardware Process stations are graphite crucibles Sublimation used to deposit material
Benefits: High material utilization Moderate temperatures and vacuum
levels required High quality films produced rapidly
Challenges: Film uniformity driven by thermal
uniformity and hardware geometry Deposition hardware costly to machine
Modeling ApproachThin-Film Processing Hardware
Fluid volume within the source modeled initially Multi-zone mesh created with 160k elements
Modeling ApproachThin-Film Processing Hardware
Modeling Difficulties: Deposition and condensation surface chemistry Low pressure physics must be accounted for at 40mTorr Flow at walls require special consideration
Sublimation and condensation are the two dominant reactions that take placeArrhenius rate equation used by Fluent
Sc: Sticking coefficient Calculated after experiments
A: Pre-exponential factor Calculated for each reaction
EA: Activation energy
β: Temperature exponentR: Universal gas constantT: Temperature
Modeling ApproachThin-Film Processing Hardware
Modeling ResultsThin-Film Processing Hardware
Cd gas molar fraction in the pocket Cd growth rate on the substrate (kg/m2s)
Vapor distribution and film uniformity can be analyzed
Modeling ResultsThin-Film Processing Hardware
Simulation-based engineering analysis provides otherwise unobtainable insight
Flow lines colored by Cd molar fraction
Experimental ValidationThin-Film Processing Hardware
Results validated by comparing modeled and deposited film thicknesses
Scanning White Light Interferometry
Sticking coefficient applied from initial experiments
Experimental ResultsThin-Film Processing Hardware
Modeled film thickness correlates strongly with experimental results
Validated model used to improve new source design before production
Hardware Design Improvement
Model used to predict film growth Same equations and boundary conditions Different geometry
Improved film uniformity Deeper pocket Shallower wells Gen 1 Gen 2
1st Generation
2nd Generation
Hardware Redesign Results
The model predicts that the 2nd Generation source should produce more uniform films
1st Generation
Contours of CdS film thickness: Each line represents a 1% change in thickness
2nd Generation
Hardware Redesign Results
Film uniformity experimentally matches predicted values Uniformity improved by over 70% with one design iteration
1st Generation 2nd Generation
Continuing Work
Modeling different thin-film material evaporation processes
CdS
CdTe
CuCl
CdCl2
Deposition Rates in (nm/s)
Continuing Work
Deposition system thermal performance Shielding and temperature control optimization
Thin Film Product Operation
New thin-film PV module design:
Designed for UV and moisture resistance
No lamination or batching required
Small factory footprint Patent pending
Source: Nordson.com
Prototype Architecture
Two panes of custom made glass 1200 x 600 x 3.2mm each
2+ encapsulating polymers with additives
Air gap between glass panes
3μm thick semiconductor film
Top Glass
Bottom Glass
Silicone PIB Low cost polymer / desiccant
CdTe Film
Desiccated gap
X-section of module edge
Modeling ApproachThin Film Product Operation
Over 3 million elements used
Convection boundary conditions obtained from 2D model
Film represented as surface
Wind velocity(m/s)
Modeling ApproachThin Film Product Operation
Radiation heat transfer must be considered
Real world solar spectrum used
Unique, wavelength-based quantum efficiency of the device accounted for
Prototype and industry-standard devices modeled and compared
Numerous convection and radiation conditions queried
Operating matrix created to analyze thermal response trends
Modeling ResultsThin Film Product Operation
Film temperature (K)
Experimental ResultsThin Film Product Operation
Both devices thermal response observed
Real-world solar and wind conditions measured at nearby station
Experimental conditions input as boundary conditions for model
Experimental results match modeled values
Conclusion
Method for modeling thin-film processing demonstrated
Method can be used to improve hardware before manufacturing
Thin-film product operation in real-world conditions modeled
Simulation is valuable for thin-film processing and product design before manufacturing
Questions?