Optical isolater monolithically integrated with DFB-laser.

In the BRS technology we developed an alternative isolator configuration, based on lateral current injection in the active core. This configuration has the advantage that the source of the non-reciprocal effect - the MO metal - no longer serves as the electrical contact, hence no absorbing ternary contact layer must be present between the cladding and the MO metal. This allows reducing optical losses in the device. We have demonstrated experimental operation of the lateral injection leading to similar gain as direct injection device operation.

In the isolator configuration studied within the ISOLASER project, the non-reciprocal - isolating - effect is caused by a magnetized ferromagnetic metal film sputtered close the guiding core of an InP-based waveguide. The source of the non-reciprocal effect is the transverse magneto-optic Kerr effect in this metal film, present when transversely magnetizing this film - perpendicular to the light propagation direction and parallel to the metal-semiconductor interface. At infrared wavelengths however, very little is reported in literature on the magneto-optic properties of ferromagnetic materials. Therefore, in order to optimise the choice of metal and, possibly even more important, to enable proper design of the isolator layer structure, these magneto-optic properties (i.e. the non-diagonal elements of the permittivity tensor) needed to be extracted. Furthermore, the complex refractive index of the metal film has also a large influence on the performance of the isolator, hence exact knowledge of this parameter is indispensable. With the experimental set-up developed for this task (see project result no.35323), the optical and magneto-optic parameters of three CoFe alloys have been determined, providing clear input for the design of the optical isolator demonstrator. The three materials are Co90Fe10, Co50Fe50 and Fe. Of these compounds, the equiatomic Co50Fe50 clearly provides the best combination of optical and magneto-optical properties.

In the isolator configuration studied and demonstrated within the ISOLASER project, the ferromagnetic metal film not only is the source of the non-reciprocal effect, but also serves as the electrical contact for the underlying semiconductor optical amplifier. Consequently, an ohmic electrical contact needs to be developed with this metal. Traditionally, the In0.53Ga0.47As compound is used for the highly doped semiconductor layer at the metal-semiconductor interface. For the isolator contact this material is not suitable as it is highly absorbing at the operation wavelength of 1300nm, hence the modal loss that needs to be compensated would enhance seriously. The quality of four alternative contact structures have been experimentally extracted and compared with the standard In0.53Ga0.47As layer. These alternatives were both structures with a bulk In0.81Ga0.19As0.41P0.59 layer and hybrid structures with a 100nm In0.81Ga0.19As0.41P0.59 layer topped with a thin (15nm and 5nm) In0.53Ga0.47As layer. Contact test structures (Cross Bridge Kelvin Resistors) have been fabricated and characterized. Furthermore, the influence of rapid thermal annealing on the contact quality has been studied. It could be concluded that the addition of a 15nm InGaAs layer on top of a 100nm InGaAsP layer results in a contact quality that is close to that of the standard InGaAs structure and this realizes an ohmic electrical contact without significantly enhancing the modal absorption. This contact scheme has been used in all isolator demonstrators realized within the project.

Once the concept of the amplifying waveguide optical isolator was demonstrated, in the frame of the ISOLASER project, improvement of the device performance could start. Four optimisation routes can be distinguished. The first one is the choice of the ferromagnetic metal (see project result no.35324). Secondly there is the development of tensile strained amplifying material with better gain performance (see project result no.35332). A third route towards optimisation is the improvement of the fabrication of the ridge waveguide isolator demonstrators. The first devices were fabricated using standard ridge waveguide laser processing, but it was directly clear that modifications were necessary to realize well-fabricated devices. The main problem was to realize laterally monomode ridge waveguides where the ridge, and only the ridge, is entirely covered with the magneto-optic metal. The solution was to use the metal layers as the etch mask, i.e. deposit magneto-optic metal stripes on a plain wafer before defining the ridge waveguides with an etch process. Optimising the processing parameters finally resulted in perfectly fabricated ridge waveguide isolators. The last possibility to improve the device performance is the optimisation of isolator design. At the time of designing the first devices the experimental gain-current relation of the amplifying material and the properties of the ferromagnetic metal were unknown, making the isolator design a difficult task. Furthermore, increased understanding of the component resulted in optimisation of the design strategy, focusing on the practical device specifications.

At the start of the ISOLASER project, the amplifying waveguide optical isolator was only a theoretical idea and there was no evidence that this was a viable concept. Following the development of tensile strained multi-quantum well material providing TM-selective gain, a first isolator demonstrator was designed, fabricated and characterized. The magneto-optic metal used on this demonstrator was Co90Fe10. On this device we were able to demonstrate, for the first time worldwide, an amplifying waveguide optical isolator. The demonstration - first qualitative and later quantitative - was realized through examination of the influence of a magnetic field on the amplified spontaneous light (ASE) emitted by the device. It was found that a change in the magnetization of the ferromagnetic metal film results in a proportional change in TM-polarized optical output power. Furthermore, the TE-polarized emission remains unaffected by the magnetic field. This is an unambiguous proof that the observed effect is the transverse magneto-optic Kerr effect, as this influences only the TM-mode of the optical waveguide.

The optical isolator basically is a semiconductor optical amplifier with a magnetized ferromagnetic metal contact close to the amplifying waveguide core. The amplifying material must provide the material gain to overcome the loss in the metal contact to end up with transparency in the forward propagation direction (but absorption in the backward direction). As the magneto-optic Kerr effect only influences TM-polarized guided modes, high TM-selective gain material must be developed. At the same time TE-gain must be suppressed as TE-light would add noise to the signal and possibly even saturate the amplifier. Strong TM-selective gain can be realized with tensile strained multi quantum well (MQW) material. Initially, we explored the InGaAsP/InP system, most common at 1300nm wavelength. In this system TM emitting lasers have been demonstrated. In order to achieve the very high gain levels needed for the isolator, it was opted that the AlGaInAs/InP system needed to be considered, as this material is known to have better gain properties at 1300nm. Optimisation of the number of quantum wells, the tensile strain in the wells and the compressive strain in the strain compensating barriers resulted in the demonstration of more than 100cm-1 effective internal gain for a current injection below 2.5kA/cm2.

Livrables

Partager cette page

Télécharger