Search for: All records

In a conventional atomic interferometer employing$N$atoms, the phase sensitivity is at the standard quantum limit:$1/\sqrt{N}$. Under usual spin squeezing, the sensitivity is increased by lowering the quantum noise. It is also possible to increase the sensitivity by leaving the quantum noise unchanged while producing phase amplification. Here we show how to increase the sensitivity, to the Heisenberg limit of$1/N$, while increasing the quantum noise by$\sqrt{N}$and amplifying the phase by a factor of$N$. Because of the enhancement of the quantum noise and the large phase magnification, the effect of excess noise is highly suppressed. The protocol uses a Schrödinger cat state representing a maximally entangled superposition of two collective states of$N$atoms. The phase magnification occurs when we use either atomic state detection or collective state detection; however, the robustness against excess noise occurs only when atomic state detection is employed. We show that for one version of the protocol, the signal amplitude is$N$when$N$is even, and is vanishingly small when$N$is odd, for both types of detection. We also show how the protocol can be modified to reverse the nature of the signal for odd versus even values of$N$. Thus, formore »a situation where the probability of$N$being even or odd is equal, the net sensitivity is within a factor of$\sqrt{2}$of the Heisenberg limit. Finally, we discuss potential experimental constraints for implementing this scheme via one-axis-twist squeezing employing the cavity feedback scheme, and show that the effects of cavity decay and spontaneous emission are highly suppressed because of the increased quantum noise and the large phase magnification inherent to the protocol. As a result, we find that the maximum improvement in sensitivity can be close to the ideal limit for as many as 10 million atoms.