Research IR+UV Control Optimal Control

Optimal Laser Pulse Control of Chemical Reactions

Introduction:

Optimal laser pulse control of chemical reactions is a hot topic in femtosecond chemistry [1]. Our goal is to employ these laser pulses to transform a reactant into a particular product, suppressing competing reactions.
This dream is becoming reality thanks to the fundamental theoretical and experimental work of the groups of H. Rabitz in 1992 and [2] and G. Gerber in 1998 [3].

Concepts and Results:

Essentially, optimal laser pulses are determined experimentally. Typically starting from a transform-limited pulse as a reference. This pulse is then shaped by means of a feed-back learning algorithm, until the desired product is obtained. The optimized pulses look usually very complex. Therefore, a major challenge has been to interpret or "decode" the mechanisms of such optimal laser pulses. After progresses for photophysical processes by the pioneers and other groups, we have succeeded in decyphering the mechanism of an optimal laser pulse which selects a specific product, among other competing ones. The details are summarized in the joint experimental and theoretical paper [4], published together with our experimental partners in Sonderforschungsbereich Sfb 450 i.e. the group of L. Woeste. The following movies illustrate some important aspects of our paper.

       
Figure 1:
Specifically, we consider the photochemical reaction of the organo-metallic compound, cymanthrene (cyclopentadienylmangenesetricarbonly CpMn(CO)_3), as shown in Figure 1. A laser pulse may ionize the parent molecule, or dissociate a carbonyl ligand. Exemplarily, the goal has been to ionize the intact molecule while suppressing the loss of ligands.

       
Figure 2:
The experimental optimal laser pulse for this purpose is shown in Figure 2. It consists mainly of two peaks separated by approx. 85 fs. Inspection of the phase of the pulse shows that the first peak consists of slightly higher frequency components in comparison with the second peak. The mean frequency corresponds to the 800 nm wave length of the reference pulse. The task for our laser control theory group has been to discover the mechanism of this optimal laser pulse. The solution was found in two steps, in cooperation with the experimental partners, and it is illustrated by the movies:

Klick to download movie I.       

Organometallic compounds such as cymanthrene have many near degenerate excited electronic states. The movie shows the corresponding potential energy curves along the distance between the metal atom and a carbonly ligand, both for the neutral system (i.e. CpMn(CO)3 , 5 lower curves) as well as for the ionized systems (CpMn(CO)3+, 3 upper curves).
Initially, the system is in the electronic ground state, represented by the wave packet in the lowest potential energy curve. The movie illustrates the effects of two sequential experimental laser pulses which correspond to the two main peaks of the optimal pulse, but with the same frequency (800&mbsp;nm) i.e. without frequency optimization. Apparently, the first "pump" laser pulse excites several near-degenrate electronic states, and the second "probe" pulse ionizes the system. The corresponding wave packet dynamics appear very complicated, representing many competing excitation pathways and processes such as ligand loss in the excited neutral and ion systems, internal conversion from one electronic excited state to another, etc.
Close inspection discovers a small contribution for one of the competing processes which would serve the goal, i.e. vibrations in the highest of the near-degenerate excited states of the neutral system (labeled c1A', blue curve), without decay. Obviously, this laser pulse has not been optimized (even though it looks similar to the optimal laser pulse), hence it is not selective. We can already understand that the optimal laser pulse should support excitation of this c1A' state while suppressing the others.

       Klick to download movie II.

For the second simulation, we apply again two sequential pump-and-probe laser pulses, but now with optimized frequency, i.e. the frequency components of the first "pump" and second "probe" pulses are slightly higher or lower than the reference pulse similar to the experimental optimal pulse. As a consequence, the "pump" laser pulse does no longer excite many near degenerate excited electronic states, but preferably just the excited c1A' state. Subsequently, the probe laser pulse prepares the ion in its second excited state labeled b2A'.
The result is that the combined pump and probe laser pulses with optimal frequencies and time delay prepare predominantly the parent molecule as an ion, without loss of ligands. This explains the equivalent pulse shape of the optimal pulse.

Acknowledgments:

We are grateful to our experimental partners in Sfb 450, for fruitful and pleasant coopeartion, and also to Dr. Chantal Daniel (Strasbourg) for expert cooperation on the quantum chemistry of our model system, cymanthrene. We should also like to thank our tutors for preparing the movies. Financial support by Deutsche Forschungsgemeinschaft through project Sfb 450 is gratefully acknowledged.

Literature:

[1] A. H. Zewail, Angew. Chem. Int. Ed. Engl. 39, 2586 (2000)
[2] R. S. Judson and H. Rabitz, Phys. Rev. Lett. 68, 1500 (1992)
[3] A. Assion, et al., Science 282, 919 (1998)
[4] C. Daniel et Science 299, 536 (2003)
Rev.: 14-May-2005