Integrated Quantum Photonics

The nano-optics section focusses its research on wavelength-scale structures to control photon propagation and the interaction of photons with electronic excitations and spins in semiconductor quantum dots. The principal long term goal is the integration of such functions to achieve the first Semiconductor Integrated Quantum Optical Circuits. Both photonic crystals and suspended nanobeam structures are employed to guide and control the photon propagation. Cavities also play a key role, to enable the regime of cavity quantum electrodynamics to be attained.

The research is carried in well-equipped laboratories, with a variety of continuous wave and ultrafast lasers, highly stable low temperature optical systems and high resolution spectrometers all available. We are utilise the excellent fabrication facilities available in the EPSRC National Epitaxy Facility in Sheffield, where we fabricate our structures using high resolution electron beam lithography and reactive ion etching techniques.

Some recent highlights from this section’s activities are described below:

Electrically-tunable photon-photon interactions

Light is an almost ideal information carrier with a very long coherence time, and is therefore of great interest for quantum computation. However, this comes at the cost of intrinsically weak photon-photon interactions, which are necessary for the deterministic operation of quantum photonic gates. One technique to address this challenge is to leverage an intermediary element which can enable an effective photon-photon interaction. Here, we couple a single QD to the single optical mode of a waveguide, and demonstrate that the QD acts as such an intermediary. We show that this effect can be tuned by electrically controlling the QD, allowing future scale-up of the device to multiple QDs.

Learn more: Optica 5(5) 644-650 (2018)

Broadband sources of indistinguishable photons

Photonic crystal waveguides (PCWs, see Figure 1) are a leading technology for photon transfer in integrated QD circuits. Advantages of PCWs over competing technologies such as photonic nanowires are the efficient coupling of QD emission and the capability to engineer the local density of optical states (LDOS) to produce broadband slow-light, providing the Purcell enhancement of QD spontaneous emission required to achieve indistinguishable photons from InAs QDs. However, photon propagation lengths inevitably suffer when operating in the slow-light regime due to enhanced scattering at low group velocities, thus propagation losses must be taken into consideration when designing integrated optical circuits to operate in the slow-light regime. Alternative cavity-based means to obtain large Purcell enhancements exist, but at the price of significantly reduced bandwidth and mode overlap compared to slow-light approaches.

Learn more: Applied Physics Letters 101 5 051116 (2012)


Figure 1 – 3D diagram of the experimental setup with laser excitation spot (red cone) and spatially selective PL collection (yellow cones) both of which can be independently positioned on the “W1”type of photonic crystal waveguide (blue) end-terminated with output couplers.

Spin photon interface

The spin of a quantum dot (QD) makes a good static qubit with microsecond regime coherence times and radiative lifetime limited optical transitions. A promising approach to build a register of qubits is to use a network of spatially separated QDs with a manageable energy spectrum, connected by photons. Conceptually, the most obvious way to think of transmitting spin information is to use the polarization state of the photon, since the spin of a recombining QD exciton and the polarization of the emitted photon are related by strict selection rules. However, this is not compatible with a planar photonic circuit, which is essential for photonic integration. The fundamental problem is that, due to strong vertical confinement, the optical dipole of a self-assembled QD lies in the xy plane, and since only the x or y linear polarization component of the left or right circularly polarized light can propagate along a waveguide, the in-plane transfer of spin information is inhibited.

To overcome this issue, we introduce a spin-photon interface based on two orthogonal waveguides (see Figure 2), where the polarization emitted by a quantum dot is mapped to a path-encoded photon. A QD located at the center of the waveguide intersection will coherently emit the x (y)-polarization component of a circularly polarized state into the waveguides aligned along the y(x) directions, respectively. By collecting both polarization components, while retaining their relative phase, the full polarization state of the photon is mapped to a path-encoded state. On recombining the light from the waveguides, the polarization state of the photon can be reconstructed at another point in the plane, hence enabling on-chip transfer of spin information.

Learn more: Physical Review Letters 110 037402 (2013)


Figure 2 – Scanning electron microscope image of prototype spin-photon interface. Two orthogonal nanowire waveguides are excited at their intersection by a laser, linearly polarized at 45˚ to the x axis. QD emission into the two waveguides is measured via output couplers, which scatter light into the z direction. OT, OR, OB and OL signify the top, right, bottom, and left output coupler, respectively.

For many purposes, path encoding with indistinguishable photons is desired which could be achieved by inclusion of an unpolarized optical cavity.  An example of such cavity is a point defect photonic crystal cavity (H1 cavity) which has two degenerate mutually orthogonal dipole modes. The two TE dipole modes form a Poincaré-like sphere with states which have a one-to-one correspondence to the in-plane QD spin states. Path encoding could be achieved by selective coupling of each of the two dipole modes into two separate propagating photon channels (photonic crystal waveguides) allowing information of the QD spin state to be transferred to the path encoded photons in waveguides via the cavity (see Figure 3).

Learn more: Optics Express 22 2376 – 2385 (2014)


Figure 3 – Photoluminescence map of the waveguide coupled H1 cavity. A contour of the device structure is overlaid.

On-chip beam-splitter

A fundamental component of an integrated quantum optical circuit is an on-chip beam-splitter operating at the single-photon level. Directional couplers are a particularly attractive means to achieve this thanks to their single mode operation, easy modeling, flexibility in the splitting ratio and low back-scattering losses (Figure 4a). They do not require tapering to interface them with waveguides or more complex structures such as nanobeam cavities or waveguide based spin-photon interfaces, allowing for low overall losses and simpler circuits designs. At the same time, a very small footprint of order 10-20 µm, comparable to recently demonstrated plasmonic circuits, is achieved.

Single photons from a waveguide embedded QD are guided to one arm of the directional coupler which acts as a beam-splitter and output into two separate waveguides. 50:50 coupling into the two output ports of the coupler is achieved by careful optimisation of the air gap between the arms and the length of the coupling region. The single photon beam splitting by the whole device is demonstrated by cross-correlating the QD signal from separate output ports (Figure 4 b).

Learn more: Applied Physics Letters 104 23 231107 (2014)

 Figure 4 (a) SEM image of directional coupler. L is the interaction length between the waveguides.(b) Normalised cross-correlation function (black line, without background subtraction) from two different output ports (inset). Dotted line indicates 0.5 limit. Red line is the best fit to experimental data yielding g(2)(0) = 0.31±0.03.

Figure 4 – (a) SEM image of directional coupler. L is the interaction length between the waveguides. (b) Normalised cross-correlation function (black line, without background subtraction) from two different output ports (inset). Dotted line indicates 0.5 limit. Red line is the best fit to experimental data yielding g(2)(0) = 0.31±0.03.

Related Research Areas:


Recent publications





2014 login problems with gmail account and sign in tips