What factor decides that when will scattering happen and when will reflection? I don't understand what is the difference between Scattering and reflection?
I have searched and found that in scattering, the atom absorb and re emits the photon, while reflection is due to particle nature(elastic collision). But I don't understand what factor decides, that when a photon hits an atom, weather scattering will take place or reflection. 
Also on Wikipedia I read that, Rayleigh scattering is elastic scattering of photon in which scattered photon have little less energy due to recoil of atom , but if scattering shows wave nature, then how recoil of atom is possible?
(I know this question is asked, but I want deep understanding of this concept, that how actually atom behaves in field of photon)
 A: Reflection is a form of scattering, but it is coherent scattering.  When scattering occurs from a lot of identical particles rigidly fixed in a plane, then all the possible ways a single photon can be scattered from the plane of particles are fixed in phase.  That is coherent scattering, and it is reflection.  If the particles instead are, e.g., in a gas, the phase of the scattered light changes continuously.  The scattered light is both temporally and spatially incoherent so we say it is scattered rather than reflected.
There is an intermediate case which is easy to observe: a rigid rough  surface covered with the particles.   Light scattered from the particles is temporally coherent because nothing is moving, but it is spatially incoherent because of the roughness of the surface (the scattering particles are not confined to a plane).  What you end up with in the reflected/scattered light is a speckle pattern.  The speckle pattern does not move because it is caused by the (fixed) locations of the scattering particles, but we would not normally call it a reflection; we call it scattered light.  
However, a clever holographer can record a hologram of the rough surface and its speckle pattern, use the reconstruction of the speckle pattern to illuminate the hologram of the surface, and obtain a nice clean time-reversed version of the original illumination that struck the rough surface. No information is lost. 
When the scattering is both temporally and spatially incoherent, it's not possible to record the holograms nor to reconstruct the time reverse of the original illumination. Almost all information about the scatterer and the illumination is lost.
A: In both cases, an electron absorbs and re-emits a photon. The difference is in the structure to which the electron belongs. In Rayleigh scattering by the atmosphere, electrons belong to molecules in a random (gaseous) structure in which each molecule moves independently from other molecules. The photon is typically re-emitted with its initial momentum, or the molecule may emit more than one photon, with some recoil to the molecule. Because the molecules in the air have no structural dependence on each other, one can regard this as a particle process. It is described as an incoherent process, leading to scattering (scattered photons have random phase shifts).
In a structured material (a solid), the photon may be transmitted (transparency) or reflected. In a metal, there are free electrons which very readily absorb photons. If the photon is transmitted in one interaction, it will almost immediately be absorbed by another electron. It will almost always end up being reflected (a small proportion will be absorbed, heating the metal).
A photon can only be absorbed by one electron, but we cannot say which electron absorbs and re-emits the electron. According to the laws of quantum mechanics, we have to sum over all the possibilities, using a quantum superposition which is equivalent to wave behaviour. The possible interaction with every possible electron is identical, and it is described as a coherent process (phase shifts for reflected photons are identical). Again, it is a particle process, but calculations in quantum mechanics follow the same maths as wave mechanics (for obscure mathematical reasons which I won't attempt to explain here). Consequently particle processes look like wave processes, and reflection takes place as though the photon were a wave.
A: Google "radiation pressure", When an EM wave interacts with a reflecting surface, current is induced into the material (aluminium for example) in such a direction that the resultant magnetic field is in the same direction as the magnetic component of the incident wave, like magnetic lines repel, this is called Lenz's law. So the aluminium experiences a repulsive force.  this is the principle of light sails. At a non a reflecting boundary only half the force is experienced, the  electrons around the molecule's feel the magnetic component of the impinging wave, vibrating as well as translating the molecules this will cause friction and heating, therefore a form of black body radiation is produced, peaked  in the infrared.  This is called down conversion. Optical light that is down converted to infrared, no transitions, very broadband.  what's more the infrared radiation exerts a force back on the boundary.  this effect has caused a lot of problems for satellites and long range space craft. some spacecraft generate their power from a hot lump of plutonium, the rejected heat has to be radiated away, this imparts unwanted thrust on the spacecraft, unless it's aimed out the back where it would be extremely useful as thrust. Google "pioneer anomaly"
