Light-Emitting Diode

A light-emitting diode (LED) is a semiconductor diode that emits light when forward current crosses a p–n junction. Electrons and holes recombine across the band gap and release photons, so the color is set by the semiconductor material rather than by the temperature of a filament. Its small size, low power use and long operating life have made it the dominant light source for general illumination, displays and indicators.

The LED's history runs from early observations of light emission in the 1900s and 1920s to the first visible red devices in the 1960s and the blue and white devices of the 1990s. The development of blue and white LEDs was recognized by the 2014 Nobel Prize in Physics. Because LEDs convert electricity into light far more efficiently than older lamps, they have become a major tool for reducing energy demand in buildings, transport and climate mitigation strategies.

This entry explains the physical principle and the role of band structure in determining color, followed by materials, device construction, light extraction and white-light generation. It then covers performance metrics, electrical behavior, reliability and history, before surveying inorganic, organic, quantum-dot and micro-LED technologies. It ends with applications, health and environmental considerations, economics, the relationship to lasers and solar cells, and the main open research problems.

Basic operating principle

An LED is built around a P–N junction in a semiconductor crystal. The n-type side contains a surplus of mobile electrons supplied by donor dopants, and the p-type side contains a surplus of holes, which are vacancies in the valence band that behave as positive carriers. With no external voltage, diffusion carries carriers across the interface until a depletion region forms, and the resulting built-in potential blocks further flow.

Forward bias reverses this situation. When the p side is connected to the positive terminal and the applied voltage is large enough, the barrier shrinks, and electrons move from the n side while holes move from the p side into a thin active region at the junction. In that region the two populations overlap, and the injected carriers recombine: an electron drops from the conduction band into an empty state in the valence band.

Recombination can release energy as light or as heat. When the transition is radiative, the energy difference is emitted as a photon, and in a well-designed device its energy is close to the band gap of the material. Non-radiative routes compete with this. Electrons can be captured at crystal defects, where the energy ends up as lattice vibrations, or they can transfer their energy to a third carrier in an Auger process. Each non-radiative event removes a photon that could have been produced, and the ratio of radiative to total recombination events largely determines how much of the electrical input becomes light.

The light itself is produced by electroluminescence, meaning that electrical energy is converted directly into emitted photons through carrier recombination. An incandescent lamp works differently. Its tungsten filament is heated to roughly 2,700 to 3,000 K and radiates a broad, continuous spectrum that approximates blackbody radiation, with most of its energy in the infrared. An LED's emission, by contrast, is concentrated in a narrow band around the energy set by the semiconductor's band gap. For visible devices the spectral width is typically a few tens of nanometres, so the light is nearly monochromatic but not perfectly so.

The current–voltage behaviour of the junction determines when emission occurs. Below the turn-on voltage the current is very small, and above it the current rises exponentially with voltage. For an idealized diode this is described by $I = I_s\left(e^{qV/(nkT)}-1\right)$, where $I_s$ is the saturation current, $n$ is an ideality factor, and $q$, $k$ and $T$ have their usual meanings. For a direct-gap material the turn-on voltage is roughly $V_{\text{on}} \approx E_g/q$, which is why devices with wider band gaps need higher forward voltages. Under reverse bias the current is only a tiny leakage, carriers are not injected into the active region, and the device emits essentially no light. This asymmetry is why an LED emits only when it is connected in the forward direction, and why excessive reverse voltage, which can cause breakdown, must be avoided.

Energy bands and emission wavelength

In a semiconductor, electrons fill a lower-energy valence band and can move through a higher-energy conduction band. The energy interval between the two is the band gap, a range of energies that electrons cannot occupy in an ideal crystal. In an LED, electrons in the conduction band fall into empty states (holes) in the valence band, and the released energy leaves the crystal as a photon. To a first approximation, the photon energy equals the band gap $E_g$. Photon energy and wavelength are linked by $E = hc/\lambda$, where $h$ is Planck's constant and $c$ is the speed of light. With $E_g$ in electronvolts and $\lambda$ in nanometres, this becomes $\lambda \approx 1240/E_g$.

Substituting typical values shows how strongly colour depends on the material. A gap of about 1.8 eV corresponds to roughly 690 nm, a deep red, while about 2.7 eV corresponds to roughly 460 nm, a blue. Shorter wavelengths therefore require wider gaps. Real emission usually falls somewhat below the nominal gap, because carriers often recombine between shallow energy levels near the band edges rather than at the extremes of the bands. For this reason, measured peak wavelengths are best read as close approximations of the gap value, not exact equivalents.

The band gap alone does not guarantee bright emission. Whether recombination is efficient depends on whether the band extrema are aligned in crystal momentum. In a direct-gap semiconductor, the conduction-band minimum and valence-band maximum occur at the same momentum, so an electron can recombine with a hole and emit a photon, which carries negligible momentum, with high probability. In an indirect-gap semiconductor, the extrema are offset in momentum, so a radiative transition needs a third participant, a lattice phonon, to supply the momentum change. This makes the process far less likely, and non-radiative pathways take over.

Gallium phosphide illustrates the problem. Its indirect gap of about 2.26 eV would correspond to green light, but pure crystals emit weakly, with conversion efficiencies of around 0.1 percent. Introducing nitrogen as an isoelectronic impurity creates localized states that capture electrons and holes and allow them to recombine radiatively. This made nitrogen-doped GaP practical for green and yellow-green indicator LEDs, although its efficiency remained limited to a few percent. The approach works by providing a recombination pathway inside the gap, not by changing the gap itself, which explains why doping can raise efficiency without shifting the nominal band-gap wavelength.

Even for a direct-gap material, an LED emits a band of wavelengths rather than a single line. Carriers are spread over a range of energies by thermal distribution and by recombination through levels of varying depth, and compositional variations in alloys add further broadening. The typical spectral width is a few tens of nanometres at half maximum. This is why LED light is described as nearly, but not perfectly, monochromatic, and it distinguishes LEDs from laser diodes, whose stimulated emission produces much narrower lines.

Semiconductor materials and colors

An LED's color is set by the band gap of its active layer, so the choice of semiconductor alloy determines which part of the spectrum a device can reach. Each device emits a single hue whose photon energy matches that gap (the active layer's gap sets the energy of the emitted light). Each material family therefore covers its own portion of the spectrum, and the families differ in how far they can be tuned.

Early arsenide and phosphide compounds

Development began with gallium arsenide. At RCA, Jacques Pankove reported infrared electroluminescence from GaAs in 1962, though GaAs was an imperfect candidate for a semiconductor. Texas Instruments then announced a commercial GaAs infrared product that emitted near 900 nm later that same year. Visible emission came from Nick Holonyak Jr. at General Electric, who in 1962 used gallium arsenide phosphide to build a red device by applying stimulated emission to that material. Replacing part of the arsenic with phosphorus widens the gap and moves the output from infrared into the visible range (the composition change shifted the wavelength of the GaAs diode into the visible). Tuning the phosphorus fraction produced red near 660 nm, orange near 610 nm and yellow near 590 nm (the composition of the n-type GaAsP layer set the color). Past a certain phosphorus fraction the band structure turns indirect, so emission is weak unless nitrogen is added as an emission center (indirect alloys were brightened by nitrogen doping).

Aluminum gallium indium phosphide: red through yellow

Aluminum gallium indium phosphide (AlGaInP) can be grown lattice-matched to GaAs wafers, and its color depends on the aluminum fraction. Its direct-gap window extends from about 1.9 eV, which is red, to about 2.26 eV, which is yellow-green, but radiative efficiency drops as aluminum content nears the direct-to-indirect crossover (efficiency falls significantly at higher aluminum composition). For this reason commercial AlGaInP devices are mostly red, orange and amber (commercial production is concentrated in these colors). Substrate choice also matters. A GaAs substrate absorbs part of the emitted light because its gap is smaller, whereas a GaP substrate, with its larger gap, does not (GaAs absorbs the light, while GaP's wider gap avoids this). Doping imposes further limits. A patent description states that p-type AlGaInP generally cannot reach carrier concentrations of about 10^18 cm^-3, whereas GaP cladding layers can exceed that level (AlGaInP carrier density is hard to push past this level).

Gallium nitride and indium gallium nitride: blue, green and ultraviolet

Gallium nitride (GaN) has a band gap of about 3.4 eV, and its recombination produces light at roughly 360 nm in the ultraviolet (GaN is a direct-gap material with a gap near 3.4 eV, which corresponds to emission near 360 nm). Alloying with indium narrows the gap and moves emission into the visible range (adding indium to GaN lowers the band gap). In principle the gap of InGaN can be tuned from about 0.69 eV in the infrared to 3.4 eV in the ultraviolet (composition control spans this entire range). In practice, blue quantum wells use indium fractions between 5% and 20% (quantum wells of this composition emit blue), and a pure blue of about 470 nm corresponds to an indium fraction near 0.2 (the 470 nm case was given as z = 0.2). Because GaN cannot be grown on a lattice-matched wafer, devices are built on sapphire or silicon carbide (lattice-matched substrates such as these are used for growth). Before the 1990s, LEDs offered a limited palette, and the GaN system completed it, allowing blue and white devices to be made (earlier LEDs lacked blue; GaN-based devices filled the gap).

Limits of each material system

Each system is constrained in a different way. AlGaInP loses efficiency near its indirect crossover and is hard to dope p-type, so it cannot reach the green end cleanly. InGaN depends on incorporating enough indium to reach longer wavelengths, which strains the crystal and complicates growth on available substrates. The aluminum-rich nitrides face doping and extraction difficulties, which limit their ultraviolet reach. These constraints are the reason no single material family covers the full visible range with uniform efficiency.

Device structure and epitaxial growth

The active layers of an LED are built as single-crystal films by epitaxial growth, most often by metalorganic chemical vapor deposition (MOCVD) or molecular beam epitaxy. For III-nitride blue and green chips, sapphire has been the workhorse substrate for about two decades; the epitaxial growth of GaN-based epi-layers on sapphire by metal organic chemical vapor deposition (MOCVD) process is still the main stream of LED chips manufacturing. Sapphire is cheap, but it conducts heat poorly, with a thermal conductivity of about 25 W/m·K. Silicon carbide and silicon are the other commercial substrates, and each is chosen by balancing cost, crystal match and heat flow against the requirements of the device.

Substrate mismatch is the central materials problem. The lattice spacing of sapphire differs from that of gallium nitride by as much as 14%, which introduces misfit dislocations in the plane of the interface. Thermal mismatch adds to the strain: the thermal expansion coefficient of sapphire is about 34% larger than that of GaN, leading to a bending state accompanied by a residual compressive strain during the cooling process. The result is a threading dislocation density in heteroepitaxial nitrides of 10^7–10^10 cm^-2. These dislocations matter because they are extended non-radiative defects that serve to reduce the internal quantum efficiency, so carriers that reach them recombine without producing photons.

Several strategies reduce this penalty. Growers deposit a low-temperature nucleation layer before the main film, and some use patterned sapphire substrates; one study found that the PSS was proved to be an efficient method to decrease the threading dislocation (TD) density in GaN epifilm with the lateral growth mode on PSS. Lateral overgrowth and compliant buffer layers are also used, the latter because a buffer layer absorbs the misfit dislocations generated by the lattice mismatch between GaN and the sapphire substrate. The cleanest route is homoepitaxy on bulk GaN, which is limited by the cost and scarcity of large GaN wafers. One comparison reported that homoepitaxial LEDs had threading dislocation densities of 2–5×10^8 cm^−2 which was one order of magnitude less than those grown on c-plane sapphire substrate, and the optical output power of the light emitting diode increased by more than one order of magnitude compared to those on sapphire substrate.

The active region is designed to confine carriers. The double-heterostructure concept, proposed in 1963 by Zhores Alferov and Herbert Kroemer, uses a double heterostructures for carrier confinement in the active region. A narrow-gap layer is placed between wider-gap layers, so the energy barriers at both interfaces hold electrons and holes together. Under forward bias, recombination takes place in and is essentially restricted to the central region. The wider-gap cladding is also transparent to the light from the central layer, so emitted photons are not reabsorbed. Modern nitride LEDs usually replace the thick central layer with one or more quantum wells: a thin InGaN well is sandwiched between higher bandgap GaN barriers. When the well is thin enough, carrier motion perpendicular to the layer becomes quantized, and multiple wells are commonly stacked to form a multi-quantum-well region. Electron-blocking layers are often placed on the p-side of the active region to keep electrons from escaping into the p-type material.

Carriers must also reach the active region uniformly from the surface. Because p-type nitrides conduct poorly, the top surface is covered by a transparent conductive layer. Indium tin oxide's main advantageous feature is the combination of electrical conductivity and optical transparency. Without such a layer, due to the presence of N-GaN series resistance, current crowing phenomenon still exist at the electrode edge of vertical structure GaN-based LEDs. Designers sometimes add a strip-shaped SiO2 current blocking layer, which is well aligned with a p-electrode, to prevent the current from crowding around the p-electrode. Metal contacts and bond pads then define the electrical path into the chip. Because indium is scarce and its price has risen, researchers have also tested graphene and carbon nanotube films as alternatives to ITO for these spreading layers.

Light extraction and packaging

Light generated in the active region must still leave the chip before it reaches a viewer, and this step often limits the overall efficiency of an LED. Semiconductors such as gallium nitride and AlInGaP have a refractive index far above that of air, so photons that meet the top surface at a steep angle are reflected back into the crystal instead of escaping. Because the escape window is narrow, total internal reflection traps much of the light, and the trapped photons are eventually absorbed inside the chip and turned into heat.

The size of the loss can be estimated from the critical angle. For a chip with an index of about 3.5 encapsulated in epoxy with an index of about 1.5, the critical angle is roughly 25°, so only light falling within narrow cones around each face can cross the boundary. An efficient conventional transparent-substrate AlInGaP LED encapsulated in epoxy may have an external quantum efficiency of only about 40%, despite having an internal quantum efficiency of nearly 100%. The gap between these two figures is the extraction loss that packaging and chip design try to close. Shaping the epoxy surface alone does not solve the problem, because the largest step-down in index occurs at the semiconductor–encapsulant interface itself.

Several strategies target this bottleneck. Shaping the chip is the oldest: grinding LEDs into hemispheres makes light strike the curved surface closer to normal incidence, which reduces reflection. However, this technique is tedious and wasteful of material. Modern approaches use patterned or roughened surfaces, photonic-crystal structures and patterned sapphire substrates to scatter trapped light into escape directions, while flip-chip mounting moves the light-emitting surface away from the wire-bonded side so that light can leave through the transparent substrate. Metallic reflectors on the mounting side redirect photons that would otherwise be lost, and dome-shaped encapsulants increase the critical angle. Higher-index domes are not a free remedy, however: optical absorption in high index glasses and mechanical stress typically degrade the performance of an LED encapsulated in such glass.

Packaging also has to protect the chip and supply power, and its materials are chosen with extraction and reliability in mind. Epoxy and silicone encapsulants are the most common, but silicone is generally preferred for high-power devices because epoxy's thermal expansion differs strongly from that of the semiconductor, and the resulting stress can damage the die during repeated heating and cooling. Lead frames serve small indicator packages, while ceramic and metal-core substrates handle higher power. Lenses formed over the package shape the emission pattern, so that a given amount of extracted light is directed toward the intended area.

Thermal design is inseparable from optical design. Heat generated at the junction must flow through the die, the substrate, the solder joints and the package to the surrounding air or to a heat sink, and the efficiency of this path depends on the thermal conductivity of each material. Rising junction temperature reduces light output and shortens operating life, so a package with poor heat flow loses both brightness and longevity even when its optics are well designed. For this reason, extraction gains and thermal management are usually optimized together rather than separately.

White light and color mixing

White light from LEDs is produced in two main ways: by adding the outputs of several colored emitters, or by converting part of a short-wavelength emission into longer wavelengths with a luminescent material. The two routes differ in how easily color can be tuned, how faithfully the light renders colors, and how the output changes over years of operation.

Additive mixing of colored chips

In an RGB design, red, green and blue chips sit in one package or luminaire, and their drive currents are balanced so that their combined output looks white. Because each channel can be adjusted separately, this approach suits color-changing fixtures. Its weakness is that the three chips do not respond to heat or age in the same way. Their emission wavelengths shift with temperature, and their light output falls at different rates, so a mix that looks neutral when new can drift toward blue, pink or yellow. Engineers counter this with feedback: sensors measure each channel and the controller corrects the currents, though this adds cost, circuit complexity and energy use. Some products add a dedicated white emitter alongside the three colors (an RGBW arrangement), which gives a cleaner white than a simple red-green-blue mixture.

Phosphor-converted white LEDs

The dominant commercial method is to coat a blue InGaN chip with a phosphor. The usual choice is cerium-doped yttrium aluminum garnet (YAG:Ce³⁺), which absorbs part of the blue emission and re-emits it as broad yellow light; the unabsorbed blue and the yellow together are perceived as white. The design is simple and efficient, but its spectrum is weak in red. The result is a cool, bluish white with a high correlated color temperature and a color rendering index that is often in the 70s or low 80s. Warm-white products therefore add red-emitting phosphors, typically manganese-activated fluorides. Research groups have also shown that a single yellow-red phosphor can reach warm tones on a blue chip: one reported phosphor with a blue chip at 440–470 nm gave a correlated color temperature below 4000 K and a color rendering index above 80. Near-ultraviolet chips paired with tricolor phosphor blends are another active research route.

Color metrics

Correlated color temperature (CCT) is expressed in kelvin and describes whether white light appears warm or cool. Lower values look warmer and more yellowish, higher values look cooler and bluer, and roughly 2500–4000 K is conventionally called warm. The color rendering index (CRI, usually written Ra) indicates how closely a source reproduces the colors of reference illumination, with 100 as the ideal. The standard index averages eight moderately saturated test colors and says little about saturated red, so a supplementary index, R9, is often reported alongside it. CCT and CRI describe different things: a light can have a pleasing color temperature yet render colors poorly, or the reverse.

Trade-offs arise because the same phosphor choice affects both efficacy and rendering. Broadening the spectrum with red emission improves CRI, but the human eye is least sensitive to deep red, so the extra red photons add little to measured lumens and luminous efficacy falls. Thermal behavior matters as well: phosphor emission decreases as temperature rises, so the thermal stability of a phosphor limits how well its color holds under real operating conditions.

Color shift over time

Color stability is one of the main reliability concerns for white LEDs. In a phosphor-converted device, the balance between blue and yellow changes as the phosphor ages, which shifts the color. Studies have also linked the color degradation of a lamp shortly after switch-on to the thermal transient behavior of the chip, since the peak wavelength drifts and the peak intensity falls with time. In one patent example, a warm white LED with a red sulfide phosphor had its color temperature rise from about 3275 K to 3575 K after 1000 hours at 85 °C and 85% relative humidity. Yellowing of silicone or plastic encapsulants, driven by heat and ultraviolet exposure, adds a further filter effect. In practice, designers manage drift through thermal design, stable phosphors, and in some systems by adding red LEDs to pull the combined color back toward its target.

Efficiency, photometric units and performance metrics

An LED's efficiency is best understood as a chain of conversion steps, and each step can lose energy. The first is the internal quantum efficiency (IQE), the fraction of injected electrons that produce a photon inside the active region. IQE falls whenever carriers are lost through defects or other non-radiative paths, so it is highly sensitive to crystal quality. The external quantum efficiency (EQE) is the product of IQE and the light-extraction efficiency, which is the fraction of generated photons that actually leave the chip. Because high-index semiconductors trap much of their light by total-internal-reflection, extraction is often the bottleneck, and a conventional chip can emit far fewer photons than it generates.

Electrical performance is described by wall-plug efficiency (WPE), the ratio of emitted optical power to electrical input power. Each injected electron gains an energy of $qV_f$ from the bias, but it emits a photon of energy $h\nu$. This gives $\eta_{\mathrm{WP}} = \eta_{\mathrm{EQE}} \cdot \dfrac{h\nu}{qV_f} \approx \eta_{\mathrm{EQE}} \cdot \dfrac{1240}{\lambda\,V_f}$, where λ is in nanometres and $V_f$ is in volts. The forward voltage therefore matters directly: any bias above the photon energy is lost as heat in series resistance and contacts. For example, a blue device at 450 nm with $V_f = 3.0\ \mathrm{V}$ has a voltage factor of about 0.92, so an EQE of 60% corresponds to a WPE of roughly 55%.

Radiometric quantities such as radiant power in watts count all wavelengths equally, whereas photometric units count only what the human eye responds to. The lumen is the photometric unit of luminous flux, given by $\Phi_v = K_m \int \Phi_{e,\lambda}(\lambda)\,V(\lambda)\,d\lambda$, where $V(\lambda)$ is the photopic luminous efficiency function and $K_m = 683\ \mathrm{lm/W}$ is the maximum luminous efficacy of radiation, reached at 555 nm. The candela is the unit of luminous intensity and equals one lumen per steradian. As a consequence, a deep-red LED carries much less luminous flux than its radiant power suggests. At 650 nm, where $V(\lambda)$ is about 0.107, the theoretical ceiling is roughly 73 lm/W, and infrared emitters contribute nothing to visible brightness.

Luminous efficacy, defined as $\eta = \Phi_v / P_{\mathrm{el}}$ and expressed in lumens per watt, is the practical figure for comparing light sources. Luminous efficiency is the dimensionless ratio of efficacy to the 683 lm/W ceiling, and its value depends on the emission spectrum as well as on conversion losses. In white LEDs, the overall efficacy combines the blue chip's wall-plug efficiency with the conversion efficiency of the phosphor, so losses in either stage reduce the final result. Laboratory demonstrations have reported very high values for individual devices; for example, a 527 nm green InGaN LED made on patterned sapphire was reported with a peak EQE of about 53% and a luminous efficacy of 329 lm/W. Such figures describe specific devices under specific conditions and should not be read as typical commercial performance.

Efficiency is not a fixed property of a device. IQE typically rises at low drive current as defect-related losses saturate, then declines at high current density, a behaviour known as efficiency droop that is discussed in the open-problems section. Rising junction temperature increases non-radiative recombination, reduces efficiency and shifts the emission wavelength, which in turn changes the luminous efficacy because the eye's sensitivity varies with wavelength. For this reason, a quoted efficacy is meaningful only together with the drive current, pulse or continuous operation, and junction temperature at which it was measured.

Measurement itself adds uncertainty. Total luminous flux is usually determined with integrating spheres, and the results depend on instrument calibration against standards traceable to national metrology institutes. Reported values therefore carry measurement-uncertainty that must be considered when comparing products from different manufacturers, especially when the test protocols differ.

Electrical and thermal behavior

An LED is a p–n junction formed in semiconductor material, so its current–voltage relation follows the rectifying pattern of a diode. Almost no electric-current flows below a threshold. Beyond it, current rises very steeply with voltage. The threshold, called the forward voltage, is set largely by the band-gap of the active layer. A recombining electron and hole release a photon of roughly the gap energy, and a wider gap means the carriers must be driven across a larger potential before they can recombine. A red device with a gap near 1.8 to 2 eV therefore turns on at roughly 2 V, while blue and white devices built on gallium nitride alloys typically need more. White LEDs, which combine a blue chip with a phosphor, are commonly specified in a range of about 2.8 to 4.0 V, and that range also reflects manufacturing tolerance from one device to the next.

The steepness of the curve shapes how LEDs must be powered. A small change in voltage produces a large change in current, so an LED connected directly to a fixed-voltage supply can draw excessive current and fail. Two common solutions are a series resistor, which is simple but wastes energy and tolerates supply variation poorly, and a constant-current driver, which regulates the current and absorbs differences in forward voltage. Forward voltage is not fixed. It falls as the junction warms, by a few millivolts per degree Celsius, and it rises at low temperature, which can be a practical problem during cold startup. Driver designers therefore leave voltage headroom so that a string of LEDs can still start and reach full current at low temperature.

Dimming is usually achieved in one of two ways. Analog dimming lowers the drive current, but at low levels it can shift the color and reduce efficiency. Pulse-width modulation (PWM) instead switches the LED fully on and off at a fixed frequency and varies the fraction of time spent on, which keeps the color stable. The switching frequency must be high enough that the flicker is not perceived by the eye. Values above about 100 Hz are often cited for avoiding visible flicker, and frequencies of roughly 500 Hz to 1 kHz are often recommended to avoid audible noise from the driver. Video and high-speed camera work may demand far higher rates. Compatibility is a separate concern. Older phase-cut dimmers, designed for incandescent and halogen loads, may not produce a signal that an LED driver can interpret, and a driver can also fail to regulate at very low load or at low dimming levels. The result can be flicker, buzzing or dropout, and the remedy is usually matching the dimmer type to the driver, for example choosing a 0–10 V or digital protocol where appropriate.

Heat is the central thermal issue. Only part of the electrical input leaves the chip as light, and the remainder appears as heat within the junction. The junction temperature depends on the power dissipated, the ambient temperature and the thermal resistance between the junction and its surroundings. Rising junction temperature has several effects. The emission peak shifts, usually toward longer wavelengths, which changes the color of the output. Nonradiative recombination becomes more likely, so the internal efficiency and light output fall. Heat also accelerates the chemical and structural changes that cause degradation, including discoloration of phosphors and encapsulants. In constant-voltage designs, the falling forward voltage of a warming device can allow its current to increase, which raises heating further, creating a feedback loop that the driver must prevent. For these reasons, the packaging and heat-sinking discussed elsewhere in this article are an integral part of electrical design, and the same power level can produce very different light output and lifetime depending on how well the heat leaves the chip. Thermal behavior is therefore closely tied to the choice of drive current, which is also discussed in the sections on efficiency and lifetime.

Lifetime, degradation and reliability

LED products seldom stop working abruptly. Their light output sinks slowly, so the industry defines lifetime as the operating time needed for output to fall to a chosen fraction of its starting value. The most common benchmark is L70, the point at which 70% of the original light remains, with L80 and L90 used for stricter applications. An LED that still glows after its rated hours is therefore not necessarily performing as specified. Lumen loss, color shift and the failure of a driver or solder joint are separate issues, and a single lifetime figure can hide any of them.

Several physical processes cause the decline. Inside the chip, defects that grow over time increase non-radiative recombination, so fewer injected carriers produce photons. In packaged white LEDs, the encapsulant and phosphor layers are more vulnerable. Silicone and epoxy binders that surround the phosphor absorb blue light and heat from the die, which can break polymer chains, oxidize the material and turn it yellow. This lowers light extraction and shifts the color. Phosphor can also discolor, and some nanophosphors are reported to degrade quickly above roughly 60–80 °C. Moisture adds another route: it can hydrolyze the phosphor and encapsulant, corrode metal lead frames, and tarnish silver-plated reflectors. In high-temperature aging tests, silver chloride formed on exposed leadframe surfaces after delamination and was identified as a main cause of lost light. Thermomechanical stress from repeated switching can fatigue solder joints and die-attach material, which raises thermal resistance and, in turn, the operating temperature. Measurements also show that devices running with power applied degrade faster than unpowered ones, because the heat accelerates every process above.

Manufacturers and standards bodies quantify these effects with two linked methods. LM-80, published by the Illuminating Engineering Society (IES), specifies how to measure lumen maintenance and color shift of LED packages, arrays and modules at several controlled temperatures. It requires a minimum test period of 6,000 hours, with 10,000 hours preferred, and readings at intervals of at least 1,000 hours. LM-80 itself does not extrapolate beyond the measured period. That role belongs to TM-21, introduced in 2011, which fits an exponential decay model to the LM-80 data and projects values such as L70. TM-21 limits any projection to six times the length of the test, so a 6,000-hour dataset supports a claim of at most 36,000 hours. Claims exceeding this limit, such as many advertised "100,000-hour" ratings, are generally regarded as unsupported by the standard. TM-21 also applies to components rather than complete luminaires, so system-level lifetime requires separate evaluation.

The most important operating variable is the junction temperature, the temperature of the semiconductor die itself, rather than the ambient air or the outer case. Higher drive current produces more heat and therefore faster degradation, while low current and moderate temperatures slow chromaticity shift to the point where it may hardly occur. Consequently, the same device can last very different periods in a well-ventilated fixture and in an enclosed, hot luminaire. Thermal design, including heat sinks, board materials and the thermal path from die to ambient, is therefore central to reliability. Because observed lifetime depends on these conditions, data-sheet values are meaningful only when the test temperature and current are stated, and comparisons between products should use identical conditions.

Historical development

The earliest observations date to the first decade of the twentieth century. Silicon carbide (carborundum) crystals in a circuit at Marconi Laboratories in Britain were found to glow, and the discovery is usually dated to 1906, with the report appearing in early 1907 Electroluminescence from a solid state diode was discovered in 1906 by Henry Joseph Round of Marconi Labs, and was published in February 1907 in Electrical World. Depending on the crystal and the voltage, the samples produced yellow, light green, orange, or blue light. The material was a poor emitter, because silicon carbide has an indirect band gap, so the output was faint.

Interest revived in the 1920s. Losev studied the light emission from zinc oxide and silicon carbide crystal rectifier diodes, which were used in radio receivers. He concluded that the glow was cold rather than thermal and was tied to the rectifying action of the junction. His 1927 patent for a "light relay" pointed toward optical signalling, an application that remains central to LED communication research.

The next major step came in the 1950s, when research moved toward gallium arsenide. Infrared emission from gallium arsenide and related alloys was documented at RCA in 1955 In 1955, Rubin Braunstein, of the Radio Corporation of America (RCA), reported on infrared emission gallium arsenide (GaAs) and semiconductor alloys. Sources differ on which milestone counts as the first practical LED. Texas Instruments researchers developed infrared devices in 1961, but because their output was invisible to the eye, the more widely cited breakthrough is the visible one that followed.

That breakthrough belongs to Nick Holonyak Jr. at General Electric. He had begun work on the visible-spectrum device around 1960, growing GaAs1-xPx alloys who first began work on the visible-spectrum light-emitting diode in 1960 while working at General Electric. By varying the phosphorus content, he shifted the emission from infrared into the red part of the spectrum. A red device made from this alloy operated at room temperature in October 1962 Holonyak created crystals of gallium arsenide phosphide to make an LED that would emit a visible red light., and this made it genuinely useful rather than a laboratory curiosity.

Commercial production followed within a decade. Until 1968, visible and infrared LEDs were so expensive that visible and infrared LEDs were extremely costly, on the order of US $200 per unit, and so had little practical use. Monsanto then produced GaAsP red indicator chips in volume, and Hewlett-Packard initially using GaAsP supplied by Monsanto. In 1972, work led by M. George Craford, a former student of Holonyak, produced a yellow device and a much brighter red one. Green emitters arrived in the 1970s, but blue remained out of reach until the 1990s After LEDs that produce green and blue light were developed (in the 1970s and '90s, respectively).

Blue emission was much harder to achieve. Gallium nitride was a promising candidate because of its wide band gap, but growing high-quality crystals and making p-type material were serious obstacles the compound semiconductor gallium nitride (GaN), which could be ideal for creating blue LEDs because it had a large band-gap energy corresponding to ultraviolet light.. Good crystals were obtained in the late 1980s and early 1990s, with This was solved independently in the late 1980s and early 1990s by Akasaki and Amano and also by Nakamura. Doping was a separate problem, since how to dope the GaN so it is a p-type semiconductor, which is crucial for creating an LED. The first blue-emitting diode from the Nagoya group was reported in 1992.

Isamu Akasaki, Hiroshi Amano and Shuji Nakamura shared the 2014 Nobel Prize in Physics for this work. Despite considerable efforts, both in the scientific community and in industry, the blue LED had remained a challenge for three decades. The prize citation described the achievement as "the invention of efficient blue light-emitting diodes which has enabled bright and energy-saving white light sources". Once blue emitters existed, white light could be produced by combining them with red and green sources or with phosphors, which underpins most of today's LED lighting.

Types of inorganic LEDs by function

Inorganic LEDs, built from crystalline semiconductor layers in a diode structure, are commonly grouped by the job they perform. Function determines chip size, drive current, package type and thermal design, so an indicator lamp and a streetlight module differ in far more than brightness.

Indicator LEDs

Indicator LEDs are the smallest and least powerful devices in the family. They are typically driven at currents of a few to about twenty milliamperes and serve as power, standby and status lights on appliances, instruments and network equipment. Many are also arranged into seven-segment or dot-matrix arrays for numeric and alphanumeric displays. Their small emitting area and modest heat load allow simple, inexpensive packages, which is why they were among the earliest mass-market LED products.

High-power illumination LEDs

High-power devices run at hundreds of milliamperes, and often at an ampere or more, and they use larger chips to deliver useful luminous flux. Most of the electrical power that is not converted to light becomes heat, so these LEDs depend on efficient heat removal through metal-core or ceramic substrates and heat sinks. Their main applications are general lighting, automotive headlamps and signal lamps, and projection, where a compact, bright source replaces bulkier lamps. Thermal management is therefore a limiting factor in their design rather than a secondary concern.

Infrared emitters

Infrared emitters operate beyond the red end of the visible range, within the electromagnetic spectrum. Near-infrared devices around 850 nm and 940 nm are common in television remote controls and in proximity and optical sensors. In fiber-optic communication, LEDs carry signals over short-reach multimode links near 850 nm and 1300 nm, while long-haul systems rely on laser diodes in the 1310 nm and 1550 nm windows, where fiber loss is lowest.

Ultraviolet LEDs

Ultraviolet LEDs cover UV-A, roughly 315–400 nm, and, more recently, UV-C, below about 280 nm. UV-A devices are used for curing inks and adhesives and for security markings that fluoresce under UV light. UV-C devices, typically emitting near 260–280 nm, are being adopted for disinfecting water, air and surfaces. Their efficiency remains well below that of visible-light LEDs, largely because aluminum-rich nitrides are difficult to dope and to extract light from.

Packaging forms

Packaging is matched to function. Surface-mount devices (SMDs) with standard footprints, such as the 3528 (about 3.5 × 2.8 mm) and 5050 (5.0 × 5.0 mm) sizes, are placed by automated machines on circuit boards in electronics assembly. Through-hole lamps, usually 3 mm or 5 mm in diameter, remain common for indicators and hand-assembled products. Multi-chip RGB packages house red, green and blue dies in one housing, allowing color mixing for displays and architectural fixtures. Chip-on-board (COB) modules mount many bare dies directly on a single substrate and cover them with a phosphor layer, producing a dense, bright source used in downlights and spotlights.

Organic, quantum-dot, perovskite and micro-LED technologies

Organic light-emitting diodes (OLEDs) use stacks of very thin films as the active part of the device, and the emissive layer is built from small carbon-based molecules or polymers. Because these films can be deposited over large areas and on flexible supports, OLEDs suit displays in which each pixel emits its own light and no separate backlight is needed. Their color and efficiency are set by the organic materials chosen, so the design work centers on molecular chemistry rather than on the crystal growth that dominates inorganic LEDs.

Quantum-dot LEDs (QLEDs) use semiconductor nanocrystals whose size sets the emission color. Smaller dots emit shorter wavelengths, so one material family can cover a range of narrow, saturated colors. The first efficient dots contained cadmium, which is toxic, and cadmium-based QDs have made remarkable progress, primarily due to their exceptional stability. Specifically, red and green QLEDs have almost met industrial standards. Cadmium-free replacements, mainly based on indium phosphide and zinc selenide compounds, have been developed to meet regulatory and environmental concerns. Blue has been the hardest color. One study reported cadmium-free blue quantum dot LEDs with a quantum yield of unity, an efficiency at the theoretical limit, high brightness and long operational lifetime, while indium phosphide-based devices have reached an external quantum efficiency of 26.6% in another report. Remaining problems include efficiency roll-off at high current, which is often linked to charge imbalance and charged QDs that induce Auger recombination.

Perovskite LEDs use metal-halide compounds whose emission can be tuned across the visible range while the emission lines stay narrow. One review states that these materials can have extremely narrowband emissions (full-width-at-half-maximum, FWHM ≤ 20 nm) and easily tunable bandgap (emission range: 410–850 nm). Efficiency has risen quickly; the current external quantum efficiency (EQEs) of PeLEDs have approached 30%, which is comparable to commercial OLEDs. Other reports have described devices with a peak external quantum efficiency of 27.8% and a turn-on voltage of 1.7 volts. Stability, however, remains the main obstacle. One review calls the poor operational stability of PeLEDs the grandest challenge toward commercial applications, and a device with a 25.6% peak efficiency has been reported with an operating half-life of only about two hours at high luminance. Hybrid tandem devices that stack a perovskite unit on an organic unit have reported efficiencies above 40%, which shows how the two technologies can be combined.

Micro-LEDs are arrays of microscale inorganic emitters, usually gallium nitride-based chips, that act as individual pixels. The inorganic chips offer high brightness, fast response and long life, and micro-LED displays have advantages such as high contrast ratio, fast response speed, wide color gamut, low power consumption and long lifespan compared with LCDs and OLEDs. Several obstacles slow their commercial spread. Red chips are typically grown on gallium arsenide wafers, while blue and green chips are grown on sapphire, so the three colors must be brought onto one backplane. This makes mass transfer, which requires transferring millions of micro-LEDs from a growth substrate to a display plane, a central bottleneck. The scale is large: about 25 million chips are needed for a 4K TV display. Efficiency also falls as chips shrink, because the main loss of EQE comes from a non-radiative recombination formed by the defects and surface energy states of sidewalls or surfaces of LED. Inspection and repair of defective pixels are additional technical hurdles.

Lighting applications

LEDs have displaced incandescent lamps and many fluorescent lamps in homes, offices, factories and public spaces. Two routes account for much of this change. Retrofit lamps are designed to fit existing sockets and fixtures, so a building can switch without rewiring. Integrated luminaires, in which the LED module, driver and optics form a single unit, are more common in new construction and in commercial projects. Adoption has been driven by energy savings and long service life, as well as by efficiency regulations in many countries. In the United States, the Energy Information Administration reported that the share of homes using LEDs as their primary light source rose from about 4% in 2015 to 54% in 2023. The U.S. Department of Energy projects that LED lamps and luminaires will account for most lighting installations by 2035.

Street, roadway and architectural lighting use LED fixtures that are often managed as networked systems rather than as standalone lamps. Because an LED emits light from a small source, its optics can direct the beam toward the road surface or building facade with less wasted light, which reduces spill into neighbouring areas and the sky. Dimming and scheduling can be linked to sensors that detect traffic, pedestrians or ambient daylight, and to central software that reports the status of each fixture. Architectural installations also exploit the ability of multi-colour LED packages to produce changing colours and dynamic effects, while the low heat output of the sources makes them practical for lighting surfaces that would be damaged by hot lamps. The same controllability raises questions about outdoor lighting's effects on wildlife, which are discussed elsewhere in this entry.

In vehicles, LEDs are used in headlamps, daytime running lights, brake and turn signals, and interior lighting. Their small size gives designers more freedom in shaping lamp housings, and their rapid switching means that a brake signal reaches full brightness far faster than a filament lamp, which must heat up before it emits its full output. The compact form and low heat output also make LEDs suitable for dense lamp clusters and for sealed assemblies in which hot filaments would be a problem.

Horticultural lighting is a specialised application in which the spectrum, rather than just the brightness, is the design target. Chlorophyll absorbs most strongly in the blue and red parts of the visible spectrum, with pronounced absorption peaks near 430 nm and 660 nm. Researchers often describe red light between about 600 and 700 nm as the most efficient driver of photosynthesis, and blue light around 400 to 500 nm is added to support normal development and desired plant form. Because narrow-band LEDs can be combined in precise ratios, growers in greenhouses, vertical farms and growth chambers can tune the spectrum for seedlings, vegetative growth or flowering. Results are not uniform across species, however. In one study of lemon balm grown under LED spectra in a plant factory, white light and red-rich spectra produced the best growth and yield, while blue-rich spectra gave lower yields despite a clear effect on photosynthetic activity. Such findings show why horticultural designers evaluate biomass and quality, not only photosynthetic efficiency.

Displays, signaling and communication

Liquid-crystal displays do not emit light themselves, so LEDs have become the usual light source behind them. In LED-backlit panels the diodes are arranged in one of two main ways. Edge-lit designs line the rim of the screen with LEDs and send their light into a light guide plate that spreads it across the panel. This allows thin housings, but light from each LED travels a long way across the guide, so turning individual LEDs off has little effect on contrast. Direct-lit designs place a two-dimensional array of LEDs directly behind the panel. Each LED, or group of LEDs, can be dimmed separately, which improves dynamic contrast. The cost is thickness: a diffuser plate must sit some distance in front of the diodes to avoid bright spots and uneven brightness, and that spacing makes the display thicker than an edge-lit one. Televisions, monitors, laptops and phones make these trade-offs differently depending on whether thickness or picture contrast matters more.

LEDs are also used where a signal must be visible at a distance, often in daylight. Traffic signals were one of the earliest large-scale applications. A single-color LED emits the required hue without a color filter, and its directional output removes the large reflector that incandescent signal heads needed. Eliminating that reflector and filter also removes a source of false light, in which ambient sunlight entered the housing and left through the lens. Lower power draw and longer service life reduced maintenance, although early outdoor LED signs failed in sunlight and heat until designers added ventilation and shading. Large video walls and signage now use modular LED panels that can form flat, curved or irregular surfaces, and temporary versions can be wall-mounted, hung or fitted to trucks. Vehicle indicators such as brake lights and turn signals use LEDs for their compact size and fast switching.

Visible light communication uses LED illumination to carry data. The IEEE 802.15.7 standard, published in 2011 and later superseded by a 2018 revision, defines the physical and medium-access layers for short-range links in the visible band, roughly 380 to 780 nm. Data are encoded by rapidly changing the light intensity, a change too quick for the eye to see, so one fixture can light a room and transmit information at the same time. The original standard supports data rates of up to 96 Mb/s. Because the modulation must not produce visible flicker, the design has to account for dimming and use coding schemes that keep the average brightness steady. The approach is widely known as LiFi. Its advantages include the use of unlicensed spectrum and immunity to radio interference, while its main constraint is the need for a clear optical path between transmitter and receiver. The 2018 revision broadened the scope to optical wireless communication across a wider range of wavelengths, from the infrared into the ultraviolet.

Infrared LEDs paired with photodiodes provide some of the most common optical functions in consumer electronics. Because the emitted light lies outside the visible range, users do not see it, yet a detector can register reflected infrared to determine whether an object is close, for example to switch off a phone's touchscreen during a call. The same pairing carries short-range data in remote controls and in simple optical links between devices, where the transmitted pulses are interpreted as commands or digital bits. These applications rely on the same electromagnetic spectrum principles as visible light, but they operate at wavelengths chosen for detection rather than for human vision.

Health, safety and environmental effects

Because an LED concentrates its output in a fairly narrow band of the electromagnetic spectrum, its photobiological safety is assessed separately from its electrical safety. The main reference is IEC 62471, which covers optical radiation from electrically powered incoherent sources, from the ultraviolet to the infrared. Products are sorted into risk classes, with exempt devices at one end and high-risk devices at the other.

The most discussed hazard for white and blue LEDs is the blue-light hazard. Wavelengths roughly between 400 and 500 nm can drive photochemical reactions in retinal tissue, and reactive oxygen species may form there under prolonged, intense exposure. The blue component of many white LEDs peaks near 440 nm. Because this damage is chemical rather than thermal, it accumulates over repeated exposure and the eye's usual aversion response may not occur in time. Severe retinal injury can produce blind spots.

Flicker and glare are a separate concern for visual comfort. Flicker can arise from poorly designed drivers or from pulse-width dimming at low settings, and its effect depends heavily on driver quality and on how the light is used. Not all biomedical uses are cautionary. Blue LEDs are used in phototherapy for neonatal jaundice, and ultraviolet LEDs are used for disinfection in clinical settings.

Ecological effects are an active area of research. Many animals rely on the daily and monthly cycles of daylight and moonlight to decide when to forage, migrate or mate, and artificial light can disturb those cues. In a UK study of roadside hedgerows, areas under LED streetlights held roughly half as many caterpillars as unlit areas, while sodium lamps produced a smaller decline. A separate analysis of moths found that white diodes had a sharper effect on abundance, development and feeding than high-pressure sodium lamps did. Collisions with buildings may kill as many as a billion migratory birds across North America each year. Common mitigation measures include shielding fixtures so light falls only where needed, dimming outdoor lighting, and choosing warmer spectra. The impacts of light pollution on ecosystems are therefore an important design consideration.

Compared with incandescent lamps, LEDs contain no mercury, deliver the same light for less electricity and have much longer rated lives. Their end of life nonetheless raises questions. Each module combines semiconductors, drivers, metals and plastics, which complicates recycling, and the growing volume of discarded fixtures adds to electronic waste.

Economics, manufacturing and intellectual property

Few lighting technologies have become cheaper as quickly as LEDs. Online retail data for household LED bulbs in the United States show annual price declines of 28% to 44% between 2011 and 2014, with higher-lumen lamps falling faster. Analysts describe this as an experience curve, in which price falls as cumulative output grows. Estimates differ: one analysis finds LED A lamp prices fell by about 18% for each doubling of cumulative production, while another reports a 20-25% decline per doubling of cumulative shipments in 2012-2013. Both figures exceed the 14-15% per doubling that earlier lighting technologies showed. A simpler empirical rule, Haitz's law, expresses the same trend as a roughly tenfold drop in cost per lumen each decade, alongside a large rise in light output per package. Public procurement can speed this up. India's UJALA programme bought bulbs in bulk through competitive tenders, and winning bids fell from Rs 310 per bulb in the first round to as low as Rs 38 in later rounds.

Chip fabrication, packaging and module assembly are concentrated in a small number of East Asian industrial clusters, where dense supplier networks and large-scale capacity reinforce one another. This concentration brings price competition with it. One market-research summary, citing the China Semiconductor Industry Association, reports that Chinese LED chip prices declined by 20% in 2023 alone, and the same summary estimates that global LED chip production capacity exceeds demand by approximately 30%. Such figures come from a commercial source and should be read as estimates rather than official statistics.

Government policy has been a major driver of adoption. Brazil and Venezuela began phasing out incandescent lamps in 2005, several years before the European Union. In the EU, the Ecodesign framework phased out incandescent lamps from 2009 to 2013 through regulation 244/2009, and the phase-out was then extended to halogen and fluorescent lamps. In the United States, the Energy Independence and Security Act of 2007 led the Department of Energy to tighten efficiency rules for general-service lamps. The rule requiring general service lamps to exceed 45 lumens per watt is beyond the reach of incandescent and halogen technology, so these lamps are effectively excluded from the market. The International Energy Agency reports that around 90 countries use minimum energy performance standards that prohibit low-efficacy lighting products, and that about 50% of global residential lighting sales use LED technology.

The commercial history of the device is also a history of patent disputes. The most publicized case involved Shuji Nakamura, who developed the gallium nitride blue LED while working at Nichia. He was paid a nominal sum for each patent filing, and in 2001 he sued his former employer, arguing that he had not been rewarded adequately for key blue-LED patents. In 2004 the Tokyo District Court awarded him a one-off payment of 20 billion yen. The case then went to the Tokyo High Court, which mediated a settlement in January 2005 of roughly 840 to 844 million yen; the figures vary slightly across reports. The settlement was the largest in Japan for an employee's invention and prompted companies to reassess their reward systems. Patent rights in the wider industry have also been resolved through licensing rather than litigation alone. In 2009 Nichia settled all patent litigation with Seoul Semiconductor across several jurisdictions, and the agreement included a cross-licence covering both LED and laser diode technologies.

Standardization has helped the market mature. Common surface-mount package footprints allow chips and modules from different suppliers to fit the same board layouts. Manufacturers also sort ("bin") output by wavelength, luminous flux and forward voltage, so that buyers can specify matched parts. These practices reduce the risk of buying from a single vendor and make it easier for new entrants to compete on price and performance.

Relationship to lasers, photodiodes and solar cells

A laser diode is built on the same kind of p–n junction as an LED, but it adds an optical cavity. The two end faces of the crystal act as mirrors, and the device is driven hard enough for stimulated emission to dominate. An incoming photon of the right energy can trigger an electron–hole pair to recombine and release a second photon with the same wavelength, direction and phase. These photons bounce between the mirrors and gain strength on each pass. Below a threshold current, a laser diode behaves much like an LED and emits spontaneous light over a broad spectrum. Above threshold, the optical output rises steeply, and the light becomes coherent, narrow in wavelength and highly directional.

An ordinary LED relies almost entirely on spontaneous emission. Each recombination event releases a photon in a random direction with a random phase, so the output is incoherent and spread over a spectral width that is typically several tens of nanometres. Some devices sit between the two extremes. Superluminescent diodes suppress the cavity feedback so that they deliver high power with a spectrum and coherence length closer to an LED's than to a laser's. The distinction is therefore less a sharp boundary than a matter of how strongly the cavity feeds back into the emission process.

The same junction can also work in reverse. Under reverse bias or at zero bias, photons whose energy exceeds the band gap are absorbed in the depletion region and create electron–hole pairs. The built-in electric field separates these carriers before they can recombine, producing a photocurrent proportional to the incident optical power. This is the internal photoelectric effect that makes semiconductor photodiodes useful as light detectors. Photodiodes are optimized for this role, with thin or well-positioned absorbing regions, antireflection coatings and materials chosen for the wavelengths to be detected. Silicon, for example, is widely used for visible and near-infrared detection.

In a solar cell, the junction operates in the power-generating quadrant, where absorbed light drives carriers through an external load and produces both voltage and current. The band gap constrains this operation in a way that parallels its role in LEDs. A gap that is too large absorbs too few photons from sunlight, while a gap that is too small lets the carriers thermalize and waste photon energy as heat, and the open-circuit voltage is tied to the gap itself. The reciprocity between emission and absorption follows from detailed balance: a material that emits light efficiently, given the right conditions, also tends to absorb it efficiently. This is why an LED can detect light, a property exploited in some low-cost optical sensing schemes, including pairs in which one LED emits and another detects. Reciprocity has limits, however. Silicon has an indirect band gap and emits light poorly, which is why it is rarely used for LEDs, yet it is an excellent absorber for photodiodes and the dominant material for solar cells.

All of these devices share the band-structure physics described in quantum mechanics and doping theory. Energy bands, carrier populations, and the behaviour of electrons and holes at a junction underlie LEDs, laser diodes, photodiodes, solar cells and transistors alike. Advances in crystal growth, doping control and junction design therefore tend to benefit several device families at once. Progress in one area also highlights the constraints of another: the same carrier dynamics that make a material a good LED can make it a poor detector, and vice versa.

Open problems and research frontiers

Efficiency droop is the fall in an LED's internal efficiency as the drive current density rises, and it limits how much light a single chip can produce. For years the leading explanation has been Auger recombination. In this process, an electron–hole pair recombines without emitting a photon and transfers its energy to a third carrier. The most prevalent mechanism is a three-particle interaction, two electrons and one hole, or two holes and one electron. In 2011, a UCSB group led by Chris van de Walle proposed that this process caused droop in nitride LEDs, and a 2013 experiment by Speck, Weisbuch and colleagues detected energetic electrons emitted from a working device, a signature of the Auger process. The debate has not fully closed, because fitting measured efficiency to the common ABC model assigns droop to Auger recombination almost by construction, whatever the actual cause is. Competing accounts point to hot electrons overflowing the active region and to uneven carrier distribution.

The green gap is the shortfall in efficiency for emitters near 530 nm and beyond. The quantum efficiency of green light-emitting diodes is still significantly lower than that of blue- and red-emitting devices. This efficiency lag is commonly known as the "green gap" in the solid-state lighting industry. The main difficulty lies in the InGaN alloy. Raising the indium content shifts emission toward green, but it also strengthens polarization-induced electric fields inside the quantum wells, which push electrons and holes apart and reduce their wavefunction overlap and lead to a reduction in the electron and hole wavefunction overlap. Indium-rich layers also suffer from lattice mismatch between GaN and InN, which creates defects, and random fluctuations in indium concentration can lower the radiative recombination rate due to random fluctuations of the Indium concentration naturally present in any InGaN alloy. The gap matters commercially because phosphor-free white light made by mixing red, green and blue chips needs a green emitter near 530 nm, which sits close to the center of the gap lying nearly at the center of the "green gap". Phosphor-converted green sources have so far outperformed direct green chips in efficiency the phosphor-converted green LEDs are more efficient than direct GaN-based green LEDs so far.

Deep-ultraviolet LEDs, which use aluminum-rich AlGaN, lag far behind their visible counterparts. Compared with InGaN-based blue LEDs, the efficiency of AlGaN-based DUV LEDs is still very low. Several problems combine. Threading dislocations are especially harmful here: AlGaN QWs conventionally use high TDD AlN templates with IQEs as low as 1%. Doping is also a barrier. In p-type AlGaN with high aluminum content, the hole concentration of p-AlGaN with a high Al content (Al > 60%) is low, which limits hole injection into the active region. Extraction is a further obstacle, since the p-GaN contact layer absorbs ultraviolet light and the intrinsic optical properties of AlGaN restrict how many photons escape the extraction of DUV photons is significantly limited by intrinsic material properties of AlGaN. These problems matter because deep-UV sources are sought for purification of air and water, sterilization in food processing, UV curing, medical-, and defense-related light sources, and they could eventually replace mercury-based lamps.

Micro-LEDs, perovskite LEDs and quantum-dot LEDs face separate challenges. Micro-LED displays require placing very large numbers of microscale chips onto a backplane, so mass transfer, yield and brightness uniformity are central engineering problems. Perovskite and quantum-dot emitters need longer operating lifetimes, and their development also includes efforts to remove toxic heavy metals from their formulations.

References

1. The wavelength range of LEDstoshiba.semicon-storage.com
2. Gallium arsenide phosphide - Knowledge and Referencestaylorandfrancis.com
3. 3.3.9: Light Emitting Diode - Engineering LibreTextseng.libretexts.org
4. Direct Band Gap Wurtzite Gallium Phosphide Nanowires - PMCpmc.ncbi.nlm.nih.gov
5. Gallium phosphideen.wikipedia.org
6. Green luminescence efficiency in gallium phosphideresearchgate.net
7. Wide bandgap semiconductor light emittersimage-ppubs.uspto.gov
8. White color light emitting deviceimage-ppubs.uspto.gov
9. 75 Years of Innovation: LED, first blue-light emitting diode - SRIsri.com
10. Light-Emitting Diodes in the Solid-State Lighting Systemsarxiv.org
11. Nick Holonyak, Jr.britannica.com
12. Vacuum light emitting diodeimage-ppubs.uspto.gov
13. Epitaxial wafer for a light-emitting diode and a light-emitting diodeimage-ppubs.uspto.gov
14. AlGaInP LEDs break performance barriers (Cover Story) - Newscompoundsemiconductor.net
15. Light emitting diode with asymmetrical energy band structureimage-ppubs.uspto.gov
16. White-light emitting devices and methods for manufacturing the sameimage-ppubs.uspto.gov
17. Light-emitting deviceimage-ppubs.uspto.gov
18. Monolithic integration of AlGaInP-based red and InGaN-based green LEDs via adhesive bonding for multicolor emissionnature.com
19. What is an LED?buildings.com
20. AlGaInP red-emitting light emitting diode under extremely high pulsed pumpingresearch.aston.ac.uk
21. Sapphire Substrate - an overviewsciencedirect.com
22. Defect analysis of the LED structure deposited on the sapphire substrate - ScienceDirectsciencedirect.com
23. Ultralow threading dislocation density in GaN epilayer on near-strain-free GaN compliant buffer layer and its applications in hetero-epitaxial LEDsncbi.nlm.nih.gov
24. Dislocation-specific lateral epitaxial overgrowth to reduce dislocation density of nitride filmsimage-ppubs.uspto.gov
25. (PDF) Characteristic enhancement of InGaN-based light emitting diodes grown on pattern sapphire substratesresearchgate.net
26. Dislocation analysis in homoepitaxial GaInN/GaN light emitting diode growth - ScienceDirectsciencedirect.com
27. Nobel Lecture: The double heterostructure concept and its applicationslink.aps.org
28. Method of fabricating semiconductor light emitting devicesimage-ppubs.uspto.gov
29. Quantum-Confined Stark Effect - an overviewsciencedirect.com
30. LED device with embedded top electrodeimage-ppubs.uspto.gov
31. (PDF) Effect of indium tin oxide (ITO) current spreading layer on the current uniformity of vertical structure GaN-based light-emitting diodesresearchgate.net
32. ito ohmic contact: Topics by Science.govscience.gov
33. Nickel Foam as a Substrate for III-nitride Nanowire Growtharxiv.org
34. Quantum Well Semiconductor Lasersfiberoptics4sale.com
35. What is Double Heterojunction (DH) LED : Structure, Working, Advantages and Disadvantagesrfwireless-world.com
36. Semiconductor Junction Structuresfiberoptics4sale.com
37. Monte Carlo simulation of indium tin oxide current spreading layers in light emitting diodes (Journal Article)osti.gov
38. AlGaInP light-emitting diodes with SACNTs as current-spreading layerlink.springer.com
39. Light emitting structures with selective carrier injection into multiple active layersimage-ppubs.uspto.gov
40. Light emitting diodes with improved light extraction efficiencyimage-ppubs.uspto.gov
41. Light emitting devices with improved light extraction efficiencyimage-ppubs.uspto.gov
42. Light Extraction - an overviewsciencedirect.com
43. Light emitting devices with improved light extraction efficiencyimage-ppubs.uspto.gov
44. Tripling the light extraction efficiency of a deep ultraviolet LED using a nanostructured p-contactnature.com
45. Blue LED (InGaN)-driven yellow and red emitting phosphor for white light generation - ScienceDirectsciencedirect.com
46. New yellow Ba0.93Eu0.07Al2O4 phosphor for warm-white light-emitting diodes through single-emitting-center conversionnature.com
47. Novel rare-earth-free yellow Ca5Zn3.92In0.08(V0.99Ta0.01O4)6 phosphors for dazzling white light-emitting diodesncbi.nlm.nih.gov
48. Blue-light-excited SrLu2Al2GaScSiO12:Ce3 + cyan-emitting garnet phosphor for high color rendering index full-spectrum WLEDs - ScienceDirectsciencedirect.com
49. Phosphor-converted white LED with low deviation of correlated color temperature and color coordinates and method of preparing the sameimage-ppubs.uspto.gov
50. Method of color mixing for an aviation anti-collision lightimage-ppubs.uspto.gov
51. blue led chips: Topics by Science.govscience.gov
52. LED drive circuitimage-ppubs.uspto.gov
53. Why LEDs Show Different Colors and How to Fix Themstarledprofile.com
54. Color Control System for RGB LED With Application to Light Sources Suffering From Prolonged Agingresearchgate.net
55. (PDF) Very high external quantum efficiency and wall-plug efficiency 527 nm InGaN green LEDs by MOCVDresearchgate.net
56. Luminous Efficacy and Efficiencyrp-photonics.com
57. Plug Efficiency - an overviewsciencedirect.com
58. Light-emitting device with low forward voltage and method for fabricating the sameimage-ppubs.uspto.gov
59. Luminous efficacyen.wikipedia.org
60. High luminous efficacy phosphor converted white LEDsimage-ppubs.uspto.gov
61. Constant power LED circuitimage-ppubs.uspto.gov
62. Adaptive startup method for constant current LED driversimage-ppubs.uspto.gov
63. Constant Voltage Vs. Constant Current LED Driversmy.avnet.com
64. 1 Junction temperature in light-emitting diodes assessed by different methodsproaudiodesignforum.com
65. Efficient LED Control: Exploring PWM Dimming - LEDYi Lightingledyilighting.com
66. LED Dimmer Compatibility: Fix Flickering & Pulsing Lightspanelsandplugs.com
67. LED Driver and Dimmer Compatibility Guide: How to Choose the Right Pair - Enoledenoled.com
68. LED Flickering: Is It the Light Source or the Driver?elisacustom.com
69. Lumen maintenanceen.wikipedia.org
70. Techniques for lumen maintenance and color shift compensationimage-ppubs.uspto.gov
71. Decroluxdecrolux.com.au
72. LM-80 vs. TM-21: Testing vs. Lifetime Projectionhi-hyperlite.com
73. Light Standard: What is TM-21? TM-21-19 or TM-21-21?ledlightexpert.com
74. LED Lighting Lifespan Guide: L70, L80, TM-21 Projections Explained for B2B Procurementksimpexp.com
75. PS-10-18: IES Position on LED Product Lifetime Prediction - Illuminating Engineering Societyies.org
76. Ageing Analysis of Light-Emitting Diodes Used in Consumer Lighting - PMCpmc.ncbi.nlm.nih.gov
77. Degradation Analysis of Silicone as Encapsulation and Molding Material in High Power LEDsintechopen.com
78. Next Generation Lighting Industry Alliance LED Systems Reliability Consortiumenergy.gov
79. Failure Mechanisms and Color Stability in Light-Emitting ...osti.gov
80. Degradation Physics of High Power LEDs in Outdoor Environment and the Role of Phosphor in the degradation processnature.com
81. (PDF) Investigation on Some Parameters Affecting Optical Degradation of LED Packages During High-Temperature Agingresearchgate.net
82. Thermal conductivity and phase transition heat transfer mechanism including optical element to be cooled by heat transfer of the mechanismimage-ppubs.uspto.gov
83. Hot light emitting diode (LED) lighting systemsimage-ppubs.uspto.gov
84. Light-emitting diodeen.wikipedia.org
85. Light emitting diodeimage-ppubs.uspto.gov
86. Nick Holonyak Jr., pioneer of LED lighting, diesece.illinois.edu
87. The First Visible LED Glowed Red - DEV Communitydev.to
88. 0% found this document useful (0 votes)scribd.com
89. Gallium arsenide phosphidebritannica.com
90. Isamu Akasaki, Hiroshi Amano and Shuji Nakamura win 2014 Nobel Prize for Physicsphysicsworld.com
91. The 2014 Nobel Prize in Physics - Press release - NobelPrize.orgnobelprize.org
92. History of the LEDen.wikipedia.org
93. The long history of light-emitting diodes - RadioLocmanradiolocman.com
94. Nick Holonyak, Jr. - Inventor of the First Visible LED - Radiant Historyradianthistory.com
95. History Of Light-emitting Diodehykolity.com
96. Nobel Prize in Physics 2014 - ChemistryViewschemistryviews.org
97. Full article: A comprehensive review on the lifetime of cadmium-free quantum-dot light-emitting diodestandfonline.com
98. Efficient and stable blue quantum dot light-emitting diodenature.com
99. Blue cadmium-free and air-fabricated quantum dot light-emitting diodespubs.aip.org
100. Efficient and stable hybrid perovskite-organic light-emitting diodes with external quantum efficiency exceeding 40 per centncbi.nlm.nih.gov
101. Highly stable perovskite light-emitting diodes: Mattercell.com
102. Diode arrayimage-ppubs.uspto.gov
103. Mass transfer, detection and repair technologies in micro-LED displayslink.springer.com
104. Mass transfer for Micro-LED display: Transfer printing techniques - ScienceDirectsciencedirect.com
105. Perovskite LEDs with an external quantum efficiency exceeding 25%communities.springernature.com
106. Efficient and stable hybrid perovskite-organic light-emitting diodes with external quantum efficiency exceeding 40 per centnature.com
107. Ultralow voltage–driven efficient and stable perovskite light-emitting diodesscience.org
108. Highly efficient light-emitting diodes via self-assembled InP quantum dotsncbi.nlm.nih.gov
109. Future trends of display technology: micro-LEDs toward transparent, free-form, and near-eye displaysnature.com
110. LED Lighting Statistics to Know in 2025ledlightingsupply.com
111. LED lighting Market Share, Size and Industry Growth Analysis 2021 - 2026industryarc.com
112. LED Lighting Market Size, Share & Growth Report 2026-2033grandviewresearch.com
113. Spectral effects of light-emitting diodes on plant growth, visual color quality, and photosynthetic photon efficacy: White versus blue plus red radiationncbi.nlm.nih.gov
114. The Impact of LED Lighting Spectra in a Plant Factory on the Growth, Physiological Traits and Essential Oil Content of Lemon Balm (Melissa officinalis)ncbi.nlm.nih.gov
115. LED Grow Lights for Horticulturewaveformlighting.com
116. LED-backlit LCDen.wikipedia.org
117. Backlights including patterned reflectorsimage-ppubs.uspto.gov
118. Backlight unit and display deviceimage-ppubs.uspto.gov
119. Methods and apparatus to construct an ultrathin display panelimage-ppubs.uspto.gov
120. LED indicator lampimage-ppubs.uspto.gov
121. LED lamp with prismatic cover lensimage-ppubs.uspto.gov
122. Information display signimage-ppubs.uspto.gov
123. LED traffic light with individual LED reflectorsimage-ppubs.uspto.gov
124. Methods and systems of large scale video displayimage-ppubs.uspto.gov
125. IEEE 802.15.7-2011 - IEEE Standard for Local and Metropolitan Area Networks--Part 15.7: Short-Range Wireless Optical Communication Using Visible Lightstandards.ieee.org
126. IEEE SA - IEEE Standard for Local and Metropolitan Area Networks--Part 15.7: Short-Range Wireless Optical Communication Using Visible Lightstandards.ieee.org
127. [PDF] IEEE 802.15.7 visible light communication: modulation schemes and dimming supportsemanticscholar.org
128. Visible Light Communication Technologies: A Review of IEEE 802.15.7 and Its Role in Next‐Generation Networks - Bhutani - 2025 - International Journal of Communication Systems - Wiley Online Libraryonlinelibrary.wiley.com
129. IEC 62471 Blue Light Hazard Explained - Custom LED Strip Manufacturer From Chinasignliteled.com
130. For visible lamp radiation, including the radiation emitted from LEDs, thereenvironmentallights.com
131. Blue Light Safety - Konica Minolta Sensingsensing.konicaminolta.us
132. Photobiological Safety (EN 62471): Blue Light Hazard Explainedfanxstar.com
133. Want to help wildlife? Turn off your lights.nationalgeographic.com
134. LEDs may cause more insect declines than other lights - The Wildlife Societywildlife.org
135. Street lighting has detrimental impacts on local insect populationsscience.org
136. How light pollution can imperil migrating birds by luring them into citiespbs.org
137. The evolving price of household LED lamps: Recent trends and historicalosti.gov
138. Recent price trends and learning curves for household LED lamps from a regression analysis of Internet retail dataeta.lbl.gov
139. India’s LED Lighting Storyenergy.prayaspune.org
140. LED Material Market Size, Growth Outlook 2034emergenresearch.com
141. The incandescent light bulb phase-out: exploring patterns of framing the governance of discontinuing a socio-technical regimelink.springer.com
142. US & EU Complete Incandescent & Halogen Lamp Phase Outslightnowblog.com
143. Lighting - IEAiea.org
144. LED firm rejects Nobel laureate's olive branchphysicsworld.com
145. IPBiz: Nichia/Nakamura settle over blue LEDipbiz.blogspot.com
146. Nichiaen.wikipedia.org
147. LEDs: Two decades of glorious growth - Newscompoundsemiconductor.net
148. Fiber delivered laser induced white light systemimage-ppubs.uspto.gov
149. Quantum well superluminescent diodeimage-ppubs.uspto.gov
150. Superluminescent diode having a quantum well and cavity length dependent threshold currentimage-ppubs.uspto.gov
151. What are Laser Diodes?techweb.rohm.com
152. LEDs and Laser Diodes: A Tale of Two Semiconductor Devicesthors.com
153. Laser diodes or LEDs what sets them apart - Utmelutmel.com
154. (PDF) Identification of Auger effect as the dominant mechanism for efficiency droop of LEDsresearchgate.net
155. Efficiency Models for GaN-based Light-Emitting Diodes: Status and Challengesarxiv.org
156. Green gap in GaN-based light-emitting diodes: in perspectiveresearchgate.net
157. 1 The unexpected origin of the quantum efficiency reduction (“greenosti.gov
158. Unraveling the "Green Gap" problem: The role of random alloy fluctuations in InGaN/GaN light emitting diodesarxiv.org
159. Status of GaN-based green light-emitting diodes<xref ref-type="fn" rid="cpb150160fn1">*</xref>cpb.iphy.ac.cn
160. Research Progress of AlGaN-Based Deep Ultraviolet Light-Emitting Diodesncbi.nlm.nih.gov
161. Research Progress of AlGaN-Based Deep Ultraviolet Light-Emitting Diodes - PMCpmc.ncbi.nlm.nih.gov
162. An elegant route to overcome fundamentally-limited light extraction in AlGaN deep-ultraviolet light-emitting diodes: Preferential outcoupling of strong in-plane emissionncbi.nlm.nih.gov
163. algan uv leds: Topics by Science.govscience.gov
164. Cause of LED efficiency droop finally revealed: Auger recombination responsiblephys.org
165. (PDF) Auger effect identified as main cause of efficiency droop in LEDsresearchgate.net