Achieving Sub-Shot-Noise Absorption-Spectroscopy with Avalanche Photodiodes and with a Charge-Coupled Device

Chemical and biological samples can be interrogated by measuring the spectral absorption of light. When low intensity is used to illuminate a sample, the signal-to-noise ratio of absorption features is limited fundamentally by the quantum nature of light. When an intensity measurement is performed on the coherent state we obtain a probabilistic measurement outcome for the intensity with a Poisson distribution. This is the shot noise limit and this uncertainty impacts on estimates of physical properties that are derived from an optical measurement. There are scenarios where it is advantageous to maximize the precision achieved whilst minimizing the intensity of the input probe. It has been shown that Fock states can enable precision in absorption estimation beyond the shot noise limit and that Fock states are the optimal quantum state for maximizing precision. In the present work we use correlated photon pairs to achieve sub-shot-noise absorption spectroscopy. The experiment uses a type II phase matched PPKTP crystal pumped by 404nm light to generate separable pairs of photons. We adjust the crystal temperature to tune the wavelength of a probe photon beam through the absorption features of a sample using the other beam either as a herald or reference. We use this setup to demonstrate evidence for sub-shot-noise performance towards the ultimate quantum limit using three forms of measurement: (i) Using two commercial avalanche photodiodes, we observe a precision advantage of 27.5 ± 8% per photon detected (ii) We incorporate an optical delay and switch into the previous scheme. This feed forward scheme yields sub-shot-noise performance per photon flux and we observe an advantage of up to 27.5 ± 8%; (iii) In our latest work we direct the two orthogonally polarized output channels of the source onto two separate pixel regions of a cooled, high efficiency, commercial CCD camera. Here we observe sub-shot-noise performance with a mean normalized variance of the detected intensity difference between the two beams of 0.40 ± 0.01. As this is b