Calibration of a single-atom detector for atomic microchips

We experimentally investigate a scheme for detecting single atoms magnetically trapped on an atom chip. The detector is based on the photoionization of atoms and the subsequent detection of the generated ions. We describe the characterization of the ion detector with emphasis on its calibration via the correlation of ions with simultaneously generated electrons. A detection efficiency of $47.8\pm 2.6\%$ is measured, which is useful for single-atom detection, and close to the limit allowing atom counting with sub-Poissonian uncertainty.

Figure 1

(Color online) (a) Detection scheme, showing the position of the detector relative to the atom chip. The geometry allows ions to be extracted from close to the chip surface. (b) Numerical simulation of the ion trajectories. With a suitable choice of electrode voltages, ions within a well-defined region can be guided into the CEM with high efficiency, while ions outside this region are not detected.

Figure 2

Setup for calibration of the ion detector. Rubidium atoms from a background vapor are photoionized within the mode of an optical resonator. The generated electrons are observed with an additional CEM and the ion-electron coincidence signal is also recorded.

Figure 3

(Color online) Ion detection rate as a function of $U_e$, with $U_t$ and $U_d$ set to their optimized values. Saturation of the detection rate is observed for large values of $U_e$. The error bars account for Poissonian fluctuations and the measured fluctuations in the ion detection rate of 2.9%.

Figure 4

(Color online) Normalized ion detection rate as a function of $U_t∕U_e$, with $U_d$ optimized. Experimental results for three different values of $U_e$ and results from the numerical simulation are shown.

Figure 5

(Color online) Normalized ion detection rate as a function of $U_t∕U_e$ and $U_d∕U_e$. The region between the dashed lines is the region predicted by the numerical simulations for which ions impact the CEM.

Figure 6

(Color online) Coincidence spectrum showing the number of electron-ion coincidences counted in 60 s within a 20 ns coincidence window as a function of the delay time of that window. The error bars correspond to the square roots of the data, assuming the detection of coincidences follows Poisson statistics. A Gaussian fit is also shown (solid line) and has a mean value of $3.45\mu \text{s}$ and a standard deviation of 50 ns.

Figure 7

(Color online) Ion detection efficiency as a function of the voltage across the ion CEM. Saturation at about 50% is observed for large values of the CEM voltage.