Coherent Control of Molecular Dynamics with Shaped Femtosecond Pulses

Coherent control of molecular dynamics deals with the steering of
quantum mechanical systems with suitably shaped ultrashort laser
fields. The coherence properties of the laser field are exploited to
achieve constructive interference for a predefined target wave
function via a phasecorrect superposition of wave functions. The
goal of coherent control, the selective preparation of a target
state, is an important prerequisite for mode-selective chemistry.
The laser pulse tailored to drive the system from the initial to
the target state as a perturbation can in general not be determined
by a quantum-mechanical calculation, since usually even the
Hamiltonian of the system is unknown. A practical alternative is to
determine the required shape of the laser field in a
feedback-controlled regulation loop which uses a signal derived
from the experiment as feedback. The loop is repeated until a pulse
that suits the requirements is obtained. Experiments in this area
have until recently mostly been limited to the wavelength regime of
Ti:Sa lasers and their fundamentals. This work deals with the
fundamentals of feedback-controlled shaping of ultrashort laser
pulses with respect to both establishment of its technical
prerequisites and its application to suitable model systems. The
feedback loop has been tested using a simple optimization
experiment with known outcome; then it was applied to experiments
of progressively increasing complexity. From the optimized pulses,
physical insight into the optimization process has been gained. In
the first part of this work, the required technology has been
implemented and standardized such that control experiments might
employ it as a standard tool. One of the technical prerequisites
was the frequency conversion of the 800 nm Ti:Sa laser pulses to a
wavelength range suited to the particular systems. To this end,
non-collinear optical parametric amplifiers have been built in
different designs that routinely produce tunable sub-20 fs pulses in
the visible. The characterization techniques for ultrashort pulses
have been implemented as well. Pulse shapers with cylindrical
instead of spherical mirrors have been implemented for the
modulation of broadband pulses, and their functionality has been
explained both theoretically and experimentally. A new liquid
crystal device, the core of our pulse shapers, has been developed
in cooperation with the group of Thomas Feurer at the Universit¨at
Jena and the Jenoptik GmbH which allows for the generation of more
complex pulse shapes than with other commercially available devices
to date. Using a pulse shaper to modulate the white light continuum
that serves as the seed for the non-collinear optical parametric
amplifier, generation of phase-locked two-color double pulses has
been achieved, with tunable wavelengths, delays, and relative
carrier phases between the single pulses. The basic principle,
phase conservation during optical parametric amplification, has been
demonstrated. With this setup, control experiments which require
pulses with the above described attributes in electronically
controllable form are possible for the first time. An evolutionary
strategy used as the optimization algorithm in the feedback loop
has been programmed and characterized both in simulation and
experiment using a simple optimization experiment, namely pulse
recompression by phase compensation. In the second part of this
work, pulse recompression of ultra-broadband spectra in the
sub-20fs regime serves as an example of utility of
feedback-controlled optimization. This experiment simultaneously
served as a further test of the feedback loop in the limit of a
physically unreachable optimization goal. It has been demonstrated
that a suitable parameterization of the electric field, implemented
by a mapping of the optimization parameters adjusted by the
algorithm to the physical parameterscontrolling the liquid crystal
mask affords a means of acquiring physical knowledge from the
retrieved optimal electric fields. A parameterization helps to
dissect the physical processes mediating the control process,
thereby assuring fast, secure convergence and robustness against
signal noise. So-called ”bright” and ”dark” pulses, i.e. pulses
that are absorbed by a medium or transmitted, respectively, have
been demonstrated for the case of the two-photon transition
Na(3s5s). The physical constraints responsible for pulses being
either ”bright” or ”dark”, namely a symmetric or anti-symmetric
spectral phase, have been incorporated in the parameterization with
the purpose of testing the concept of parameterization for such
studies. An example of mode-selective preparation of vibrational
states in a polyatomic molecule is the control of the ground state
dynamics in polydiacetylene. In a Raman step with a shaped Stokes
pulse, the population of the backbone vibrations of polydiacetylene
in its ground state could be controlled. A consecutive probe pulse
in a CARS (coherent anti-Stokes Raman scattering) arrangement
generates an anti-Stokes signal which, once frequency-resolved,
served as feedback. Of the three or four modes, respectively,
accessible within the pulse bandwidth, single modes as well as
combinations of modes could be excited with high selectivity.
Again, suitable parameterizations helped to identify one of the
processes responsible for the control as a Tannor-Rice scheme.
Since both the amplitude and the phase of each mode could be
influenced, the focusing of a wave packet at a predefined time, or,
equivalently, the generation of local modes represents the control
of a unimolecular reaction. Starting from the control of a
unimolecular reaction, the possibilities of controlling a
bimolecular reaction were addressed. The NaH2 collision complex was
chosen as a suitable system for the control of bimolecular
reactions generally and a conical intersection in particular. First
timeresolved experiments have been presented.