DuPont Home
DuPont Photovoltaic Solutions

DuPont Delivers Technical Topics at IEEE Photovoltaic Specialists Conference

Abstract: Determination of Moisture Ingress through Various Encapsulants in Glass/Glass Laminates

By Jane Kapur, Kristof Proost, C. Anthony Smith
Wednesday, June 10 at 10:00am, Poster Session, Area 7: Modules and System Components, #484

» Download White Paper 

Encapsulant materials can provide protection and electrical isolation of the solar components in photovoltaic (PV) modules from the environment. However, some photovoltaic devices are sensitive to low levels of moisture and the ingress of water into a module can decrease its performance significantly during the lifetime of a module.  In glass/glass PV modules, the moisture penetrates through the encapsulant to the module’s metal components and degradation can occur. In this study, we have developed and validated methods to determine moisture ingress in situ in a laminate. Results indicate that water permeability and equilibrium moisture level (which are temperature dependent) can both affect the corrosion of the metal. We have measured the moisture ingress through a developmental encapsulant material from the glass edge towards the center by an in-situ Fourier transform infrared (FTIR) spectroscopy technique after damp-heat exposure. The FTIR measurements were performed on glass / encapsulant / glass laminates that were weathered at various times at elevated temperatures and humidity. The moisture level in the encapsulant can be determined by integration of the IR band between 1880 and 1990 nm. This peak surface was compared to a calibration curve, which was obtained using laminates with known encapsulant moisture levels (determined by Karl- Fischer titration). The moisture migration through an encapsulant material from the edge was also measured using an ASTM D7191 moisture analysis method [1]. The measurements were made on Al foil/encapsulant/Al foil laminates that were exposed to 85°C and 85%RH (damp heat). The experimental data from both methods were well-correlated using a Fickian diffusion model.