Status : Verified
| Personal Name | Imperial, Bruce Brandon C. |
|---|---|
| Resource Title | Estimation of persistent organic pollutants (PBDE/PFOS) in the Philippine semiconductor industry using artificial neural networks |
| Date Issued | 17 December 2025 |
| Abstract | Persistent organic pollutants (POPs) are substances that are resistant to degradation and bioaccumulate in living organisms and the environment. Semiconductors are among the products that still contain trace amounts of POPs, specifically perfluorooctane sulfonate (PFOS) and polybrominated diphenyl ethers (PBDEs). This study aims to estimate PBDE/PFOS emissions from semiconductor exports of the Philippines from 2025 to 2034 using an artificial neural network coded in MATLAB. Semiconductor imports, producer price, competitive industrial performance index, volume of net sales and production indices, plant capacity utilization, production turnover, and GDP from 2006 to 2024 were used to train, validate, and optimize the artificial neural network. Neural networking training determined that the optimum setup to utilize is 0.4 as the learning rate and 0.5 as the momentum coefficient. Comparison of the training output to the actual historical sales in the Philippines from 2006 to 2024 demonstrated high accuracy. Semiconductor exports were predicted to increase in 2025 before taking a downward trend and flattening in 2030 to 2034. PBDE and PFOS emissions initially followed the same trend as unit production, before the implementation of stricter thresholds on POPs result in large decrease of estimated emissions, particularly in PFOS. It is recommended to collect data more specific to the semiconductor industry, adding more relevant input factors, investigating other data normalization methods, and further optimization of the MATLAB neural network in order to further improve the accuracy and reliability of the results. |
| Degree Course | MS Environmental Engineering |
| Language | English |
| Keyword | semiconductors, persistent organic pollutants, artificial neural networks, MATLAB |
| Material Type | Thesis/Dissertation |
Preliminary Pages
685.76 Kb
Category : F - Regular work, i.e., it has no patentable invention or creation, the author does not wish for personal publication, there is no confidential information.
Access Permission : Open Access