The emergence of quantum physics from the page to the lab and the world at large is an exciting development of recent years. The prospects of absolutely secure communication and efficient simulation of physical systems have spurred great human effort into understanding these possibilities and turning them into realities. Photons are the most easily manipulated quantum particles and are a promising candidate for implementing these technologies. Limitations of photons include the difficulty of keeping objects that move at the speed of light, and producing strong interactions between particles that do not normally interact. The work presented in this thesis is motivated by the possibility of overcoming these limitations. The ability to faithfully store and reproduce a quantum state is essential for many quantum information technologies. Quantum memories for light have been developed over the last two decades to provide this ability. The group at the Australian National University developed the gradient echo memory (GEM): A quantum state of light can be controllably stored and released from an atomic ensemble by the use of additional optical fields and magnetic field gradients. This scheme was previously shown to preserve the quantum characteristics of the light. We used the GEM scheme with a cold rubidium ensemble to create the first optical memory that simultaneously beat the no-cloning limit, a benchmark for many of the technologies relying on quantum memories, and the loss rate for a delay line composed of optical fibre. We also created an analogue to a pulsed optical resonator using GEM with a warm rubidium vapour. This was done by replacing the circulating optical field of a resonator with light stored in the memory, and replacing the coupling of light into and out of that circulating mode with storage and recall from the memory. The bandwidth and repetition rate of this resonator were rapidly tunable as they were controlled by external optical and magnetic fields. We worked on implementing GEM with strings of thousands of atoms strongly coupled to the evanescent field of an optical nanofibre. This raised new possibilities for creating a true random access memory that would allow a more flexible use of the multi-mode capacity of GEM. We developed the theory for a novel type of stationary light in the gradient echo memory. Our stationary light scheme relies on the destructive interference of counter-propagating optical fields throughout the memory. The optical intensity scales with optical depth, as with other forms of stationary light. However, as the destructive interference could be set up over a much greater distance, more of the optical depth is available for generating stationary light. Finally, we studied how a control-phase gate for single-photon optical states could be implemented using a nonlinear interaction with stationary light. The stationary light generated by one state modulates the phase of another state stored in the memory. The second state modifies the stationary light, also producing a back-action on the first state and generating the required cross-phase shift.
The emergence of quantum physics from the page to the lab and the world at large is an exciting development of recent years. The prospects of absolutely secure communication and efficient simulation of physical systems have spurred great human effort into understanding these possibilities and turning them into realities. Photons are the most easily manipulated quantum particles and are a promising candidate for implementing these technologies. Limitations of photons include the difficulty of keeping objects that move at the speed of light, and producing strong interactions between particles that do not normally interact. The work presented in this thesis is motivated by the possibility of overcoming these limitations. The ability to faithfully store and reproduce a quantum state is essential for many quantum information technologies. Quantum memories for light have been developed over the last two decades to provide this ability. The group at the Australian National University developed the gradient echo memory (GEM): A quantum state of light can be controllably stored and released from an atomic ensemble by the use of additional optical fields and magnetic field gradients. This scheme was previously shown to preserve the quantum characteristics of the light. We used the GEM scheme with a cold rubidium ensemble to create the first optical memory that simultaneously beat the no-cloning limit, a benchmark for many of the technologies relying on quantum memories, and the loss rate for a delay line composed of optical fibre. We also created an analogue to a pulsed optical resonator using GEM with a warm rubidium vapour. This was done by replacing the circulating optical field of a resonator with light stored in the memory, and replacing the coupling of light into and out of that circulating mode with storage and recall from the memory. The bandwidth and repetition rate of this resonator were rapidly tunable as they were controlled by external optical and magnetic fields. We worked on implementing GEM with strings of thousands of atoms strongly coupled to the evanescent field of an optical nanofibre. This raised new possibilities for creating a true random access memory that would allow a more flexible use of the multi-mode capacity of GEM. We developed the theory for a novel type of stationary light in the gradient echo memory. Our stationary light scheme relies on the destructive interference of counter-propagating optical fields throughout the memory. The optical intensity scales with optical depth, as with other forms of stationary light. However, as the destructive interference could be set up over a much greater distance, more of the optical depth is available for generating stationary light. Finally, we studied how a control-phase gate for single-photon optical states could be implemented using a nonlinear interaction with stationary light. The stationary light generated by one state modulates the phase of another state stored in the memory. The second state modifies the stationary light, also producing a back-action on the first state and generating the required cross-phase shift.
Quantum memories for light lie at the heart of long-distance provably-secure communication [1], while containing the potential to help break current encryption methods [2], and allow better measurement of quantities than ever before [3]. Demand for a functioning quantum memory is therefore at a premium. Unfortunately, the same properties of light that make it such an effective carrier of quantum information make it difficult to store. Furthermore, by the laws of quantum mechanics, storage must be achieved without measurement to preserve the quantum state. A quantum memory needs to have an efficiency approaching unity without adding noise to the state, and storage times from milliseconds to seconds. Ideally it would also have a high bandwidth and be able to store many pieces of information simultaneously. Many different techniques are currently being developed and much experimental progress has been made over the past few years, with: efficiencies approaching 90% [4]; storage times of over seconds [5]; bandwidths of gigahertz [6, 7]; and over 1000 pieces of information stored at one time [8]. These results were, however, achieved using different memory schemes in different storage media. The challenge now is to reproduce these results with one memory. This thesis focuses on extending the gradient echo memory (GEM) scheme, which shows great promise due to the high efficiencies achieved (87%) [4]. GEM has also been used to demonstrate temporal compression and stretching of pulses, as well as a capacity to arbitrarily resequence stored information [9] and the interference of initially time-separated pulses [10]. Firstly, we demonstrate the noiseless nature of GEM storage in a warm vapour cell to prove that the output from the memory is the best-possible copy of the input allowed by quantum mechanics. We show GEM’s ability to coherently and precisely spectrallymanipulate stored information by having fine control over the memory’s frequency gradient, with potential applications for dynamic conditioning of information inside quantum networks [11]. We demonstrate cross-phase modulation of a stored light pulse with an additional optical field, a process with applications in quantum computing [12]. We also carry out storage of different spatial modes and arbitrary images, demonstrating the potential for orders of magnitude improvement in storage capacity. We then switch from warm vapour cells to cold atomic ensembles to improve the storage time of GEM, seeing a maximum coherence time of 350 μs (seven times that of the warm vapour system) and achieving efficiencies of up to 80%, on a par with the highest efficiency achieved with a cold atomic ensemble [13]. In the process we developed an ultra-dense cold atomic cloud with potential applications in a range of quantum optics experiments. Cold atoms, and the small volumes they occupy, also allowed us to develop an alternative to using magnetic field gradients for our alkali-atom memories in the form of a light-field gradient. This holds promise for extremely fast gradient switching and fine control over the gradient. We also present a digital locking code with application in a range of quantum optics experiments.
The non-interacting and high-speed nature of light makes it an ideal carrier of information that is essential for transmission of quantum information. Indeed, many proposals and demonstrations of quantum cryptography rely on the use of fibre-optic networks. Construction of a memory that can store light and preserve its quantum properties will be useful in a range of quantum information systems such as secure quantum communication and quantum computation. This is why a quantum memory for light is a remarkable objective. The key to quantum memory is to store the probability amplitude of the possible outcomes of measurement but without measurement. An important criterion for a quantum memory is that the efficiency of the recall must exceed 50%. This is the crucial no-cloning limit for security of information, since it guarantees that nobody can access the information by copying it. This benchmark is important because any kind of deterministic amplification of quantum information is fundamentally impossible. On-demand retrieval of information and ability to controllably manipulate the quantum information are also important for quantum applications. When light is absorbed by atoms, it is actually possible to reverse the absorption process. In our memory system: light is absorbed by an ensemble of atoms and, using careful conditioning and control, we can cause the stored light to be regenerated and released at a later time. This is done by applying a gradient of magnetic field along the atomic ensemble that is the basis for our optical memory. To recall the light we flip the sign of the gradient field. This kind of reversible absorption is called photon echo, hence the name of our scheme: The Gradient Echo Memory (GEM). This simple protocol is used in our experiment and can be applied to a range of different atomic systems. We have extended the GEM protocol and experimentally implemented a memory using three-level atoms. We used an off-the-shelf Rb vapour cell operating above room temperature as the memory medium. In this realisation, we broke the efficiency record with 87% recall of the input light pulse. Moreover, through complete state tomography of coherent states we have demonstrated the ability of our memory to noiselessly store quantum states of light. We have also demonstrated that the memory can store a string of pulses and then recall the pulses ondemand in arbitrary order allowing re-sequencing of the stored information. Furthermore, we have shown that pulses could be time-compressed, time-stretched or split into multiple smaller pulses and selectively recalled in several pieces. This technique enables the construction of an optical random-access memory for quantum information. Moreover, the scheme to manipulate the spectral properties of optical data, stored inside the memory, has been introduced. We have also investigated the possibility of obtaining large nonlinear phase shifts between single photons inside the memory. Such strong interactions can be used for the implementation of universal quantum gates.
Devices that harness the surprising properties of quantum systems at scale hold the possibility of revolutionary advances in information technology. Algorithms for quantum computers can take advantage of quantum parallelism to solve problems that are intractable with classical hardware, including prime factorization and simulating chemical processes. A quantum communication network that carries qubits instead of bits can leverage the irreversibility of projective measurement to guarantee secure communication. Because quantum states are fragile, building such devices requires exquisite control over many isolated quantum systems. One of the most critical capabilities is the interaction of stationary and flying qubits. In this thesis we consider how to engineer atom-light interactions as an interface for networked quantum systems. Quantum atom-light couplers require the efficient and reversible interaction of atomic and optical qubits, but the natural interaction of atoms and photons is weak. It is this very lack of interaction between light and its environment that makes light such an excellent carrier of quantum information. There are three promising strategies for stronger coherent atom-light interactions. First, the light field may be confined to a cavity so that it interacts with the atom over a longer time. Second, the light field may be focussed to a small spot with high aperture optics and matched to the natural atomic radiation pattern. Third, we may pass the light field through an ensemble of many atoms and store the qubit in a distributed state. In practice, atom-light couplers employ some combination of all three strategies. We will focus our research on the second two approaches: first by coupling one and two trapped ions with high aperture optics, and then operating a quantum optical memory with an ensemble of approximately 10 billion neutral atoms. Atoms that have been ionized and trapped in a common electromagnetic potential are an advanced few-qubit computation platform. Qubit states are associated with an electronic excitation at each atom, controlled coherently with lasers, and made to interact in two (or many) qubit gates by mutual Coulomb repulsion. One path to quantum advantage with ion-trap quantum processors is to scale small qubit registers by distributing entanglement with photonic interconnects. We consider the feasible efficiency of free-space atom-light couplers for atomic dipole transitions, and derive the atomic image by common free-space couplers. We couple a trapped-ion register with high-aperture lenses and operate a single trapped ion as a photon source. We model the dynamic character of the source and show spin-orbit coupling in the single-photon spatial mode. Although the total collection efficiency is only approximately 0.01, the source has exceptional photon-number purity A = (1.9 +/- 0:2) * 10^-3 such that the higher-order quantum nature of the field persists even after most of the field has been discarded. We derive an efficient quantum non-Gaussian witness and surpass it, the first such demonstration for a trapped-atom single-photon source. We entangle two trapped atoms by single-photon detection, and observe an interference pattern in the spatial mode of the bipartite state. Constructive interference between the entangled components enhances the emission probability by up to 29%, an example of collective enhancement by a two-atom ensemble. The efficiency of free-space atom-light couplers is limited by our capability to engineer high-precision, high-aperture optics. We fabricate ultra-precise hemispheric mirrors with numerical aperture 0.996 by single point diamond turning. The mirrors are amongst the smoothest hemispheric surfaces ever manufactured with root-mean-square (RMS) errors consistently below 25 nm. The smoothest of our mirrors has a RMS error of 14 nm and peak-to-valley error of 88 nm. A mirror with this surface is capable of suppressing or enhancing the spontaneous emission of an atom into free space by 96% in a proposed quantum electrodynamics experiment. We show how these mirrors, with a simple modification, can shape the spatial mode of a trapped atom by similar vacuum-mode engineering. We derive a near-hemispheric mirror coupling scheme that should be 72% efficient with mirrors as precise as ours. We design an ion trap for use with such high-aperture optics. With careful conditioning and control, ensembles of many atoms can be made to store and release photonic qubits on demand. Such optical quantum memories leverage the collective interaction of a light field with many billions of atoms to achieve storage and recall efficiencies approaching unity. We show how stationary light fields for photonic phase gates can be generated in optically deep ensembles, an effect which has since been observed. We implement the gradient echo memory (GEM) scheme in an ultra-high optical depth cold-atom ensemble. The performance of a practical quantum memory is contingent on the chosen qubit encoding. We extend the GEM protocol to allow the simultaneous storage of frequency separated signals and demonstrate that this `dual-rail' memory is suitable for high-fidelity frequency qubits. Dual-rail signals are recalled with 35% efficiency, 82% interference fringe visibility and 6 degrees of phase stability. We describe how the fidelity of the scheme is limited by frequency-dependent polarization rotation and how this may be addressed in an improved configuration. Finally, we demonstrate single-rail storage by GEM with 87% efficiency and 1 ms memory lifetime. Our memory surpasses the no-cloning limit for up to 600 us of storage, out-performing an optical fibre delay line by a factor of six. This is the first quantum memory to beat this important benchmark.
Gradient echo memory (GEM) is a protocol for storing optical quantum states of light in atomic ensembles. The primary motivation for such a technology is that quantum key distribution (QKD), which uses Heisenberg uncertainty to guarantee security of cryptographic keys, is limited in transmission distance. The development of a quantum repeater is a possible path to extend QKD range, but a repeater will need a quantum memory. In our experiments we use a gas of rubidium 87 vapor that is contained in a warm gas cell. This makes the scheme particularly simple. It is also a highly versatile scheme that enables in-memory refinement of the stored state, such as frequency shifting and bandwidth manipulation. The basis of the GEM protocol is to absorb the light into an ensemble of atoms that has been prepared in a magnetic field gradient. The reversal of this gradient leads to rephasing of the atomic polarization and thus recall of the stored optical state. We will outline how we prepare the atoms and this gradient and also describe some of the pitfalls that need to be avoided, in particular four-wave mixing, which can give rise to optical gain.
Gradient echo memory (GEM) is a protocol for storing optical quantum states of light in atomic ensembles. The primary motivation for such a technology is that quantum key distribution (QKD), which uses Heisenberg uncertainty to guarantee security of cryptographic keys, is limited in transmission distance. The development of a quantum repeater is a possible path to extend QKD range, but a repeater will need a quantum memory. In our experiments we use a gas of rubidium 87 vapor that is contained in a warm gas cell. This makes the scheme particularly simple. It is also a highly versatile scheme that enables in-memory refinement of the stored state, such as frequency shifting and bandwidth manipulation. The basis of the GEM protocol is to absorb the light into an ensemble of atoms that has been prepared in a magnetic field gradient. The reversal of this gradient leads to rephasing of the atomic polarization and thus recall of the stored optical state. We will outline how we prepare the atoms and this gradient and also describe some of the pitfalls that need to be avoided, in particular four-wave mixing, which can give rise to optical gain.
This thesis presents progress towards two complementary goals. The first part discusses our aim of building high-efficiency quantum memories for quantum optical states and exploring the potential of the memory for performing operations as well as for passive storage. The second part presents our work towards developing a compact source of quantum light by fabricating optical resonators from non-linear optical crystals. Our high-efficiency memories operate by storing the information carried by a weak laser field in the coherence between the hyperfine-split electronic energy levels in a vapour of atomic rubidium. We analyse the performance of a three-level gradient echo memory protocol that is used to store and retrieve optical information from the atomic coherence using an off-resonant Raman transition. Using a warm rubidium vapour, we find that the memory preserves the quantum state of the input light better than any classical means of storage could achieve. We also present results showing that trapped, cold atomic vapours provide an increased storage lifetime and analyse the contribution that four-wave-mixing makes to the retrieved light. Both the warm- and cold-atom implementations of the memory are shown to store and recall light with over 80% efficiency. We also present a number of extensions to the basic memory protocol. We demonstrate that the light-atom coupling can be controlled by the bright coupling field that is used to complete the Raman transition. This control is used to perform a beam-splitting operations between two optical pulses that are separated in time. Using frequency-multiplexed states, we show that beam-splitting operations are also possible between optical modes that are separated in frequency. This operation can be extended to perform arbitrary operations by using a series of memories. Finally, we present a high-fidelity memory for dual-rail q-bits encoded across two frequency modes. The resonators that were fabricated to explore compact sources of quantum light demonstrated many of the properties that are required to obtain the stable operation of a quantum source. We demonstrated that non-linear optical crystals can be machined into monolithic resonators that use total internal reflections to confine light. Two geometries were tested, whispering gallery resonators and square travelling wave cavities. Because both types of resonators use total internal reflection instead of dielectric coatings, they feature very low loss rates. Lithium niobate resonators of both types exhibited cavity decay rates that were dominated by bulk absorption in the crystal. The resonators were shown to be frequency tunable using an applied voltage and could be coupled to free-space modes with a tunable coupling rate. We present preliminary results of non-linear frequency conversion in the resonators. We also investigated calcium fluoride and magnesium fluoride resonators for use as high-finesse reference cavities. We report a finesse of 5000 in a calcium fluoride whispering gallery resonator. A square calcium fluoride monolithic resonator was also produce to have a low insertion loss when coupled to a Gaussian free-space mode. We demonstrate a power coupling efficiency of 94% to that resonator. We also present a proposal to use a square lithium niobate resonator as a tunable coupler for a high-finesse calcium fluoride whispering gallery resonator. The coupled resonator system could be used as a compact high-efficiency optical quantum memory.