Nonlinear Fiber Optics - 4 ed. Agrawal

8.4. Soliton Effects 313
ton laser indicated that the noise was more than 50 dB above the shot-noise level [179].
Pulse-to-pulse timing jitter was also found to be quite large, exceeding 5 ps at 1.6-W
pump power. Such large noise levels can be understood if one considers the impact
of the Raman-induced frequency shift (see Section 5.5) on the performance of such
lasers. For synchronous pumping to be effective, the round-trip time of the Raman
soliton inside the laser cavity must be an integer multiple of the spacing between pump
pulses. However, the Raman-induced frequency shift changes the group velocity and
slows down the pulse in such an unpredictable manner that synchronization is quite difficult
to achieve in practice. As a result, Raman soliton lasers generate pulses in a way
similar to single-pass Raman amplifiers and suffer from the noise problems associated
with such amplifiers.
The performance of Raman soliton lasers can be significantly improved if the
Raman-induced frequency shift can be eliminated. It turns out that such frequency
shifts can be suppressed with a proper choice of pump power and laser wavelength. In
a 1992 experiment, a Raman soliton laser was synchronously pumped by using 200-ps
pump pulses from a 1.32-μm Nd:YAG laser. This laser was tunable over a wavelength
range of 1.37–1.44 μm [180]. The Raman-induced frequency shift was suppressed in
the wavelength range of 1.41–1.44 μm for which the Raman-gain spectrum in Figure
8.2 had a positive slope. Measurements of noise indicated significant reduction in both
intensity noise and timing jitter [181]. Physically, Raman-gain dispersion is used to
cancel the effects of the last term in Eq. (2.3.43) that is responsible for the Ramaninduced
frequency shift.
It is difficult to operate a soliton laser at wavelengths below 1.3 μm because most
fibers exhibit normal GVD for λ < 1.3 μm. A grating pair is often employed to solve
this problem, but its use makes the device bulky. A compact, all-fiber, laser can be
built if a suitable fiber exists that can provide anomalous dispersion at the Raman-laser
wavelength. Such fibers have become available in recent years in the form of the socalled
microstructured or holey fibers. In a 2003 experiment [182], a 23-m-long holey
fiber was employed within the cavity of an all-fiber Raman laser to generate 2-ps output
pulses at the 1.14-μm wavelength. The laser was pumped synchronously with 17-ps
pulses from a Yb-doped fiber laser. The Raman gain was provided by 15 m of standard
silica fiber with 3% Ge doping.
In the wavelength region near 1.6 μm, one can employ dispersion-shifted fibers
that permit the control of cavity dispersion while also acting as a Raman-gain medium.
Such a fiber was used in a 2004 experiment to realize a compact, synchronously pumped,
Raman laser that was also tunable from 1620 to 1660 nm [183]. The 2.1-km-long fiber
loop inside a ring cavity had its zero-dispersion wavelength at 1571 nm with a dispersion
slope of about 0.1 ps 3 /km. This laser also employed as a pump a gain-switched
DFB semiconductor laser emitting 110-ps pulses at a 54.5 MHz repetition rate that
were amplified to an average power level of up to 200 mW. Such a Raman laser produced
pulses shorter than 0.5 ps over its entire 40-nm tuning range.
8.4.3 Soliton-Effect Pulse Compression
In some sense, Raman solitons formed in Raman amplifiers or lasers take advantage of
the pulse compression occurring for higher-order solitons (see Section 5.2). Generally