Phosphors consist of an inorganic host material doped with an optically active element. Common hosts are garnets, which have the chemical formula A3B5Oi2 where A and B are chemical elements and O is oxygen. Among the large group of garnets, yttrium aluminum garnet (YAG), Y3Al5Oi2, is a particularly common host material. Phosphors having YAG as a host material are called YAG phosphors. The optically active dopant is a rare-earth element, a rare-earth oxide, or

another rare-earth compound. Most rare-earth elements are optically active. Rare-earth light - emitting elements include cerium (Ce) used in white-light YAG phosphors, neodymium (Nd) used in lasers (Nd-doped YAG lasers), erbium (Er) used in optical amplifiers, and thorium (Th) oxide used in the mantle of gas lights.

The optical characteristics of YAG phosphors can be modified by partially substituting Gd for Y and Ga for Al so that the phosphor host has the composition (Y1_xGdx)3(Al1_yGay)5O12. The emission spectra for a Ce-doped (Y1_xGdx)3(Al1_yGay)5O12 phosphor with different compositions are shown in Fig. 21.5 (Nakamura and Fasol, 1997). The figure reveals that the addition of Gd shifts the emission spectrum to longer wavelengths whereas the addition of Ga shifts the emission spectrum to shorter wavelengths.

Fig. 21.6. Chromaticity points of YAG:Ce phosphor, and the general area (shaded) accessible to white emitters consisting of a blue LED and YAG:Ce phosphor (adapted from Nakamura and Fasol, 1997). Also shown is the planckian locus with color temperatures.

The chromaticity points of the YAG:Ce phosphors are shown in Fig. 21.6. The shaded region
reveals the chromaticities that can be attained by mixing light from a blue LED source with the light of YAG:Ce. The figure reveals that such white emitters can have a very high color temperature.

An alternative to YAG phosphor is TAG phosphor, which is based on terbium aluminum garnet, Tb3Al5O12. Both YAG and TAG crystallize with the garnet structure. Additionally, the Y3+ and Tb3+ ionic radii are very close (rY3+ = 1.02 A and rTb3+ = 1.04 A). Consequently, the X - ray diffraction pattern of YAG does not strongly change as Y in YAG is substituted with Tb, even for Tb mol fractions of 30% (Potdevin et al., 2005). Although TAG phosphors may have slightly lower radiative efficiency than YAG phosphors, TAG phosphors represent a viable alternative to YAG phosphors (Kim, 2005).

21.2 White LEDs based on phosphor converters A white LED lamp using a phosphor wavelength converter and a blue GaInN/GaN optical excitation LED was first reported by Bando et al. (1996) and reviewed by Nakamura and Fasol (1997). The GaInN/GaN LED used for optical excitation (“optical pumping”) was a device reported by Nakamura et al. (1995). The phosphor used as a wavelength converter was Ce-doped YAG with chemical formula (Y1-aGda)3 (Al1-bGab)5 O12 : Ce. The exact chemical composition of the host (YAG) and the dopants (e. g. Ce) is usually proprietary and not publicly available.

Fig. 21.7. (a) Structure of white LED lamp consisting of a GaInN blue LED chip and a phosphor, (b) Wavelength - converting phosphorescence and blue luminescence (after Nakamura and Fasol, 1997).

The cross-sectional structure of a white LED lamp is shown in Fig. 21.7 (a). The figure shows the LED die emitting in the blue and the YAG phosphor surrounding the die. The YAG phosphor can be made as a powder and suspended in epoxy resin. During the manufacturing process, a droplet of the YAG phosphor suspended in the epoxy is deposited on the LED die, so that the resin fills the cup-shaped depression in which the LED die is located, as shown in Fig. 21.7 (b). As indicated in the figure, a fraction of the blue light is absorbed by the phosphor and re-emitted as longer-wavelength light.

The emission spectrum of the phosphor-based white lamp thus consists of the blue emission
band originating from the semiconductor LED and longer-wavelength phosphorescence, as shown in Fig. 21.8. The thickness of the phosphor-containing epoxy and the concentration of the phosphor suspended in the epoxy determine the relative strengths of the two emission bands. The two bands can thus be adjusted to optimize the luminous efficiency and the color-rendering characteristics of the LED.

The location of the white lamp in the chromaticity diagram is shown in Fig. 21.9. The location suggests that the emission color is white with a bluish tint. A bluish white is indeed confirmed when looking at the lamp.

Fig. 21.9. Chromaticity coordinates of a commercial phophor-based white LED manufactured in 2001 by Nichia Corporation (Anan, Tokushi­ma, Japan). Also shown is the planckian locus and associated color temperatures.

First-generation white LEDs from Nichia Corporation were improved in terms of their color

rendering capability by adding an additional phosphor that, when excited by 460 nm blue light, has a peak emission wavelength of 655 nm and a full-width at half-maximum of 110 nm (Narukawa, 2004). As a result, the emission can be enhanced in the red range, as shown in Fig. 21.10. Furthermore, by using an optimized phosphor mix, the pronounced notch in the first - generation white LED is reduced. The second-generation white LED lamps from Nichia Corporation (Narukawa, 2004) render red colors better than the first-generation and have a lower color temperature that can range between 2800 K (warm white) and 4700 K depending on the phosphor mix.

Note that the downside of adding red phosphors is a reduced luminous efficiency: The large Stokes shift of red phosphors reduces the efficiency (excitation 460 nm; emission 655 nm) Furthermore, it is well known that red phosphors excitable at 460 nm are comparatively inefficient. Thus, although color rendering capabilities are improved, they are improved at the expense of luminous efficiency.

A concern with white sources is spatial color uniformity. The chromaticity of the white source should not depend on the emission direction. Color uniformity can be attained by a phosphor distribution that provides an equal optical path length in the phosphor material independent of the emission direction (Reeh et al., 2003).

Spatial uniformity can also be attained by adding mineral diffusers to the encapsulant (Reeh et al., 2003). Such mineral diffusers are optically transparent substances, such as TiO2, CaF2, SiO2, CaCO3, and BaSO4, with a refractive index different from the encapsulant. The diffuser will cause light to reflect, refract, and scatter, thereby randomizing the propagation direction and uniformizing the far-field distribution in terms of chromaticity (i. e. spectral composition).

21.3 Spatial phosphor distributions The spatial phosphor distribution in white LED lamps strongly influences the color uniformity and efficiency of the lamp. One can distinguish between proximate and remote phosphor distributions (Goetz, 2003; Holcomb et al., 2003; Kim et al., 2005; Luo et al., 2005; Narendran et al., 2005). In proximate phosphor distributions, the phosphor is located in close proximity to the semiconductor chip. Proximate phosphor distributions are shown in Fig. 21.11(a) and (b). In remote phosphor distributions, the phosphor is spatially removed from the semiconductor chip. A remote phosphor distribution is shown in Fig. 21.11(c).

Photographs of the different phosphor distributions are shown in Fig. 21.12. The proximate phosphor distribution shown in Fig. 21.12(a) was introduced by Nichia Corporation during the 1990s. The phosphor particles are dissolved in the encapsulation material that is dispensed into the reflector cup. Gravity, buoyancy, and friction generally lead to a distribution of phosphor particles that favors larger phosphor particles to move downward, thereby bringing them closer to the chip surface.

Another proximate phosphor distribution, called the conformal phosphor distribution, is shown in Fig. 21.12 (b). Conformal phosphor distributions are accomplished by wafer-level phosphor dispensation thereby lowering the manufacturing cost as compared with a lamp-level phosphor dispensation. Conformal phosphor distributions provide a small emission area and high luminance, which is particularly relevant for imaging-optics applications. Imaging-optics applications (e. g. automotive headlights) frequently require the use of lenses. Optical design considerations show that point-like sources, i. e. sources with a small emission area, are desirable for these applications.

A general drawback of proximate phosphor distributions is the absorption of phosphorescence by the semiconductor chip. Phosphorescence emitted toward the semiconductor chip can be absorbed by the chip, e. g. by the metal contacts covering the chip. The reflectivity of the semiconductor chip and metal contacts is generally not very high.

This drawback can be avoided by remote phosphor distributions in which the phosphor is spatially distanced from the semiconductor chip (Kim et al., 2005; Luo et al., 2005; Narendran et al., 2005). In such remote phosphor structures, it is less likely that phosphorescence impinges on the low-reflectance semiconductor chip due to the spatial separation between the primary emitter (semiconductor chip) and the secondary emitter (phosphor). The probability that phosphorescence impinges on the semiconductor chip is greatly diminished if the distance between chip and phosphor is equal to or greater than the chip’s lateral dimension, i. e. d > a, as shown in Fig. 21.11(c). As a result, higher phosphorescence efficiency is enabled. Ray-tracing simulations and experiments using a remote blue phosphor pumped by a GaInN emitter have indeed demonstrated phosphorescence efficiency improvements of 75% and 27%, respectively (Kim et al., 2005; Luo et al., 2005). Narendran et al. (2005) reported an average of 61% improvement in light output by using the “scattered photon extraction” (SPE) method. At low currents, the SPE packages exceeded 80 lm/W, compared to 54 lm/W for a typical conventional package.