Towards developing a predictive approach to assessing electron beam instability during Transmission Electron Microscopy of drug molecules

During drug development control of polymorphism, particle properties and impurities are critical to ensuring a good quality, reproducible and safe medicine. A wide variety of analytical techniques are employed in demonstrating to regulators control over the drug substance and product manufacture, storage and supply. Transmission electron microscopy (TEM) offers the opportunity to analyse in detail pharmaceutical systems at a length scale and limit of detection not readily achieved by many traditional techniques. However, the use of TEM as a characterization tool for drug development is uncommon due to possible damage caused by the electron beam. This work outlines the development of a model, using molecular descriptors, to predict the electron beam stability of active pharmaceutical ingredients (API). For a given set of conditions and a particular imaging or analytical mode, the total number of electrons per unit area which causes observable damage to a sample in the TEM can be defined as the critical fluence (CF). Here the CF of 20 poorly water-soluble APIs were measured using selected area electron diffraction. Principal component analysis was used to select the most influential molecular descriptors on CF, which were shown to be descriptors involving the degree of conjugation, the number of hydrogen bond donors and acceptors and the number of rotatable bonds. These were used to generate several multiple linear regression models. The model that provided the best fit to the measured CF data included the ratio of the number of conjugated carbons to non-conjugated carbons, the ratio of the number of hydrogen bond donors to acceptors and the ratio of the number of hydrogen bond acceptors to donors. Using this model the CF of the majority of compounds was predicted within ±2 e-/Ų. Molecules with no hydrogen bond acceptors did not fit the model accurately possibly due to the limited sample size or the influence of other parameters not included in this model, such as intermolecular bond energies. The model presented can be used to support pharmaceutical development by quickly assessing the stability of other poorly soluble drugs in TEM. Provided that the model suggests the API is relatively stable to electron irradiation, TEM offers the prospect of determining the presence of crystalline material at low levels at length scales and limits of detection unobtainable by other techniques. This is particularly so for amorphous solid dispersions