Researchers at the Georgia Institute of Technology (Atlanta, GA) described an ultrasimple method for ultrafast laser diagnostics during a technical session at the annual Conference on Lasers and Electro-Optics (CLEO; San Francisco, CA) in May. The new method is based on frequency-resolved optical gating (FROG). It uses the FROG algorithm and was developed by a member of the original FROG team.1 The FROG hardware requirements have been drastically simplified. But the name of the new technique—grating-eliminated, no-nonsense observation of ultrafast incident laser light e-fields—and its acronym (GRENOUILLE) have become much more complicated for those of us who don't speak French. Grenouille is the French word for frog.
The FROG technique allows the complete measurement of a pulse's intensity and phase with little additional experimental complexity because instead of taking spectra and performing autocorrelation of a pulse separately, FROG places the spectrometer behind the autocorrelator and takes the spectrum of the autocorrelation. Rick Trebino, a professor at Georgia Tech and coinventor of FROG and GRENOUILLE, said that about 50 or 60 groups around the world use FROG to measure pulses, often in the process of setting up amplifier systems.
A primary strength of FROG lies in providing an overdetermination of pulse data that facilitates cross-checking of results. "When you get a pulse measurement using FROG, there are always independent checks," Trebino said. "If the measurement is not correct, you run into some trouble: the algorithm doesn't converge or some checks don't work out." One of the things that researchers realized when they started using FROG, he added, "was that making an autocorrelation was a little harder than we had thought." It's easy to measure an autocorrelation that's contaminated by systematic error, but if there's a problem in the autocorrelator component of FROG, it'll show up when you run the algorithm."
The difficulty was largely caused by a need to satisfy four alignment degrees of freedom. Three degrees of freedom resulted from the need to split the pulse in two halves and then to align the two halves with each other in space and time within the autocorrelator crystal. And the fourth degree of freedom concerned angular alignment of the signal within the crystal. Typically the FROG hardware that required all of this adjusting and tweaking occupies about 1/5 or 1/6 of an optical table, Trebino said. And subsequent pulse-measurement algorithms that have improved upon FROG's performance have also generally increased the degrees of alignment freedom and hardware requirements to a degree that rivaled the complexity and size of the laser system that was being measured.
GRENOUILLE has tackled this problem head on by eliminating the need for both the autocorrelator and the spectrometer in the FROG process. And the idea for GRENOUILLE arose as a direct result of Trebino's recent move from a government laboratory to a university. "I started teaching for the first time last year, so I had to go through a basic optics textbook, and, as I was reading, I came across a Fresnel biprism," he said. "It occurred to me that if you send your beam through the biprism, the beam is split into two and then recombined such that both halves of the original beam are perfectly overlapped in time and in space. So it saves you three degrees of freedom right away." He eliminated the fourth degree of freedom also by line focusing the beam with a cylindrical lens before it entered the biprism.
The typically thin autocorrelator crystal was next to go, and the need for a spectrometer went with it. Thin crystals (on the order of 10 µm for 10-fs pulses) provided the necessary phase-matching bandwidth for second-harmonic generation, but thin crystals also limited signal strength and introduced concerns related to careful handling as well as precision in the dimensions of the crystal itself.
While increasing signal capacity and relieving logistical concerns, on the other hand, a thick crystal introduces an angular dependence into the phase-matching process—which is also what a spectrometer does, Trebino said. Upon realizing this, he asked himself, "Instead of using a thin crystal to do our second-harmonic generation as in an autocorrelator and then placing a spectrometer after that to make it into a FROG, what if we just replace the thin crystal and the spectrometer with a thick crystal?" So Trebino, graduate student Patrick O'Shea, and research associate Mark Kimmel tried the idea using a simple apparatus that occupied a surface area of about one square foot and it worked. To date they have been able to measure simple or complex pulses from ~30 fs to ~1 ps.
REFERENCE
- R. Trebino et al., Rev. Sci Instr. 68, 3277 (1997).