Modeling of silicon pillars that use solar energy to produce hydrogen

With global energy consumption rising and reserves of conventional energy sources such as coal and oil diminishing, we will inevitably have to switch more and more to renewable energy sources. The sun is an energy source with great potential in this respect: the global energy consumption is almost negligible compared to the amount of solar energy that irradiates the land masses on earth. However the sun is not always shining, while there is always energy consumption. Before solar energy can replace most of the fossil fuels, a way has to be found to store the energy harvested from the sun on a large scale. A promising option for this is storing the energy in a chemical bond: the solar energy can be used to electrolyze water to produce hydrogen and oxygen. They can be converted back to water to produce the electricity when needed.

In practice the splitting of water requires about 1.8 V. A pn-junction in silicon will produce about 0.4-0.7 V when irradiated. This means that several of these elements have to be connected in series to produce the necessary voltage.

Oxygen is produced at the anode, hydrogen at the cathode. Ion transport through the water that is electrolyzed is necessary to complete the electrical circuit. However the produced oxygen and hydrogen must not be mixed because it forms a flammable mixture. Therefore there must be a membrane in the water that permits ions but not gases to pass. To reduce the distance over which the ions have to be transported and to have large electrode areas, the device layout in the picture has been devised, which should have as small a scale as possible.


Junctions with a limited doping level will lead to the generation of a voltage, while highly doped junctions lead to a conductive contact between the two regions. The latter can be used to connect two junctions in series. Good modeling of light absorption and the resulting voltages and currents in a micropillar with a specific dopant distribution will aid in finding the right dopant distribution to reach as high a voltage as possible with as little as possible junctions. This modeling will be the assignment for the student. Commercial software is available for calculating the light absorption and resulting voltages and currents. However full 3d simulation of such a structure will require too much computational power, and therefore reasonable simplifications have to be made.

The student will use his knowledge of optics and semiconductor physics and combine analytical and computational modeling to explain experimental data and reach an advise on a good device layout.


If you are interested in this modeling assignment and expect that it fits your background, please contact Pieter Westerik, PhD student at the MCS group: p.j.westerik