Molecular control uses tailored light to steer the dynamics of molecules — breaking a chosen bond, selecting a reaction product, exciting a vibrational or rotational state, aligning or orienting the molecule, telling enantiomers apart. It is one of the historical roots of quantum optimal control, and several of the field’s ideas and algorithms were born here [1, 2, 3].

Origins

Two schemes started the field in the 1980s. Tannor and Rice proposed timing a pump and a dump pulse so that a wave packet reaches the desired reaction channel [4, 5]. Brumer and Shapiro proposed interfering several quantum pathways that lead to the same final state, so that the relative phase of the fields decides the outcome [6]. Optimal control theory generalized both: the field is the solution of a variational problem [7, 8, 9]. Monotonically convergent algorithms of the Krotov type were developed for these problems [10, 11]. See Problem formulation & figures of merit.

Learning control in the laboratory

Molecular Hamiltonians are rarely known well enough for open-loop design. Judson and Rabitz proposed letting the experiment itself evaluate candidate pulses, with a learning algorithm choosing the next ones [12]. Femtosecond pulse shapers based on spatial light modulators made this practical [13]. Notable results:

  • Feedback-optimized phase-shaped pulses changed the branching ratio of a photodissociation reaction [14].
  • Optimally tailored strong-field pulses selectively broke and rearranged chemical bonds [15].
  • Shaped pulses controlled the flow of energy in a biological light-harvesting complex [16].
  • The spectral phase of a femtosecond pulse controls two-photon transitions through interference, so that shaped pulses can either excite efficiently or leave the transition dark [17].

Why learning control works so often is explained by the structure of control landscapes, which for controllable systems are free of traps [18]. See Closed-loop & in-situ optimization and Gradient-free & global optimizers.

Rotation, alignment and chirality

Molecular rotation is well suited to control: its Hamiltonian has few, accurately known parameters [19]. Strong non-resonant laser pulses align and orient molecules [20]. For chiral molecules, phase-controlled microwave pulses that drive a closed loop of three rotational transitions detect enantiomers [21]. They can also transfer one enantiomer selectively to a higher rotational state [22].

Ultracold molecules

At ultracold temperatures, molecules can be controlled state by state. A single STIRAP step transferred weakly bound KRb molecules to their rovibrational ground state, producing a dense gas of polar molecules [23]. Coherent control of photoassociation, which forms molecules from atom pairs with shaped light, has been reviewed in [24]. Optimal control has also been used to design laser cooling of internal molecular degrees of freedom [25]. See Adiabatic passage.

Limitations

  • Polyatomic molecules have many coupled degrees of freedom. Accurate potential surfaces and dipole moments are rarely available, which restricts open-loop design.
  • In condensed phases, fast dephasing limits coherent control to short times.
  • Pulses found by learning control are often hard to interpret, and strong fields compete with ionization.