Raman interactions in alkali vapours are used in applications such as atomic clocks, optical signal processing, generation of squeezed light and Raman quantum memories for temporal multiplexing. To achieve a strong interaction the alkali ensemble needs both a large optical depth and a high level of spin-polarisation. We implement a technique known as quenching using a molecular buffer gas which allows near-perfect spin-polarisation of over in caesium vapour at high optical depths of up to a factor of 4 higher than can be achieved without quenching. We use this system to explore efficient light storage with high gain in a GHz bandwidth Raman memory.
Raman interactions in alkali vapours are used in applications such as atomic clocks, optical signal processing, generation of squeezed light and Raman quantum memories for temporal multiplexing. To achieve a strong interaction the alkali ensemble needs both a large optical depth and a high level of spin-polarisation. We implement a technique known as quenching using a molecular buffer gas which allows near-perfect spin-polarisation of over in caesium vapour at high optical depths of up to a factor of 4 higher than can be achieved without quenching. We use this system to explore efficient light storage with high gain in a GHz bandwidth Raman memory.
Quantum memories are key components in quantum information networks. Their ability to store and retrieve information on demand makes repeat-until-success strategies scalable. Warm alkali-metal vapours are interesting candidates for the implementation of such memories, thanks to their long storage times and experimental simplicity. Operation with the Raman protocol enables high time-bandwidth products, which allows for multiple synchronisation trials of probabilistically operating quantum gates via memory-based temporal multiplexing. This makes the Raman memory a promising tool, whose broad spectral bandwidth facilitates direct interfacing with other photonic primitives, such as single photon sources. Here, such a light-matter interface is implemented in a warm caesium vapour. Firstly, we study the storage of polarisation-encoded information in the memory. High quality polarisation preservation for bright coherent state input signals can be achieved, when operating the Raman memory in a dual-rail configuration inside a polarisation interferometer. Secondly, heralded single photons are stored in the memory. To this end, the memory is operated on-demand by feed-forward of source heralding events, which is a key technological capability. Prior to storage, single photons are produced in a spontaneous parametric down conversion source, whose bespoke design spectrally tailors the photons to the memory acceptance line. The faithful retrieval of stored single photons is found to be currently limited by noise in the memory, with a signal-to-noise ratio of 0.3 in the memory output. Nevertheless, a clear influence of the input's quantum nature is observed in the retrieved light by measuring signal's photon statistics. Finally, the memory noise processes are examined in detail. Four-wave-mixing noise is determined as the sole important noise source for the Raman memory.
Within any practical proposal for utilising quantum mechanics for information processing whether for the efficient computation of problems intractable by classical methods, or for secure communication; at some stage one must have the means to transfer quantum information between remote nodes of a network. For this, light is the obvious choice. To realise this vision one requires the means to overcome the ``scaling problem'' intrinsic to many photonic-based quantum technologies due to probabilistic operations. It has been identified that optical quantum memories which facilitate the storage and retrieval of quantum states of light are an enabling technology. Another requisite technology to ensure the scalability of a quantum network is the means to interface dissimilar material nodes, which in practice means the translation of quantum light in bandwidth, frequency and temporal shape. The first part of this thesis presents experimental, theoretical and numerical investigations of noise reduction in the Raman memory protocol in thermal caesium vapour, by means of a cavity. To do this, I develop a theoretical description of the cavity memory interaction, along with a model of the atom-cavity system to enable meeting the required resonance conditions. This is followed by a proof-of-concept experimental demonstration, showing suppression of noise in the retrieved state. To conclude this part, I investigate the optimisation of this system and provide a numerical framework for its design, and propose a route towards realising the Raman memory as a practical quantum memory. The second theme is an exploration of the practical application of the Raman memory as an interface for temporal modes of light. I perform a preliminary investigation, and develop characterisation tools, to experimentally verify the modal structure of the memory interaction. This work provides the basis for deploying the Raman memory as a temporal-mode selective device for GHz bandwidth quantum states of light.
Photonic quantum information processing has emerged as a powerful platform for realising quantum-enhanced technologies. In order to be scalable, many of these technologies depend on the availability of a suitable quantum memory for the coherent storage and on-demand retrieval of photonic quantum states. In this thesis, I investigate broadband light storage in a room-temperature Raman memory, implemented both in free space and, for the first time, inside a low-finesse optical cavity designed for low-noise operation. The ability of the Raman memory to preserve phase coherence was tested by storing coherent polarisation states in two spatially separate atomic ensembles. Polarisation storage with a fidelity of up to 97 ± 1% was demonstrated by performing full process tomography on the system. The Raman memory was then interfaced for the first time with a spontaneous parametric downconversion (SPDC) source of heralded, GHz-bandwidth single photons. The memory performance was characterised by measuring the second-order autocorrelation of the retrieved fields. While the SPDC input photon statistics showed a clear influence on the statistics of the retrieved field, four-wave mixing (FWM) noise, stimulated by spontaneous Raman scattering, prevented the preservation of non-classical photon statistics during read-out. Suppressing this source of noise represents the last remaining challenge for realising a broadband single-photon Raman memory suitable for quantum information applications. To this end, I demonstrate a novel cavity implementation of the Raman memory which reduces the FWM contribution relative to the signal field by re-distributing the density of states into which the noise photons can be scattered. Cavity-enhanced memory operation was investigated using weak coherent input states, showing a significant improvement of the signal-to-noise ratio compared to the free-space memory implementation. This proof-of-principle demonstration suggests that cavity Raman memories may offer a practical route towards low-noise, high-bandwidth quantum storage at room temperature.
Abstract Quantum memories are essential for quantum information processing. Techniques have been developed for quantum memory based on atomic ensembles. The atomic memories through optical resonance usually suffer from the narrow-band limitation. The far off-resonant Raman process is a promising candidate for atomic memories due to broad bandwidths and high speeds. However, to date, the low memory efficiency remains an unsolved bottleneck. Here, we demonstrate a high-performance atomic Raman memory in 87 Rb vapour with the development of an optimal control technique. A memory efficiency of above 82.0% for 6 ns~20 ns optical pulses is achieved. In particular, an unconditional fidelity of up to 98.0%, significantly exceeding the no-cloning limit, is obtained with the tomography reconstruction for a single-photon level coherent input. Our work marks an important advance of atomic memory towards practical applications in quantum information processing.
Broadband quantum memories hold great promise as multiplexing elements in future photonic quantum information protocols. Alkali-vapor Raman memories combine high-bandwidth storage, on-demand readout, and operation at room temperature without collisional fluorescence noise. However, previous implementations have required large control pulse energies and have suffered from four-wave-mixing noise. Here, we present a Raman memory where the storage interaction is enhanced by a low-finesse birefringent cavity tuned into simultaneous resonance with the signal and control fields, dramatically reducing the energy required to drive the memory. By engineering antiresonance for the anti-Stokes field, we also suppress the four-wave-mixing noise and report the lowest unconditional noise floor yet achieved in a Raman-type warm vapor memory, (15±2)×10^{-3} photons per pulse, with a total efficiency of (9.5±0.5)%.
This thesis is about the experimental implementation of a high-speed and robust quantum memory for light. A novel far off-resonant Raman approach to ensemble-based quantum memories in a room-temperature environment is developed and demonstrated. Storage and retrieval of sub-nanosecond, weak coherent light pulses at the single-photon-level with total efficiencies exceeding 30% and storage times of up to 4 μs are achieved. The coherence of the memory is shown by directly interfering a copy of the incident signal with the retrieved signal from the memory. The unconditional noise floor of the memory is found to be low enough to operate the memory in the quantum regime at room temperature. Multiple readout of a single stored excitation is demonstrated, suggesting that 100% readout is possible in different temporal modes. Furthermore, first results regarding the storage and retrieval of polarisation encoded qubits are obtained. This and the memory’s ability to operate in the quantum regime at room temperature with a low unconditional noise floor illustrate its potential usefulness for real world applications.