Silicon nanoparticles have important applications, including nonvolatile floating-gate memory devices. To prevent device performance variations, particle size and oxide thicknesses need to be controlled with a high degree of precision. Additionally, producing well-ordered, two-dimensional arrays of nanoparticles may require the exploitation of self-assembly techniques and colloidal forces, which in turn requires that silicon nanoparticles first come into contact with liquids. Until recently, aerosol silicon nanoparticle collection into liquid was assumed to be an inert process. Once formed, the silicon nanoparticle colloid was assumed to be inert. In fact, silicon nanoparticles produced in the aerosol phase by dilute silane pyrolysis and size classified with a differential mobility analyzer undergo a size reduction upon collection in ethylene glycol, water, and ethanol. Unclassified polydisperse silicon aerosol nanoparticles with an average diameter of 11 nm become monodisperse when collected in a colloid and have a final particle diameter of 2-5 nm. Further evidence suggests that silicon nanoparticles collected in ethanol react with the ethanol to produce tetraalkylorthosilicate-like species. Collections of aerosol silicon nanoparticles in degassed water do not show measurable differences between the aerosol and colloidal size distributions. This reduced reactivity to the solvent indicates that the presence of dissolved oxygen in the solvent may be responsible for the reactivity between the silicon nanoparticles and the solvent. (C) 2004 Elsevier Inc. All rights reserved.