First accuracy evaluation of the NPL-CsF2 primary frequency standard

An accuracy evaluation of the caesium fountain NPL-CsF2 as a primary frequency standard is reported. The device operates with a simple one-stage magneto-optical trap as the source of cold atoms. Both the uncertainty in and magnitude of the cold collision frequency shift are reduced by taking advantage of the dependence of the cross section on the effective collision energy in an expanding atomic cloud. The combined type B uncertainty (typically 4 x 10(-16)) is dominated by an estimate of the frequency shift due to the distributed cavity phase. When operated at single density, the short-term fractional frequency instability of NPL-CsF2 is 1.7 x 10(-13) at 1 s and limited by the noise of the room temperature quartz-based local oscillator. During a typical frequency measurement campaign, the fountain is operated in an alternating mode at high and low density in order to measure and correct for a residual collision shift. This increases the effective fractional frequency instability to 5.4 x 10(-13) at 1 s; consequently the averaging time required for the type A uncertainty level to match that of the type B is 20 days.