INTRODUCTIONA critical question arising during aerosol deposition to surfaces is: How many arriving particles eventually stick to the collector surface? The sticking probability π s of a single particle-surface collision event is either 1 or 0, dictated by the specific contact physics. It is instructive to think that the transition from 1 to 0 occurs whenever some parameters λ i (for example the impact velocity), affecting the energetics of the collision reaches a critical value λ i , c r i t if all other parameters involved are kept constant. When considering an ensemble of collision events, each with its own set of λ i values however, it is convenient to introduce the notion of a sticking fraction S ≡ π s , representing the overall fraction of impacting particles captured by the collector. In this way it is the ensemble averaging of the various collision events that gives rise to S values between 1 and 0 and sometimes, S itself is called the sticking probability or sticking coefficient. For clean convex collectors exposed to a unidirectional aerosol flow S can be also defined in terms of the ratio of the area over which particle capture occurs to the total projected area of the collector.METHODSIn the asymptotic limit of high particle Stokes number, Stk simple, analytic expressions can be derived for the sticking fraction S of particles inertially impacting on clean convex collectors, based on the commonly used sticking criterion that requires the normal impact velocity of the particle V n to be less than a critical value V c r i t specific to the particular particle-surface combination, see e.g. [1]. For this criterion the impact velocity at the stagnation point of the collector V i , s p , (which is practically the same as the far upstream fluid velocity U ∞ ) the collector geometrical shape and the critical velocity, V c r i t , suffice to determine S. For finite Stk it becomes necessary to trace particle trajectories in order to compute the area of the collector that receives particles.RESULTS-DISCUSSIONFor high Stk, V n /V i , s p follows a cosine distribution and the following expressions based on the normal impact velocity criterion (V n >V c r i t ) were explicitly derived: These are compared to available experimental data obtained using isolated cylindrical collectors and granular bed filters in Figs. 1a & 1b.Inertial impaction in a granular bed can be adequately correlated in terms of an effective particle Stokes number StK e f f , considering a single unit spherical collector exposed to a uniform particle stream of an effective velocity U e f f ≡ A u s where A is a hydrodynamic factor depending on bed porosity and Reynolds number and u s is the superficial bed velocity. Since particle rebound appears to start for Stk * e f f ( 1 from Fig. 1b) larger by at least an order of magnitude from the value at which inertial impaction sets in (i.e. Stk e f f Stk c r i t see Fig. 6 in), it is reasonable to assume that the stagnation point impact velocity does not differ much from the free stream flow velocity U e f f . Accordingly, the ratio of Stk e f f to Stk * e f f where particle rebound sets in can be taken to correspond to the ratio V i , s p /V c r i t in Eq. (1). As seen from Fig. 1b the theory describes these data very well. Incidentally, ref. [5] gives a least square empirical fit of the data with the exponent of the power law in Fig. 1b equal to 1.968, which is very close to the derived value 2 of Eq. (2). We do not observe here the deviation seen at higher impact velocities in the cylindrical collector case of Fig. 1a, presumably because the data do not extend to such sufficiently high impact velocities.For sufficiently high impact energy however, the data for cylindrical collectors exhibit a deviation from the simple asymptotic theory which can be explained including in the analysis, the influence of the tangential component of the impact velocity V t , as well as the finiteness of the particle Stokes number. In this case, a new oblique impact rebound criterion is derived motivated by a scaling analysis: V t /V n ≥ const which for elastic-frictional particle-collector contact becomes V t /V n ≥ const 1 μ (E*/G*) 1 2 with E* and G* being effective Young's and Shear moduli respectively, see e.g. [1]. The new oblique impact rebound criterion amounts to the existence of a critical impact angle beyond which no deposition is possible irrespective of the value of V n /V c r i t . The range of Stk numbers in the experiments of [4] is from 0.28-1.5 and the S data are well accounted for by our theoretical estimates for this range of Stk.CONCLUSIONSThe augmented theory is in excellent agreement with the available experimental data on the onset of particle rebound and should increase significantly the reliability of inertial deposition rate predictions for clean collectors.