Report

# Dynamic Stark Control of Photochemical Processes

See allHide authors and affiliations

Science  13 Oct 2006:
Vol. 314, Issue 5797, pp. 278-281
DOI: 10.1126/science.1132289

## Abstract

A method is presented for controlling the outcome of photochemical reactions by using the dynamic Stark effect due to a strong, nonresonant infrared field. The application of a precisely timed infrared laser pulse reversibly modifies potential energy barriers during a chemical reaction without inducing any real electronic transitions. Dynamic Stark control (DSC) is experimentally demonstrated for a nonadiabatic photochemical reaction, showing substantial modification of reaction channel probabilities in the dissociation of IBr. The DSC process is nonperturbative and insensitive to laser frequency and affects all polarizable molecules, suggesting broad applicability.

Molecular catalysts increase chemical reaction rates by applying forces that modify potential energy barriers along a reaction coordinate. Because electrical forces underlie all of chemistry, such barrier manipulation is also possible by application of a laser field. The duration of modern ultrafast laser pulses is on the time scale of chemistry itself, and therefore precise control over the form and delay of these pulses offers access to different portions of a potential energy surface as a reaction occurs. DSC is a technique that uses nonresonant infrared laser fields to dynamically alter a potential energy landscape during a photochemical reaction. The application of this field modifies the potential surface via the Stark effect, enhancing or inhibiting a specific reaction channel. Importantly, it does so without inducing any real electronic transitions to other potential surfaces, which can lead to chemical reactions other than the one of interest. DSC will be general because the nonresonant Stark effect is independent of the laser frequency and is applicable to all quantum systems.

The control of chemical reactions by lasers is an area of great interest (13). Quantum control can be viewed as chemistry where light is used as a photonic reagent (4). By contrast, we experimentally demonstrated that the nonresonant dynamic Stark effect participates in quantum control by altering reaction barriers, as if it were a photonic catalyst. Because all strongfield approaches tacitly contain the dynamic Stark effect, DSC can be considered as a fundamental element of the quantum control toolbox (5). However, by exclusively utilizing the dynamic Stark effect, DSC avoids the ionizing fields that can produce numerous competing processes, such as Coulomb explosion (6), enhanced ionization (7), and nonadiabatic multi-electron ionization (8). Perturbative coherent control approaches do not modify potential energy surfaces but rather use interference between two or more real electric dipole transitions (1). Strong dipole coupling of states in nonperturbative laser fields creates light-induced potentials that lead to phenomena such as static and transient bond hardening or softening (3, 9). Optical pulses shaped by feedback-learning algorithms (10) can optimize, via multiphoton transitions, the yield of a chosen product, as demonstrated in the strong-field ionization-fragmentation of polyatomic molecules (11, 12). Theoretical studies of laser catalysis (via electronic transitions) of ground-state collision processes investigated the line shape for the reaction probability as a function of optical frequency (13,14) or, alternatively, considered resonantly induced couplings during the collision (15). Moving wave packets between potential energy surfaces can likewise be used to avoid ground-state barriers (2).

To demonstrate DSC, we applied it to an important class of photochemical reactions: nonadiabatic processes. These processes, such as internal conversion or intersystem crossing, entail charge rearrangements that occur along a reaction path at the intersections of potential energy surfaces and act as triggers of the ensuing chemistry. Nonadiabatic processes are of paramount importance in the biological mechanisms of vision and photosynthesis and underlie the photochemistry of almost all polyatomic molecules (16). Chemical branching ratios in nonadiabatic processes are very sensitive to the intersection geometry, and therefore the dynamic modification of these processes is an important application of DSC. We specifically applied DSC to the canonical example of a nonadiabatic process, photodissociation of IBr (1719). The reaction is initiated by absorption of a visible photon, making the transition from the ground state 1(X) to 2(B) (Fig. 1). The nonadiabatic intersection between states 2(B) and 3(Y) leads to two chemically distinct, neutral atomic channels: $Math$ and I + Br*(2P½). These Br and Br* atomic products have different chemical reactivity and play an important role in the destruction of ozone (20), especially in the marine boundary layer (21). An infrared DSC field was used to modify the curve-crossing barrier at a specific time, thus promoting the yield of one chosen product over another.

DSC of IBr photodissociation was demonstrated by using a molecular beam technique. Briefly described, a seeded supersonic jet of IBr in argon was produced by expansion through a 200-μm-diameter glass capillary nozzle (Fig. 2). Glass and polytetrafluoroethylene parts were used exclusively to avoid the decomposition of IBr. The reaction was initiated by a 100-fs laser pulse centered at 520 nm, above the dissociation limit for both channels. To achieve DSC, we focused a time-delayed 1.7-μm infrared pulse, 150 fs in duration, to nonperturbative intensities (22). To maximize the effects of DSC, we set the laser intensity to just below the threshold for ionization (i.e., below 1013 W/cm2).

Importantly, a third weak laser field was used as the final probe of free neutral product formation. About 60 ps after the pump and catalysis pulses, free neutral ground-state iodine atoms produced in the reaction were selectively detected via (2+1) resonance-enhanced multiphoton ionization (REMPI) by using a 304.5-nm pulse of 0.4-nm linewidth. Conservation of energy and momentum dictate that the iodine fragments from the Br* channel have a lower velocity than those from the Br channel. Therefore, measurement of the iodine atom kinetic energy distribution permits unambiguous determination of the Br*/Br product branching ratio. This approach avoids uncertainties in the relative ionization cross sections of the chemically distinct Br and Br*. The velocity distribution of the iodine fragments was measured by using velocity map imaging (VMI), a charged-particle spectrometry technique that maps velocity to position, permitting the measurement of three-dimensional fragment velocity distributions (23).

A typical fourfold symmetrized velocity-mapped image was obtained under control-free conditions (i.e., 520-nm initiation and 304.5-nm I-atom probe beams only) and is presented along with its associated Abel inverted reconstruction and speed distribution (Fig. 2). The field-free branching ratio is Br*/Br = 3.5, in good agreement with previous measurements (24). Changes in the branching ratio due to the application of the infrared (IR) control field are easily and accurately determined from the VMI reconstructed speed distribution.

The mechanism underlying DSC's ability to reshape the potential energy barriers is the nonresonant dynamic (or AC) Stark effect (25). This effect is similar to the well-known static (or DC) Stark effect but with several important distinctions. Infrared laser frequencies in the nonresonant 1.5- to 2-μm region are large with respect to rotational and vibrational frequencies but small with respect to electronic transition frequencies, and, hence, no single photon transitions can occur. In this situation, the dynamic Stark effect can be insensitive to the laser frequency, following instead the intensity envelope of the pulse (26). This property produces a quasi-static energy-level shift that reversibly follows the envelope of the laser pulse, avoiding all real electronic transitions (including ionization). The reaction Hamiltonian returns to its field-free form after the DSC pulse has passed, and, in this sense, DSC modifies the propagator during the propagation. Ultrafast lasers can produce vastly higher fields and more rapidly varying intensity envelopes than can DC electric field sources; therefore, they have the capacity to strongly shape a specific potential energy landscape on ultrafast time scales.

State intersection problems can be considered in two representations, the diabatic and the adiabatic, which are connected by a unitary transformation. In the limit of slow velocity along the reaction coordinate, the system follows the adiabatic potential. In the limit of high velocity, it follows the diabatic potential. The Landau-Zener (LZ) formalism offers insight into how DSC affects the curve-crossing probability (27). The simple LZ formula gives the probability for nonadiabatically hopping from one surface to another: $Math$, where R is the reaction coordinate, v is the reaction coordinate velocity, V23 is the coupling between the channels, and the Vi terms are the diabatic potential energy surfaces. In IBr, the two diabatic curves correlate to atomic states that are not dipole coupled, and estimates suggest that the dipole coupling between these diabats is small compared with all other couplings (28). The parameters on which P depends are all influenced by the dynamic Stark effect. A key effect is the differential Stark shifting of the diabatic potential energy surfaces, which displaces the crossing point so that the coordinate velocity at the intersection is altered. DSC can also be understood in terms of the adiabatic potentials as $Math$. The adiabatic potential barriers rise and fall as the diabats are Stark shifted, altering the wavepacket velocity and hence the hopping probability. Therefore, the simple LZ formula suggests that one can expect sensitive catalytic control by altering the barrier at the intersection.

The raw I-atom speed distributions were plotted as a function of control pulse delay, Δt (Fig. 3). A key point is that at all delay times the recoil velocities of the I atom fragments remain essentially constant (i.e., follow vertical lines), demonstrating that IR control field has not excited new electronic states. There are two time delays when the reaction is critically sensitive to the control field: during initiation and during traversal of the crossing point. If the applied control pulse is simultaneous with the initiation pulse, the reaction begins on a Stark-shifted surface. The Stark lowering of the ground state has an effect equivalent to spectrally red-shifting the initiation wavelength. The result is that, as the control pulse fades, the wavepacket velocity is transiently reduced at the crossing point, decreasing the hopping probability and enhancing the Br channel. Application of the control pulse during traversal shifts and lowers the adiabatic barrier, thereby increasing the Landau-Zener hopping probability and thus production of Br*. Alternatively, in the diabatic picture, the crossing point shifts in the field, leading to a modified velocity at that point.

DSC is demonstrated by the variation of the overall integrated Br*/Br branching ratio as a function of IR pulse delay, Δt (Fig. 4). At early and late times, the speed distribution is identical to that of the molecule under field-free conditions. This result demonstrates the truly reversible nature of the DSC interaction: There are no real electronic or ionizing transitions due to the application of the control pulse. Application of the control field at Δt = 0 fs results in a 60% enhancement of Br yield at the expense of the Br* channel. Conversely, during control at Δt = 300 fs (traversal of the crossing), the reaction favors Br* production by more than 30%. The peak-to-valley contrast is over 90%, and, importantly, control is exerted on 100% of the reacting population. The experimental result is compared with numerical simulations based on a three-state split-operator approach, described elsewhere (29). The maximal Stark shifts used in the simulations are +0.125 eV for the diabatic Y3Σ(O+) state and –0.022 eV for the diabatic B3Π(O+) state. The simulation results are in good agreement with experiment, and although the polarizabilities of the (relativistic) IBr states are difficult to calculate (30), DSC will always be possible if there are differential polarizabilities between states.

We have shown that the nonresonant dynamic Stark effect can be used to dynamically alter a potential energy barrier in a photochemical reaction, promoting the formation of a given product. Variants of DSC that incorporate Raman pumping will be applicable to ground-state reactions. Pulse-shaping methods from the quantum control toolbox will also prove useful. For example, implementing DSC with adaptive-feedback techniques will lead to the design of custom-shaped Stark-control laser pulses. As well, it will be possible to use interference effects in DSC to alter, for example, excited-state lifetimes (29). The frequency independence, the avoidance of excited state chemistry, and the universal applicability of the nonresonant dynamic Stark effect should prove important for scaling DSC to larger and more complex systems.