Intelligent routes to the controlled synthesis of nanoparticles.

We describe an autonomous 'black-box' system for the controlled synthesis of fluorescent nanoparticles. The system uses a microfluidic reactor to carry out the synthesis and an in-line spectrometer to monitor the emission spectra of the emergent particles. The acquired data is fed into a control algorithm which reduces each spectrum to a scalar 'dissatisfaction coefficient' and then intelligently updates the reaction conditions in an effort to minimise this coefficient and so drive the system towards a desired goal. In the tests reported here, CdSe nanoparticles were prepared by separately injecting solutions of CdO and Se into the two inlets of a heated y-shaped microfluidic reactor. A noise-tolerant global search algorithm was then used to efficiently identify-without any human intervention-the injection rates and temperature that yielded the optimum intensity for a chosen emission wavelength.