Development and optimization of silicon based light sources for integration into a sensor platform

Abstract

We have characterized electroluminescent devices based on silicon rich oxide and/or silicon rich nitride. We have discussed the photoluminescence and structural characterization of the active layers and the electrical and electroluminescent characterization of full devices.

We have noted that the electroluminescence can appear in the form of discrete points scattered across the active area of the devices, in the form of emission along the rim of the active area, or homogeneously distributed across the area. These different kinds of emission have been related to the optical and electrical properties of the devices.

In the two former cases, the electroluminescence comes with high current densities,of the order of 1 A/cm2, and low efficiencies of the order of 10-8. On the other hand, the homogeneous emission comes with lower current densities, of the order of 0.01 A/cm2, and better efficiencies, in the range 10-7–10-5.

We have concluded that the homogeneous emission is optimal in terms of efficiency. Furthermore, a simple model has been proposed to explain the appearance and occasional coexistence of the different kinds of emission.

The effect of a nitride layer on top of the SRO has been explored, concluding that it helps in achieving a uniform conduction that favors the homogeneous emission in the active layer.

The conductivity states of the active layer associated with the different kinds of emission have been related with its CV behavior. The results of the study show that the homogeneous emission corresponds to well behaved CV curves, whereas the emission through points does not.

The injection mechanisms in PECVD and ion implanted samples have been studied, concluding that no single emission mechanism can account for the injection at all regimes in the studied range of electric fields. Fowler-Nordheim or trap assisted tunneling have been found to play a significant role in PECVD samples. In implanted samples, Fowler-Nordheim dominates at low fields, whereas Poole-Frenkel is more likely to be the dominant mechanism at higher fields.

Comparison of the photoluminescence and electroluminescence spectra of bilayers SRO/SRN, allows us to conclude that each layer contributes a different band in the total emission, which results in a wider distribution of the energy across the visible spectrum.

The comparison between the photoluminescence and electroluminescence has revealed massive differences in their spectra, which have been attributed to interference effects.

A computer software based in the Crawford method for the study of the interference effects in multilayer stacks has been presented. The program has been used to quantitatively study the interference effects in the emission of our devices. We can conclude that the photoluminescence and electroluminescence spectra are the same despite their apparent difference. Our analysis has also made it apparent that a quantitative understanding of the interference effects in the system is important in order to draw valid conclusions regarding the origins of the luminescence.

We have presented the design, fabrication and characterization of a CMOS compatible optical transceiver, and two main challenges in the integration of the emitter, waveguide and detector have been successfully overcome, namely achieving a reasonably flat and uniform silicon oxide trench and a good detector.

In the end, the transceiver has not worked as expected, most likely due to a poor SRN emitter. More work is required in order to better control the fabrication process of the SRN layers. However, we believe the basic design to be valid, given the low electrical coupling detected between the emitter and the detector components of the transceiver.

Aquesta tesi presenta un estudi de les propietats òptiques de capes d'òxid de silici enriquit en silici (SRO) i nitrur de silici enriquit en silici (SRN) que han sofert un procés tèrmic d'alta temperatura. Aquest procés indueix la creació de nanoaglomerats de silici en la matriu dielèctrica. Aquestes nanoestructures de silici presenten una superior eficiència en l'emissió respecte al silici en bloc, i a més a més emeten en el visible en comptes de l'infraroig. Això és interessant per a l'obtenció de dispositius fotònics integrats basats en silici que poden ser fabricats monolíticament en un procés compatible amb la tecnologia CMOS que domina la indústria microelectrònica.

A més a més, hem estudiat les propietats òptiques i elèctriques de dispositius metall-aïllant-semiconductor en les quals l'aïllant és una capa d'SRO o SRN amb nanoaglomerats de silici. N'hem mesurat paràmetres d'interès com ara l'eficiència de conversió d'energia elèctrica-òptica o la potència òptica, i n'hem estudiat els mecanismes d'injecció que hi tenen lloc. S'han identificat tres tipus diferents d'emissió: per punts, per la vora del dispositiu, i emissió homogènia, i hem determinat que l'emissió homogènia és la més adecuada pel que fa a l'eficiència dels dispositius.

Hem desenvolupat un programa que permet el càlcul de les interferències òptiques que tenen lloc als sistemes multicapa que conformen els dispositius estudiats, i que distorsionen l'espectre observat respecte al que les capes realment emeten. L'habilitat de poder calcular aquests efectes ens permet, en molts casos, eliminar l'efecte de les interferències i determinar l'autèntic espectre d'emissió de les capes i per tant estar en millors condicions d'assignar l'emissió als mecanismes correctes.

Finalment, hem proposat un prototip per a un transceptor en el qual l'emissor, la guia d'ones i el detector estan integrats monolíticament en un procés CMOS. Hem fabricat el dispositiu i l'hem caracteritzat. Tot i que no hem aconseguit acoblament òptic entre l'emissor i el detector, creiem que el disseny bàsic queda validat, ja que els principals obstacles en l'obtenció del dispositiu han sigut superats amb èxit.