Is the photon recoil equal to the photon momentum?

It is well known that an isolated atom recoils with the momentum of the photon when it absorbs a photon from an infinite plane wave. When the transverse electromagnetic field has a finite extent, the photon wave vector in the longitudinal direction is smaller. When an isolated atom absorbs a photon from a finite laser beam, is the atomic recoil smaller than that for an infinite plane wave? We show that it is smaller and it has a peculiar dependence on the transverse variation of the field. For atom interferometers that measure the photon-recoil, the difference in the size of the photon recoil is safely below the current measurement accuracy for the laser beams that are used. A related problem is the photon-recoil frequency-shift for microwave atomic clocks. For a cesium clock, the usual photon-recoil frequency shift of h$^{2}$k$^{2}$/2m gives a fractional frequency shift of 1.5$\times $10$^{-16}$, which is not far below the accuracy of current clocks. We show that the frequency shift does not have the usual form and has a similar behavior as for the transverse field variations above. It depends in unusual ways on the interrogation time, and the size of the atomic wavefunction at the first interaction.