Computational methods to support the development of sustainable processes for the synthesis of ammonia and other chemical products
Abstract/Contents
- Abstract
- The current processes for producing chemicals and fuels are unsustainable, and at the same time, renewable electricity sources are becoming increasingly inexpensive. For these reasons, there is a possibility that mounting political and economic driving forces will cause a major overhaul of the chemicals industry. Such a shift would require the identification of new catalyst materials for a wide variety of chemical processes. The challenge is that the space of all possible materials is intractably large to search through using physical experiments. Traditionally, to discover a new catalyst material, first, a fairly large set of candidate materials is identified by the research team based on a literature review and past experience. Second, high-throughput experiment is used to identify the top-performing materials among the candidates. Computational tools may be able to accelerate both of these two steps. Ab-initio computational models rely on some simplifying assumptions, and this can make it difficult for such models to identify the specific material that will perform optimally for a given application; however, the trends identified by ab-initio modeling are quite robust. This makes ab-initio models very useful for reducing the dimensionality of the search space and identifying regions of the search space that are likely to contain the best-performing materials. Data-driven statistical models fit to experimental data have the opposite set of advantages and disadvantages. The well-known "curse of dimensionality" refers to the fact that applying statistical models to high-dimensional search spaces is very challenging. However, once the search space is sufficiently low-dimensional, these statistical models can be paired with optimization schemes such as Bayesian optimization to find the exact top-performing material within the smaller search space very efficiently. It follows naturally that the materials discovery process could be greatly accelerated by using ab-initio modeling to reduce the dimensionality of the problem and identify promising regions of the search space and using statistical modeling to identify the best-performing materials within this reduced search space. In this thesis, density functional theory (DFT) is the ab-inito modeling method that is frequently used. We begin by explaining the assumptions that DFT makes, how the effects of the modeling errors are mitigated in our analysis, and how DFT can be used to decrease the dimensionality of problems in heterogeneous catalysis. First, we apply these methods to the thermochemical nitrogen reduction reaction in order to shed light on the problem posed by oxygen poisoning the industrial catalyst at ambient temperature. We then present DFT-based analysis that identifies a promising active site motif for the electrochemical nitrogen reduction reaction. We also shed light on the physics behind adsorbate-adsorbate interactions and its implications on selectivity in electrochemical nitrogen reduction. We then show how the same methods can be applied to electrochemical carbon dioxide reduction. First, we use microkinetic modeling to show the impact of two important variables on the selectivity of the reaction. Second, we show an example of a theory-experiment collaboration in which a DFT-based analysis leads to a strategy for the electrochemical synthesis of acetylene from carbon dioxide. Finally, we show how DFT and statistical models can be applied more broadly. We begin by showing how the same type of DFT analysis used in heterogeneous catalysis extends to homogeneous catalysis. Second, we show how statistical models can be paired with DFT-calculated results to reduce the computational cost of creating micro-kinetic models. Finally, we show how statistical models can be paired with physical experiments to reduce the number of experiments required to identify top-performing materials in a search space
Description
| Type of resource | text |
|---|---|
| Form | electronic resource; remote; computer; online resource |
| Extent | 1 online resource |
| Place | California |
| Place | [Stanford, California] |
| Publisher | [Stanford University] |
| Copyright date | 2020; ©2020 |
| Publication date | 2020; 2020 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Author | Rohr, Brian Andrew |
|---|---|
| Degree supervisor | Cargnello, Matteo |
| Degree supervisor | Noerskov, Jens |
| Thesis advisor | Cargnello, Matteo |
| Thesis advisor | Noerskov, Jens |
| Thesis advisor | Jaramillo, Thomas Francisco |
| Degree committee member | Jaramillo, Thomas Francisco |
| Associated with | Stanford University, Department of Chemical Engineering. |
Subjects
| Genre | Theses |
|---|---|
| Genre | Text |
Bibliographic information
| Statement of responsibility | Brian Andrew Rohr |
|---|---|
| Note | Submitted to the Department of Chemical Engineering |
| Thesis | Thesis Ph.D. Stanford University 2020 |
| Location | electronic resource |
Access conditions
- Copyright
- © 2020 by Brian Andrew Rohr
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...