Keren Bergman

Assistant Professor of Electrical Engineering
Ph.D. 1994, Massachusetts Institute of Technology

A temporal soliton is a solitary wave that preserves its shape as it propagates, due to compensation between nonlinearity and dispersion. We can routinely generate solitons in optical fiber that balance Kerr nonlinearity against group-velocity dispersion in the anomalous dispersion regime. This anomalous regime occurs at the minimum loss region in standard silica fiber near a wavelength of 1550 nm, which is also the region where efficient, nearly quantum noise limited optical amplification is provided by Erbium-doped fiber amplifiers. The fundamental nature of solitons and their potential applications in long-distance communications and fiber optic networks is the focus of extensive research in our laboratory.

Recent explosive advances in the manufacturing of optical fibers and related components have driven the performances of fiber optic-based devices and systems to rapidly approach fundamental physical limitations. The thrust of our research effort addresses questions regarding the physical nature of nonlinear noise processes in optical fibers. Experimentally, we exploit nonlinear optical processes to achieve quantum noise reduction for applications toward sensitivity enhancement in photon number and phase measurements. Using our new scheme for generating amplitude-squeezed solitons in an asymmetric fiber Sagnac loop interferometer, we measured a record reduction in photon number fluctuations by direct detection. The same scheme was also shown to generate a significant classical noise reduction and is fundamentally limited by Raman effects in fiber. Our work aims to probe the Raman noise and dispersion-dependent limitations on squeezing and applications of soliton number states to sensors and quantum nondemolition (QND) measurements.

Our further research on the nonlinear propagation of chirped soliton pulses in optical fibers, motivated by the recent advancements in dispersion-managed communications systems, has led to the observation of new phenomena in low-dispersion fibers that are both of practical and fundamental interest. In particular, when a prechirped pulse is injected into low-dispersion anomalous fiber with a sufficiently strong chirp, the pulse will break up into a train of solitons, which propagate away from the center-position of the original pulse. In this break-up process, the ejected solitons are ordered according to height: the taller soliton pulses propagate faster than the smaller ones. We have shown the first numerical and experimental evidence demonstrating explicitly this break-up phenomenon. The break-up occurs in a well-determined manner: one-solitons are ejected away from the center-position, with the tallest ones moving fastest.

Solitons are also critical to the design of short-pulse lasers. In a mode-locked laser, both group-velocity dispersion and Kerr nonlinearity are present in the cavity elements, causing perturbations to both the shape and spectrum of the pulse as it propagates through the cavity. We can obtain soliton formation in a specially designed laser cavity with minimal perturbations where the pulse is essentially a soliton of the average group-velocity dispersion and Kerr nonlinearity. In our extremely compact cavity the perturbations to the pulse with each round trip are small enough that fundamental solitons propagate in the cavity. We demonstrate the presence and influence of solitons by showing that the output pulses remain transform-limited despite changes in average cavity dispersion, pulse width, spectral bandwidth, and pulse energy.

We pursue the applications of solitons in fiber optic networks. In particular, we are studying a novel optical packet switched network architecture that enables scalable data communications at ultrahigh throughput bandwidths with minimal latency. The network was designed for an optical implementation that employs soliton bits modulated in both the time and wavelength domains. The network is bufferless and features radically new traffic control that minimizes the logic in data packet routing decisions. We are studying the network with numerical simulations as well as by experiments that demonstrate the feasibility of key technologies in a test-bed environment.