We clarify the difference between the SiH4 consumption efficiency eta and the SiH4 depletion fraction D, as measured in the pumping line and the actual reactor of an industrial plasma-enhanced chemical vapor deposition system. In the absence of significant polysilane and powder formation, eta is proportional to the film growth rate. Above a certain powder formation threshold, any additional amount of SiH4 consumed translates into increased powder formation rather than into a faster growing Si film. In order to discuss a zero-dimensional analytical model and a two-dimensional numerical model, we measure eta as a function of the radio frequency (RF) power density coupled into the plasma, the total gas flow rate, the input SiH4 concentration, and the reactor pressure. The adjunction of a small trimethylboron flow rate increases eta and reduces the formation of powder, while the adjunction of a small disilane flow rate decreases eta and favors the formation of powder. Unlike eta, D is a location-dependent quantity. It is related to the SiH4 concentration in the plasma c(p), and to the phase of the growing Si film, whether the substrate is glass or a c-Si wafer. In order to investigate transient effects due to the RF matching, the precoating of reactor walls, or the introduction of a purifier in the gas line, we measure the gas residence time and acquire time-resolved SiH4 density measurements throughout the ignition and the termination of a plasma. (c) 2015 AIP Publishing LLC.