Control over the nucleation of new phases is highly desirable but elusive. Even though there is a long history of crystallisation engineering by varying physicochemical parameters, controlling which polymorph crystallises or whether a molecule crystallises or forms an amorphous precipitate is still a black art. Although there are now numerous examples of control using laserinduced nucleation, a physical understanding is absent and preventing progress. Here we show that the proximity of a liquid-liquid critical point or the corresponding binodal line can be used by a laser-tweezing potential to induce concentration gradients. A simple theoretical model shows that the stored electromagnetic energy of the laser beam produces a free-energy potential that forces phase separation or triggers the nucleation of a new phase. Experiments in a liquid mixture using a low-power laser diode confirm the effect. Phase separation and nucleation through a laser-tweezing potential explains the physics behind non-photochemical laser-induced nucleation and suggests new ways of manipulating matter.The nucleation of new phases such as crystals from solution is of enormous technological importance but poorly understood. Although the vast majority of pharmaceutical products and most fine and speciality chemical products are made in crystalline form, industrial crystallisation has changed little in the past 350 years and suffers from an embarrassing lack of control with sometimes unexpected and severe financial consequences.1 Therefore, a deeper understanding of and control over (crystal) nucleation is of great importance.Two decades ago, it was shown that nanosecond pulsed lasers can nucleate crystals in a supersaturated solution through a non-photochemical process.2 Most significantly, it was shown that the crystal polymorph could be selected by the polarisation state of the laser.3 Subsequent work has shown that laser pulses can induce nucleation of various crystals, 4,5 liquid crystals, 6,7 and bubbles. 8 It was initially suggested that the Kerr effect was responsible for laser-induced nucleation 2 but subsequent work has shown that the data are inconsistent with it. 9 Most disconcertingly, the reproducibility of polarisation-control over polymorph selection has been questioned while pulsedlaser induced nucleation appears to require impurity particles.10 However, a laser-induced nucleation experiment using CW lasers did reliably show polarisation-control over polymorph selection.11 Thus, it is fair to say that a physical understanding of these phenomena is still sorely lacking.In a supersaturated solution, the crystal is the thermodynamically most stable state but is difficult to access in the absence of heterogeneous nucleation sites. In the simplest case, random fluctuations lead to the formation of a critical nucleus and ultimately a macroscopic crystal. As suggested by Frenkel for proteins 12 and more generally in non-classical nucleation theories, 13-15 the concentration fluctuations leading to nucleation may well be enhanced by...