Do cool white LEDs emit UV light?

I have seen some internet electronic forums claming that low quality cool white LEDs emit more UV light than CFLs. Is there any truth in this claim?

• As several people have noted, UV from phosphor LEDs is unlikely. HOWEVER - the blue light from the LED which excites the phosphor can be a notional eye hazard at quite low power levels. I had a formal eye hazard assessment carried out for some cool white LEDS (6000 K) which were very high efficiency (as the time) and rated at 65 mS, Vf = 2.95V so power was only 190 mW (and 25+ lumen at about 70 degrees double half-cone angle). These were assessed as being formally hazardous in the deepest blue part of the test ranges. Actually achieving eye damage would have required gross stupidity or ... Mar 4 '17 at 9:59
• ... having somebody tape an LED to your eye BUT it was interesting to see what low levels are required. Mar 4 '17 at 10:00

Is there any truth in this claim?

Truth, No.

It is possible there is a tiny bit of UVA and UVB. Amounts so small they would be very difficult to measure.

Occasionally there are some papers that report a couple of spikes of UV from the yellow phosphors used to convert blue to white. The reason they are reported is they are hoping to find phosphors that convert down to UV wavelengths and when they find a hint of UV it will likely be reported. Sample of such a paper. I do not know any of these phosphors that have been reported are used in commercial lighting.

I have some measurements from a horticulture research project at the University of Florida. I used a StellarNet BLUE-Wave Spectrometer,

• 200-1150nm wavelength
• Signal to noise = 1000:1

a decent \$15,000 spectrometer, to measure the Photon Radiance of one of my LED grow lights. This fixture uses the standard color LEDs which includes the deep blue (Cree XP-E) that white LEDs are made from.

I was able to measure a well formed peak as low as 310nm, two wavelengths adjacent at 288 and 289nm and one spike at 280nm.

The UVA and UVB wavelengths were very low reading, but well above the noise level. So there are probably some UV emitting from a cool white LED, but they cannot be read with my spectrometer.

I do not use "low quality" LEDs, only Cree and Phillips Luxeon.

The day before I had measured a fixture with Luxeon LXML-PWC2 4000K LEDs. Not a cool blue. The lowest measurable wavelength for the 4000K was at 409nm @ 0.00077778. For comparison the peak at 310nm on the graph below, was 0.0034645 (4.45x larger)

These are the readings I got from the spectrometer from 280 to 400nm.
first column is wavelength, second is relative to micromoles/m2/s. By relative, I mean, I forgot what aperture was used. Readings were taken October 7, 2016 in a grow tent at my house. The current flowing through the LEDs was 350 mA.

 280.00  1.1290E-002
281.00  0.0000E+000
282.00  0.0000E+000
283.00  0.0000E+000
284.00  0.0000E+000
285.00  0.0000E+000
286.00  0.0000E+000
287.00  0.0000E+000
288.00  1.1274E-002
289.00  9.0670E-003
290.00  0.0000E+000
291.00  0.0000E+000
292.00  0.0000E+000
293.00  0.0000E+000
294.00  0.0000E+000
295.00  0.0000E+000
296.00  0.0000E+000
297.00  0.0000E+000
298.00  0.0000E+000
299.00  0.0000E+000
300.00  0.0000E+000
301.00  0.0000E+000
302.00  0.0000E+000
303.00  0.0000E+000
304.00  0.0000E+000
305.00  0.0000E+000
306.00  0.0000E+000
307.00  1.4569E-002
308.00  1.4059E-002
309.00  2.4614E-002
310.00  3.4645E-002
311.00  2.5260E-002
312.00  6.2953E-003
313.00  3.6099E-003
314.00  0.0000E+000
315.00  0.0000E+000
316.00  0.0000E+000
317.00  7.2641E-003
318.00  1.4983E-002
319.00  9.9858E-003
320.00  3.7277E-003
321.00  5.7936E-003
322.00  4.5364E-003
323.00  2.1041E-003
324.00  1.3479E-002
325.00  9.7610E-003
326.00  2.4578E-003
327.00  3.9432E-003
328.00  7.8247E-003
329.00  3.6119E-005
330.00  0.0000E+000
331.00  0.0000E+000
332.00  0.0000E+000
333.00  5.3355E-003
334.00  6.3256E-003
335.00  1.6233E-003
336.00  9.1491E-003
337.00  5.3881E-003
338.00  7.1098E-004
339.00  3.9042E-004
340.00  9.4603E-004
341.00  0.0000E+000
342.00  5.8628E-003
343.00  1.9500E-003
344.00  0.0000E+000
345.00  0.0000E+000
346.00  0.0000E+000
347.00  1.8100E-003
348.00  8.4942E-003
349.00  1.1109E-003
350.00  0.0000E+000
351.00  0.0000E+000
352.00  0.0000E+000
353.00  0.0000E+000
354.00  0.0000E+000
355.00  7.3287E-004
356.00  6.0877E-003
357.00  1.8842E-003
358.00  0.0000E+000
359.00  0.0000E+000
360.00  0.0000E+000
361.00  0.0000E+000
362.00  2.4273E-003
363.00  4.5285E-003
364.00  3.7282E-003
365.00  1.7190E-003
366.00  4.4021E-003
367.00  1.1359E-002
368.00  1.2333E-002
369.00  1.9218E-003
370.00  0.0000E+000
371.00  1.8477E-003
372.00  5.1049E-003
373.00  8.0459E-003
374.00  1.9174E-003
375.00  0.0000E+000
376.00  0.0000E+000
377.00  0.0000E+000
378.00  0.0000E+000
379.00  1.3191E-003
380.00  8.7230E-003
381.00  6.5227E-003
382.00  3.3852E-003
383.00  4.2065E-003
384.00  6.1745E-003
385.00  8.8286E-003
386.00  4.2646E-003
387.00  0.0000E+000
388.00  5.2739E-003
389.00  8.0110E-003
390.00  4.6509E-003
391.00  5.4275E-004
392.00  3.7508E-003
393.00  1.0168E-002
394.00  1.0653E-002
395.00  9.2924E-003
396.00  7.9877E-003
397.00  1.0248E-002
398.00  1.2157E-002
399.00  1.1787E-002
400.00  1.2336E-002


This graph was made from the reading above plus the wavelengths from 400 to 1099nm.

Zoom

Photo of the tent with the white 4000K fixture and the spectrometer in my home lab.

Blue and White LEDs made from Blue do not have any significant level of UV.

However high levels of Blue can cause damage to eyes and also Art Galary paintings. This can occur due to the relative percentage level of blue and the higher energy of shorter wavelength visible Blue` LEDs compared to Metal Halide makes LED rejected by Lighting Experts for Art galleries to prevent energetic photon related aging.

Summary: CFLs can emit a small amount of UV light. Only very specialist white phosphor LEDs will emit any UV at all. These are specialised and unlikely to be encountered in practice - certainly not at the low cost / low quality end of the market.

However, even correctly operating good quality white phosphor LEDs can emit blue (near to UV) light at levels which are potentially eye-hazardous.
I have 200 mW DC in LEDs which are rated as marginally eye hazardous in the blue end of the spectrum. ANY say >= 500 mW LED is liable to be potentially eye hazardous to some degree.

__________________________

As several people have noted, obtaining UV from phosphor LEDs is extremely unlikely - and will only occur wity the very specialist UV emitter versions that Tony mentions. You are extremely unlikely to ever encounter these.

HOWEVER - the blue light from the LED which excites the phosphor in most white LEDs can be a notional eye hazard at quite low power levels.
I had a formal eye hazard assessment carried out for some cool white LEDS* (6000 K) which were very high efficiency (as the time) and rated at 65 mA, Vf = 2.95V so power was only 190 mW (giving 25+ lumen at about 70 degrees double half-cone angle). (* Nichia - 'as good quality as any').

The regulations break the areas of interest into a number of wavelength ranges. (3 AFAIR). These LEDs were assessed as being formally hazardous in the deepest blue part of the test ranges. Actually achieving eye damage would have required gross stupidity or having somebody tape an LED to your eye BUT it was interesting to see what low levels are required.

It depends. Both cool and warm white LEDs can be built from blue LEDs + luminescent material or UV LEDs + luminescent material.

Those made from blue LEDs are in the far majority because they are much cheaper. Their UV emission is zero. The problem with these is the luminescent material is aging and so they are getting bluer with time, because a higher fraction of the light is passing the luminescent material unaltered.

Those made from UV LEDs don't have that aging problem because all light passing the luminescent material unaltered is invisible. They are only getting darker with time. So, yes, these LEDs do emit UV. And they are not cheaper but more pricey because of their better color stability over time.

BUT, in a light bulb there is a glass in front of the LEDs. And ordinary glass filters UV nearly perfectly. (You can't get a suntan behind a window.) That's the same as for CFLs. If you need the UV, e.g. in a solarium lamp, the CFLs or LEDs need to use quartz glass or a transparent plastic for the casing.

Not likely. The doping required to get an LED to emit UV is considerably more expensive than the doping needed for visible light. You can buy a UV LED, but no manufacturers, not even the cheap ones, would include it in an LED product unnecessarily.