ePubs

The open archive for STFC research publications

Full Record Details

Persistent URL http://purl.org/net/epubs/work/67430253
Record Status Checked
Record Id 67430253
Title Towards Rationalising the Structure - Property Relationships of Energetic Materials
Contributors
Abstract An energetic material is a substance containing a large amount of stored chemical energy that can be released quickly upon initiation, for example, an explosive, propellant or pyrotechnic. This initiation event can occur via several methods, including from imparted shock, friction or spark, and so a material’s sensitivity to those stimuli is an important safety metric to consider. Of particular importance is a material’s response to a mechanical impact, called its impact sensitivity, as accidental initiation during storage or transport would likely be via this mechanism. Experimental measurement of impact sensitivity uses a drop weight test, the outcome of which is affected by variables including temperature, sample purity and grain size, and are also subject to the decision of the person carrying out the test. Therefore, the ability to predict impact sensitivity, for example, by a computational method, would be very powerful. Previous work in the group has allowed for predictive methods for impact sensitivity based on the vibrational up-pumping model, and using density functional theory (DFT), to be developed. The vibrational up-pumping model is a physics-based description of how energetic impact initiation could occur in the short time- and length-scale regime, wherein mechanical impact energy is absorbed into the low energy (lattice) vibrational modes, or phonon modes. After equilibration, this energy is transferred to higher energy vibrational modes by phonon-phonon coupling, causing bond excitation and breakage, therefore leading to the decomposition of the material and release of stored energy. The two-phonon density of states can then be calculated from the vibrational spectrum; this is a measure of how many phonon-phonon coupling combinations can occur based on the density of vibrational modes in the low energy phonon region. This is integrated over the modes which can couple to absorb the transfer of energy, which can predict how sensitive the material is to impact. This method has been implemented to predict the impact sensitivity of a range of different materials, including molecular crystals, salts, co-crystals and coordination polymers, where the crystal structure of the material is known. Any model designed for prediction, by definition, connects the material’s structure to the property being predicted, which in this case is impact sensitivity. This has been extremely effective, however, the necessary step forward, particularly in the field of energetic materials, is to be able to use the model in reverse, i.e., to be able to predict the design for a structure with a given property. This would provide a practical design tool for safer synthesis of new energetic materials and is the over-arching aim of this thesis. Chapter 3 of this thesis explores implementing this method in a new regime for a previously studied material, acting to showcase the predictive power of the vibrational up-pumping model. The Chapter describes prediction of changes to the sensitivity of pentaerythritol tetranitrate (PETN) when it is subjected to hydrostatic pressures up to 11 GPa. It was observed experimentally that under pressure, the sensitivity of PETN increased, so computational studies have been carried out at varying pressures to determine the cause of this increased sensitivity. Investigations include measuring changes in molecular structure, crystal structure and the vibrational density of states as potential causes, which show that changes in the density of states lead to a peak in sensitivity of the material at around 4 - 5 GPa, similar to where initiation of the samples occurred experimentally. While the vibrational up-pumping method has been successfully applied in new predictive ways as shown in Chapter 3, it is limited by two main issues. Chapter 4 of this thesis introduces an alternative method, again utilising the crystal structure of the material, addressing the first of these issues with the vibrational up-pumping method, which is the high-performance computer time required to carry out these calculations. In the methodology previously implemented in the group, some detail in the vibrational spectra is lost by only sampling at the gamma point of the Brillouin zone during vibrational frequency calculations. This is because sampling at multiple points quickly becomes very expensive using DFT. Interatomic potentials based on DFT calculations but trained using machine learning models have become more widespread in recent years. This Chapter describes the use of one such potential, MACE, and its application to calculating vibrational density of states for the full BZ, which can be done in just a few hours. This provides an avenue to quicker, and just as accurate, predictions of impact sensitivity, as investigated in this Chapter. Chapter 5 of this thesis addresses the second of these issues, namely the requirement of knowledge of the crystal structure of the material. In the field of energetic materials this is potentially dangerous, since crystal structure must be measured experimentally, requiring the synthesis of the new material to be carried out without knowledge of its sensitivity. This Chapter explores how a method could be developed to predict sensitivity which removes the need to know the material’s crystal structure. The method introduced here investigates the use of classification models, due to the variability of experimental data. Machine learning methods have additionally been employed to allow the important structural features of the molecule which affect the material’s sensitivity to be determined. In summary, this thesis explores new methods and variations on existing methodologies to predict impact sensitivity of energetic molecular crystals, which are designed to provide an insight into how the structure of these materials influences their properties. These methodologies provide routes to design new materials; by identifying the structural features which most greatly affect the sensitivity of the material, a design tool can be developed which can produce new molecules with desired sensitivity properties.
Organisation ISIS , ISIS-PEARL , ISIS-TOSCA , STFC
Keywords
Funding Information
Related Research Object(s):
Licence Information:
Language English (EN)
Type Details URI(s) Local file(s) Year
Thesis PhD, University of Edinburgh, 2026. doi:10.7488/era/7454 2026