Publications

PhD Dissertation (Available here)

Title: Super-Resolution Imaging, Orientation Analysis and Image Processing with Machine Learning in Sum Frequency Generation Spectroscopy. Summary: I have focused on three main areas: super-resolution SFG imaging, molecular orientation analysis of a rubrene surface, and the application of machine learning for improved chemical image analysis.

• Super-resolution SFG Microscopy: A significant portion of this work is dedicated to the development and implementation of a super-resolution SFG microscopy technique. This method utilizes the principle of ground state depletion (GSD), where a donut-shaped pump mid-infrared (IR) beam is employed to suppress the SFG process in the peripheral regions of the probe IR focal spot. This effectively reduces the signal-emitting area, leading to an enhanced spatial resolution beyond the diffraction limit. Simulations supported the feasibility of this approach. The developed raster scanning microscope, incorporating a vortex-phase-plate to shape the pump IR beam, demonstrated a resolution improvement of more than 3 times in a proof-of-concept experiment on a hydrogen-terminated silicon surface. The dissertation suggests that further refinements in optics and detection techniques could lead to true sub-micron resolution.

• SFG Studies of Rubrene Single-Crystal Surface: The surface specificity and chemical sensitivity of SFG spectroscopy were utilized to analyse the orientation of molecules on a rubrene single-crystal. Spectroscopic measurements were performed under different azimuthal orientations and polarisation combinations. Spectral curve fitting and orientation analysis indicated a 45° tilt from the surface normal for the phenyl rings of the molecules on the rubrene surface. The study also observed evidence of electronically resonant SFG, suggesting complex pathways due to the visible and SF photon energies matching prominent electronic transitions in rubrene.

• Neural Networks for Chemical Image Analysis: The dissertation explores the application of neural networks (NNs), a machine learning approach, to improve chemical assignment for pixels in SFG imaging. Trained through supervised learning on SFG spectra with known chemical labels from self-assembled monolayers (SAMs), the NNs were deployed to predict chemical labels for new pixel spectra. Tests revealed accuracy above 90% under high spectral noise conditions, demonstrating the potential to overcome the pixel binning requirements traditionally encountered in spectral fitting for chemical analysis, thereby preserving spatial resolution. The NN-generated chemical maps showed qualitative agreement with results from Target Factor Analysis (TFA), suggesting these techniques can be used to corroborate each other.

B.S. Thesis  (Available here)

Title: Vibrating Sample Magnetometery: Analysis and Construction

Summary: In this project, I designed, constructed, and tested a Vibrating Sample Magnetometer (VSM) to measure the magnetic properties of materials. My goal was to build a functional, low-cost, and precise system that could be used for research and instructional purposes.

I began by exploring the theoretical foundations of magnetism, studying various types of magnetic materials and comparing existing magnetometry techniques. I chose the VSM due to its robustness and accuracy in capturing magnetization, especially in ferromagnetic samples. To create the vibration mechanism, I used a PASCO mechanical wave driver and measured the resulting vibrations with a Linear Variable Differential Transformer (LVDT). I then developed a detailed mathematical model of the system using transfer functions and Bode plots to analyze its frequency response and optimize performance.

For the detection mechanism, I modeled the magnetic field generated by a vibrating sample and designed detection coils that would best capture the resulting flux changes. I carefully selected the geometry and positioning of the coils to maximize sensitivity based on electromagnetic field gradients. After constructing the full system—which included the vibration mechanism, detection coils, an electromagnet, and a LabView-based data acquisition interface—I tested it using a steel ball bearing and an iron strip as sample materials. I was able to successfully measure magnetization and observe clear hysteresis behavior.

The VSM I developed worked well for strongly magnetic materials, capturing key magnetic features such as saturation and coercivity. However, the device’s sensitivity to weakly magnetic samples was limited, suggesting that future improvements—such as integrating a lock-in amplifier—could significantly enhance its performance. Overall, this project provided me with a comprehensive understanding of magnetic measurements and hands-on experience in building and characterizing a working magnetometer from scratch.