Our group develops and utilizes new light sources and techniques to follow the motions of electron, holes, and nuclei in molecular and condensed matter systems on ultrafast time scales. Developing new technologies and physics ideas go hand in hand with gaining insight into ultrafast dynamics. We have built two novel instruments in the lab, one for Cavity-Enhanced Ultrafast Optical Spectroscopy, and another for XUV Time-Resolved Photoelectron Spectroscopy (TRPES). Both are based upon our home-built frequency comb lasers.

Frequency Combs

Frequency comb lasers, recognized with the 2005 Nobel prize in physics, have revolutionized atomic clocks and precision measurement. However, their enormous potential for ultrafast time-resolved measurements has been largely unexplored. The exquisite coherence of the pulse train of a frequency comb enables the signals from successive pulses to be coherently added and stored. This leads to orders of magnitude improvement in the attainable signal to noise of time-resolved experiments, enabling new and exciting directions that simultaneously push the boundaries of our ability to probe matter in the time and frequency domains. Now mature in the visible and near-infrared, recent breakthroughs have pushed frequency comb technology into more spectroscopically interesting regions of the electromagnetic spectrum. Mid-IR combs enable the probing and excitation of vibrational dynamics and UV combs enable the probing of electronic dynamics.


A frequency comb is formed by a train of well controlled femtosecond pulses.

Our frequency comb lasers have been based on Ytterbium (Yb) doped optical fibers. The small quantum defect, high doping, and the availability of double clad large-mode-area photonic crystal fibers makes Yb ideal for scaling to high average power. We have built in our lab an 80 W Yb:fiber laser for generating high order harmonics at high repetition rate. A detailed “how-to” paper is here.


Our home-built Yb:fiber amplifiers.

To make widely tunable frequency combs for cavity-enhanced optical spectroscopy (see below), we have also started branching out into using Erbium-doped fiber lasers frequency shifted in highly nonlinear fibers. Using this backbone, we can make multiple shifted branches and then make widely-tunable mid-IR combs via difference frequency generation (DFG).

Left: A nonlinear Erbium-doped fiber amplifier, compactly fit into a small box. Right: Tunable frequency comb light spanning 1-2 microns. In this particular spectrum, the dispersive wave has been optimized for seeding our Yb-doped amplifiers.

Cavity-Enhanced High Harmonic Generation

High harmonic generation (HHG) can be used to generate ultrashort pulses of extreme ultraviolet (XUV) and soft x-ray light. These photon energies are ideal for performing photoelectron spectroscopy from surfaces and probing absorption edges of soft matter. However, due to limitations in laser power, most high-harmonic generation systems run at repetition rates around 1-10 kHz. We generate harmonics at repetition rates around 100 MHz, using the technique of cavity-enhanced high-order harmonic generation. In the cavity enhancement technique, a frequency comb laser is resonantly enhanced in an optical cavity whose repetition rate and carier-envelope offset frequency are matched to that of the laser. With kiloWatts of intracavity laser power, efficient HHG can be done at the full repetition rate of the mode-locked laser.

These sources are now among the world’s brightest XUV sources. In addition, the higher repetition-rate is particularly critical when recording photoelectron signals from surfaces, where space-charge puts severe limitations on the number of electrons per pulse that can be extracted from the sample. At synchrotron beamlines, nano-Ampere photocurrents parsed by sophisticated electron energy analyzers enable detailed studies of the electronic structure of solids and surfaces in energy, momentum, and spin. However, historically photoemission studies using XUV light produced from lasers have been limited to the pico-Ampere range, drastically limiting what can be studied. Using frequency comb methods, our system at Stony Brook can now produce space-charge-free photoemission currents also in the nA-range. For the purpose of surface photoemission, our system can be thought of as “table-top synchrotron” with ~1000 times better time resolution. A recent paper showing the performance of the source appears here.

The Stony Brook Time-resolved photoelectron spectroscopy beamline. Synchrotron-level fluxes, but with pulse durations ~1000 times shorter, are delivered to carefully controlled samples under ultra-high vacuum conditions. TM = Toroidal mirror. GJ = gas jet. VPD = Vacuum photodiode. PD = photodiode.

Time- and Angle-Resolved Photoemission

An electron’s momentum parallel to the surface is conserved in the photoemission process. This allows photoemission to report on both the energy and momenta of electrons in solids, and angle-resolved photoelectron spectroscopy (ARPES) using synchrotron radiation has become and indispensable tool for determining the electronic structure of matter. ARPES can be used for determining the band structure of solids, the electronic structure of more complicated “quantum materials” with strong electron correlation, and also reconstruct orbital shapes for molecules on surfaces.

With support from the Stony Brook Foundation’s Discovery Prize, we are upgrading our photoelectron spectrometer to one that measures photoelectron emission angles and energies in parallel. Stay tuned for this exciting upgrade, which will further increase our system’s performance by orders of magnitude. Combining modern electron analyzers with our “table-top synchrotron”, we will be able to record frames of time-resolved ARPES spectra from excited states at synchrotron data rates.

Charge Transfer Processes at Molecule/Surface Interfaces

Photo-induced charge transfer processes at molecule-semiconductor interfaces are central to a wide range of technological applications including solar energy conversion, heterogeneous photocatalysis, and organic optoelectronics. In general, these photoreactions involve a number of elementary steps including charge carrier (e−/h+) excitation and thermalization, interfacial charge transfer, adsor- bate intramolecular relaxation, and subsequent reaction. The overall efficiency of photoconversion is determined by the complex interplay between these elementary processes, and identifying the rate limiting steps can ultimately provide the insight needed to improve efficiency.

Although modern time-resolved spectroscopic techniques have focused on some individual aspects of semiconductor photoconversion, the dynamics of molecular photoreactions on semiconductor surfaces remains largely unexplored.


An illustration of TRPES from the photooxidation of acetone on a titania surface.

Cavity-Enhanced Ultrafast Spectroscopy

Ultrafast spectroscopy is widely applied to a range of systems from molecules in solution to isolated molecules in the gas phase. However, measurements in the gas phase are almost always done using photoionization, due to the very high sensitivity of charged particle detection, whereas measurements in solution are restricted to measuring absorbed or nonlinear scattered light. Photoionization of gas phase clusters is also problematic due to cluster fragmentation. Comparison of gas phase, cluster, and solution phase data is thus often highly nontrivial.

We have developed techniques using frequency combs and optical resonators to achieve a large sensitivity in ultrafast spectroscopy, so that optical signals can be recorded from dilute molecular beams. We have demonstrated transient absorption spectroscopy with detection limits as low as ΔOD = 2×10¯¹º, and also published schemes for using multiple combs for recording cavity-enhanced multi-dimensional spectroscopy signals. This allows the direct comparison of solution-phase and gas-phase data, and more importantly allows ultrafast vibrational spectroscopy to be applied to gas-phase and cluster systems.


Simple illustration of cavity enhanced transient absorption spectroscopy. Power enhancement of the pump comb allows excitation of the sample at high repetition rate. The probe cavity enhances the absorption sensitivity.

A: The cavity-enhanced ultrafast spectrometer. Two ring cavities are mounted on an vibrationa-isolated platform inside the large rectangular vacuum chamber. B: The inside of the vacuum chamber with the pump beam at ~50 W intracavity power visible. C: A transient absorption signal from molecular iodine with noise at the OD = 10¯¹º level.

Vibrational Dynamics of Clusters

While hydrogen bonding is a ubiquitous theme in Chemistry, the hydrogen bond is actually a complicated subject. Many interactions contribute to hydrogen bonding, including electrostatic forces, polarization, dispersion forces, and also a component of covalent bonding.

The complexities of the hydrogen bond play a major role in the difficulties in the theo- retical description of liquid water. Water, nature’s most important solvent, makes extensive use of the hydrogen bond and plays a central role in chemistry and biology. Water mediates the interactions between molecules, altering energy levels of solvated species, modifying po- tential energy profiles along reaction coordinates , and facilitates efficient proton transport through ion channels and interfaces

Using cavity-enhanced techniques, we can follow vibrational energy flow in small gas-phase hydrogen bonded clusters, such as water. These studies allow us to build up the liquid “one molecule at a time” and gain a very systematic understanding of the vibrational couplings between bonds.


a) A linear spectrum of D2O clusters in a supersonic expansion recorded with cavity ring down spectroscopy in a slit jet expansion, taken from J.B. Paul, J. Chem. Phys. 109, 10201 (1998). b) Oversimplified illustration of the modes vibrational modes of the water trimer (the real normal modes are of course more complicated). Time resolved dxperiments measure the coupling between the free OD stretch and bonded OD stretch in small D2O clusters. The spectra of each cluster is well separated from its neighbors, so mass selection is not be necessary.