Scott Egbert, PhD

A Quick Introduction

Welcome! I’m Scott, a research scientist at IMRA America’s Boulder Research Lab in Longmont, Colorado. I develop stabilized lasers and precision optical systems, combining optical references, electro-optic modulation, RF electronics, and controls. I especially enjoy taking demanding R&D systems, understanding what limits their performance, and turning them into compact hardware that other people can use.

At IMRA, my work centers on developing compact dual-wavelength Brillouin lasers (DWBLs) and transferring millimeter-wave stability to other frequencies of interest. So far, I have integrated these low-noise optical references with electro-optic and Kerr frequency-division systems spanning microwave, millimeter-wave, and terahertz frequencies. I enjoy bringing a systems approach to my research, balancing optical performance, thermal behavior, packaging, controls, and stability for the intended application.

Before joining IMRA, I earned my PhD in Mechanical Engineering at CU Boulder in Greg Rieker’s Precision Laser Diagnostics Lab, where I developed portable frequency-comb spectrometers and high-temperature molecular absorption databases. Explore my work at IMRA and PhD research page for examples of my work, or take a look at my resume. I’m always happy to connect with people working on interesting optical systems.

My Work at IMRA

Research Scientist · Boulder Research Lab

I develop stabilized optical references and the systems that transfer their spectral purity to microwave and millimeter-wave domains. My work spans laser stabilization, modulation, frequency division, phase-noise metrology, and deployable packaging. A central thread has been converting IMRA’s dual-wavelength Brillouin laser (DWBL) from a laboratory-scale reference into portable hardware that our collaborators can successfully operate.

Why develop millimeter-wave sources?

Commercial millimeter-wave generation often starts with a lower-frequency oscillator and multiplies it upward. Even ideal frequency multiplication raises phase noise by 20log₁₀(N), while amplifiers, mixers, and other stages can further increase noise. Maintaining spectral purity becomes increasingly difficult as the carrier frequency rises, which presents an additional barrier to leveraging the benefits of millimeter-waves.

DWBLs provide a photonic alternative. Two telecom lasers generate correlated Stokes waves in a shared Brillouin fiber cavity, so common-mode cavity and fiber noise is strongly suppressed in the optical frequency difference between the two laser tones. A high-speed photodiode can convert that spacing directly into a millimeter-wave signal, or frequency-comb techniques can divide it to lower frequencies, improving the stability by the same 20log₁₀(N) and facilitating the development of millimeter-wave technology.

The examples below show several ways our group and collaborators have used the low noise of DWBLs in our research.

Wireless communication

High-capacity wireless links are one promising application for low-noise photonic millimeter-wave sources. The constellation diagrams below are from a 300 GHz signal transmitted over a 214-meter link.

Molecular frequency references

For applications that require long-term frequency stability of a millimeter-wave carrier, the DWBL spacing can be referenced to a molecular rotational transition. This example compares a free-running DWBL with one stabilized to carbonyl sulfide (OCS) absorption at 316 GHz.

Resonant tunneling diode injection

A DWBL can also transfer its spectral purity to another oscillator. Injection locking a resonant tunneling diode (RTD) combines the low-noise optical reference with the RTD’s much higher millimeter-wave output power.

Building a portable Brillouin laser

I led the redesign and ruggedization of IMRA’s passively stabilized DWBL, reducing its volume from 570 to 15 liters while improving its phase-noise performance. Much of the work involved testing which laboratory-scale components could be simplified or removed while preserving the system-level phase noise and long-term stability.

The portable source keeps the same basic optical architecture as the laboratory system. Two pump diodes drive a shared Brillouin fiber cavity, and optical feedback from the Stokes waves stabilizes the pumps. The common cavity strongly suppresses differential noise between the two optical tones, which is what gives the millimeter-wave difference frequency its low phase noise.

The original system relied on a large vacuum-insulated fiber spool for thermal isolation. I replaced it with compact nested metal enclosures and a noise-canceling spool design. Short-term phase-noise performance was preserved, but the smaller package made long-term drift from HVAC and diurnal temperature changes more visible. We used an EO-comb heterodyne beat at MHz frequencies to track the DWBL spacing and feed that signal back through the existing thermal controls.

Measuring the source’s phase noise

A major struggle when improving the DWBL was characterizing the phase-noise. A conventional two-source comparison measures the combined noise of both references. Not only do you not know which source has the cleaner signal at any given Fourier frequency, the noisier source can mask an improvement made to the companion device.

Our team implemented the first millimeter-wave cross-correlator to isolate the phase noise of one DWBL at a time. Using two other DWBL as independent local oscillators, the cross correlation suppresses noise that is uncorrelated between the channels. With this technique can isolate the phase noise of one source, measure changes directly, and iterate much faster. The compact source now reaches below −160 dBc/Hz at 1 MHz offset on a 260 GHz carrier.

Transferring optical stability to new frequencies

300 GHz through Kerr optical frequency division

We integrated a compact 3.3 THz DWBL prototype with a Kerr microresonator oscillator, dividing the optical frequency separation to the 300 GHz resonance of the resonator and improving the phase noise by the predicted 20log₁₀(11), achieving 135 attosecond RMS timing jitter from 1 kHz to 1 MHz [5]. The results have been accepted in Nature Photonics for publication.

Low-noise X-band synthesis

We also implemented and characterized an electro-optic frequency-division scheme that used an EO comb to divide the DWBL spacing to baseband with a microwave synthesizer. Feed-forward noise cancellation then suppressed the synthesizer phase noise without requiring a feedback loop. The system produced tunable 8–16 GHz output with up to 30 dB lower phase noise than the original synthesizer [6].

Deployment and user handoff

We developed a control GUI, diagnostics, and standard operating procedures for the portable DWBL systems so collaborators could operate and troubleshoot the hardware independently. The systems are now deployed with collaborators in North America, Europe, and Asia.

For the frequency-comb and spectroscopy work that led me here, see my PhD research. You can also view my resume or get in touch.

References

1. K. Maekawa et al., “Single-carrier 220-Gbit/s sub-THz wireless transmission over 214 m using a photonics-based system,” Optics Letters (2024).
2. J. Greenberg et al., “Dual wavelength Brillouin laser terahertz source stabilized to carbonyl sulfide rotational transition,” Nature Communications (2025).
3. J. Greenberg et al., “Terahertz amplification by injection locking of waveguide resonant tunneling diode,” IEEE Journal of Quantum Electronics (2026).
4. B. M. Heffernan et al., “Brillouin laser-driven terahertz oscillator up to 3 THz with femtosecond-level timing jitter,” Nature Photonics (2024).
5. S. C. Egbert et al., “Attosecond-timing millimeter waves via Kerr optical frequency division,” accepted in Nature Photonics (2026).
6. J. Greenberg, S. C. Egbert et al., “Tunable microwave frequency synthesis with optically-derived spectral purity,” Nature Communications (2026).

PhD Research at CU Boulder

In Greg Rieker’s Precision Laser Diagnostics Lab at the University of Colorado Boulder, I developed frequency-comb spectroscopy systems for high-speed and high-temperature gas measurements, along with the molecular databases needed to interpret them. The work combined stabilized fiber combs, nonlinear optics, free-space beam delivery, opto-mechanics, FPGA-based phase correction, and scientific software.

The examples below focus on two parts of that work: building a portable mid-infrared dual-comb spectrometer and improving high-temperature H₂O spectroscopy for HITRAN.

Building a Portable Mid-IR Dual-Comb Spectrometer

Working closely with Nazanin Hoghooghi, Scott Diddams, and Peter Chang at NIST Boulder, I developed and integrated a GHz-repetition-rate mid-infrared dual-comb spectrometer for high-speed, multi-species absorption measurements. In our first shock-tube experiments, we resolved formaldehyde formation and decomposition into carbon monoxide on microsecond time scales.

After the initial shock-tube measurements, I led the optical and mechanical redesign of the laboratory system into our first portable mid-infrared dual-comb spectrometer, reducing the footprint by more than 80% for rack deployment.

Each channel begins with a stabilized near-infrared fiber comb and amplifier chain. We condition the femtosecond pulses in free space and use intra-pulse difference-frequency generation (IP-DFG) in PPLN to produce broadband 3–5 μm light. The two mid-infrared combs are then combined and delivered to the experiment. My SolidWorks redesign packaged the fiber hardware, nonlinear conversion, and free-space optics into rack-mountable modules while preserving the alignment and stability needed for dual-comb measurements. I also hired, trained, and mentored three interns who supported the redesign and deployment effort.

Improving Spectral Databases

Accurate spectral databases are just as important as the spectrometer itself. I collected 58 high-resolution H₂O datasets spanning 300–1300 K to improve near-infrared absorption models used for combustion diagnostics and other high-temperature measurements. Dual-comb spectroscopy was especially useful because each measurement captured nearly 200,000 spectral points over a broad bandwidth while still resolving individual absorption features.

The improved measurements exposed a different bottleneck: analysis. I had increased the measured bandwidth by roughly 3× while reducing noise by about 10×, but the community Fortran fitting workflow was difficult to scale to the dense H₂O spectrum. The example below shows the original interface I was working from.

To make that analysis tractable, I developed an object-oriented Python library and GUI around the trusted Fortran solver. The workflow reduced processing time from months to about two weeks and cut database errors from 23% to 1%. I ultimately processed 24,118 parameters across more than 6,000 transitions; the resulting measurements and fits contributed to HITRAN2024. A small sample of the database improvements is shown below.

This work shaped how I approach optical systems today: the laser hardware, controls, metrology, and analysis tools all have to be designed around the measurement you ultimately want to make.

Connect

Please reach out if you would like to talk about stabilized lasers, frequency combs, phase-noise metrology, optical systems, or related work. My email is scottegbert219@gmail.com.

Elements

i = 0;

while (!deck.isInOrder()) {
    print 'Iteration ' + i;
    deck.shuffle();
    i++;
}

print 'It took ' + i + ' iterations to sort the deck.';