Thesis
Application of computer vision in industrially important reactions
- Creator
- Rights statement
- Awarding institution
- University of Strathclyde
- Date of award
- 2022
- Thesis identifier
- T16386
- Person Identifier (Local)
- 201953378
- Qualification Level
- Qualification Name
- Department, School or Faculty
- Abstract
- As an emerging digital technology, computer vision is already widely applied across a range of sectors in our life. In this thesis, computer vision methods are developed for and applied to kinetic analysis of chemical transformations of immediate relevance to challenges in industrial catalysis, and amide bond formation. All computer vision-enabled colour analysis reported herein is achieved using a program (Kineticolor) that is currently being developed within the Reid Research Group. Pd Catalysis - With the aid of computer vision, quantifiable information on catalyst colour kinetics, degradation and activation, is accessible without defaulting to more traditional analytical methods. In the more complex Miyaura borylation chemistries, compelling correlations (R2 ≥ 0.8) were found between colour values and the concentration of the desired borylation product. These promising results demonstrate the possibility of determining the success or otherwise of a reaction based solely on the information derived from camera footage. Amide Bond Formation - Similarly, it is found that the amide bond formation reactions mediated by selected coupling reagents exhibit moderate to high correlation between the colour values and off-line HPLC analysis. Further investigations of solid-phase peptide synthesis indicate the possible usefulness of this computer vision technique in optimising the coupling time, which can potentially realize the self-optimisation of solid-phase peptide synthesis. Dual real-time monitoring and colour data analysis - In the simple crystal violet experiment, comparable results were found between the computer vision analysis and in situ UV-Vis monitoring, giving excellent linear relationships. In addition, most colour parameters were proven to be reliable and reproducible by statistical analysis. Furthermore, a prediction model based on colour analysis can be achieved with the help of machine learning. Overall, this computer vision monitoring method can provide reliable colour analysis reflecting the reaction kinetics, presenting a great potential to expand the toolbox of non-invasive analytical techniques.
- Advisor / supervisor
- Reid, Marc
- Jamieson, Craig
- Resource Type
- Note
- This thesis was previously held under moratorium from 10/10/22 until 10/10/24.
- DOI
- Embargo Note
Relations
Items
| Thumbnail | Title | Date Uploaded | Visibility | Actions |
|---|---|---|---|---|
|
|
File | 2022-10-26 | Embargo |