The dynamics of collapsing bubbles close to a flat solid is investigated with respect to its potential for removal of surface attached particles. Individual bubbles are created by nanosecond Nd:YAG laser pulses focused into water close to glass plates contaminated with melamine resin micro-particles. The bubble dynamics is analysed by means of synchronous high-speed recordings. Due to the close solid boundary, the bubble collapses with the well-known liquid jet phenomenon. Subsequent microscopic inspection of the substrates reveals circular areas clean of particles after a single bubble generation and collapse event. The detailed bubble dynamics, as well as the cleaned area size, is characterised by the non-dimensional bubble stand-off γ=d/Rmax, with d: laser focus distance to the solid boundary, and Rmax: maximum bubble radius before collapse. We observe a maximum of clean area at γ≈0.7, a roughly linear decay of the cleaned circle radius for increasing γ, and no cleaning for γ>3.5. As the main mechanism for particle removal, rapid flows at the boundary are identified. Three different cleaning regimes are discussed in relation to γ: (I) For large stand-off, 1.8<γ<3.5, bubble collapse induced vortex flows touch down onto the substrate and remove particles without significant contact of the gas phase. (II) For small distances, γ<1.1, the bubble is in direct contact with the solid. Fast liquid flows at the substrate are driven by the jet impact with its subsequent radial spreading, and by the liquid following the motion of the collapsing and rebounding bubble wall. Both flows remove particles. Their relative timing, which depends sensitively on the exact γ, appears to determine the extension of the area with forces large enough to cause particle detachment. (III) At intermediate stand-off, 1.1<γ<1.8, only the second bubble collapse touches the substrate, but acts with cleaning mechanisms similar to an effective small γ collapse: particles are removed by the jet flow and the flow induced by the bubble wall oscillation. Furthermore, the observations reveal that the extent of direct bubble gas phase contact to the solid is partially smaller than the cleaned area, and it is concluded that three-phase contact line motion is not a major cause of particle removal. Finally, we find a relation of cleaning area vs. stand-off γ that deviates from literature data on surface erosion. This indicates that different effects are responsible for particle removal and for substrate damage. It is suggested that a trade-off of cleaning potential and damage risk for sensible surfaces might be achieved by optimising γ.