First Week of the Program Left to right: Paul Kigaya, Sarah Darbar, Marianna Marquardt, Sarah Miller, James Rush, Adrian Macias, Brandon Cohen, Eliza Ballantyne
Dinner in Paris
2023 Projects
Eliza Ballantyne: Exploring Nanorod and Liquid Crystal Samples with Polarized Light
Whiteboard Video:
The goal of my project this summer was to use polarized light to explore nanorod and liquid crystal samples. I worked with Jeff Audibert and Vitor Brasiliense in the PPSM lab at École normale supérieure Paris-Saclay. I started by working with the nanorod samples. Because they fall below the diffraction limit, these nanorods appear as tiny dots instead of thin rods in our dark field images and we can’t determine their orientation with visible light. We can get around this limitation with a Scanning Electron Microscope, but it would be costly and time consuming. Instead, we tackled the problem by tracking how the nanorods responded to polarized light. We used a birefringent crystal to split the beam of light collected from the sample into twin s and p images, light parallel and perpendicular to the optical axis. By adjusting the polarization of the light entering the sample, we could measure the intensity of light as a function of the angle of a polarizer. Overall, the s and p images followed a predictable, sinusoidal pattern. Dust or large agglomerates of nanorods were made of enough particles that the intensity was constant throughout the cycle, but single nanorods flickered as we varied the orientation of the polarization of the incident light. If a nanorod were oriented more in the s or p direction, its response to rotating the polarizer before the sample would be much more dramatic than other parts of the sample. By measuring the intensity of each nanorod for many different polarizer orientations, we can get the same orientation information we would have gotten from the SEM, only much faster. Towards the end of my internship we worked on comparing results from a scanning electron microscope with our predictions made using polarized light. Taking on the optical characterization efforts, I also measured the complete Mueller matrices of nanorod and liquid crystal samples. Neither of the samples is expected to generate circularly polarized light, but this analysis prepares us for future experiments with more complex, chiral shapes. By placing polarizers and quarter wave plates before and after the sample and observing how the intensity of our sample changed, we could predict how the polarization of light would change by going through the sample. I empirically measured background Mueller matrices, the Mueller matrix of our system of lenses, mirrors, and condensers, as well as Mueller matrices for individual nanorods and liquid crystals. In future work, it would be interesting to compare the Mueller matrix of liquid crystals when it interacts with a laser controlled by a spatial light modulator. Throughout my internship, I practiced aligning optical systems, doing calibration measurements such as determining magnification of our set up or checking the stability of a laser, and using code in MATLAB and python to interface between cameras and control boards to automate data collection. I wrote algorithms to analyze the images we collected pixel by pixel and produce Mueller matrices or polar graphs representing the pixel's response to changing polarized light. It was really cool this summer to see concepts I’ve learned about in my undergraduate classes applied to real world problems. I gained a lot of confidence working with different optical components and I especially appreciate the chance I had to learn from my mentors and other students.
Brandon Cohen: Implementation of Random Illumination Microscope
Whiteboard Video: This summer I worked at Université Paris-Saclay under the guidance of Guillaume Dupuis and Sandrine Lévêque-Fort implementing a random illumination microscope (RIM). RIM is a new and exciting microscope due to its simplicity of design, its ability to resolve thicker samples, and, most importantly, its ability to achieve super-resolution. With the use of RIM, we are able to achieve super-resolution by doubling the resolution meaning we can see organelles of cells that are past the diffraction limit such as mitochondria. The basic idea of RIM is to illuminate the sample using random patterns which in turn will cause the sample to give off random speckle images which can be put into an algorithm for reconstruction. In order to reap the benefits of RIM, my project involved installing a spatial light modulator (SLM), learning the software that communicates with the SLM, designing the microscope, synchronizing the camera and SLM, and optimizing the algorithm for reconstruction. I began this long journey by unboxing and building the SLM, an integral part of the setup that gives rise to random speckle due to the images presented on the surface of the SLM. Afterwards, I started learning how the SLM’s software worked and how to build .REPZ files which are the files that the SLM’s hardware reads. The .REPZ file was very complicated in the beginning as it is composed of three parts: sequences, images that you want to be displayed on the SLM, and running orders. After getting acquainted with the many manuals of the SLM, I learned how each part worked together to produce images on the SLM and I started building small .REPZ files. The next major part of the project was removing the previous set-up on the optical table. This was very fun and enlightening because, by the time I was done with this, I felt like I had a good grasp on the many parts that make up an optical design. To start my set-up, I first built the ~561 nm laser stand and screwed it into the table. I then added mirrors, a diverging lens, and a converging lens which allowed the collimated beam to hit the SLM. The next part of the design was adding 2 converging lenses to create a real Fourier Plane. The last part of the design was using a mirror to direct the collimated beam into the back focal plane, but before it entered the back focal plane, the zero-order was blocked due to the only information encoded in the zero-order was noise. I also calibrated the microscope with the excitation beam to optimize the angle of unmodulated illumination. After fully building the microscope, I added the camera. I then developed a Python script that created .PNG files with different grid sizes where each grid was either white (1) or black (0). The .PNG files were very reminiscent of QR codes especially when the grid sizes were very big. When I started sending images to the SLM, I experienced one of my first major problems because the camera only showed a Fourier Plane which was not expected. To solve the problem, we realized that the length of the beam was traveling too far so we redesigned the set-up to shorten the distance. The next major problem came with the synchronization of the camera and the SLM. In order to solve this problem, creativity was required as I found a library called pyautogui to press the buttons on the screen to take the pictures and change the images on the SLM. Once I solved these problems, I started taking images and sending the stack of images to AlgoRIM which is the algorithm that reconstructs the images. Lastly, I tested grid sizes of 1x1, 2x2, 4x4, 8x8, 16x16, 32x32, 64x64, and 128x128 to see which grid sizes give the best-reconstructed images. I found the best images came from the smallest grid sizes with the best reconstructed images coming from 2x2 and 4x4 grids. This experience in Paris has been quite an illuminating one as I have learned so much about optics as well as French culture. I will definitely continue to use the skills I learned during my time in France. Lastly, I would like to thank Guillaume and Sandrine for opening their lab up to me to experience the world of SUPER RESOLUTION. Top
Sarah Darbar: Numerical Simulation of Attosecond Electron Wavepacket Interferometry
Whiteboard Video:
The shortest pulses of light ever produced have durations that are on the order of a few tens of attoseconds (1 as = 10−18s) and are created from high-order harmonic generation. When exposed to a gas, attosecond light pulses will ionize atoms/molecules and free photoelectrons which behave quantum-mechanically. During the summer, I worked with Gabriel Granveau and Dr. Charles Bourassin-Bouchet in the Charles Fabry Laboratory of the Institut d’Optique Graduate School to develop numerical and analytical tools to characterize the propagation of these photoelectron wavepackets. We were particularly interested in modeling their propagation based on a Time of Flight (ToF) experiment where the photoelectrons transverse 2 meters from initial ionization to detection at a microchannel plate. In this sort of experiment, a time varying voltage can be applied during the wavepacket’s propagation to produce matter-wave interference phenomena. It is unknown, however, if the quantum signatures of these interference effects are measurable at the detector which is a classical distance away. Therefore, it became my job to model the wavepacket’s propagation in the ToF experiment to better understand this inquiry. Many challenges presented themselves when trying to accomplish this. Firstly, we were unsure if it would be more useful to use classical/semi-classical approximation methods rather than treating the problem in a fully quantum-mechanical way. I initially started out with numerically simulating the wavepacket’s spread in time using the stationary phase approximation at the beginning of my internship. This could be characterized as a classical approach to the problem since it purely concerned itself with the classical trajectory of the wavepacket’s components. Although this turned out to be a promising method, I moved onto a semi-classical technique to investigate if more detailed information about the problem could be gained. In this way, I next analytically calculated the wavepacket’s spread by modeling the potential voltage and the photoelectron’s energy and temporal profiles as Gaussian distributions. This turned out to be too approximative of a method, so I concluded my research experience with modeling the wavepacket’s propagation with the sudden approximation, a fully quantum mechanical approach. Both the analytical and numerical calculations of the wavefunction’s evolution was rigorous and tedious in this methodology, but the sudden approximation was undoubtedly a more exact treatment of the problem compared to previous attempts. I would like to conclude by thanking Jin, Charles and Gabriel for not only being such amazing colleagues but for becoming my friends as well. I would also like to thank the University of Michigan and the National Science Foundation for giving me this wonderful opportunity to study abroad during the summer. I have had so much fun doing cool science as well as exploring France!
James Rush: Realization of plasmonic structures in polymeric thin film by holographically assisted polymeric templates
Whiteboard Video: This summer, I worked in the LUMIN laboratory with Professor Diep Lai and his two PhD students Long Ngo and Truong Pham at Ecole Normale Superieure Paris- Saclay. The goal of my project was to investigate and eventually realize the creation of plasmonic structures on thin film. This meant that I would be working at this goal by breaking it down in ‘chunks’. First, I would have to show that gratings could be created on my polymer (DR1-PMMA) through the method put forward by my professor, and fully investigate the limitations of said technique. To this end, I found moderate success. I was able to successfully simulate the structures I could make by creating my own python code and following the theory presented in previous reports in this area. I was also able to create several different photonic crystals on this polymer, with some of these being of better quality than those created in the past. Having done that, I began examining how metallic nanoparticles might behave in polymer. To do that, I had to learn about the Finite Difference Time Domain method that is commonly used to solve maxwell’s equations. I used Lumerical’s implementation of this algorithm, and watched tons of instructional videos made by them, and read much of the documentation given on their website. I also had to learn the theory of Mie Scattering, which was necessary to simulate the behavior of spherically shaped nanoparticles efficiently and accurately. Through this method, I was able to find how the absorption spectra of metallic nanoparticles change with particle size, and the refractive indices of nanoparticles and their background material. With that done, I moved back into the lab to further characterize the polymer. I learned to ‘corona pole’ the polymer, which aligned the molecular dipoles in a small region of the polymer by heating it and applying a strong electric field. After I did this, I was able to observe the second harmonic generation inside the polymer, and observe how the gratings I fabricated changed the second harmonic response of the material. I also investigated how light created by second harmonic generation and reabsorbed by the sample changed it over time and discovered a near sinusoidal oscillation in the power transmitted through the sample over time. If I had more time with this project, I would have connected the two parts of my experiment and created nanostructures using nanoparticle-infused polymer. But even though I didn’t get that far in the project, I’m still glad that I took part in it because I learned a ton, met plenty of interesting people, and had a lot of fun along the way. This experience has definitely opened my eyes to how fascinating optics can be as a branch of science, and how much it can be applied to other fields in and out of physics.
Adrian Marcais: Low One Photon Absorption Direct Laser Writing
Whiteboard Video: This summer I worked in the Lumiére, Matiére et Interfaces (LuMIn) Laboratory at École Normale Supérieure Paris-Saclay under the supervision of Professor Ngoc Diep Lai, PhD student Gia Long, and Masters student Pham Quang Truong. I worked towards fabricating a CPC structure on a sample of spin coated DR1-PMMA using Low One Photon Absorption Direct Laser Writing (LOPA-DLW). The CPC structure was to be used to manipulate the properties of a laser beam when shone onto the structure, namely the beam shape. The samples were prepared by spin coating a layer of DR1-PMMA onto a thin glass substrate. The LOPA-DLW set up consisted of an array of mirrors, lenses, and filters that focused a 532 nm laser beam onto a small point only a few micrometers thick. The sample is then placed on a Piezoelectric Translator (PZT), where the light is focused onto the sample, which “pushes” the DR1-PMMA away from the focusing spot via the mass transfer effect. The PZT moves the sample according to instructions supplied via Lab View, this creates a structure with the shape of the laser beams path across the sample.
To fabricate a photonic structure, the optimal parameters for that structure must be found. Given a specific periodicity, I had to find the parameters (such as the laser power and PZT velocity) that produced the best structures. I began by fabricating a 1D structure of lines 10 micrometers long with a period of 2 micrometers. After a few attempts, I found the optimal parameters of the fabrication to be a laser power of 0.8 mW, and a PZT velocity of 3 micrometers per second. I then moved onto the fabrication of a 2D pillar array. First, I varied the periodicity and laser power while holding the PZT velocity constant. This was done so I could determine what period I should work with first, then work my way towards smaller periods. I found the structure with a period of 4 micrometers to be the best fabrication and began working my way towards a period of 2 micrometers. In the end, I found the optimal parameters for a 2D pillar array with a period of 2 micrometers to be a velocity of 2 micrometers per second, and a power of 1 mW.
Lastly, I fabricated the CPC structure. The CPC structure is fabricated a little differently than the rest of the structures. Instead of the laser being continuously focused on the sample while the PZT moves, it is instead focused on a single spot, then the beam is blocked by a shutter. During this time the PZT moves the sample, then the shutter reopens allowing the beam to shine on a new spot. The PZT moves in a way such that it creates a spiral of “dots” around the starting point. The dots produced by the laser are so big that they overlap with one another creating a continuous spiral. The parameters for this fabrication are the step size – the distance between two consecutive points – and laser power. After a few fabrications, I found the optimal parameters for the CPC structure with a radius of 40 to be a laser power of 3 mW and a step size of 0.4 micrometers.
Unfortunately, due to time constraints, I wasn’t able to achieve the beam shaping. However, I still learned a tremendous amount during my time at ENS Paris-Saclay and would like to thank Professor Lai, Mr. T, and Long for being patient, helpful and welcoming to me.
Paul Kigaya: Advanced Optimization of Laser Wakefield Acceleration
Whiteboard Video:
During the Summer of 2023, I came to Paris to partake in an internship at Universite Paris- Saclay under the supervision of Francesco Massimo and Brigitte Cros. During this internship, I worked on Optimizing a physical process known as Laser Wakefield Acceleration using computational methods. At first, I started by understanding the physical concepts that made Laser Wakefield Acceleration work, and then after I learned how to create simulations to mimic this process with Python using open-source code such as the Particle in Cell code SMILEI, and scripts created by Francesco Massimo and I. By simulating the experiment, I was able to explore the parameters of the physical process and hence optimize it to achieve a desired outcome. Throughout this internship, I have learned various computational methods and physical processes that have furthered my understanding and will help me in my future career. Laser Wakefield acceleration is a technique at the interface of the fields of particle plasma physics and particle accelerator physics. The technique involves using an intense laser pulse in a plasma to create a plasma wave in its wake, whose field is called a Wakefield. The longitudinal Wakefields can then be used to accelerate charged particles, such as electrons, to extremely high energies over a very short distance. The purpose of this is to create an electron beam whose electrons are uniform in that they have a low deviation between each other and move in the same direction. To optimize laser Wakefield acceleration researchers have implemented various computational methods through simulations to obtain cost-effective predictions and results. The technique used for these simulations is the Particle in Cell (PIC). The PIC method uses a numerical method to solve problems involving the motion of charged particles within an electromagnetic field. Three optimization techniques used this summer were Bayesian Optimization, Particle Swarm Optimization, and Random Scan Optimization. Bayesian Optimization is a type of numerical computational method based on Bayes theorem used to optimize black-box functions which are functions that are observed entirely based on their input and output. Particle Swarm Optimization (PSO) is a computational optimization technique inspired by the behavior of swarms or flocks in nature. Imagine that each bird in a flock looking for food above a 100m^2 field represented a particle. The algorithm then uses the exploration of these particles (birds) in each space (100m^2 field) to obtain an optimal solution for a given problem (finding food). The Random Scan technique, as the title states, is a technique based on the PIC method which randomly selects points within given intervals and predicts where the optimal solutions are based on previous points. Once obtaining optimal points from all three techniques we used an unknown function (delta) to further optimize our electron beam. We then plotted this function on a graph and referred to the width and peak of the figures to determine whether our beam was efficient enough. In conclusion, we agreed that to give a concrete optimum for our electron beam multiple simulations (10-20 runs) with many iterations would need to be completed.
Top Sarah Miller: Electroluminescence of a 2D Semiconductor using STM
Whiteboard Video:
This summer I worked on a project at the Photophysique et photochimie supramoléculaires et macromoléculaires (PPSM) laboratory with Jeff Audibert and Vitor Brasiliense at ENS Paris-Saclay. My project focused on building an optical set-up that allows switching between a confocal path and a TIRF path for writing and reading patterns on a surface. I conducted research at ISMO CRNS - University of Paris-Sacaly in Eric LeMoal’s lab on the electroluminescence of a two-dimensional (2D) semiconductor in a scanning tunneling microscope. The motivation of the research is a combination of fundamental research and engineering. We have compared electroluminescence, which is not fully understood, to photoluminescence in monolayer tungsten diselenide (WSe2), a 2D semiconductor that belongs to the class of transition metal dichalcogenides (TMDs). Monolayer TMDs have strong light-matter interaction as a result of direct band gap emission and exhibit excitonic effects at room temperature. Due to these properties, they have been studied for optoelectronic uses such as photodetectors, LEDs, and other novel quantum devices. Investigating these properties also serves a huge fundamental interest to understand the excitonic dynamics relating to the excitation and emission of light. Photoluminescence is the emission of light that is induced by the absorption of light. This has been well-studied in TMDs. Electroluminescence is the emission of light as a result of electronic excitation, such as electric current or electric field. Both types of luminescence occur as a result of radiative recombination of excitons, which are quasiparticles that consist of an electron and hole, held together by strong Coulomb interactions. The excitation mechanism for the electroluminescence of monolayer TMDs is still a matter of debate and may be a result of a combination of energy transfer, and electrical charge carrier injection into electronic bands, resulting in the creation of excitons (and other excitonic species). Previous work in the lab has demonstrated scanning tunneling microscopy (STM) coupled to an optical microscope as a method to induce electroluminescence and probe the excitonic and optoelectronic properties of 2D semiconductors (in MoSe2 and WS2). When STM is coupled to an optical microscope, one is able to locally probe excitons, resolving the luminescence both spatially and angularly. The goal of this internship was to compare the excitation and emission mechanism of STM-induced electroluminescence to photoluminescence in WSe2 to work towards creating a model of excitation. We collected data on the spectra, Fourier plane, and photon count while varying the tunneling current and bias voltage. Additionally, we repeated the experiment on a sample that has been annealed, to investigate the role of defects in the material. Ultimately, we were able to create a preliminary model of electroluminescence in WSe2, based on our findings, which was very exciting and fulfilling. I want to express my gratitude to Eric and the entire NanoPhys group at ISMO for being so welcoming and for all of their guidance and support. I truly had an amazing experience this summer.
Top Marianna Marquardt: Towards Local Probe Microscopy at Femtosecond Timescale
Whiteboard Video:
My project is called Towards Local Probe Microscopy at Femtosecond Timescale. The main goal of this project is to combine local scanning-tunneling microscopy (STM) with ultrafast vibrational sum frequency generation (SFG) to probe dynamic information at a femtosecond time resolution and nanometer length scale. Scanning tunneling microscopy is a way to learn about the 3D topography of a sample down to the atomic scale. When a metallic tip approaches a sample surface in the nanometer range, and a voltage difference is applied between them, a quantum mechanical process occurs called electron tunneling. The current depends exponentially on the tip-surface distance, which gives high sensitivity to topography. By maintaining a constant current of tunneling electrons and scanning the tip across the surface, we can recover the topography. The second method, SFG, is a non-linear optical process that gives access to the vibrational properties of molecules only located at surfaces. SFG combines two laser pulses that overlap spatially and temporally. The result is the emission of light at the sum frequency of the two incoming lasers. To learn about the ultrafast mechanism occurring in the molecule, we perform a pump-probe experiment. We use a third laser pulse, called the pump, which perturbs the molecular system and we probe the vibrational dynamic by SFG with a 100 femtosecond time resolution. However, SFG is limited to micrometer resolution, and STM doesn’t give optical information. So, the goal of the project is to benefit from the nanometer scale resolution of the STM and the femtosecond resolution of SFG. I followed along with many of these experiments throughout my internship this summer, looking at self-assembled monolayers of organic molecules and ordered 2D arrays of semiconductor quantum dots samples. To bring these two experiments together, I worked to design and implement an optical system to bring femtosecond pulses into the STM. It was also necessary to enclose the laser. Infrared lasers for SFG have high absorbances in humid air, so I designed boxes and tubes to enclose the laser and optics in and run dry air through for experiments. I did this using 3D modeling software such as Fusion 360 and TinkerCad. There are multiple other experiments that happen in the same area in the lab, so it was necessary to route the laser around existing equipment. I also implemented the optical setup inside the ultra-high vacuum STM chamber to focus the beam down onto the sample and collect the resulting SFG light to analyze. This design was especially challenging because it required precision as we cannot easily remove the STM from the ultrahigh vacuum to make modifications. Separately, I was also involved in using a modeling software called COMSOL to analyze the local amplification of the electric field under the STM tip when doing STM-SFG experiments. We see high local amplification in the region just underneath the STM tip, which enhances resolution. This project will be picked up this fall by a Ph.D. student with the ultimate goal of running simultaneous STM and SFG experiments on octadecanethiol samples! Thank you to the University of Michigan and Institut Polytechnique for setting up this incredible program!