2DPCs have grabbed the attention of many research groups because, with their lattice dimensions of the order of the wavelength, they enable extensive control of the photonic states. Right from the early stages of development in this domain, laser emission has been demonstrated by incorporating III–V active materials within the PC, showing very promising performances like compactness ( ~µm2) and low thresholds (~µW)

In this project, we will make use of these outstanding properties to fabricate flip-flop devices. In order to implement the basic idea of the connected lasers configuration described previously (WP2), the first step is to obtain laser modes emitted in only one given direction. Indeed, usual PC cavity lasers can be considered simply as Fabry Perot lasers. This means that when they are coupled to a waveguide, the light will be channelled equally in two opposing directions of propagation. This is, of course, incompatible with the design of the flip flop we have chosen if we want to obtain a non-zero contrast! However, recently, it has been demonstrated that it is possible to obtain whispering gallery-like mode behaviour with 2DPCs giving us the possibility to implement the desired flip-flop design. Two configurations will be undertaken: microcavity and slow-mode cavity.

a) Microcavity based lasers
A schematic of the flip-flop is depicted below.

The device consists of two ultra-small cavities made in an InP membrane containing an active material emitting at 1.55µm (quantum wells or quantum dots). The two cavities are connected through a SOI wire running below the membrane. Each cavity is formed by a missing hole in a square lattice of holes.

b) Slow-mode cavity lasers

The second type of PC-based flip-flop is represented in the figure below. The device is now composed of two “slow-mode” cavities made in an InP membrane containing an active material emitting at 1.55µm (quantum wells or quantum dots). Once again, the two cavities are connected through a SOI wire running below the membrane. Each cavity is formed by a square mesa whose size is chosen to enable a WG-mode at the desired wavelength. Additionally the mesa is patterned with a perfectly periodic square lattice of holes.

Just as for the microcavity-based flip-flop, performances should lead to breakthroughs in terms of the switching speeds and switching powers.

Both the micro-disk and photonic crystal lasers are based on dielectric cavities and their dimensions are thus limited by the wavelength of the light. Metallic cavities can confine the light to volumes with considerably smaller dimensions. Although it is commonly believed that the high losses of the metals are prohibitive for laser operation in small metallic cavities, it has been demonstrated recently that metallic-coated nano-cavities with modal volumes smaller than dielectric cavities can have moderate quality factors and that indeed laser operation is possible. The structure of the demonstrated laser is shown in Figure 9. It consists of an InP/InGaAs/InP pillar with a conventional heterostructure surrounded by a thin insulating silicon nitride layer and then encapsulated in gold. The InGaAs layer forms the laser gain medium, with electrons injected via the top of the pillar and holes via the p-InGaAsP layer and a large area lateral contact.

The initial demonstration of the metallic nano-cavity laser has been at cryogenic temperatures. To be useful in many signal processing applications, operation at room temperature is required. Obtaining room temperature operation of these lasers is a key objective for this proposal. Another objective is the further miniaturization of the laser concept from what has been demonstrated. This miniaturization will require the realization of true Plasmon gap mode metallic waveguide cavities. In such cavities the laser modal and active regions can in principle be shrunk down to the scale of a few tens of nanometers, or less, in two dimensions. Finally the integration of such lasers in a membrane environment needs to be investigated. To make use of such lasers they will have to be efficiently coupled to dielectric waveguides.