Vast majority of white LEDs are phosphor based and not color-mixing
Are you talking about single-colour LEDs? Since white LEDs use a "phosphor", just like fluorescent lights, that convert blue or UV light. In fact I thought that the phosphors used by LEDs today are much better than those used by fluorescent lights.
Maybe you're talking about RGB lights, which combine separate reg, green and blue LEDs so that the colour mix can be changed. But even then I thought, unlike lasers, LEDs are not using single frequencies but relatively wide bands (but will still leave gaps between the red, green and blue colours).
Wow. I didn't know that. So I guess the junction actually emits UV which is converted to visible spectrum? Does that mean they're bad for your eyes if some UV leaks out?
How I thought it worked is that different doping materials are blended in the junction to get a mix of frequencies. Maybe that's impossible.
And yes, sure for RGB clusters the spacing is obvious even to my eyes.
So no UV, although there are some concerns about the hazards of blue light. But if you ask me, you get much more blue light onto your retinas by just going outside on a sunny day than you'd ever get from normal LED lighting.
As far as I can see from distributors, most white LEDs today are using a blue light emitter, so there's no issue with UV. When using white LEDs that use an UV emitter, some glass covering is needed to catch the remainder, from what I've heard from a friend (I've never actually built an LED lamp myself). But it appears that blue LEDs are much more efficient than UV ones and that probably explains why normally white LEDs use blue emitters.
More information here (confirming what I just wrote): https://en.wikipedia.org/wiki/Light-emitting_diode#White
PS. I've been looking into these things to some extent because of all the issues with commercial LED lamps (short life, AC or PWM flickering, colour rendering), and have been thinking of mixing more LEDs together (various single colour ones together with various phosphor-employing ones) so as to mix a more ideal spectrum myself, but I haven't gotten around to actually executing that plan.
The junction usually emits violet, but some blue ones exist too.
It's wasteful to emit UV. Fluorescent lights do it because it's not practical to tune them into visible light, but leds do not share that problem.
On the other hand I find it terrifying how many people will spend their lives cowering in quiet ignorance never venturing anything, and never learning anything, all for the fear of looking foolish.
And trust me, it would be much better with LEDs though obviously not perfect. Sometimes the colors just aren't different enough to easily tell apart. It seems like an issue that should be fixed.
But still beware, the CRI printed on the box (Ra) is kind of flawed since it is measured on unsaturated colors and does not include information about color renditions of red objects. You would have to look at the R9 value as well.
Sunlight, candle-light and resistive filament lights (incandescent) have a broadband spectrum. Through most of history we've viewed the world lit by these sources.
Moving to fluorescent lights causes spectral quantisation as the frequencies are due to electronic band transitions in the gas. The phosphor absorbs and re-emits energy in different bands that fills out the spectrum but it's still really strange, with missing bits. When fluorescent lights first became popular people complained of headaches from the strange spectrum. They are disfavoured in certain industrial applications because they make some colours harder to see.
Semiconductor lighting is even worse in this regard. Radiative re-emission in valance holes emits line spectrums. Only by combining these do we trick the eye into seeing "white light". Recently we've got good at blending doping agents to give mixed spectrums that approximate daylight.
If these forms of electronic lighting have holes in their spectra, perhaps that's why an analysis of a corpus of photos shows a change in spectrum over time?