First, I would like to congratulate those who have been working on blue GaN blue LEDs for so long, in some cases neary 50 years, for their progress so far. These days, there is virtually no reason at all to buy a tungsten-filament incandescent bulb (at least of the common globular shape), and even compact-fluorescents (CFL's) are hard to justify.
Ever since 2008, I have been considering isotopes, and how to use them to improve processes and devices. It's been a lonely task, because virtually every chemist or physicist views "isotopes" as merely atoms with a different number of neutrons, and thus a different atomic weight. Yes, they are indeed that, but they are so much more.
I scan the literature, recently phys.org, to figure out how to modify the makeup of isotopes in ordinary-isotope materials to enhance characteristics and behaviors. Until a few weeks ago, I simply didn't realize how much trouble that the GaN blue LED industry was having doping crystalline GaN with magnesium to provide 'holes'.
I should give credit where it is due: It was only a few weeks ago that I was reading the Quora.com system, and I saw that on October 1, 2015, a person named Karan Mehta had answered a question about "What was so difficult in making a blue LED?" https://www.quora.com/What-was-so-difficult-in-making-a-blue-LED I had long wondered how blue LEDs were made, but I wasn't really aware of how 'difficult' it was. My first exposure to 'blue LEDs' was in about 1985, when I had purchased a silicon carbide blue LED from Digi-Key, as I recall for $10. Nice color, not especially bright.
The third paragraph of Mehta's answer said:
"2)P-type doping: An LED uses a diode to inject charge carriers into quantum wells. A diode has a p-type material in contact with an n-type material. However, for a long time, no one could find a suitable p-type dopant for GaN. Nakamura managed to use magnesium as a p-type dopant by finding the right conditions (temperature and pressure). Even today, Mg doping is not ideal, due to the high activation energy. Only 10% of the Mg is activated. So, if your Mg concentration is 1e19, only about free 1e18 holes are generated. "
Aha! I saw his statement that "Only 10% of the Mg is activated". I wasn't quite sure what "activated" meant, but I assumed that meant, "did the job". But THAT 'rang a bell', so to speak.
I've long been aware that elemental (stable) magnesium isn't merely "magnesium". Magnesium in nature consists of 78.99% Mg-24 isotope, 11.01% Mg-26, and 10.00% Mg-25. https://www.webelements.com/magnesium/isotopes.html
Moreover, I was well aware that it was only the Mg-25 isotope whose nucleus posses 'nuclear spin': Mg-24 and Mg-26 have both an even number of protons, and an even number of neutrons. But Mg-25 is different: its nucleus contains an odd (not even) number of neutrons, and so it has a slight 'wobble'. The unpaired neutron can be thought as orbiting around the positively-charged rest of the nucleus, so that rest of the nucleus behaves like a positive electric charge, itself spinning around the center-of-mass of the whole structure. And as every physicist should know (my degree is in Chemistry, from MIT), a charge travelling in a circle causes a magnetic dipole to exist.
From reading Mehta's description, I concluded that the problem is that not all of the magnesium 'worked'. I'll let you guess which one did. It seems fairly obvious to me.
Zinc, similarly, is made up of isotopes. https://www.webelements.com/zinc/isotopes.html Only 4.1% of natural, stable zinc is Zn-67 and it has a nuclear 'spin'. The rest is Zn-64, Zn-66, Zn-68, and Zn-70, and none of them have nuclear 'spin'. And I notice that some early work on GaN LEDs used Zinc as a p+ dopant. It worked, I suppose, but somehow it was abandoned early on, since magnesium worked better. Why? Could that be because 10% is greater than 4.1% ? Well, THAT can be fixed!
Doing some more research, I also notice that the radius of gallium atoms is 130 picometers. https://www.webelements.com/gallium/index.html The radius of zinc atoms is 135 picometers. https://www.webelements.com/zinc/ And the radius of magnesium atoms is 150 picometers. https://www.webelements.com/magnesium/ So I can certainly understand the difficulty they had packing a 150 picometer-radius magnesium atom into a position for suitable for a 130 picometer gallium atom. They must have used a shoe-horn to pack the magnesium into the spot! Zinc's 135 picometers looks far more easily matched!
Merry Christmas. And you're welcome!