Efficiency is a system property, not a single number

Growers who compare horticultural lighting usually run into a vocabulary problem before they run into a physics problem. Fixture documentation quotes figures taken at different points along the power path — the LED emitter, the driver output, and the wall plug — and it labels them with words that sound interchangeable but are not. Getting the measurement plane right is what makes an energy comparison meaningful. Getting it wrong produces the familiar complaint that a “high-efficiency” fixture did not move the electricity bill as far as the brochure implied.

Two definitions carry most of the weight. **LED efficiency is the ratio of optical output power to electrical input power**, written in W·W⁻¹ or as a percentage. **Efficacy is the number of photosynthetic photons emitted per second divided by the electrical input power**, written in µmol·J⁻¹ and also printed as µmol/J or PPE, short for photon efficacy. Efficiency compares watts with watts at the emitter. Efficacy compares photons with joules at the boundary of the luminaire, and it is the metric that horticultural qualification programmes actually regulate.

Both definitions are set out in the peer-reviewed review *From physics to fixtures to food: current and potential LED efficacy*, published in **Horticulture Research** (Nature portfolio): https://www.nature.com/articles/s41438-020-0283-7

A third quantity sits one level above both: photon delivery per day, in mol·d⁻¹. That is what the crop actually experiences, and it is the bridge between a fixture rating and a lighting plan. The sections below work through the three layers, the loss mechanisms that separate them, and the cost arithmetic that turns watts and photons into money.

Three layers of efficiency, and why they are not interchangeable

Layer one is the emitter. Device-level efficiency describes how much optical power a given LED die or package produces for the direct current pushed through it. It is a component figure, measured at a defined junction temperature and drive current, and it appears in component datasheets and binning tables.

Layer two is the luminaire. Luminaire efficacy, the µmol·J⁻¹ figure on a qualification listing or a test report, is measured at the wall plug. It therefore already contains everything that happens between the mains connection and the photons leaving the housing. Comparing a component efficiency figure with a luminaire efficacy figure means comparing two different boundaries, and the gap between them is the loss chain described in the next section.

Layer three is delivery. Daily photon delivery is not a fixture property at all; it is efficacy multiplied by wall-plug power multiplied by hours of operation. Two identical fixtures can deliver very different daily totals depending on the photoperiod chosen and the output level used. **The daily light integral (DLI) is the accumulated photosynthetic photon flux density over one day**, usually expressed in mol·m⁻²·d⁻¹, and it is the quantity that lighting schedules are written against.

Keeping the layers apart matters because the errors compound. A device number used as a fixture number flatters performance. A fixture number used as a daily total ignores the schedule. A daily total quoted without the photoperiod it assumes cannot be checked. Any PPFD figure quoted for a fixture is likewise only interpretable together with the mounting height above the canopy and the daily duration of operation; a PPFD value without those two conditions describes an installation, not a luminaire.

Where the joules go: the wall-plug loss chain

Because luminaire efficacy is measured at the plug, every stage between the plug and the canopy can reduce it. The chain runs in a fixed order.

  • **AC to DC conversion in the driver.** Mains power is rectified and converted into the current the LEDs need. Driver conversion losses are part of the luminaire efficacy figure, not additional to it.
  • **Conversion inside the emitter.** Not every charge carrier that crosses the junction produces a photon, and the fraction that does is the device efficiency discussed above.
  • **Secondary optics.** Lenses, reflectors and diffusers shape the beam, and they also absorb and scatter light. Optical loss sits inside the luminaire efficacy figure as well.
  • **Thermal derating.** Junction temperature rises with drive current and with poor heat sinking. Photon output falls as the junction warms, so a fixture measured hot will show lower efficacy than the same fixture measured cold.
  • **Ageing.** Light output declines over service life, so the efficacy measured at hour zero is not the efficacy delivered at hour 20,000.

Two cautions follow. First, the sources consulted for this article do not list typical percentage splits for driver, optical and thermal losses individually, and no such percentages are quoted here; anyone who publishes those splits should also state the conditions under which they were measured. Second, rated life statements are comparable only when the hours and the measurement basis are stated with them. A life claim without rated hours is not a specification, and the specific hours applicable to a given product must be taken from that product’s own documentation.

The external benchmark: what a qualifying efficacy threshold means

Rather than inventing an internal standard, it is more useful to compare against a published one. The DesignLights Consortium (DLC) publishes horticultural technical requirements that set a minimum photon efficacy threshold for LED luminaires seeking listing. In the V4.0 requirements, **the minimum PPE threshold for LED luminaires is 2.5 µmol·J⁻¹**. DLC describes that value as an 8.7% increase over the threshold in Hort V3.0, and as more than 45% above the highest-efficacy non-LED reference in the same comparison, the 1000 W double-ended HPS fixture.

Source: DesignLights Consortium, *Horticultural Technical Requirements V4.0* — https://designlights.org/our-work/horticultural-lighting/technical-requirements/hort-v4-0/

A threshold is a floor, not a ranking. A luminaire that just clears 2.5 µmol·J⁻¹ and one that clears it comfortably are both listed; the threshold states the entry requirement, not the operating cost. It is also a plug-level number, so it carries the driver and optical losses described above, which is precisely why comparing it against a bare LED device figure is invalid. The same logic applies in reverse: a device-level efficiency quoted as a percentage cannot be used to argue that a finished luminaire meets the threshold, because the boundary is different.

Cost per mole of light: why high efficacy is not automatically low cost

Efficacy tells you how many photons you get per joule. It does not tell you what a photon costs, because photons are paid for twice: once in capital and once in electricity. The useful frame is **cost per mole of photons, which is the total annualised cost of owning and running a fixture divided by the moles of photons it delivers in a year**. Nelson and Bugbee developed and applied that framework in a PLOS ONE economic analysis of greenhouse lighting.

Source: PLOS ONE, *Economic Analysis of Greenhouse Lighting: Light Emitting Diodes vs. High Intensity Discharge Fixtures* — https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0099010

In that study, **the cost per mole of photons for LED fixtures was reported as about 2.3 times higher than for HPS fixtures, a gap the authors attributed to high capital costs**. That result is specific to the conditions of the study and to the period in which it was carried out — the analysis was published in 2014 — so it is a historical data point rather than a description of current pricing. What survives is the framework, which remains the reason the metric exists: it forces capital expenditure and electricity into the same units.

Written out, the arithmetic is straightforward:

cost per mole of light = (annualised capital cost + annual electricity cost) ÷ annual moles of photons delivered

where annual electricity cost = wall-plug watts ÷ 1000 × hours of operation per year × electricity tariff, and annual moles of photons = efficacy in µmol·J⁻¹ × wall-plug watts × seconds per year ÷ 10⁶.

The structure of the result is what matters. When capital dominates — low utilisation hours, short ownership horizons, or a large price gap between fixture classes — the fixture with lower efficacy can still win on cost per mole. When electricity dominates — long photoperiods, high tariffs, long service life — efficacy gains translate more directly into operating savings. Electricity tariffs, utilisation hours and maintenance costs are site-specific and are not quoted here; they must be entered as variables for the specific project.

One related caution belongs here even though the mechanism does not: reducing output through dimming does not automatically reduce electricity consumption in proportion to the percentage of dimming, so dimming should not be treated as a substitute for buying efficacy in the first place. The signalling and protocol-level reasons behind that statement are a separate subject.

Comparing efficiency metrics like for like

The table below sets the common metrics side by side, including what each one leaves out.

Metric Unit What it measures Where it is published What it does not cover
Emitter (device) efficiency W·W⁻¹ or % Optical power out of the LED against DC power into it Component datasheets and binning tables Driver, optical and thermal losses at luminaire level
Luminaire efficacy (PPE) µmol·J⁻¹ Photosynthetic photons leaving the luminaire per joule drawn at the wall plug Test reports and qualification listings Whether photons reach the canopy
Qualification threshold (DLC Hort V4.0) µmol·J⁻¹ Minimum photon efficacy required of a listed LED luminaire DLC Horticultural Technical Requirements V4.0 Performance above the floor; spectrum fit
Daily photon delivery mol·d⁻¹ Efficacy × wall-plug watts × hours of operation Calculated from the rows above Not a fixture rating; depends on the photoperiod used
Cost per mole of photons $·mol⁻¹ Annualised capital plus electricity divided by moles delivered Economic analyses such as Nelson & Bugbee (2014) Tariffs, utilisation and maintenance are site-specific

Putting the numbers to work

A practical specification sequence falls out of the three layers. Start with the delivery target: decide the DLI target for the crop and growth stage, and the photoperiod that will be used to reach it. That converts into a required rate of photons per second, which is the number a fixture has to supply after accounting for geometry losses at the chosen mounting height and for the reflectance of the growing space. Only then compare luminaire efficacy, and only among fixtures measured on the same basis. Finally, run the cost-per-mole arithmetic with your own tariff and utilisation hours, because that is the step where rankings frequently change.

For growers working at greenhouse scale, Michigan State University Extension maintains a resource on LED lighting for greenhouse crops that covers the system-level context, including the relationship between fixture output and the crop canopy: https://www.canr.msu.edu/resources/led-lighting-for-greenhouse-crops

The credibility test for a manufacturer is whether all three layers are visible in the documentation: photon efficacy measured at the plug, the conditions under which it was measured, and the driver, optical and thermal design that sustains it over the service life. Koray Lights has manufactured LED luminaires for horticultural and commercial applications for 17 years, and it publishes its framework for the six elements of grow light on its own site at https://koraylights.com/6-indispensable-elements-of-grow-light/, where high efficiency and energy saving is presented as one element of a system rather than as a single headline figure.

What to take away

Efficacy in µmol·J⁻¹ is the number to compare between luminaires, because it is measured at the wall plug and therefore includes driver and optical losses. Device-level efficiency figures answer a different question and cannot be substituted for it. Daily photon delivery, not instantaneous output, is what the crop integrates, and it depends on the photoperiod and mounting height as much as on the fixture. Cost per mole of light is the metric that decides whether a purchase actually saves money, and its outcome depends on the balance between capital cost and electricity cost at a specific site. Any of these numbers used without its stated conditions is not evidence of anything.

Frequently asked questions

What is the difference between LED efficiency and efficacy?

LED efficiency is the ratio of optical output power to electrical input power, expressed in W·W⁻¹ or as a percentage. Efficacy is the number of photosynthetic photons emitted per second divided by the electrical input power, expressed in µmol·J⁻¹. Efficiency is measured at the emitter, while efficacy is measured at the luminaire boundary, so the two values describe different points on the power path.

Is a higher µmol·J⁻¹ rating always cheaper to run?

No. A higher efficacy rating reduces the electricity required to deliver a given number of photons, but the total cost of light also includes capital cost. Where utilisation hours are low or the price difference between fixture classes is large, a lower-efficacy fixture can have a lower cost per mole of photons. The comparison has to include the local tariff, the hours of operation and the service life.

What does the DLC 2.5 µmol·J⁻¹ threshold mean for growers?

It is the minimum photon efficacy that an LED luminaire must reach to be listed under the DesignLights Consortium horticultural technical requirements V4.0. It functions as an entry requirement rather than a performance ranking, and because it is measured at the wall plug it already includes driver and optical losses.

Why can an LED fixture have a higher cost per mole of light than an HPS fixture?

Because the two fixture classes differ in the split between capital and running cost. In the 2014 PLOS ONE economic analysis by Nelson and Bugbee, the cost per mole of photons was reported as about 2.3 times higher for LED fixtures, attributed to high capital costs under the conditions of that study. The framework remains useful because it puts capital and electricity into the same units.

Does dimming reduce energy use proportionally?

Not necessarily in direct proportion to the dimming percentage, so dimming should not be treated as a replacement for choosing an efficient luminaire. The protocols and signal behaviour that govern dimming performance are a separate topic with their own engineering considerations.