How Are Bandgaps Tuned and Modified in LEDs?

[johnuk]

I know the width of possible band gaps for something like an LED depends on the P and N material present, and that they can change them by altering the density (doping) of the charge carriers in the two.

Do they also have ways to implant or strip carriers from the layers other than doping?

I've heard of things like ion implantation. How do they do that? Are there others as well?

All the best shipmaties!
John

 

[zomgineer]

The band gap is the energy difference between the valence band and conduction band. There is an empty region between the two bands where there are no available states for electrons to occupy. It's fairly constant for a given type of semiconductor.

What you really change with doping (aka ion implantation) is the Fermi level. You can think of the Fermi level as the midpoint of the probability distribution that defines the likelihood of an electron occupying some state. If you dope Silicon with an N-type dopant (donor ion), the Fermi level moves closer to the conduction band which means the free roaming electrons become more likely. The positve charge of the ions that give up their electrons balance the negative charge but they're stuck in place so they can't carry current.

In LED's or photodetectors, the goal is to implant some atom that will have an energy state somewhere in the bandgap. That way, when electrons jump into or out the state there will be an associated photon of a desired wavelength (color). A professor once told me that they don't really have it down to an exact science. The quantum models are not perfect so what many companies actually did is to have grad students cook up lots of batches of LED's with different dopants to get the colors right.

 

[johnuk]

Interesting, interesting...

I know that commercially the error in the wavelength gets 'fixed' by simply sorting them into different bins, like ball bearings are 'fixed' for size by sorting them by size after production.

In my possibly over simplified take on things, I thought the light emitted was due purely to the band gap being direct and the size of the gap. ---------->Wiki explanation; Implications for radiative recombination

The method you described sounds like the dislocation one described there.

I have no idea what they use commercially as their primary choice. I'd have thought the simple band gap route would be cheaper and easier. Perhaps the dislocation methods makes tuning the emission wavelength easier.

 

[zomgineer]

I thought the light emitted was due purely to the band gap being direct and the size of the gap.

No, it's not just the bandgap. Generally, you have to add dopants to the semiconductor that create electron steps inside the bandgap. Also, don't forget that these are diodes! Diodes are either made by doping one side of a semiconductor or by joining different materials together.

 

[johnuk]

I understand diodes and depletion zones between two different materials, like Schottkys, and that dopants are added to silicon to make it P or N and increase the current carrying capacity.

Adding things that offer steps within it would seem to be drifting from the idea of creating a direct gap, where the large separation means the energy has to come out as EM.

Here's Shuji talking about his work on the first blue LEDs, seems like he's a lot more concerned about the intrinsic band gap than doping the materials.

Why did you decide to use gallium nitride?

Shuji Nakamura: At that time, in 1989, there were two materials for making blue LEDs: zinc selenide and gallium nitride. These had the right band gap energy for blue lasers. But everybody was working on zinc selenide because that was supposed to be much better. I thought about my past experience: if there’s a lot of competition, I cannot win. Only a small number of people at a few universities were working with gallium nitride so I figured I'd better work with that. Even if I succeeded in a making a blue LED using zinc selenide, I would lose out to the competition when it came to selling it.

That sounds almost insurmountable. How did you get around that defect problem?

Shuji Nakamura: Well,first I needed a MOCVD reactor. MOCVD stands for "metal organic chemical vapor deposition." Since I had money now, I bought a commercial reactor and used it to grow gallium nitride crystals, but I couldn’t get them to grow on the substrate. So I spent two years modifying my commercial reactor and succeeded in making what I called the two-flow MOCVD reactor. Usually a MOCVD has only one gas flow. That’s a reactive gas that blows parallel to the substrate. I added another subflow, with an inactive gas blowing perpendicular to the substrate. That suppressed the large thermal convection you get when you’re trying to grow a crystal at 1,000 degrees. Using this two-flow MOCVD I succeeded in 1991 in making the highest quality of gallium nitride crystals in the world. The dislocatoin density was still 1010. But there’s another measure of crystal quality, which is hole mobility, and I achieved a hole mobility of 200. That was a world record. The highest hole mobility ever achieved with gallium nitride was 100.