There are many key elements on which the broadband information networks must rely on. Wavelength Division Multiplexing (WDM) technology is one of the most important one. The choice of material and technological solutions for WDM depends on many factors: performance, costs, reliability etc. InP is the most promising material in this regard because of its integrability with other optoelectronic devices. In what follows we will discuss the possibility to obtain wave-guide-like structures using porous InP.
(100) and (111) oriented S-doped, n-InP, n=1018 cm-3 wafers have been used. The samples were covered with a layer of photoresist in which by standard lithography parallel 10 Ám broad stripes have been opened (Figure 5.7a). Consequently, the samples were anodized in 5 % HCl aqueous solutions at potentiostatic conditions (U = const.). The etching conditions were specially chosen in order to obtain Current-line Oriented Pores, i.e. U=3-9 V. The etching time was between 0.1 and 1 min. After the etching the samples have been investigated in cross section and top view using a Philips XL series Scanning Electron Microscope working at 10 and 15 kV. Note that the polymer mask was intentionally not removed from the sample after the etching, in order not to affect the structure.
Figure 5.7 shows the results of the experiments done with samples patterned with a polymer photoresist (U=6 V for t=0.2 min). Figure 5.7a and b shows a general top overview and a magnified top view under the photoresist respectively. In order to see the structure under the photoresist the accelerating voltage of the SEM was increased from 10 to 15 kV. Figure 5.7c and d show the cross section overview (c) and a magnified cross section situated directly under the photoresist (d). Figure 5.7e shows the cross section view of a region between two stripes of photoresist.
From Figure 5.7e one can observe that the pores nucleate only on the surfaces not covered by photoresist, i.e. between two stripes, and grow radially, also under the photo resist, away from the nucleation region. It is evident that the pores do not expose any crystallographic characteristic of the single crystalline substrate. The direction of pores changes gradually from perpendicular to the substrate surface (in the center of the nucleation region), to parallel to the surface (at the edges of the nucleation region, i.e. near the photoresist).
The pores growing parallel to the surface are clearly visible in Figure 5.7b and d. The radial growth of current-line pores is an additional provement of the fact that such pores grow perpendicularly to the equipotential lines of the electric field in the anodized substrate. Making an analogy between a light wave passing thought a small aperture and the current flow through a region of uncovered InP surface surrounded by two stripes of photoresist, the equipotential lines of the electric field will behave similar to the wave front of the light passing thought the aperture, i.e. will move radially outward of the slit, exposing a semispherical shape. The bulk wall visible between the two porous regions is a hint that the Current-line Oriented Pores can not intersect, therefore they will stop to grow or will change their direction of growth when the pore wall becomes equal to the double width of the space charge region. Thus, in contrast with crystallographically oriented pores, the current line oriented pores can not intersect.
Figure 5.7: Curro pores suitable for integrated wave guide structures. U = 5 V, t = 10 sec, 5% HCl; a) Top view showing the overview of the patterning; b) High magnification top view between two waveguide structures; The pores growing parallel to the surface of the sample as well as the wall between the neighboring waveguides are indicated by arrows; c) Cross section view of two neighboring wave-guide-like structures; d) Cross section view; Higher magnification between two waveguide structures; The pores growing parallel to the surface are also easily visible; e) Cross section view; An overview showing clearly the radial distribution of the pores;
A careful investigation of Figure 5.7d and e reveals that the diameter of the pores increases slightly as the pores grow deeper into the substrate (or under the photoresist). In this way the porosity of the porous layer increases as well. Considering the porous layer as an effective medium with a porosity dependent refractive index, it is straightforward that the (effective) refractive index of the porous structure will decrease as the porosity increases. As a result, a gradient in the refractive index in the depth of the structure is obtained. Recently, layers exposing a porous gradient have been proposed as waveguide-like structures in Si . In InP such structures are of interest as well, taking into account the integration possibility of passive (e.g. wave guides) and active elements (e.g. LEDs) on the same chip. The main idea for waveguiding as well as for optical fibers is based on the reflections occurring at the interface between a high refractive index core and a low index cladding. The light is kept inside the core as a result of the total internal reflection (TIR) effect, a consequence of Snell's law, which occurs for angles larger than a critical angle when the light passes from a high to a low refractive index material.
A more pronounced difference in the degree porosity and thus in refractive index is shown in Figure 5.8. In this case a layer of crystallographically oriented pores is first formed with a low degree of porosity and then the radial growth of the Current-line Oriented Pores is allowed (higher porosity). The low porosity layer can be considered to be the core whereas the Current-line Oriented Pores supply the cladding layer of a waveguide-like structure. Taking into account that the degree of porosity of porous layer made of crystallographically oriented pores is less than 15 %, while that of current-line-pore layers is at least two times higher, in a very rough approximation the refractive indices of the core and cladding layers shown in Figure 5.8 differ by 15 % and 30 % respectively, from the refractive index of the bulk InP (n=3.1). The simulations made on the structures shown in Figure 5.8 show that such a waveguide will be multimode, and in order to make it monomode it is necessary to decrease the dimensions of the core to nearly 1 Ám.
Figure 5.8: Curro pores suitable for integrated waveguide structures: higher core-shell contrast. U = 4 V, t = 1 min, 5 % HCl; a) Cross section view; b) Cross section - higher magnification. n3 < n 1 < n 2.
Guiding of light can be realized also using other principles than TIR. Recently optical fibers with a cladding based on photonic crystals (PC) have been reported. Taking into account that PCs are Bragg-reflectors in three dimensions it is probably possible to use the porous Bragg-like structures for wave guiding purposes as well. An example of a waveguide-like structure with a Bragg-like cladding is presented in Figure 5.9a. However, the periodicity of the Bragg structure changes by moving from the center to the sides of the structure. Nevertheless, by proper design it will be possible to improve the light confinement within the core of the waveguide structure. A 3D collage picture of the waveguide structure based on porous InP is presented in Figure 5.9b.
Figure 5.9: a) Bragg structure around the core was obtained by modulating periodically the anodic current density; b) A 3D collage picture of waveguide-structure on InP.
Thus, we showed that waveguide like structures can be fabricated on InP substrates using the possibility for controlling spatial distribution of the degree of porosity.