From many points of view microelectronics is one of the most important field of the modern industry. Computers, internet, mobile phones etc. technologies wouldn't be possible without the invention of transistor by Bardeen and Brattain [1, 2] and the idea of making integrated circuits (IC) proposed by Kilby .
Although the first transistors were demonstrated on Germanium (Ge), the modern microelctronics technology is dominated by Silicon (Si). There are many reasons which "force" the industry to prefer Silicon in 95 from 100 cases. Here are some of them:
However, advantages are always accompanied by disadvantages. Here are some Si disadvantages with respect to other semiconductors:
Concerning the fist "disadvantage" mentioned above, an interesting discovery was made simultaneously but independent by Lehmann and Canham at the beginning of the nineties of the last century [4, 5]. They observed that the emission of light from Si can be considerably improved if Si is made porous. "Porous" in this context means that billions of nanometer-size holes, similar to those in cheese, are made in a normal single crystalline Si substrate. Thus, a device based on porous Si will emit efficiently light, whereas one based on a normal single crystalline Si substrate will not.
A real investigation boom began immediately after this discovery was published. This was triggered by the hope that porous Si will make possible the integration of passive optical devices like gratings, waveguides etc., on the same substrate with porous Si light emitting devices, i.e. LEDs. Such an integration would make true an old dream - to produce integrated optoelectronic circuits on cost-effective and well investigated Si substrates.
Unfortunately, this dream did not become reality even after 14 years of research and trials. The main problem up to now is the stability of the emitted light from porous structures. This means that the emission degrades with time and therefore it is not good enough for commercial applications.
Nevertheless, two important conclusions should be made at this point:
Did Lehmann and Canham use mechanical nanometer drilling machines or something similar for obtaining their porous structures? The answer is definitely "No". Both of them used the well known electrochemical etching or dissolution.
Electrochemical dissolution is a result of anodic current flow at solid-liquid interfaces. The exact definition of anodic currents will be given in the next chapters. In more simple terms, in order to obtain the porous structures, Lehmann and Canhamm simply applied a positive potential on a Si electrode and forced a current to flow through the Si/HF interfaces.
The positive potential will force the positive charge carriers (i.e. holes, see next chapters for definition of carrier types in semiconductors) from the semiconductor to move towards the Si/HF interface. As we will see later, a hole is nothing else than a missing electron in a bond between two atoms of the semiconductor. Thus, a surface atom surrounded by many holes will be less strongly bonded to the substrate and consequently much easier removed, i.e. dissolved, from the surface of the substrate as compared to an atom without holes in its surrounding. As a rule of thumb, it can be said that a semiconductor region with a higher density of holes will have a higher dissolution rate during the anodization as compared to a region with lower hole density.
However, it is reasonable to assume that if an anodic current is forcing the holes to go towards the Si/HF interface, then the density of the holes should be constant over the whole surface of the electrode. In such a case the dissolution rate should the same over the whole sample and the dissolution should be isotropic.
However, as was demonstrated by Lehmann and Canham for certain anodization conditions, the dissolution can be anisotropic, resulting in porous structures like those presented schematically in Figure 1.1.
Figure 1.1. What are pores and how they can be obtained? A Si sample is immersed in a vessel containing a HF solution. A closed circuit is made between the Si sample and a Pt electrode. A voltage source is applying a positive potential on the sample, thus forcing a current to flow. One possible result of the current flow is the formation of small holes in the substrate. Such holes are called pores.
At this point, it is important to note that the first observations of porous silicon layers go 40 years back and were done by Uhlir and Turner . They observed that as a result of anodic current flow through Si/HF interfaces the colour of the samples changed. However, they interpreted the colour change (brown to black) as precipitation of silicon related species from the solution used. Therefore, it can be considered that Lehmann and Canham did nothing else than to prove that the result of anodization is not a deposition but a selective dissolution process of Si. The selective dissolution generates small pores in the initial substrate, leaving the pore walls in their initial single crystalline form.
After the first experiments on porous Si, it was clear that the morphology of porous structures strongly depends upon the experimental conditions, i.e. doping level of the substrate, electrolyte concentration, current density, applied voltage, temperature of the electrolyte, electrochemical cell design etc. These findings urged the scientific community to adopt a standard classification for the different types of electrochemically obtained pores. According to the International Union of Pure and Applied Chemistry, for short IUPAC, three categories of pores (not only electrochemically obtained) have to be distinguished if taking into account only the average pore diameter and average distance between pores:
Figure 1.2: Three types of pores observed in Si; a) Nanopores - sponge-like structure; b) Mesopores - tree-like structure; c) Macropores - smooth pore walls.
Figure 1.3: Current voltage characteristics of the Si/aqueous HF interface with back side illumination.
In Si, the bridging between the IUPAC pore types and the etching conditions can be done easily by means of the so called current-voltage (I-V) characteristics of the Si/HF interface. Schematic examples of typical Si/HF IV curves for p- and n-type samples are presented in Figure 1.3. The I-V curves for n- as well as for p-type Si can be divided in two main regions:
The amazing thing about the pore formation region in Si is that for each of the three type of pores, a separate sub-region can be found on the I-V curve. As a rule of thumb, by increasing the current through the Si/HF interface from 0 to jPSL, first macropores, then meso and near the jPSL peak nanopores are found.
On the other hand, electropolishing region means that isotropic dissolution of the substrate without pore formation is taking place. Pore formation and electropolishing regions are separated by the so called jPSL/UPSL peak (PSL = Porous Silicon Layer). The second peak on the I-V curve is called - oxidation peak jOx/UOx.
At its turn, the electropolishing region has two sub-regions: non-oscillating and oscillating. The non-oscillating sub-region is mainly between jPSL/UPSL and jOx/UOx, whereas oscillations are found after the oxidation peak jOx/UOx. "Oscillating" in this context means that the measured voltage across the Si/HF interface oscillates in time if the current applied to the sample is kept constant (galvanostatic regime), or vice versa, the measured current oscillate if the applied voltage is kept constant (potentiostatic regime).
Going back to the pore formation region, it should be noted that there are many theories explaining the improvement of light emission from nanoporous Si. The most accepted explanation is based on quantum size effects. If the thickness of the pore walls reaches values below several nanometers, quantum effects can change the band structure of the nanoparticles (pore walls), e.g. by transforming it from an indirect into a direct band. Therefore, nanoporous samples eventually are able to emit strong visible photo- and electroluminscence.
On the other hand, mesoporous and macroporous materials do not exhibit efficient luminescence. However, due to the fact that a porous skeleton has a three dimensional configuration (3D) the total surface of the structure increases considerably as compared with the surface initially exposed to the electrolyte (see Figure 1.4). These huge surface could be used in different applications were a large surface in a small volume is required. Surface increase is valid for nanopores as well, however the mesoporous and macroporous materials are mechanically more stable as compared with nanoporous ones. The smoothness of macropores is of great advantage for applications were a high uniformity of the structure is required, for example in photonic crystal applications.
Figure 1.4: a) Initial surface of a Si sample; b) Increased surface of a porous sample.