Journal article 7 views
An In‐Line Machine Vision–Based Profilometry Tool for Non‐Destructive Thickness Assessment of Perovskite Films
,
Franklin Jaramillo
Advanced Electronic Materials, Volume: 12, Issue: 11, Start page: e00877
Swansea University Authors:
Ershad Parvazian, Trystan Watson
-
PDF | Version of Record
© 2026 The Author(s). This is an open access article under the terms of the Creative Commons Attribution License.
Download (2.16MB)
DOI (Published version): 10.1002/aelm.202500877
Abstract
Perovskite technology offers an advantage over conventional photovoltaics technologies related to its capacity to be manufactured on flexible substrates using continuous printing techniques, delivering an improved cost–benefit ratio. However, producing perovskite solar cells (PSCs) via these approac...
| Published in: | Advanced Electronic Materials |
|---|---|
| ISSN: | 2199-160X 2199-160X |
| Published: |
Wiley
2026
|
| Online Access: |
Check full text
|
| URI: | https://cronfa.swan.ac.uk/Record/cronfa72153 |
| Abstract: |
Perovskite technology offers an advantage over conventional photovoltaics technologies related to its capacity to be manufactured on flexible substrates using continuous printing techniques, delivering an improved cost–benefit ratio. However, producing perovskite solar cells (PSCs) via these approaches demands rigorous quality control of printed films, as film uniformity influences device performance. This quality control process often involves characterization techniques, such as profilometry that is destructive, making them unsuitable for in‐line real‐time or large‐scale inspections. Here, we present a versatile in‑line inspection framework that combines a low‑cost webcam‑based transmitted‑light imaging setup with a deep‑learning regression model to estimate the thickness of printed MAPbI 3 and FAPbI 3 films in real time. The resulting prototype achieves coefficients of determination (R 2 ) of 0.985 and 0.986 and root‑mean‑square errors of 18 and 25 nm, respectively which is comparable to conventional invasive profilometry. When applied to 25 cm 2 flexible mini‑modules, an inverse correlation between layer thickness variability and power conversion efficiency (PCE) was observed, and integrating the prototype into in‐line inspection opened new possibilities for data‐driven decision‐making. This in‐line, high‐spatial‐resolution inspection framework enables the fully automated and high‐throughput manufacturing of large‐area perovskite solar modules, by allowing real‐time optimization of key operational parameters. |
|---|---|
| Keywords: |
deep learning, In-line characterization, machine learning, machine vision, perovskite, roll-to-roll manufacturing, thickness inspection |
| College: |
Faculty of Science and Engineering |
| Funders: |
University of Antioquia; Colombian Government through the SGR project BPIN. Grant Number: 2022000100012; EPSRC Centre for Doctoral Training in Technology Enhanced Chemical Synthesis. Grant Number: EP/X025217/1. |
| Issue: |
11 |
| Start Page: |
e00877 |