Research

Exoplanets & Planetary Science

Dynamical Origins of Short-Period Giants

2021-Present

Since the discovery of 51 Pegasi b in 1995, exoplanet scientists have been puzzled by the existence of gas giants on close-in orbits known as hot Jupiters. Various observational properties can be used to constrain their origins, such as their system architectures, and eccentricity and stellar obliquity distributions. Dedicated observational campagins over the past two decades have revealed that hot Jupiters are often isolated, lacking nearby planetary companions, and have large eccentricities and spin-orbit misalignments consistent with high-eccentricity migration triggered by violent dynamical evolution. However, a growing sample of hot Jupiters have been discovered to host nearby companions on low-mutual inclination and aligned orbits, suggesting that they have arrived at their current orbits via more quiescent mechanisms such as migration through the protoplanetary disk or perhaps in situ formation.

Curiously, the population of wider-orbiting counterparts to hot Jupiters, known as warm Jupiters, appear to have properties consistent with quiescent origins, and their somewhat less massive counterparts, known as hot sub-Saturns, may have properties intermediate to hot and warm Jupiters. In this context, my PhD dissertation research aims to investigate the origins of short-period giant planets through a combination of observational and numerical studies, which are detailed below.

Stellar Obliquity Measurements Via The Rossiter-McLaughlin Effect

Stellar obliquity, or the angle between the stellar spin axis and the orbit normal axis of its planet(s), is a unqiue and informative indicator of a system's dynamical history. Despite decades of effort, it remains unclear whether the primary mechanism driving spin-orbit misalignment is a universal process that takes place during formation with the protoplanetary disk present or an evolution-specific process that occurs in post-disk phase, perhaps due to the violent dynamics inherent to high-eccentricty migration. In the latter case, misalignment may largely be confined hot Jupiters and hot Saturns. Compact multi-planet systems offer a unique opportunity to differentiate between these competing hypotheses, as their tightly-packed configurations preclude violent dynamical histories, including high-eccentricity migration, allowing them to trace the primordial disk plane.

The Rossiter-McLaughlin (RM) effect (Holt 1893, Rossiter 1924, McLaughlin 1924), which takes advantage of the time-resolved spectral distortion from an actively transiting or eclipsing body, presents a reliable method to determine a system's projected spin-orbit angle. With additional information on the stellar inclination, the true 3D stellar obliquity can be derived as well.

TOI-5143 (Radzom et al. 2024b, in prep.)

Our NEID RM measurements of TOI-5126 and TOI-5398.

TOI-5143 is a hot Jupiter in a compact multi-planet system recently discovered by NASA's Transiting Exoplanet Survey Satellite (TESS) mission and confirmed by Quinn et al., in prep., and is only the third of such systems to have its stellar obliquity measured. We employ Bayesian Inference and the Doppler Shadow technique to measure its stellar obliquity via the RM effect, finding it is well-aligned. Our result continues the trend of alignemnt seen in previous spin-orbit measurements for other hot Jupiters with nearby companions (e.g., WASP-47 b and WASP-84 b), as well as the current census of RM measurements for compact multi-planet systems. This reinforces the hypothesis that planetary systems are largely primordially aligned, with misalignments being acquired in the post-disk phase due to dynamical evolution. Further, our measurement demonstrates that some hot Jupiters must have quiescent origins, having avoided achieving the large eccentricities that would have otherwise destroyed their companions and/or excited their obliquities during their path to their close-in orbits.

TOI-5126 and TOI-5398 (Radzom et al. 2024 AJ 168 116)

Sub-Saturns, with masses intermediate of Neptune and Jupiter, represent an interesting population to study both the origins of giant planets and the mechanisms driving misalignment. Using the radial velocity anomaly of the RM effect, we measured the stellar obliquity for two close-in sub-Saturns in compact multiple-transiting systems: TOI-5126 b and TOI-5398 b. Both are spin-orbit aligned, joining a fast-growing group of just three other compact sub-Saturn systems, all of which exhibit spin-orbit alignment. Our obliquity measurements allow the first statistical analyses to be run on the stellar obliquities of close-in sub-Saturns, and we perform a random shuffling procedure to verify that the preferential alignment of compact sub-Saturns is statistically significant (to 2.6σ). Our results strongly suggest that sub-Saturn systems are primordially aligned and become misaligned largely in the post-disk phase, as appears to be the case increasingly for other exoplanet populations.

N-body Simulations of Short-Period Giant Systems

In light of recent companion searches that have revealed a number of hot Jupiters with nearby companions (e.g., WASP-47 b; Becker et al. 2015, Kepler-730 b; Canas et al. 2019, TOI-1130 c; Huang et al. 2020, WASP-148 b; Hebrard et al. 2020, TOI-2494 b, TOI-5143 c; Quinn et al., in prep., WASP-132 b; Hord et al. 2022) and TTV signal searches (Wu et al. 2023), it is plausible that hot Jupiters in compact configurations are more common than we thought.

To this end, I am leading a project to test the viability of quiescent formation pathways to produce hot Jupiters, such as disk migration and in situ formation, using N-body simulations. In this work, we study and describe the dynamical evolution of compact multi-planet systems containing a close-in gas giants at a range of orbital periods. We find that quiescent formation pathways for such giants naturally reproduces several known observational properties, including the higher nearby companion rate for warm Jupiters compared to hot Jupiters. This is illustrated qualiatively in the figure on the right, which shows the period ratios of our simulated Jupiters with their nearest low-mass companion after 10 Myr of dynamical evolution, where we see that hot giants are much more isolated than warm giants. Based on resultant orbital architectures, we posit that a large fraction of these companions to hot Jupiters are too distant and/or exhibit too high of mutual inclinations to be detected efficiently, causing hot Jupiters to appear more lonely than they truly are. Our paper (Radzom et al.) has been submitted to ApJ.

THUNDER: A Titan orbiter mission concept for NASA's New Frontiers program

As part of NASA's 2023 Planetary Science Summer School (PSSS), I co-developed THUNDER (Titan's Hydrocarbons: Unveiling New Dimensions of Evolutionary pRocesses), a New-Frontiers-class orbiter mission concept to Saturn's largest moon, Titan. As Science Objective Lead, I took charge of one of the mission's three science objectives: To determine whether Titan's major liquid hydrocarbon bodies are connected and exchanging material with each other through a subsurface reservoir. I also acted as Deputy PI and Science Chair on the concept.

Galaxy Evolution & AGN

X-ray Sources in the SSA22 Field

2019-2022

Image Credit: NASA/Goddard Space Flight Center Conceptual Image Lab

Supermassive black holes lie at the center of most galaxies, and many actively accrete matter to become Active Galactic Nuclei (AGNs). Due to the presence of surrounding gas, optical campaigns to detect AGNs are biased to unobscured, or broad-line AGNs (BLAGNs). Observations in the X-ray (2-8 keV) with Chandra's ACIS instrument can rectify this and provide an unbiased probe of these objects, thanks to the high column density-penetrating nature of hard X-rays. Combined with tools like the luminosity function, astronomers have learned that AGNs are a strongly evolving population that have been active much more recently in cosmic history than was previously thought. Further, obscured or non-BLAGNs tend to make up the faint end of the X-ray luminosity function (XLF) while BLAGNs dominate the bright end at higher redshift.

The focus of my undergraduate astronomy research was investigating the differential evolution of these two AGN types with newly obtained high-resolution spectroscopy from the Deep Imaging Multi-Object Spectrograph on Keck. I led a project with collaborators Anthony J. Taylor and Dr. Amy Barger to combine these spectra with existing X-ray and optical photometry in the Hawaii Survey Field SSA22, which is one of the deepest and most well-studied Chandra fields (thanks to the famous protocluster at z=3.09). In total, we bring the completeness of the field to 62% by updating spectroscopic redshifts for nearly half of the available X-ray sources, and derive optical classifications for most from our data (these are roughly encoded in the luminosity distribution on the left).

After applying an incompleteness correction and computing photometric redshifts for objects without redshifts, we constructed the XLF in three redshift intervals for ~160 AGN candidates in the field across z=0-4 (see figure below). We found good agreement with previous works, with the XLF shifting the higher X-ray luminosity with increasing redshift. Additionally, we remove all candidate members of the z=3.09 galactic protocluster and reconstruct the XLF to demonstrate that this overdensity did not significantly affect our result. We also use our classifications to study the XLF for BLAGNs and non-BLAGNs separately, which reveals that the unobscured BLAGNs become increasingly important at higher redshift, peaking in number density near z=2. As a consequence of this, the faint-end slope of our AGN XLF flattens with redshift. While we find that both BLAGNs and non-BLAGNs evolve in luminosity and density with increasing redshift, non-BLAGNs exhibit the strongest evolution and peak in number density near z=1. The XLF of these obscured AGNs tend to have a lower break luminosity and higher normalization across z=1-4. This work (Radzom et al. 2022) was published in ApJ in November 2022 (doi:10.3847/1538-4357/ac9bfe).

The XLF of AGNs (black squares) plotted against comparable results from Aird et al. (2015; A15).

The XLF of BLAGNs (purple triangles) and non-BLAGNs (green circles) plotted against our full AGN XLF results (black squares).

Engineering Applications to Quantum Computation

Polarization Alignment in Birefringent Optical Fibers

2018-2020

My first research experience was in Dr. Saffman's lab at UW - Madison, which focused on approaches to quantum computing using neutral atoms. The most substantial but rewarding project I worked on was developing a device and an associated procedure to align laser beam polarization to the optical fiber axis in birefringent fibers. Single-mode fibers are ubiquitous in optics labs (thanks to their ability to reduce system complexity), but are subject to a variety of environmental stresses (e.g., bending, heating). These stresses induce unwanted drifts in the polarization state of the light traveling through the fiber over time. Such misalignment renders the large optical set-ups needed for neutral atom quantum experiments useless until realignment is achieved, as polarization is a fundamental parameter in quantum logic. Birefringent polarization-maintaining (PM) fibers are able to mitigate these effects and maintain the input linear polarization state due to the presence of two orthogonal stress rods with differing refractive indicies internal to the fiber. The difference of these refractive indicies is known as birefringence. How this birefringence can be used to maintain polarization alignment is complicated (and fascinating!), but is broadly explained by the fact that birefringence induces a well-defined phase delay for the two orthogonal states over the length of the fiber.

The alignment provided by these birefringent fibers is good but not perfect, and is further complicated by the fact that these fibers are often connectorized, featuring a key-hole output at one end. If not carefully executed, the process to attach (usually glue) this key-hole onto the fiber can create additional sources of stress which substantially reduce the effectiveness of the birefringent design. This misalignment can be corrected for by once again inducing a well-defined perturbation onto the fiber. Several methods exist for this purpose, including the famous elasto-optic (E-O) method, but many are too complicated or carry too high of a potential to damage the fiber to be practical.

In collaboration with Dr. Matthew Ebert, Dr. Trent Graham, and Dr. Mark Saffman, I designed and implemented a novel polarization alignment scheme which is safe, systematic, and simple to perform. Briefly, we fabricate a controllable signal that is proportional to the polarization misalignment and we manipulate the light using a polarimeter until alignment is achieved. More specifically, we attach the fiber to a polarimeter that outputs a real-time 'differential signal' which is proportional to the difference in the power between the two orthogonal polarization modes (but normalized to the total power). This voltage signal yields the polarization angle of the light incident to the fiber (via a cosine relation). Since the output angle is dependent on the input laser's frequency (due to the fiber's birefringence), it becomes possible to decouple the power from the frequency using a frequency ramping procedure. This is because once the frequency scan begins, a periodic 'spike'-like signal is output by the polarimeter. One can flatten this signal via iterative adjustment of the input and output polarization (using the polarimeter) while also ensuring that the sum of the power in each polarization mode is maximal. A flat signal indicates that the power difference between both modes is no longer frequency-dependent (i.e., the polarization state is decoupled fro the frequency). Polarization alignment is guarenteed to be maintained throughout the fiber in this case since it is directly proportional to the now-flattened differential signal, while maximal transmission is retained. I built the polarimeter using laboratory optics and custom-machined parts, but any polarimeter which separates the orthogonal components of linearly polarized light will work for this procedure (see Figures to the left).

As one can see, the perturbation required for our alignment procedure is simply a frequency scan, which presents no real stress on the fiber. Additionally, the method is repeatable and easy to perform as the frequency ramping is over some well-defined range and all of the necessary optical components are readily present in most laboratories. We also found that introducing a quarter waveplate (QWP) greatly reduced the polarization drift over time, which we quantified by running a series of experiments. A schematic of our optical set-up prior to feeding the light into the polarimeter is shown on the right.

In order test the alignment stability and sensitivity associated with this method, we induce an external thermal stress on an OZ Optics PM fiber and measure the resultant polarization drift. Particularly, we used a make-shift oven to heat the PM fiber roughly 5°C over 15 minutes and we measure the mean and the variance of the alignment drift during the experiment. We repeat this for several angular offsets from nominal alignment (±0.25,±0.5,±1, and ±2 degrees) to show that the drift is minimized when our frequency-ramping procedure is carried out. We find that our procedure is capable of aligning the polarization axes to within 0.5 degrees without the QWP included along the beam path and 0.25 degrees with it. In addition to these experiments, we also ran test to show that:

(1) OZ Optics PM fibers maintain polarization better than equivalent Thorlabs fibers
(2) the E-O method is not as effective for a given fiber
(3) longer fibers (up to 10 meters) show lower overall drift
(4) higher beam coupling (>50%) minimizes drift
(5) equivalent results are obtained regardless of the alignment axis (i.e., slow or fast)

The primary "best-case" results of this project are represented in the figure on the left, which displays the mean drift and variance as a function of offset angle for the 10-meter OZ PM fiber experiments both with (blue x) and without (blue sqaures) the QWP. These results are currently drafted in a manuscript which may be submitted to the Review of Scientific Instruments journal, and the full procedure is written up on our lab's wiki page.

Temperature Regulation System for an Atomic Trapping Chamber

2017-2018

The first-ever research project I worked on was also in Dr. Saffman's lab at UW - Madison, but was dedicated to a specific experiment. Particularly, I designed and implemented a high-precision temperature control system for our new Rubidium atomic trapping chamber. But first, what is an atomic trapping chamber and why is temperature regulation important? In order to build a quantum computer using neutral atoms, ensembles of neutral atoms must be encoded and trapped (via laser cooling) in special excited states known as W-sates. Maintaining this array of neutral atom quantum bits or qubits results in an optical lattice that can serve as the basis for a quantum computer. If reliably trapped, one can perform quantum logic operations by exciting the atoms to a high-lying (n~100) so-called Rydberg state where they can undergo coherent interactions. However, great care must be taken to mitigate the rate of logical errors (i.e., incorrect atomic states) for this trap to hold. Theoretically, the potential fidelity (error rate) for two-qubit entanglement via Rydberg interactions exceeds >0.9999, but previous attempts by any laboratory have only reached 0.8. Maximizing the fidelity of the initial W-state preparation is possible by either reducing system noise, minimizing polarization crosstalk (i.e., polarization drift, see above project), or otherwise improving the stability of atomic cooling in the trap.

My work for this project addressed the first of these points, for which I was charged with creating a temperature regulation system for our group's newest atomic trap. The environmental perturbations caused by large temperature fluctuations (>1°C) throughout the previous fiberoptic system essentially made our optically trapped atoms moving targets. Therefore, I first developed a see-through prototype of a temperature-stable chamber which involved three ultra low-noise water-cooled fans (seen in the picture on the top right). Proof-of-concept tests with this prototype showed promise, so I drafted a full plan for a negative feedback PID temperature loop. I ordered the appropriate electronics (PID controller, fuses, heat sink, temperature probes, etc.) and machined an electronics box I designed on SolidWorks. After careful assembly, I ran another set of preliminary tests with the system to optimize the set-up, and integrated real-time temperature readout and control via lab computers. Finally, I installed the temperature regulation system into the new atomic chamber. The system is capable of consistently maintaining ~mK-level stability within the chamber, far exceeding initial requirements and expectations and representing an improvement of over two orders of mangitude relative to the previous ambient trap. The lab still uses this system to this day for Rubidium experiments.