We investigated deformation of InP that was introduced by thin, narrow, dielectric SiN x stripes on the (100) surface of InP substrates. Quantitative optical measurements were performed using two different techniques based on luminescence from the InP: first, by degree of polarization of photoluminescence; and second, by cathodoluminescence spectroscopy. The two techniques provide complementary information on deformation of the InP and thus together provide a means to evaluate approaches to simulation of the deformation owing to dielectric stripes. Ultimately, these deformations can be used to estimate changes in refractive index and gain that are a result of the stripes.

Stress effects in InP and GaAs induced by SiNx stripes of different widths • Stress mapping by photo-luminescence (degree of polarization)• SiNx films grown under different built-in stress (tensile or compressive)

17 18 We investigate the crystal lattice deformation that can occur during the etching of structures in bulk InP using SiNx hard 19 masks with Ar/Cl2/CH4 chemistries in an inductively coupled plasma (ICP) reactor. Two techniques were used: degree of polarization (DOP) of the photo-luminescence (PL), which gives information on the state of mechanical stress present 21 in the structures, and spectrally resolved cathodo-luminescence (CL) mapping. This second technique also provides 22 elements on the mechanical stress in the samples through analysis of the spectral shift of the CL intrinsic emission lines. 23 Preliminary DOP mapping experiments have been conducted on the SiNx hard mask patterns without etching the 24 underlying InP. This preliminary study demonstrated the potential of DOP to map mechanical stress quantitatively in the 25 structures. In a second step, InP patterns with various widths between 1 and 20 µm, and various depths between 1 and 6 26 µm, were analyzed by the 2 techniques. DOP measurements were made both on the (100) top surface of the samples and 27 on the (110) cleaved cross section. CL measurements were made only from the (100) surface. We observed that inside 28 the etched features, close to the vertical etched walls, there is always some compressive deformation, while it is tensile 29 just outside the etched features. The magnitude of these effects depends on the lateral and depth dimensions of the 30 etched structures, and on the separation between them (the tensile deformation increases between them due to some kind 31 of proximity effect when separation decreases). 32 33 Key words: indium phosphide, photonic devices, plasma etching, mechanical stress 34 35 36 INTRODUCTION 37 38 Plasma etching is broadly used nowadays in nanotechnology platforms, both in research, development and production, 39 for the fabrication of a large number of components and integrated circuits. A good example of a family of such 40 components is the photonic devices that can be fabricated on InP. One major drawback of plasma etching is the creation 41 of structural and chemical damage in the vicinity of the etched surfaces. The defects introduced change the electro-42 optical properties of the semiconductor, thus ultimately compromising the device performance. Different studies have 43 already reported on the damage created on flat surfaces exposed to reactive plasmas [1-3], either by identifying the 44 structural and chemical defects, or by analyzing the changes in material electronic and optical properties due to damage. 45 The degree of damage depends on the plasma etching parameters and notably it has been demonstrated that ion 46 bombardment is mainly responsible for defect generation. Phenomena such as ion channeling combined with diffusion 47 can create defects as deep as hundreds nm below the etched surface [4-7]. In GaAs etched with SiCl4 under a self-bias 48 voltage of 80 V damage was found at depths of the order of 150 nm below the etched surface [7]. In a recent study, we 49 have examined the impact...

In-plane micro-photoluminescence (µ-PL) and micro-reflectivity measurements have been performed at room temperature by optical excitation perpendicular to the surface of two different structures: a complete vertical surface emitting laser (VCSEL) structure and a VCSEL without the upper p-type distributed Bragg reflector (P-DBR). The two structures were both laterally oxidized and measurements were made on top of oxidized and unoxidized regions. We show that since the photoluminescence (PL) spectra consist of the cumulative effect of InGaAs/AlGaAs multi quantum wells (MQWs) luminescence and interferences in the DBR, the presence or not of the P-DBR and oxide layers can modify significantly the spectrum. µ-PL mapping performed on full VCSEL structures clearly shows oxidized and unoxidized regions that are not resolved with visible light optical microscopy. Finally, preliminary degree of polarization (DOP) of the PL measurements have been made on a complete VCSEL structure before and after an oxidation process. We obtain an image of DOP measured by polarization-resolved µ-PL. These measurements allow us to evaluate main components of strain.