Author ORCID Identifier



Date of Graduation


Document Type


Degree Type



Eberly College of Arts and Sciences


Forensic and Investigative Science

Committee Chair

Luis Arroyo

Committee Member

Tatiana Trejos

Committee Member

Casper Venter

Committee Member

Edward Sisco


Stimulant drugs comprise one of the top drug categories abused in the United States. Due to its accessibility, low price, and manufacturing simplicity, methamphetamine is frequently placed within the top 10 seized drugs in the country. As of March 2023, methamphetamine is the most seized controlled substance in the United States, with 34,291 kg. In 2022, the United States seized over 79,000 kg of methamphetamine. One reason for the proliferation of methamphetamine is related to the production itself, which does not require large warehouses but can be manufactured in houses using relatively accessible materials and small containers. When a clandestine laboratory is investigated, law enforcement and CSIs must be able to identify what drug is being produced and what hazards are associated with the production method being utilized by the clandestine laboratory.

Due to shifting manufacturing routes by underground chemists, it has become difficult for forensic scientists to identify illicit substances and their respective precursors reliably. Indeed, rapid analytical tools that facilitate the identification of legal and scheduled drugs are highly desirable for first responders, health personnel, and forensic chemists. This thesis addresses this deficit by investigating Raman Spectroscopy and Direct Analysis in Real-Time Mass Spectrometry (DART-MS) to determine ways to improve mixture identification. This research focuses on methamphetamine and its precursors, ephedrine, and pseudoephedrine. However, the scope was expanded to include several other drugs and cutting agents of concern in the United States.

This research compared three Portable Raman instrumentations for detecting methamphetamine and its precursors in binary mixtures. From a practical perspective, the TacticID GP from B&W Tek (Newark, DE) and the Mira XTR DS from Metrohm USA (Riverview, FL) were determined to be suitable for on-site detection due to their simple operation and color-coded results that provide immediate safety information for the results, in case the user is not familiar with the compound. The mixture analysis function allowed for better identification of the controlled substance due to the controlled substance being the minor component in most cases. The iRaman Prime from B&W Tek (Newark, DE) had limitations for the mixtures. Software used to compare the collected spectrum to the library does not include the mixture analysis function to help identify complex samples. There are other software present; however, the software requires an additional understanding of statistical analysis that first responders may not be equipped with.

This research also sought to improve Raman’s detection of mixtures using machine learning, specifically convolutional neural networks or CNN. The iRaman Prime from B&W Tek (Newark, DE) was used for this purpose, which had the most difficulty identifying mixtures due to the built-in software available. Using CNN, the ability to identify both components in the mixtures improved to 94.0 % compared to 71.5 % using cosine similarity. However, the algorithm had difficulty identifying the drugs and adulterants in the authentic samples. The difficulty is because the authentic samples consisted of more complex samples, with more than two compounds present. Further research can be done to train the algorithm with more complex samples or include the class of compounds to give an overall result for compounds not in the training set.

Lastly, the utility of DART-MS was investigated for methamphetamines. The Data Interpretation Tool (DIT) v. 2 from NIJ/NIST was used to see how well DART-MS could identify multiple components in mixtures. Authentic samples from the Maryland State Police Forensic Sciences Division were used as more complex samples to compare these instruments with more realistic ones. The DIT and DART-MS identified 82.5 % of the binary mixtures. The DIT also successfully identified at least one controlled substance in the samples containing controlled substances. This thesis demonstrates that the combination of Raman Spectroscopy with CNN and DART-MS with DIT improves their respective instrument's ability to detect mixtures, making them better equipped for use in clandestine operations and regular forensic casework.