Nanoparticle formation, growth and transport are important topics in several contexts, such as cloud formation, particle synthesis and additive manufacturing. This thesis approaches the subject with a broad perspective from molecular to the micro- scale, utilizing theoretical analysis, computational simulation as well as experiment observations. First, general dynamic equations are non-dimensionalized and applied to simulate aerosol formation and growth in a constant rate reaction reactor. Dimensionless equations lead to results that are independent of condensing species formation rates. The effect of particle sink processes (e.g. evaporation, wall loss, loss to preexisting particles and dilution) and acid-base reactions are systematically investigated. Errors involved with common methods used for deducing particle growth rates from experimental observations are discussed. The results suggest the maximum overestimation error for true particle growth rates occurs when particle nucleation and growth are collision controlled. Second, tandem mobility-mass spectrometry is utilized to understand sorption of organic vapors onto cluster ions. It is found that cluster structure, polarity and the molecular structure of the condensing vapors all influence uptake by cluster ions, qualitatively in agreement with previous activation efficiency measurements for condensational particle counters. Third, nanoparticle transport in an aerosol deposition device is probed with fluid dynamics and particle trajectory simulations. To facilitate particle trajectory simulations, a neural network based drag law is developed that can be applied over a wide range of Knudsen and Mach numbers. Simulation results reveal both particle impaction speeds and particle focusing effects are size dependent, with optimal particle sizes for maximizing particle impaction speed and focusing. With a newly developed framework, mass, momentum and kinetic energy fluxes from particles to the substrate are calculated. It is shown the kinetic energy flux can be above 104 W m-2 for modest aerosol concentrations due to particle focusing. Finally, classification and prediction of different types of lung cell are performed with machine learning algorithms, using the volatile organic compound profiles of different cell populations. These profiles are obtained by a proton transfer reaction mass spectrometer with high resolution. Proper data processing procedures are found to be the key to differentiate cell populations with the measured profiles.