Focused laser-like light that covers a wide range of frequencies is highly desirable for many scientific studies and for many applications, for instance quality control of manufacturing semiconductor electronic chips. But creating such broadband and coherent light has been difficult to achieve with anything but bulky energy-hungry tabletop devices.
Now, a Caltech team led by Alireza Marandi , a professor of electrical engineering and applied physics at Caltech, has created a tiny device capable of producing an unusually wide range of laser-light frequencies with ultra-high efficiency—all on a microchip. The work has potential in areas ranging from communications and imaging to spectroscopy, where the light would aid the detection of atoms and molecules in various settings.
The researchers describe the new nanophotonic device and approach in a paper that appears in the journal Nature Photonics. The lead author of the paper is Ryoto Sekine (PhD '25), who completed the work while a graduate student in Marandi's lab.
"We are showcasing that with a single nanophotonic device and low input energies in the femtojoule range, you can actually cover a broad section of the electromagnetic spectrum, from visible wavelengths to the mid-infrared. This is something that had never been done," Marandi says.
The Caltech device uses a technology that has been around since 1965: an optical parametric oscillator (OPO). Essentially, an OPO is a resonator, a tiny engineered light trap that takes incoming laser light at an input frequency and uses a special nonlinear crystal—here, lithium niobate—that, with careful engineering, can generate light of different frequencies.
In general, OPOs start from a laser source with a narrow frequency range and generate outputs at different frequencies but still in a narrow range. Typically, they have been used as laser-like sources with widely tunable, or adjustable, output frequencies.
A Light Comb
However, in this work, Marandi and his colleagues have engineered their OPO at the nanoscale on a chip to generate what is known as a frequency comb, a spectrum of evenly spaced laser-like light across a wide range of frequencies with very little input energy. The frequency comb covers a surprisingly broad spectral range, providing sharp, stable lines from the visible light that we can see all the way to the longer mid-infrared wavelengths.
Two scientists earned a share of the 2005 Nobel Prize in Physics for their work developing the frequency comb technique. Unlike conventional lasers, which emit a single color of light, frequency combs act like a ruler for light across a range of frequencies. These combs have been used to improve everything from the precision of atomic clocks and measurements made with light to environmental monitoring.
But, Marandi says, "there have been two main challenges with frequency combs: One is that the sources are too big, and the second is that it's challenging to make them in different desired spectral windows. Our work offers a path toward solving both of these problems."
The new device's key advances are what Marandi describes as dispersion engineering—shaping how different wavelengths of light travel through the device, ensuring that they stay together rather than spreading out—and a carefully designed resonator structure. Together, these allow the device to broaden the spectrum efficiently and to maintain coherence while requiring an extremely low threshold, or energy at which it begins working.
A Surprisingly Broad Coherent Spectrum
Marandi says that he and his team were surprised by the device's performance. "We turned it on and cranked up the power, and when we looked at the spectrum, we saw that it was extremely broad. We were particularly surprised that the super-broad spectrum was actually coherent. This was against the textbook descriptions of how OPOs work," he says.
That sent the researchers back to their simulations and to theory to try to figure out how that could be. In simulations, raising the energy of incoming light above the threshold caused the spectrum to become incoherent—that is, of various wavelengths and not locked in phase, which means no frequency comb is generated. But back in the lab, the spectrum was coherent when operating farther above threshold.
"It took us maybe six months to discover that there is this new regime of OPO operation in which the OPO is far above its threshold and the coherence is reestablished," Marandi says. "Because the threshold of this OPO is orders of magnitude lower than previous OPOs, and the dispersion and the resonator are engineered unlike the previous realization of OPOs, we could observe this phenomenal spectral broadening, which is orders of magnitude more energy efficient than other spectral broadening schemes."
The researchers say the work could reshape how frequency comb-based technologies, currently found in table-top setups, could transition to integrated photonic devices. One of the main techniques used to make stable frequency combs requires significantly broadening their spectrum. The energy required for such broadening has been one of the bottlenecks preventing integration of frequency comb technologies on chip.
Beyond that, the bulk of photonic technologies, including most well-developed lasers and detectors used for measuring molecules, operate in the near-infrared or visible range. OPOs that start from near-infrared lasers as the input frequency and then efficiently convert the light, outputting coherent light in the mid-infrared range, could allow researchers, for example those working with spectroscopy, to access a wealth of information at lower frequencies. At the same time, such a device could enable accessing the higher-frequency range for atomic spectroscopy.
The paper is titled "Multi-Octave Frequency Comb from an Ultra-Low-Threshold Nanophotonic Parametric Oscillator." Additional authors are former Caltech graduate students Robert M. Gray (PhD '25) and Luis Ledezma (PhD '23) as well as current Caltech graduate student Selina Zhou and former postdoctoral scholar Qiushi Guo. The device nanofabrication was performed at the Kavli Nanoscience Institute at Caltech. The work was supported by funding from the Army Research Office, the National Science Foundation, the Air Force Office of Scientific Research, DARPA, the Center for Sensing to Intelligence at Caltech, and JPL, which is managed by Caltech for NASA.
