Short answer
Most white LEDs are a blue-emitting chip coated with phosphor that converts part of the blue into a broad band of longer wavelengths; the mixture looks white. The result is a sharp blue peak, a dip in the blue-green and, unless extra red phosphor is added, a shortfall in deep red — which is why cheaper LEDs make reds, skin and wood look flatter than daylight or halogen.
An LED chip emits a fairly narrow band set by its semiconductor. White lighting LEDs almost all start with a blue chip; a phosphor layer absorbs part of that blue and re-emits it as a broad band of longer wavelengths, and the unabsorbed blue plus the phosphor emission add up to white. Changing the phosphor mix and thickness sets the colour temperature: more conversion gives a warmer light. The U.S. Department of Energy describes this as the phosphor-conversion route, alongside two others: mixing separate coloured LEDs, and hybrids of the two. In the CIE's representative LED illuminant LED-B3, sampled at 10 nm, the single highest point of the spectrum is the blue pump at 450 nm.
Between the chip's blue peak and the start of the phosphor band there is a trough, typically somewhere around 480–500 nm, where little power is emitted. Turquoise, teal and some greens depend on that region and can look slightly lifeless under a basic LED. At the other end, a single yellow-green phosphor tails off before the deep red. Raising CRI and especially R9 means adding a red-emitting phosphor, and red photons sit where the eye is less sensitive, so each watt yields fewer lumens. That is the efficiency trade-off the DOE notes: better colour fidelity usually costs some efficacy. On the CIE data, LED-B3 puts about 17% of its power above 640 nm, against about 30% for D50 daylight.
This is why 'CRI 80' and 'CRI 90' LEDs of the same colour temperature can look so different on a face: the extra ten points are mostly bought in the deep red.
The alternative to phosphor is to mix narrow-band red, green and blue emitters. That gives control — the colour can be tuned electronically — but three narrow peaks leave wide gaps, and surfaces whose reflectance falls in a gap render poorly or oddly. The CIE's RGB illuminant LED-RGB1 shows the effect: a saturated red seen under it gains about a quarter more chroma than under daylight in the calculation on the R9 page, because its red emitter sits where that surface reflects most. Hybrid products add a red or amber LED to phosphor-white emitters to fill the red. Multi-channel fixtures raise a further problem recently analysed in the lighting literature: the same white point can be reached with many channel mixes, each rendering colours differently.
Some LEDs use a violet rather than blue chip, with phosphors covering blue through red. Their spectra lack the blue spike and extend further into the violet, which matters for anything with optical brighteners: white paper and laundered cotton fluoresce blue under violet light and look whiter, while under a blue-pumped LED, which emits very little below about 420 nm, the same whites can look slightly yellow. Tunable-white fixtures blend a warm and a cool phosphor LED; the colour temperature slides between them, but the spectrum at any setting is a mix of two shapes, so rendering can dip in the middle of the range. The CIE's LED illuminant series, added in CIE 15:2018, includes blue-pumped, hybrid, RGB and violet-pumped examples for exactly this reason.
A good datasheet gives CCT, Ra, R9 and increasingly TM-30 Rf and Rg; a thorough one includes the SPD plot. The DOE's guidance treats CRI 80 as a minimum for interior lighting and 90 or more as excellent fidelity. Check two further things. Duv, the distance from the blackbody locus, tells you whether a nominally neutral white is tinted green (positive) or pink (negative). And binning: LEDs from one production run vary slightly, so fittings from different batches can differ visibly side by side even when every figure on the box matches.
Why: Phosphor LED with a weak deep-red tail (low R9) despite an acceptable Ra.
Fix: Specify R9 (or TM-30 Rf,h1 and Rcs,h1) as well as Ra; high-R9 products add a red phosphor.
Why: Positive Duv or different production bins.
Fix: Buy one batch for one sightline and check Duv on the datasheet; aim for a white on or slightly below the blackbody locus.
Why: Blue-pumped LED emits almost no violet, so optical brighteners in the fabric do not fluoresce.
Fix: Expect it; if it matters, a violet-pumped LED restores the effect.
Each statement is labelled by kind — established fact, a standard’s requirement, observed market data, a convention, or Colourwise’s own interpretation or analysis — with the strength of the evidence behind it.
FactStrong evidence
White LED light is produced by phosphor conversion of a coloured (usually blue) LED, by mixing separate monochromatic LEDs such as red, green and blue, or by hybrids of the two.
Source: LED Basics
ConventionModerate evidence
The U.S. Department of Energy's guidance treats a CRI of 80 as a minimum for interior lighting and 90 or above as excellent colour fidelity, and notes that higher fidelity involves cost and efficiency trade-offs.
Caveat: Guidance for general interiors; colour-critical tasks need more than an Ra figure.
Source: LED Basics
Colourwise analysisStrong evidence
In the CIE's LED-B3 illuminant sampled at 10 nm, the spectral peak is the blue pump at 450 nm and about 17% of the 380–730 nm power lies above 640 nm, compared with about 30% for D50.
Based on: Calculated by Colourwise from the CIE LED-B3 and D50 tables at 10 nm.
Caveat: LED-B3 is one representative spectrum; commercial LEDs vary widely.
Source: CIE datasets (colour-matching functions, illuminants)
FactModerate evidence
CIE 15:2018 added a series of LED illuminants spanning roughly 2,700–6,600 K, covering phosphor-converted blue, hybrid, RGB and violet-pumped types.
Source: Standard illuminant (Wikipedia); CIE datasets (colour-matching functions, illuminants)
FactLimited evidence
Multi-primary LED systems can reach one chromaticity with different channel mixes, and the resulting colour rendition varies between those mixes.
Caveat: Cited from the paper's bibliographic record and title; the size of the variation it reports was not reviewed here.
Reviewed 29 September 2026. Colourwise summarises its sources in its own words and does not reproduce standards text or proprietary colour data. Spotted an error? Tell us.