Velocity distributions of atomic beams by gated optical pumping

In evaluating the accuracy of a cesium-beam frequency standard, accurate measurement of the atomic velocity distribution is important. In frequency standards which employ atoms with thermal velocities, the measured atomic resonance frequency differs from the true resonance by several parts in 10¹³ due to the second-order Doppler shift. To achieve the frequency accuracy goal for NIST-7 or 1.10^-14, the uncertainty in the second-order Doppler shift must be no more than a few parts in 10¹⁵. This requirement establishes an upper bound on the uncertainty of the mean-square atomic velocity of about 1 percent. We present the results or experiments designed to measure the velocity distribution of NIST-7 using two independent techniques: gated optical pumping, and Ramsey fringe inversion. We show that these techniques yield velocity distributions and corresponding second-order Doppler shifts that agree within the stated tolerances