Active pharmaceutical ingredients (APIs) are typically required to be sufficiently hydrophobic to pass through lipid membranes and ultimately reach their targets, yet aqueously soluble enough to dissolve in the body. Approximately 40% of potential API candidates are currently abandoned due to poor systemic exposure at an estimated industry wide cost in the hundreds of millions of dollars. Packaging the ingredient into an amorphous solid is an attractive option for increasing the solubility and bioavailability of APIs, though the higher free energy comes at a price; amorphous APIs are typically metastable and can spontaneously crystallize, which significantly alters the bioavailability. A common compromise is to crystallize the API into a high energy state to balance the drug stability and bioavailability, but designing and manufacturing high energy API crystals is challenging, since there are many variables that determine the final structural form upon API crystallization. Since the API crystal structure, distribution of crystal structures, and percent crystallinity of the final dosage form is critical to the bioavailability and shelf life, having a rapid measurement throughput for crystal screening is highly desirable. X-ray diffraction is the current gold standard for determining crystal structures, but is expensive and slow. Other higher speed measurements (e.g. Raman, brightfield, calorimetry, and others) can be used to discriminate among previously diffracted structures in the late stages of the drug design process. Second harmonic generation (SHG) microscopy has previously been shown to be sensitive to crystals of smaller than 1 micron, can discriminate different crystal structures, and can quickly determine percent crystallinity with many orders of magnitude higher sensitivity than the common methods of X-ray powder diffraction or calorimetry. In the work here, instrumentation and signal processing techniques were developed to increase the information content and measurement throughput of SHG microscopy. A high throughput (GByte/second), multi-channel, computerized platform and optical circuitry interfacial electronics were developed, with parallelized computer algorithms implemented capable of realtime analysis of the datastream. From the synchronous high speed analog to digital conversion datastream, access was available to the full distribution of photomultiplier voltage measurements. Through a statistical derivation, the relationship connecting the photon counting and signal averaging techniques was developed, and resulted in a high speed algorithm that recovered 90% to approaching 100% of the theoretical maximum Poisson signal/noise. The instrumentation was expanded on to perform high frame-rate imaging by scanning the beam in a Lissajous trajectory to achieve up to kHz regime frame-rate imaging.
PhD. Candidate, Analytical Chemistry Department / Purdue University