We will now try to find some answers to our fifth question: How can we change the wavelength of the light produced by radiative recombination?  
The recipe coming to mind is: Mix two similar (direct) compound semiconductors with different bandgaps.  
Luckily, most IIIV compounds are completely miscible in ternary or even quaternary crystals.  
In other words: From the 2 compounds GaAs and AlAs we can make ternary GaAl_{1x}As_{x} for 0 £ x £ 1, from GaAs and InP we can produce quaternary Ga_{1x}In_{x}As_{1y}P_{y}.  
This gives a lot of options. What happens upon mixing, which changes of properties are useful and which are not? Are there guidelines or do we have to try it out?  
Generally, all properties of interest as given in a table in subchapter 5.1.1 will change while x and y run through the accessible range, but not necessarily smoothly or monotonously with the composition.  
Here we focus on just a few of the especially
important properties:


The two last properties will be of overriding technical importance as soon as we learn how to make heterostructures, i.e. combinations of two different semiconductors.  
There are some standard diagrams showing major properties of the most important combinations.  
The first and most important one shows the bandgap vs. the lattice constant plus information about the band type. It is shown below, with the IIVI compounds included for good measure:  


Here is another picture of the same thing including more materials  

There is a tremendous amount of information in this diagram (note that "Xgap" and Lgap" both denote indirect band gaps a the respective positions in the band diagram):  
Most IIIV compounds radiate at wavelengths above the visible region, i.e. in the infrared. However, adding some Al to GaAs producing Al_{x}Ga_{1x}As, will shift the wavelength into the red region of the spectrum  here are our red luminescence diodes and Lasers!  
Very fortunate: GaAs and AlAs have almost the same lattice constant; we can thus combine any combinations of these materials without encountering mechanical stress.  
Very unfortunate: There are no IIIV compounds in the diagram that emit blue light  this is a severe problem for many potential applications. While SiC could be used to some extent, it was only with the recent advent of GaN that this problem was solved. SiC and GaN crystals, however, are not of the "zincblende" type common to all the IIIVs in the diagram but have a hexagonal unit cell. They therefore do not easily mix with the others!  
If we want to radiate at 1.3 µm or 1.5 µm  infrared wavelength of prime importance for optical communications  we should work with combinations of InAs, GaAs, and AlSb.  
Most interesting: The IIVI compounds are all direct semiconductors and span a much larger range of wavelengths than the IIIV's. The fact that they are not much used for products tells us that there must be big problems in utilizing these compounds for mass products.  
Let us now look a bit more closely at some other properties for the more important systems.  
The following diagrams show the direct and indirect bandgap and the refractive index for Ga_{1x}Al_{x}As as a function of x.  


Mixing does not only affect the band gap and the lattice constant, but also the quantum efficiency of light production. The next figure shows the system GaAs  GaP.  
The quantum efficiency decreases rapidly as the systems approaches the indirect bandgap region.  
If an isoelectronic center  N in this case  is added, the GaP side obtains a strong radiative recombination channel via bound excitons and the quantum efficiency is two orders of magnitude larger.  

Next the lattice parameters of various mixtures as a function of x are shown.  
This is easy to calculate; for complete mixing (no precipitation etc.), the lattice parameter changes linearly with the composition index x between the values for x = 0 and x = 1.  


Finally, the technically most important systems are listed together with some key properties  
We see that all kinds of ternary and quaternary compounds are used, and that the external or total efficiency  the relation of light out to total power in  is relatively small in most cases. The external efficiency should not be confused with the quantum efficiency (relation of light produced to total power minus ohmic losses), since some of the light produced may never leave the device  remember the fourth question!  
Also remember that the total efficiency of a light bulb is just a few percent. The semiconductor values don't look so bad in this context, and that we can get up to 30 % in extreme cases (and beyond!) is encouraging. Note that the somewhat exotic exciton process can account for an efficiency of 15 %!!  

© H. Föll (Semiconductor  Script)