MICHIEL DE DOOD - MASTERS THESIS

Most electronic circuits that we use in our everyday life are fabricated from silicon. As a result, a lot is known of Si as a semiconductor and the microfabrication technology needed to fabricate devices is well develloped and understood. However, the next challenge in semiconductor industry is to make devices that have optical functionality and devices that combine optical and electronic function on a single chip; so called opto-electronics. Silicon seems to be a logical choice based on its success in electonics. However, beacuse of its nature (indirect bandgap semiconductor) Si happens to be a very poor light emitting material at room temperature. One way to resolve this drawback might be to develop an efficient light source based on silicon that can be integrated with existing silicon technology.

Erbium (Er) ions might be a good candidate for this purpose since these ions can be incorporated in silicon and they are able to emit light at a wavelength of 1540 nm (0.8 eV), corresponding to the standard telecommunication wavelength for long distances. If Er is succesfully incorporated in Si in its trivalent state, the Er³⁺ ions might emit light via the intra 4f shell transition from the first excited state (⁴I_13/2) to the ground state (⁴I_15/2). Unfortunately, if one studies the light emission of Er in crystalline silicon very little or no light is emitted at room temperature in a typical photoluminescence (PL) experiment. Efficient light emission can be obtained at low temperature (~15 K). Heating the sample to room temperature quenches the luminescence intensity by more than three orders of magnitude at room temperature.

To understand this phenomenon, the temperature dependence of the Er luminescence was investigated as function of temperature in the range between 12 and 150 K. Samples were made by doping a p-type Si (100) wafer with 5x10¹⁸ Er/cm³ using 1.5 MeV Er ion implantation. The Er doped layer was co-implanted with N ions to a concentration of 6x10¹⁹ N/cm³. The samples were annealed in a vacuum furnace to recrystallize the silicon matrix (the ion implantation made the top layer of the silicon amorphous) and to optimize the Er light emission.

The PL of the sample was investigated as function of temperature. Up to a temperature of 75 K only a weak intensity quenching is observed. For temperatures above 75 K a strong intensity quenching sets in. The decrease in intensity is accompanied by a decrease in the measured luminescence lifetime with rising temperature. The experimental results were interpreted using an Auger energy transfer model. This model assumes that the Er is excited by electon-hole (e-h) pairs in the silicon that are generated by the laser light in the PL experiment. Once excited the Er can transfer its energy back to the silicon host and form e-h pairs. The strong temperature dependence stems from the fact that the 0.8 eV of energy stored in the Er is too small to generate an e-h pair in Si directly, since this requires about 1.1 eV. The difference in energy can be bridged by using a number of phonons that supply the energy. The occurence of back-transfer, proposed in this model, is indeed observed in spectral response measurements on Er implanted silicon solar cells at room temperature. These solar cells do generate an electrical current under illumination with 1540 nm light reproducing the spectral shape of the Er absorption spectrum and form a direct evidence of the backtransfer introduced in the model. An efficient light source at room temperature using Er doped Si can only be obtained if the non-radiative backtransfer process can be reduced considerably.

The results of my masters thesis are published in the following articles:

• P.G. Kik, M.J.A. de Dood, K. Kikoin and A. Polman, Appl. Phys. Lett. 70, 1721 (1997)
  Excitation and deexcitation of Er³⁺ in crystaline silicon. Download PDF (75 kB)
• M.J.A. de Dood, P.G. Kik and A. Polman, MRS Proc. 422, 219 (1996)
  Incorporation, excitation and de-excitation of erbium in crystal silicon. Download PDF(174 kB)

Oil fields are used for the production of crude oil. However, most fields also produce water. It is a well known that oil is lighter than water and one can thus imagine that a pipeline transporting oil has a thin layer of pure water in the bottom. Depending on the conditions, this water phase can be highly corrosive to the steel pipeline. In principle, this can be avoided provided that the flow velocity of the liquids is high enough such that the oil and water completely mix. This does not lead to corrosion since the oil phase preferentially wets the steel.

To characterize the two-phase fluid flow of oil/water mixtures with small volumes of oil, an experimental device was constructed at the Shell Research and Technology Centre Amsterdam (SRTCA). The device consists of a rectangular channel bend into an annular ring. The fluid is set into motion using a wheel that forms the top of the ring. A direct translation from this device to an actual pipeline is impossible. However numerical flow simulations might be used to obtain such a translation.

As a first test, a single phase system using only water was studied to compare the experimentally measured flow velocities with numerical results. For low flow velocities the numerical model involves solving the Navier-Stokes equations, while at higher flow velocities the model should incorporate the effect of turbulence. The measured and simulated flow velocities were in perfect agreement with literature (only the laminar case without any turbulence is considered in literature). The transition from a laminar flow to a turbulent flow occurs around a Reynolds number (density*velocity*diameter pipe/viscosity of the fluid)of 2100 for a straight pipe. This transition was observed for a Reynolds number of 11000 experimentally, which agrees with predictions of the transition for similar flows through helices and spirals.

The next step towards two-phase flow which can be measured and simulated relatively easy is the flow of small polystyrene beads in a model liquid of lower density. This case resembles the flow of small water droplets in oil. Experimentally the amount of particles in the liquid was measured at different positions in the experimental device. This was compared with calculations in which individual particle tracks were calculated. A reasonable agreement was found between experiment and simulations.

Finally, the simulation of a separated two-phase flow in which the water is flowing at the bottom and the oil is flowing above the water (the density of oil is lower than that of water) is more complicated but still possible. When increasing the flow velocity from a stratified flow (i.e. water/oil interface is straight) the flow starts to develop towards a stratified wavy flow, where waves develop at the oil/water interface. This situations can be correctly modelled. At even higher flow velocities small water (or oil) droplets are sheared from the waves and the flow develops into a fully dispersed regime. At this point simulations become extremely difficult. However, from the simulations done in the stratified and stratified wavy regime one can see that droplets of water in the oil phase are immediately transported to regions of high shear in the corners of the experimental device. This is due to a secondary flow induced by the curvature of the device rather than by the flow velocity of the liquid. As a result, the behaviour of our device in the dispersed flow regime is very different from that in a straight pipeline, which does not display secondary flow and has no regions of high shear that breaks up the droplets in smaller, more stable, droplets.