565

Incoherent Li
10. Incoherent Light Sources

Part C 10
10.1 Incandescent Lamps ............................. 565
Since the invention and industrialization of
10.1.1 Normal Incandescent Lamps .......... 565
incandescent lamps at the end of the 19th
10.1.2 Tungsten Halogen Lamps .............. 566
century electrical lighting has become a com-
modity in our daily life. Today, incoherent 10.2 Gas Discharge Lamps ............................ 566
light sources are used for numerous appli- 10.2.1 General Aspects ........................... 566
cation areas. Major improvements have been 10.2.2 Overview of Discharge Lamps ......... 567
achieved over the past decades with respect 10.2.3 Low-Pressure Discharge Lamps ...... 567
to lamp efficiency Fig. 10.1, lifetime and color 10.2.4 High-Pressure Discharge Lamps ..... 570
properties. 10.3 Solid-State Light Sources....................... 574
In the following chapters an overview of var- 10.3.1 Principle of Electroluminescence .... 574
ious lamp types and their properties is given. 10.3.2 Direct Versus Indirect
They are subdivided by light generation mech- Electroluminescence ..................... 575
anism: thermal emission of radiation close 10.3.3 Inorganic Light-Emitting Diodes
to thermal equilibrium (incandescent lamps), (LEDs) ......................................... 575
atomic and molecular emission in gas dis- 10.3.4 Organic LEDs ................................ 578
charge lamps, and emission from solid-state 10.4 General Light-Source Survey .................. 581
light sources (LEDs).
References .................................................. 581

10.1 Incandescent Lamps

The incandescent lamp is the oldest electrical light improve lifetime is the reduction of the tungsten evapo-
source still in widespread use. It can be found in almost ration rate by the addition of rare gases (Kr, Xe). Alterna-
any application, especially where comparatively small tively, halogens are used in a so-called regenerative cycle
lumen packages are required and where simplicity and
compactness are favored.
Efficacy (lm / W)
10.1.1 Normal Incandescent Lamps 200 LP sodium
175 HP sodium
Incandescent lamps produce light by the electrical heat- HP metal halide
ing of a metal wire to such a high temperature that 150 LP fluorescent
radiation in the visible part of the spectrum is emit- LP compact fluorescent
125
HP mercury
ted [10.1]. The metal wire is mounted in a glass bulb White LED
100
filled with an inert gas (Fig. 10.2). Incandescent/ Halogen
According to Planck’s law the filament must be 75
heated up to at least 2400 K for a white emission color.
50
In view of the efficiency for converting electrical en-
ergy into visible light, even higher filament temperatures 25
would be favorable. Unfortunately, increasing the tem- 0
perature reduces lamp life due to enhanced evaporation 1870 1890 1910 1930 1950 1970 1990 Year
1880 1900 1920 1940 1960 1980 2000
rates of the metal. Since tungsten has a high melting
point and a low vapor pressure it permits high operating Fig. 10.1 Temporal development of the luminous efficacy
temperature and consequently higher efficiencies can be of electrical light sources (LP = low pressure, HP = high
attained than by any other metal. A possible measure to pressure)
566 Part C Coherent and Incoherent Light Sources

10.1.2 Tungsten Halogen Lamps

Part C 10.2

Bulb In a normal incandescent lamp, tungsten evaporates off

the filament and condenses on the bulb wall, result-
Fill gas ing in so-called blackening. Halogen lamps comprise
a halogen, i. e. iodine, bromine, chlorine, added to the
Filament
normal gas filling. These form volatile tungsten com-
Support wires
pounds at the glass wall, which are transported back
to the hot filament. Here, the tungsten halides are de-
Lead-in wires composed enabling a so-called chemical transport cycle
(Fig. 10.3).
By reduction of the net tungsten evaporation rate,
Stem the filament of halogen lamps can be operated at higher
Fuse temperature compared to standard incandescent lamps.
Thus, luminous efficiencies can be increased for reduced
lamp size. This allows their application in compact re-
flectors. Halogen incandescent lamps are available up to
2000 W with luminous efficacies up to 25 lm/W. This
Lamp cap
value has recently been improved to 35 lm/W by coat-
ing the glass bulb with infrared reflective multilayers,
and lifetimes up to 2000 h were obtained.

Fig. 10.2 Schematic drawing of an incandescent lamp

effectively to transport tungsten back to the filament. An

increase in luminous efficiency is achieved by coiling the
tungsten wire. A coiled filament allows higher operation W + 2Br
temperatures for a specified lifetime (typically 1000 h). Filament WBr2
W
Most common operation parameters yield luminous ef- Bulb wall
ficiencies between 8 lm/W and 17 lm/W at filament
temperatures of 2400 K and 3100 K, respectively. These 2Br
values correspond to an energy efficiency of a few per-
1500 K
cent. Incandescent lamp wattages range up to 2000 W.

Fig. 10.3 Principle of the chemical transport cycle in tung-

W + 2Br WBr2
sten halogen lamps

10.2 Gas Discharge Lamps

10.2.1 General Aspects negative electrons. In a neutral nonconductive gas the
number of charge carrying particles is extremely small.
A gaseous discharge is obtained by driving an These particles can be released from the fill gas or the
electric current through a gas, typically present cathode surface by energetic collisions. Many physical
between two electrodes. Alternatively, electrodeless factors influence the properties of a gas discharge, the
microwave-excited discharges and pulsed dielectric-bar- most important ones being the type and pressure of
rier discharges (e.g. used for plasma displays) are known the gas, the electrode material, the operating tempera-
as incoherent light sources. ture of the electrodes, the shape and surface structure
The actual carriers of the electric current in the of the electrodes, the distance between the electrodes,
gas are electrically charged particles, positive ions and the geometry of the discharge vessel, and the cur-
 Incoherent Light Sources 10.2 Gas Discharge Lamps 567

rent density. For the purpose of light generation, two

Part C 10.2
Temperature (K)
main types are distinguished: low-pressure and high-
pressure discharge lamps. For lighting applications, 105
both are operated in the arc discharge mode,
which is

characterized by high current densities > 1 A/cm2 . te
104
To limit the discharge currents, electronic ballasts are
used [10.2].
In a low-pressure discharge lamp (gas pressure typ- 103
tg
ically less than 100 Pa) the electrons have a mean
free path length larger or of the order of the ves-
102
sel diameter (e.g. a few cm). Due to low collision
rates with the neutral gas atoms they gain high en-
ergies (> 1 eV) from the applied electrical field and 10
effectively excite the cold atoms by inelastic collisions. 10–1 1 10 102 103 104 105 106
Pressure (Pa)
The electrons and atoms are not in thermal equilibrium
(Fig. 10.4). Fig. 10.4 Relationship between electron temperature, gas
In high-pressure discharge lamps the operating pres- temperature and gas pressure
sure is typically in the range between 10 kPa and 10 MPa.
Here, collisions between electrons and atoms or ions are rendering properties of high-pressure discharge lamps
much more frequent, resulting in a thermal equilibrium are fair to excellent, depending on the type of fill-
that is characterized by equal particle temperatures. ing [10.3].
In low-pressure discharges atomic line radiation
is emitted preferably from resonance transitions of 10.2.2 Overview of Discharge Lamps
the element with the lowest excitation potential (e.g.
Hg: 185 nm and 254 nm, Na: 589 nm). In high-pressure In Fig. 10.5 the most relevant types of discharge lamps
lamps various contributions to the spectrum are ob- are displayed. The light sources are distinguished with
tained: broadened atomic lines (resonance-, van der respect to their emission spectra and application fields.
Waals- and Stark-broadening), molecular radiation In the case of low-pressure mercury and xenon excimer
bands and quasi-continuous emission due to free– lamps luminescent materials are applied for conver-
free (Bremsstrahlung) and free–bound (recombination sion of ultraviolet (UV) radiation into visible light.
of electrons with ions and atoms) transitions. As The temporal improvement in efficiencies is depicted
a result of these quasi-continuous spectra, the color in Fig. 10.1.

Mercury Sodium Rare gas Sulphur

Low pressure High pressure Low pressure Low pressure High pressure
p < 1 mbar p > 1bar Ne microwave
580 – 720 nm S2
Hg/Ar Hg / Ar Na/ Ar / Ne 74 nm
Hg/Ne • p ≈ 20 bar (Phosphors)
• p ≈ 200 bar (short arc) Na 589 nm
185 + 254 nm
p ≈ 0.5 bar
(Compact) DBD, PDP
Metal halide lamps High pressure
Fluorescent Xe/ Ne
• 3line radiators
lamps 147 + 172 nm
NaX / TIX / InX, X = I, Br
Phosphors
• Multi-line / Molecular Na/ Hg / Xe
Phosphors NaX / TIX / REX3
RE = Dy, Ho, Tm, Sc High pressure
SnX2 Xe

Fig. 10.5 Overview of gas discharge lamps

568 Part C Coherent and Incoherent Light Sources

10.2.3 Low-Pressure Discharge Lamps

Part C 10.2

Energy level above stable level

Ionization
The most widely applied radiators in low-pressure dis- level
charge lamps are Hg and Na. These elements are the 10
best choice with respect to efficiency for the conversion 9
of electrical input power into radiation.
577 nm
8
Low-Pressure Mercury Lamps 1208 nm 313 nm Various
1014 nm
The working principle of a low-pressure Hg (also known 7
546 nm
excited
408 nm levels
as a fluorescent lamp) is given in Fig. 10.6. Fluorescent 436 nm 366 nm
lamps are generally designed in the form of a linear or 6
Metastable
bent tubular bulb with an electrode sealed in at each end level
5
(electrodeless versions are available where the electrical
energy is coupled inductively into the discharge ves- 4
sel via metal coils). The discharge vessel is filled with 185 nm
an inert gas (typically Ar) and a few mg Hg. Since the 3
major part of the emission of Hg atoms (97%) at low 254 nm
pressures (e.g. 5 Pa) is in the ultraviolet, the inner sur- 2
face of the bulb is coated with a fluorescent powder or
1
phosphor, which converts the UV radiation into visible
light. The composition of the phosphor determines the 0 Stable level
spectral power distribution and the color of the emitted
light. Fig. 10.7 Simplified energy-level scheme and radiative
The generation of UV photons is due to transitions of transitions of the Hg atom
Hg atoms between the excited state levels 1P 1 , 3P 1 and
the ground state level 1S 0 . About 64% of the electrical of geometries, e.g. cylindrical or circular or U-shaped
input power is converted into photons at a wavelength tubes. The latter are also known as energy-saving or
of 185 nm and 254 nm. According to the transitions in- compact fluorescent lamps. Depending on the con-
dicated in Fig. 10.7 only 3% are emitted in the visible struction of fluorescent lamps, the lifetime ranges
part of the spectrum. As a result of the Stokes shift, the from 5000 to 25 000 h.
overall efficiency is only 28%. This value corresponds
to a luminous efficiency of 100 lm/W. Fluorescent Coatings
Typical lamp parameters for fluorescent lamps are: The most important component of a fluorescent lamp is
electrical input power up to 140 W, luminous efficiency the fluorescent powder. It is coated onto the inner side
up to 100 lm/W, color temperature between 2700 K of the glass tube. The powder consists of one or several
and 8000 K. Lamps are available in a large variety luminescent materials (phosphors), which are in general


inorganic compounds doped by transition metals e.g.

Electrode Ultraviolet Visible radiation

radiation

Fluorescent powder

Electrons Mercury atom

Fig. 10.6 Sketch of a low-pressure mercury discharge (fluorescent) lamp

 Incoherent Light Sources 10.2 Gas Discharge Lamps 569



Mn2+ , Mn4+ or rare-earth ions e.g. Tb3+ , Eu3+ . The

Part C 10.2
Heat Heat
composition of the fluorescent powder is optimized to
Excitation Emission
convert the absorbing atomic resonance lines of mer-
cury low-pressure discharges (185 nm and 254 nm) into
the desired lamp spectrum. In some cases the fluores-

cent composition
 also includes so-called sensitizers e.g. S A
Ce3+ if the first dopant, i. e. the activator, does not suffi- Energy
ciently absorb the mercury lines. Figure 10.8 displays the transfer
principle of photon down-conversion in a luminescent
material.
The first step of light conversion is the absorption of Emission
incident photons either by the activator or the sensitizer.
In the case of absorption by the sensitizer the energy is Fig. 10.8 Simplified mechanism of light generation in a flu-
subsequently transferred to the activator. The excited ac- orescent powder. The abbreviations “S” and “A” indicate
tivator ion, for instance Eu3+ , decays to the ground state sensitizer and activator, respectively
by emitting a photon with a wavelength according to the
energy gap between the excited and ground state. The a trichromatic phosphor blend results in higher lamp
energy difference between the emitted and the absorbed efficacy and improved color rendition. A typically ap-
photon (Stokes shift), which is about 50%, is dissipated plied phosphor blend comprises BaMgAl10 O17 : Eu,
as heat in the fluorescent powder. LaPO4 : CeTb, and Y2 O3 : Eu. These so-called color 80
Nowadays, fluorescent compositions with quantum lamps (indicated with an a in Table 10.1) exhibit a spec-
efficiency close to unity are commercially available tral power distribution as depicted in Fig. 10.9.
for many lamp application areas. Table 10.1 gives Lamps with an even higher color rendition
an overview of most commonly applied luminescent are obtained by a tetra- or pentachromatic phos-
materials. For illumination purposes, halophosphate phor blend. Color 90 lamps additionally contain
phosphors, invented in the 1940ies, are still widespread phosphors with an emission-band position where
since they yield a white emission spectrum with the trichromatic phosphor blend does not radiate,
a reasonable efficiency and color rendering. However, viz. in the yellow (Y3 Al5 O12 : Ce) and deep red
(GdMgB5 O10 : Ce, Tb, Mn) spectral range.

Table 10.1 Overview of the most commonly used luminescent materials in fluorescent lamps
Composition of Emitted peak Color point Application area
luminescent materials wavelength (nm) x, y a (after CIE 1931)
LaB3 O6 : Bi, Gd 311 − Medical lamps
LaPO4 : Ce 320 − Tanning lamps
BaSi2 O5 : Pb 350 − Tanning lamps
SrB4 O7 : Eu 368 − Black light lamps
Sr2 P2 O7 : Eu 420 0.167, 0.014 Reprography lamps
BaMgAl10 O17 : Eu 453 0.150, 0.070 Color 80 lamps b
Zn2 SiO4 : Mn 530 0.256, 0.700 Decoration lamps
LaPO4 : Ce, Tb 543 0.343, 0.585 Color 80 lamps a
CeMgAl11 O19 : Tb 543 0.350, 0.582 Color 80 lamps a
GdMgB5 O10 : Ce, Tb 543 0.346, 0.531 Color 80 lamps a
Y3 Al5 O12 : Ce 560 0.453, 0.523 Color 90 lamps a
Ca5 (PO4 )3 (F, Cl) : Sb, Mn 575 0.356, 0.377 Halophosphate lamps
Y2 O3 : Eu 611 0.643, 0.344 Color 80 lamps a
GdMgB5 O10 : Ce, Tb, Mn 630 0.602, 0.382 Color 90 lamps a
Mg4 GeO5,5 F : Mn 660 0.700, 0.287 Decoration lamps
(a at 254 nm excitation, b see text)
 570 Part C Coherent and Incoherent Light Sources
Part C 10.2

Emission intensity (arb. units) Bend isolation

Vλ
1,0

0,8

0,6 Discharge tube

0,4 Dimple filled

with sodium
0,2 Outer bulb

0,0
300 400 500 600 700 800
Wavelength (nm)

Fig. 10.9 Spectrum of a color 80 fluorescent lamp

with a blend of BaMgAl1017 : Eu, LaPO4 : Ce, Tb, and Electrode
Y2 O3 : Eu and the eye sensitivity curve Vλ
Getter
Due to the tunability of the spectra, fluorescent
lamps have entered many other application areas be- Lamp cap
yond illumination, such as medical and cosmetic skin
treatment, decoration, reprography, horticultural light- Fig. 10.10 Sketch of a low-pressure sodium lamp
ing. The latter, for instance, comprises a dichromatic
phosphor blend consisting of BaMgAl10 O17 : Eu and tion is limited to street and outdoor lighting where
Y2 O3 : Eu. The spectrum of this mixture is adjusted ac- high efficiencies are of importance. In addition, an ad-
cording to the action spectrum for photosynthesis of vantage is their long lifetime, which are of the order
green plants. of 20 000 h.
When increasing the Na vapor pressure, as realized
Other Low-Pressure Discharge Lamps in high-pressure Na lamps, the broadened spectral lines
In low-pressure sodium lamps the discharge has the improve color rendition significantly (see below).
same origin as in mercury lamps. Sodium emits almost Besides low-pressure Hg and Na lamps other types
monochromatic yellow light, viz. a doublet consisting of radiators are used for special applications: low-
of 589.0 nm and 589.6 nm lines (Na D-lines). How- pressure neon discharges and xenon excimer discharges.
ever, the melting point of Na is higher than Hg. Thus The former are applied in automotive brake lights and
the optimal operating temperature is at about 530 K. advertisement lighting. Excimer radiation from xenon
At these temperatures the highly reactive Na metal re- is used in plasma displays and flat backlights, photo-
quires a chemically stable wall material such as quartz copier lamps and UV purification. The major advantage
(SiO2 ) or alumina (Al2 O3 ). To reduce thermal losses the of these discharges is their fast switching behavior: in
discharge vessel is mounted inside an outer evacuated contrast to Hg and Na, Xe does not need to be evap-
bulb Fig. 10.10. orated. Fast switching is achieved with pulsed barrier
Low-pressure sodium lamps exhibit the highest discharge resulting in the formation of Xe excimers.
ever achieved luminous efficiency due to the favorable Atomic and molecular radiation at 147 nm and 172 nm
position of the Na lines with respect to the spectral is observed, respectively.
sensitivity of the human eye, which peaks at a wave-
length of 555 nm. At a discharge power of 200 W 10.2.4 High-Pressure Discharge Lamps
luminous efficiencies of 200 lm/W (172 lm/W sys-
tem efficiency) are obtained. A disadvantage is the As the pressure is raised in a low-pressure discharge,
low rendition of colors due to missing emission at the rate of elastic collisions increases. Although each
other wavelengths, e.g. in the blue, green and red collision transfers little energy to heavy particles the
parts of the spectrum. For this reason their applica- plasma is heated effectively due to the high particle
 Incoherent Light Sources 10.2 Gas Discharge Lamps 571

densities. To sustain central plasma temperatures in the towards the spectral lines of longer wavelengths. When

Part C 10.2
range 4000–10 000 K a temperature gradient is estab- fully run up the lamp establishes a typical mercury pres-
lished. The heat flux along this gradient to the discharge sure 200–1000 kPa and a bluish-white light is produced
wall is a loss, which limits the radiation efficiency of
such arcs to around 60%. Fortunately, selective emis-
sion of the plasma allows for efficient radiation in the
visible part of the spectrum. High-pressure discharge
lamps containing metals like Hg or Na are known.
Besides, other efficient radiators are added to the
lamp filling in so-called metal halide lamps. Each combi-
Support for
nation of metals used generates radiation with a specific discharge tube
color and radiant efficiency. In theory at least, no fewer Discharge tube
than 50 different metals can be used to dose metal halide
lamps. As a result, lamp manufacturers have introduced
Outer bulb
various combinations on the market with a wide range
of applications. Main electrode
Auxiliary electrode
A technical breakthrough in high-pressure sodium
Resistor
and metal halide lamps has been achieved by the in-
troduction of polycrystalline alumina (sintered Al2 O3
ceramics) as a wall material. Compared to quartz, a ma-
jor advantage is its chemical resistance against the hot
sodium vapor that is present in high-pressure sodium
lamps. In addition ceramic materials offer the possibility
to achieve the high wall temperatures needed for effec-
tive evaporation of the salt fillings. In quartz envelopes,
the maximum temperature is limited by recrystallization
of the wall at about 1370 K. This value can be greatly Fig. 10.11 Sketch of a high-pressure mercury lamp. The
exceeded by using Al2 O3 instead of SiO2 . On the other inner surface of the outer bulb may be coated with phosphor,
hand, vacuum tight sealing of electrodes and the design which additionally converts UV light into visible light
of the lamps’ end construction is more complicated with
ceramic materials. This fact is due to their inconvenient
Relative spectral power
thermal expansion behavior and brittleness. In the light-
ing industry, this is still a current field of research and High-pressure
development of modern light sources. mercury
with fluorescent
coating
High-Pressure Mercury Lamps T = 3400 K
The construction of a high-pressure mercury lamp is Ra = 52
shown in Fig. 10.11. The discharge vessel is filled with
mercury, and argon is used as a starting gas [10.1]. Ther-
mal isolation of the discharge vessel is achieved by use
of an evacuated outer bulb. The lamp is switched on us-
ing a high-voltage ignition pulse, typically of the order
of several kV. After ignition, the voltage across the dis-
charge drops down to about 20 V, which is mainly the
sum of the cathode and anode fall voltage. At this stage,
the lamp is operating as a low-pressure discharge emit-
ting UV radiation mainly at 254 nm. Thermal losses in
the arc result in an increase of the wall temperature and
thus increasing evaporation of the liquid mercury with 400 500 600 700
time. With a further increase in mercury vapor pres- Wavelength (nm)
sure the radiated energy is concentrated progressively Fig. 10.12 Spectrum of a high-pressure mercury lamp
 572 Part C Coherent and Incoherent Light Sources

a significant part of the visible radiation is generated

Part C 10.2

not only by pressure-broadened atomic mercury lines

but also by Hg2 molecules. UHP lamps are available on
the market in the power range 100–250 W. A key issue
for UHP lamps is the maintenance of stable properties
of the discharge and electrodes for more than 5000 h.
This is achieved with a chemical cycle in which evapo-
rated tungsten is transported back onto the electrodes via
compounds including oxygen and bromine. In addition,
UHP lamps have to be operated with special electronic
gear to control electrode erosion and arc jumping.

Metal Halide Discharge Lamps

Fig. 10.13 UHP lamp system (lamp, reflector and electronic
Metal halide lamps are similar in construction to high-
driver) for application in projection systems and beamers
pressure mercury lamp. The major difference between
the two types is that the discharge tube of the former
(Fig. 10.12). At these pressures a part of the emission is contains metal halides in addition to the mercury, and
still in the UV. By coating the inner surface of the outer argon as the starting gas. The metal halides determine
bulb with a fluorescent powder, e.g. YVO4 : Eu, that the spectral properties of the lamps. As mentioned above
converts this UV radiation into visible light, the lumi- mercury serves as a buffer gas that determines the elec-
nous efficiency can be further increased. In addition, the trical properties, e.g. lamp voltage. The metals are dosed
spectral composition of this light considerably improves in the form of halides because of their volatile character.
the color-rendering properties. High particle densities are needed for efficient radiation.
Besides its radiative properties Hg also serves as Furthermore, the metal halides are less aggressive with
a buffer gas in today’s high-pressure discharge lamps; respect to wall corrosion compared to pure metals.
the electrons are elastically scattered by mercury atoms, There are three main groups of metal halide lamps:
which results in a low arc electrical conductivity Ac- multi-line radiators, molecular radiators and three-band
cording to Ohm’s law, electrical field strengths of the color radiators. In the first type rare-earths and associ-
order of several 10 V/mm are present in the discharge ated elements such as Dy, Ho, Tm or Sc are added to the
column, which enables an input electrical power at mod- lamp filling. These species have a large number of transi-
erate discharge currents. The lamp current must not be tions at low mean excitation energy. Consequently, such
too large with respect to electrode stability: electrode plasmas efficiently emit multi-line radiation in the vis-
temperatures are in the range 2400–3500 K when the ible spectrum. Furthermore, NaI and TlI are often added
electrons are generated thermionically. It should be em- to such lamp fillings to improve luminous efficiency and
phasized that both properties – the volatile character color properties. The second type is based on molecular
enabling high atom densities and the large cross section radiators such as SnI2 or SnCl2 , which produce a quasi-
for elastic electron scattering – are unique to mercury. continuous spectrum. Finally, the third class is based on
High-pressure mercury lamps are used in various NaI, TlI and InI3 . Such spectra consist of three color
indoor and outdoor applications. bands in the yellow, green and blue, respectively.
The power range is 50–1000 W with luminous ef- Typical spectra of these three metal halide lamp
ficiencies around 50–60 lm/W. Color rendition is only types are shown in Figs. 10.14a–c. Metal halide lamps
moderate to poor. Color temperatures are 3400–6000 K. are used for a variety of indoor and outdoor lighting ap-
A recent development of a high-pressure mercury plications. The electrical input power ranges from 35 W
discharge is the so-called UHP lamp (ultra high perfor- to more than 1000 W. In general, metal halide lamps
mance lamp), which was invented by Philips in the 1990s have high luminous efficiencies of up to 100 lm/W
(Fig. 10.13 [10.4]). This lamp has a very high operating at good to excellent color rendering. Lifetime is up
mercury pressure (about 2 × 107 Pa) and a very short to 20 000 h, depending on the type of filling used.
arc length of about 1 mm. This point-like light source A speciality of metal halide lamps is its halide cycle
with its broadened white spectrum is used in projec- (Fig. 10.15), which is quite similar to the aforementioned
tion systems and beamers. In contrast to conventional tungsten–halogen cycle. When a metal halide lamp first
high-pressure mercury lamps at these high pressures starts, the spectrum is initially that of mercury vapor,
 Incoherent Light Sources 10.2 Gas Discharge Lamps 573

Part C 10.2
a) Relative spectral power c) Relative spectral power
Rare-earth 3line radiator
metal halide T = 4500 K
Dy, Ho, Tm Ra = 65
(Na, Tl)
T = 4200 K
Na Ra = 85

Tl

Dy
Li

400 500 600 700

Wavelength (nm)
400 500 600 700
Wavelength (nm)
b) Relative spectral power
Fig. 10.14 (a) Spectrum of a NaI/TlI/DyI3 /HoI3 /TmI3
Tin halide lamp for sports and flood lighting. (b) Spectrum of
Snl2 / SnCl2
T = 5600 K a SnI2 /SnCl2 lamp for studio and theater lighting. (c) Spec-
Ra = 84 trum of a three-band color (NaI, Tl, InI3 ) metal halide lamp.
The lithium line is due to the presence of impurities in the
wall material (quartz)

contribution of atomic radiation takes place. The metals

then move nearer the cooler discharge wall where they
recombine with the halogen atoms to form once again the
halide compound. The whole cycle then repeats itself.

400 500 600 700

Wavelength (nm) 2500 K Tl+
TlI
2800 K Na+ 5000 K 1000 K
since the halides remain unevaporated at the relatively NaI
cool wall of the discharge tube (typical wall tempera- 2900 K In3+
ture of 1000 K). As the wall temperature increases, the InI3
halides melt and partly vaporize. The vapor is carried
into the hot region of the arc by diffusion and convection,
where the halide compounds are dissociated into the Discharge
halogen and metal atoms. As indicated in Fig. 10.14, dif- tube wall
ferent halides dissociate at different temperatures. The
metal atoms are excited at the hot plasma center, which Fig. 10.15 Schematic diagram of the halide cycle of a three-band
typically has a temperature of 5000 K. Here the main color metal halide lamp (NaI, TlI, InI3 )
 574 Part C Coherent and Incoherent Light Sources

As mentioned above a recent improvement of metal In high-pressure sodium lamps, the broadening of
Part C 10.3

halide lamps was achieved by the introduction of a ce- the D-lines is used to improve color properties compared
ramic wall material (polycrystalline alumina) instead to a low-pressure Na discharge. Luminous efficiencies
of quartz. Such lamps show improved color stability of up to 130 lm/W are obtained. A technical break-
and long lamp life. However, these lamps are not suit- through has been achieved with the introduction of
able for optical applications, e.g. in projection systems. a ceramic wall material (polycrystalline Al2 O3 ). This
The reason is the translucent behavior of polycrys- withstands the hot and highly aggressive Na vapor,
talline Al2 O3 (PCA) due to multiple scattering of light resulting in a much longer lamp life compared to
at grains and pores. Nevertheless, the total transmis- quartz (SiO2 ).
sion for visible light is nearly 95%. The application Another class of high-pressure discharge lamps is
of PCA to high-power metal halide lamps is also lim- represented by the sulphur lamp, invented by Turner
ited due to cracking of the material induced by thermal et al. at Fusion Systems in 1994. The advantage of
shock and thermal expansion. In the lighting indus- such a lamp is their long life due to electrodeless op-
try, the applicability of ceramic wall materials is still eration. It is operated with a microwave at a wavelength
a current field of research and development of mod- of 2.45 GHz, the electrical input power being of the
ern light sources. Furthermore, upcoming legislation order of 3 kW. A spherical bulb having a diameter of
has initiated several activities on the replacement of about 30 mm is filled with sulphur powder. Under con-
mercury in this type of discharge by nontoxic mater- tinuous operation, sun-like white light is emitted by
ials [10.3]. S2 molecules, having a partial pressure of about 600 kPa.
The high burner efficiency of about 170 lm/W is pos-
Other High-Pressure Discharge Lamps sible since at these high pressures UV radiation from
Besides high-pressure mercury and metal halide lamps, the sulphur molecules is reabsorbed by the plasma.
other high-pressure light sources are also known. The Thus, emission mainly occurs in the visible part of the
so-called blended-light lamp is derived from the con- spectrum. A technical drawback of microwave lamps
ventional high-pressure mercury lamp. The principal is the low energy efficiency of the generator, typically
difference between the two is that, whereas the latter is about 65%. Therefore, the overall efficiency of the sys-
dependent on an external ballast to stabilize the lamp tem is reduced to about 100 lm/W, being of the order of
current, the blended-light lamp has a built-in ballast conventional discharge lamps.
in the form of a tungsten filament connected in series Nevertheless, electrodeless discharge lamps are still
with the discharge tube. Consequently, the light from a hot topic in research and development. A general ad-
the discharge and the filament combine (or blend). The vantage of these systems is their potentially long life
improved spectral quality is traded off against reduced and large luminous flux. The latter, for instance, might
total efficiency of the system. be distributed by light pipes for use in industrial and
general lighting applications.

10.3 Solid-State Light Sources

10.3.1 Principle of Electroluminescence by the eye response curve, the use of lumens per Watt
(lm/W) gives the optically relevant light output in rela-
In solid-state lamps the electric current is converted into tion to the invested electrical power. The most important
light by recombination of excited states of atoms into loss factors are non-radiative recombination through
the ground state via emission of a photon, i. e. radia- phonons or second-order processes converting the elec-
tive recombination. The nature of the excited state may trical power primarily into heat. An important additional
differ for different types of electroluminescence (EL) loss factor is the outcoupling of the light. Since most EL
principles. Typically the applied voltage has to be at materials have a higher refractive index than air, part of
least as large as the emission of the material to fulfill the the light is trapped inside the device by total internal re-
energy-conservation law. flection. Due to absorption losses, this light is partly lost,
Besides radiative recombination, many other decay which reduces the overall efficiency. In addition, more
mechanisms are possible in electroluminescence that specific loss channels will be discussed for the different
reduce the overall efficiency, which is typically meas- mechanisms.
ured in wopt,out /wel., in , given as a percentage. Weighted
 Incoherent Light Sources 10.3 Solid-State Light Sources 575

General illumination was not feasible using EL de- type of EL is found in classical thin-film (TF) or pow-

Part C 10.3
vices for a long time, since the total emitted power or der EL devices, also called TFEL or alternating-current
lumen package was fairly low and/or the efficiencies thin-film EL (ACTFEL) devices. Here carriers are ac-
were low. This has changed in recent years with the ad- celerated by means of an external electrical field. These
vent of InGaN technology, which turned out to be very accelerated electrons collide with ions in the matrix.
stable against high power densities and the resulting heat By impact ionization, these ions or luminescent centers
development. Thus electroluminescence has started to are excited and the radiative recombination of the gen-
become the third class of illumination devices, creating erated electron–hole pairs releases a photon. Therefore
a new market for lamps, gears and luminaries. the theoretical efficiency of this process is limited to
about 40%. These devices are typically operated using
10.3.2 Direct Versus Indirect high-voltage high-frequency AC bias and the brightness
Electroluminescence as well as the spectrum is a function of voltage and
frequency. Typically, a few hundred cd/m2 maximum
Electroluminescence (EL) in general appears in two dis- brightness are achieved in state-of-the-art indirect EL
tinctly different mechanisms: direct EL and indirect EL. devices. Because efficiency and brightness are limited,
Direct EL describes the direct recombination of elec- we will not consider indirect EL in much detail, since
trons and holes in a suitable semiconductor structure. the respective devices have achieved little importance
This material can be inorganic, like in most conventional as light sources. They are mainly used in displays, in-
light-emitting diodes (LEDs), which were initially built dication or signaling and decorative lighting. It is not
from GaP and GaAs or more recently and very suc- expected that dramatic improvements will occur in the
cessful from InGaN, or organic such as, e.g., Alq3 (q = near future. The interested reader is referred to detailed
8-hydroxychinolinate). In this version, charge carriers descriptions in [10.5].
are injected from contacts via direct transport with ulti-
mately very little kinetic energy. The carriers meet in the 10.3.3 Inorganic Light-Emitting Diodes
recombination zone, wherein excitons are formed, which (LEDs)
successively decay radiatively or non-radiatively; the
first process releasing a photon, the latter creating heat. The first practical inorganic semiconductor LEDs were
Ideally, the injection barrier from the contacts to the con- conceived at Texas Instruments in 1961 by Biard and
duction levels of the materials is low and thus the energy Pitman. These devices emitted in the near infrared be-
necessary to generate recombination and thus light is cause the material used was InGaP with a band-gap
simply the energy of the emitted light, resulting from the of 1.37 eV. One year later the first LED in the visi-
effective band-gap of the material. Thus the theoretically ble was built by Holonyiak and co-workers at General
achievable efficiency of direct EL generation is unity. Electric. They used a different composition to fabricate
In contrast, the indirect EL process results from a device that emitted red light. From that time on, the use
creating excited states through impact ionization. This of more refined processing and other materials (such as

Fig. 10.16 Schematic of a classical indicator and high-power illumination LED

 576 Part C Coherent and Incoherent Light Sources

be built from defect-rich layers. Additionally, with the

Part C 10.3

advent of bright and stable blue LEDs it became pos-

sible to generate white light from solid-state devices,
either by color conversion of blue LEDs using phos-
phors or by externally mixing the light from blue, green
and red LEDs.
The principal of LEDs is explained in great detail
in the semiconductor device literature [10.6–8]. We will
limit ourselves here to the use of LEDs for light gen-
eration and the respective relevant aspects. Due to the
importance of color conversion in generating white light
from LEDs, we will discuss in more detail the nature of
the phosphors being used.

Pure-Color LEDs
White light is, generally speaking, always a mixture of
Fig. 10.17 AlInGaP LED shaped as a truncated inverted several pure colors. At least a combination of blue and
pyramid (TIP) for efficient light extraction orange light is needed, which yields a white appearance,
however a very low color rendering (CRI, color ren-
e.g. AlGaAs) has led to the development of the common dering index), i. e. many colors are poorly reproduced
LED, mainly used for indicator lights as, for example, under illumination of such a light source. Increasing
shown in Fig. 10.16. the number of colors by combining red, green and blue
For several decades, LEDs were not attractive for (RGB) LEDs already enables much higher CRI values
illumination purposes, mainly for the two following rea- and thus a high-quality white light. In the limit of many
sons. First, the optical output power (i. e. the lumen different visible LEDs, the black-body radiation spec-
package, defining the amount of light emitted from a sin- trum of 5500 K (essentially the spectrum of sunlight)
gle LED) was fairly limited, since the packaging and is generated, which has, by definition, a CRI of 100.
the materials in use did not allow high current dens- However, due to the finite size of the LED dies and
ities in the small chips. Since the efficiency was far their package, it is rather difficult to achieve homoge-
less than 100%, a lot of heat would have been gener- neous mixing of the discrete LEDs, at least in a small
ated, destroying the semiconductor chip. Secondly, and package. The following paragraph will briefly describe
more important, no real blue LED material was avail- the technology of pure-color LEDs, which are already
able. For several years II–VI compound semiconductors widely used in all applications, where colored light is
such as ZnSSe/ZnTe were researched as possible can- necessary, like e.g. signaling. The advantage here is
didates for blue with limited success. However, in 1993 clearly the much higher efficiency, since they replace
Nakamura et al. achieved the breakthrough for solid- filtered white light, which has very low efficiency and
state lighting by proving the successful use of InGaN as lifetime.
a material in LEDs [10.6]. Since the material is chem- Two major materials are used in colored LEDs: AlIn-
ically very stable, it is on one hand difficult to achieve GaP for red, orange and yellow and InGaN for green and
high-quality epilayers for high-efficiency LED opera- blue (see Fig. 10.18).
tion, on the other hand, the high stability allows very The quaternary mixture of AlInGaP is the prime
high current densities without disintegration of the LED candidate for red to yellow emission colors. It is typic-
material. Therefore the lumen package can be increased ally grown on GaAs wafers, since the lattice matching
to values relevant for lighting applications. More impor- of substrate and epilayer guarantees a low number of
tant, however, is the insensitivity of InGaN to defects. defects, which is essential for high efficiency and long
Although the defect density in common InGaN LEDs lifetime. After deposition of the various layers, the wafer
is much higher than what would be acceptable in other is bonded to a GaP substrate and the original GaAs sub-
material combinations, such as e.g. AlGaAs (109 /cm3 strate, which absorbs in the spectral region of the LED
vs. 106 /cm3 ), the low surface recombination velocity emission is removed. This increases the efficiency, since
of InGaN and other properties allows efficient LEDs to half of the light is emitted towards the substrate and
would be lost if the substrate was absorbing. A lot of
 Incoherent Light Sources 10.3 Solid-State Light Sources 577

light is still trapped inside the die due to total internal

Part C 10.3
Emission intensity (arb. units)
reflection. Since the refractive index of the material is
larger than one, all light emitted under an angle larger 0.35 Vλ
than the Brewster angle is reflected back into the crys- 0.30
tal. Due to some residual absorption, e.g. at the metal
0.25
contacts, multiple internal reflections further diminish
the efficiency. By shaping the semiconductor chip in the 0.20
form of a truncated inverted pyramid, the highest effi- 0.15
ciency of a red light source has been obtained by an LED
from Lumileds [10.9] (see Fig. 10.17). 0.10
This LED is estimated to have close to 100% internal 0.05
efficiency and an outcoupling efficiency of about 55%, 0.00
resulting in 102 lm/W wall plug efficiency. This value 400 500 600 700
demonstrates the potential of LED technology as a future Wavelength (nm)
lighting source, since this is the most efficient red-light Fig. 10.18 Spectra of blue and green InGaN and red AlIn-
source available. The lumen package of AlInGaP LEDs GaP LEDs and the eye sensitivity curve Vλ
is, however, limited due to thermal quenching of the
emission. For power densities beyond 0.1 W/cm2 , the dency in new values literally being published week by
efficiency and also the lifetime of AlInGaP LEDs drops week.
significantly, since the chip temperature increases. As a summary, the efficiencies as a function of wave-
For the higher energy emission from green to blue length are plotted in Fig. 10.19 together with the eye’s
and beyond, InGaN has become the material of choice. sensitivity curve for the two material systems.
After several years of intense research, it became pos-
sible to create working LEDs in the InGaN material White LEDs
system. The major difficulty to overcome was the prob- As mentioned above, combining the light of several
lem of p-doping of GaN. This is achieved by doping colored LEDs can create white light. Several different
Mg into the InGaN lattice, which happens at tem- concepts have been used, which will be briefly reviewed
peratures above 1100 ◦ C in the deposition process. By here (see Fig. 10.20). All of them have in common that
varying the amount of In in the composition of the a volume of light mixing has to be provided, since the dif-
LED, the emission can be tuned from UV to green. ferent colors have to be directionally randomized before
However, due to segregation effects and subsequent de- emission to avoid inhomogeneous illumination colors.
velopment of lattice defects, the efficiency drops with
Power conversion efficiency (%)
increasing wavelength, i. e. for high In contents. Com- 80
paring the different materials classes, it turns out, that InGaN AllnGaP
the defect density of InGaN LEDs is much higher
than what would be tolerable in more conventional
III–V compound semiconductors. This is mainly due 60
to the lack of a lattice-matched substrate material. So Na low-
far mostly Al2 O3 (sapphire) with a lattice mismatch pressure
of approx. 15% is used. Nevertheless, the devices, 40 lamps
with unsurpassed stability and high efficiency, prove
that the InGaN system is obviously much less sensi- Fluores-
cent lamps
tive to defects than other material combinations. The 20
reason for that is not yet entirely clear. However, it
has been measured that InGaN has the lowest surface Incandes-
recombination velocity of all III–V composite semi- cent lamps
0
conductors, thus giving excited carriers more chances
of radiative recombination. Current efficiencies range 400 500 600 700
Peak wavelength (nm)
from 20% (20 lm/W) for blue to 12% (60 lm/W) at
green wavelengths, with continuously increasing ten- Fig. 10.19 Efficiency of InGaN and AlInGaP LEDs as
a function of emission wavelength
 578 Part C Coherent and Incoherent Light Sources
Part C 10.3

Red + Green + Blue LEDs UV LED + RGB phosphor Blue LED + Yellow phosphor

UV LED Combined Phosphor Combined

Red peak spectrum spectrum emission spectrum
Blue peak
Phosphor Blue
Green peak emission LED
spectrum

470 525 590 630 (nm) 410 470 525 590 630 (nm) 470 525 590 630 (nm)

Fig. 10.20 Summary of different methods of generating white light with LEDs

This is typically achieved through multiple reflections in the entire spectrum to realize illumination-grade
of the different LEDs in an appropriate housing. A com- warm-white LEDs. The red luminescent materials used
mon method is the use of a poly(methyl methacrylate) in combination with yellow YAG : Ce in so-called
(PMMA) plate, where the light from different LEDs is red-enhanced LEDs are (Ca, Sr)S : Eu [10.10, 11] and
coupled. Through total internal reflection, the colors are (Ca, Sr)2 Si5 N8 : Eu [10.12]. Red-enhanced LEDs show
randomized and thus homogeneously mixed to the de- high CRIs above 90 for low color temperatures between
sired white. A second way to achieve a white color from 2500 K and 4000 K.
polychromatic LED modules is to use a diffuser plate in White LEDs with high color rendering at arbitrary
front of the LEDs, again to randomize the spatial dis- color temperatures between 2500 K and 10 000 K are
tribution of the different colors emitted by the LEDs. obtained by applying blue LEDs with a combination of
Finally tube-like hollow reflectors are being used for the green and red phosphors. Mueller–Mach and Mueller
same purpose, however in these systems the de-mixing e.g. found that SrGa2 S4 : Eu (535 nm) and SrS : Eu
and thus non-white appearance is often obvious at large (615 nm) are an appropriate combination for trichro-
angles. matic LEDs [10.13].
To date, the most common method of achieving Compared to dichromatic YAG : Ce LEDs, LEDs
white emission from LEDs is by partially converting with a trichromatic spectrum yield higher color ren-
blue light from the semiconductor to lower-energy emis- dering while the luminous efficacy is still almost as
sion by means of phosphors, which are deposited onto high.
the LED chip. This implies that the phosphor must
show strong absorption of the blue light and a high 10.3.4 Organic LEDs
photochemical stability due to the high excitation dens-
ity of up to several 100 W/cm2 . The widely used Organic LEDs became an important area of research and
combination of blue InGaN LEDs (430–480 nm) with development after the publication of direct EL observed
(Y, Gd)3 (Al, Ga)5 O12 : Ce (YAG : Ce) thereby domi- from a device incorporating only organic charge-transfer
nates the white-LED market today [10.7]. However, materials and emitters by Tang and van Slyke from Ko-
more advanced concepts are being realized, includ- dak in 1987 [10.14]. The authors were working with
ing blue LEDs in combination with yellow and red organic photoconductors, which have been used for sev-
phosphors or red and green phosphors. The former eral decades in copiers and printers. Applying a forward
concept aims at the enhancement of red radiation bias to an appropriate device built with these mater-
 Incoherent Light Sources 10.3 Solid-State Light Sources 579

ials, light was generated from directly injected carriers.

Part C 10.3
Cathode metal Cover lid Organic layers
A second important date was the publication of direct layer
organic EL observed in a single-layer polymer-based Indium tin
oxide anode Seal
device by Friend et al. in 1990, making a new class of Getter
materials available: semiconducting polymers [10.15].
Inert Gas
These marked the beginning of intense R&D in or-
ganic EL.
The general build-up of an organic LED (OLED) is Front glass substrate
represented in Fig. 10.21. The stack typically consists
of a substrate (mostly glass, however plastic is consid- Contact metal (+) Contact metal (–)
ered) with a thin layer (typically 100 nm) of indium–tin
oxide (ITO). This is the most prominent representative
Light
of transparent conducting materials, which is needed to
be able to couple the light out. On top of the ITO layer Fig. 10.21 Schematic build-up of an organic LED
the organic material is deposited, either as a multilayer
stack in small-molecule OLEDs or as a polymer coat to 10−8 cm2 /Vs for nearly insulating ones. The hole mo-
in polymeric OLEDs. The top contact, usually the cath- bility is typically larger (10−3 cm2 /Vs) than the electron
ode, is built by evaporating reactive materials such as mobility (10−5 cm2 /Vs). The mobility can be higher in
e.g. Ca, Mg or Ba to facilitate injection of electrons certain materials classes (up to 10−1 cm2 /Vs), but those
into the molecular materials. This is often followed by are not typically used in EL devices.
a less reactive metal, such as Al, to prevent fast oxi- The materials fabricated in OLEDs depend on which
dation of the top contact. Since many of the organic type of system is being used.
materials and especially the reactive cathode metal are In small-molecule devices the organic layers are usu-
water- and oxygen-sensitive, a nearly hermetic sealing ally brought onto the substrate in a layer of multiple
is required to achieve a practical lifetime of the de- materials having defined functions for each material.
vices. This is achieved by encapsulating the glass with A typical device layout is shown in Fig. 10.21. On top
a second glass or metal can, which often also contains of the ITO a hole conductor is deposited. This is in most
additional getter materials to absorb residual water and cases a material with high electron affinity to facilitate
oxygen. injection of holes from the anode and high hole mo-
In general two types of organic direct EL are distin- bility to achieve good hole transport. The next layer is
guished: polymeric and small-molecule organic LEDs. than a matrix material with a large band-gap and high
Both systems rely in principle on the same basic mech- triplet energy, doped with an emitter material. These
anisms and, although there are distinctive differences, two functions can also be combined into the same ma-
especially in the case of conjugated polymers, the device terial. On top of this layer a hole-blocker is deposited,
set-up is identical for the two versions. Whereas small- to prevent the fast-moving holes from flowing directly
molecule OLEDs are deposited using high-vacuum to the anode. The hole-blocker has to have a very large
sublimation or gas-phase transport, polymeric OLEDs band-gap and a large electron affinity to efficiently pre-
are typically spin-coated or in the case of displays, vent holes from flowing through the structure. Finally
ink-jetted. an electron conductor/injector is deposited on top to
The conduction mechanism in organic materials is facilitate electron injection into the device.
different to inorganic semiconductors. In the molecu- In polymeric OLEDs all these functions are com-
lar solid, after charge injection, the molecule becomes bined in the material, a conjugated or non-conjugated
oxidized and reduced, respectively, depending on the polymer. Thus the layer sequence is typically reduced to
nature of the charge injected. If an electron is injected two layers. First, a hole-injection layer, which also serves
from the molecule to the electrode, a hole is injected as a smoothing layer for the roughness of the substrate.
into the molecule, which oxidizes the molecule. The This layer is made insoluble and is finally coated with
charge then moves under the influence of the applied the light-emitting polymer. The total thickness is typ-
electric field across the molecule and by hopping from ically 100–200 nm. Charge transport of electrons and
the originally excited molecule to the neighboring one. holes as well as emission is provided from the different
Thus the conductivity of organic layers is rather low, functional groups of the polymer. A schematic com-
on the order of 10−3 cm2 /Vs in good materials, down
580 Part C Coherent and Incoherent Light Sources

Additionally, the coupling of two neighboring molecules

Part C 10.3

E = 0 (Vacuum)
can lead to the formation of cooperative electronic states,
which exhibit a new excited state, only present in the
LUMO Ef, coupled molecules. This phenomenon is called excimer
Xe cathode
in the case of identical molecules and exciplex, if two dif-
ferent molecules are involved. The energy of this excited
E state without ground state is typically lower than the ex-
cited states of the single molecules. Therefore, this effect
Ef,
anode can be used to generate other wavelengths than those
Xh HOMO available from the constituent molecules only. How-
ever, the efficiency of singlet excimers and exciplexes is
typically very low.
d1 d2 A crucial problem in organic LEDs (as well as in
Ef, inorganic ones) is the outcoupling efficiency of the gen-
LUMO 2 cathode erated light. Due to the fact that many of the materials
used have a high refractive index, light generated within
LUMO 1
the film and emitted at an angle larger than the critical
angle of the material suffers from total internal reflection
E
and the light is guided through the structure. Multiple
reflections at partially absorbing interfaces reduce the
amount of light coupled out. An estimate of the out-
Ef, coupling ratio, using classical ray optics, gives a value
anode
of 1/n 2 , where n is the refractive index of the material.
HOMO 2
For most materials, this value amounts to 20%, i. e. 4/5
HOMO 1
of the light generated does not leave the structure, ren-
Fig. 10.22 Schematic band diagram of polymeric (single- dering OLEDs often much less efficient than what could
layer) and small-molecule (double-layer) OLEDs, respec- be achieved with improved outcoupling. Intense work is
tively being conducted to identify the most promising meth-
ods to improve light outcoupling from thin-film organic
parison of the different electronic structures of small LEDs. Recent measurements, using an optically coupled
molecules and polymer OLEDs is shown in Fig. 10.22. hemispherical lens on top of an OLED substrate indicate
In the recombination zone, electrons and holes meet that the outcoupling can at least be doubled by appropri-
and create excitons on the excited molecules. These ex- ate means, improving the efficiency by the same factor
citons can have either a total spin of 1 or 1/2, thus of two. This topic is certainly one of the most important
creating a singlet or a triplet state. The singlet state is areas of OLED research.
optically allowed and therefore able to generate a pho- Many of the materials used in today’s OLEDs are
ton, whereas the triplet state is a forbidden transition, sensitive to moisture and oxygen. Irreversible oxidation
thus no emission of light is possible from these exci- occurs at the surface, rendering the material insulat-
tons. By statistics, the probability of creating a singlet is ing and thus creating a nonemissive (black) spot. These
only 1/4 for all injected pairs, limiting the principle ef- black spots are progressive, i. e. once formed they have
ficiency of organic EL devices to 25%. However, using a tendency to grow until the OLED is rendered useless.
other emitters, which have strong spin–orbit coupling These black spots often start at pinholes in the electrode
it is possible to overcome the selection rules and the materials, which in the case of many cathode materials
probability of forming a radiatively decaying exciton are themselves often the most critical materials with re-
theoretically increases to unity, thus opening the road to spect to oxidation within the whole OLED stack. To save
very efficient organic LEDs. the OLED from these failure mechanisms, near-perfect
Since the delocalization of the electron density encapsulation is necessary, which often includes a getter
across the molecule depends on the steric configuration material put in a sealed package, in close proximity to
of the molecule, and the electrical properties depend on the OLED cathode to capture any remaining oxygen and
the relative position between the molecules in the layer, water molecules before damage to the OLED occurs. In
the film formation is highly important in organic LEDs. the near future, it is hoped that thin-film packaging, i. e.
 Incoherent Light Sources References 581

the deposition of a nearly impermeable material on top sources. A diffuse, large area, flat lamp that can even-

Part C 10
of the OLED will solve this problem. tually become flexible, color-tuneable and transparent
Although this is one of the most often encountered is envisioned with OLEDs, creating new potential ap-
failure mechanisms of OLEDs, it is not the only one plications for light sources in the future. To generate
by far. Also, many OLED materials are intrinsically un- white light from OLEDs several possibilities can be
stable, limiting the useful life of OLEDs and creating used. External mixing of colors (e.g. from pixelated
the need for more research and development in the field sources), intrinsic broadband (white) emitters, mixed
of organic LED materials. emitters, stacked emitters or phosphor-converted blue
For application as light sources OLEDs have some or UV-emitting OLEDs (using e.g. the phosphors de-
way to go to achieve high efficiency at high bright- scribed above) can be fabricated; the different options
ness and a long lifetime. However, since all of these having different advantages and disadvantages depend-
properties have been shown in single devices, it is ing on the applications for the light source. Most
thought possible that OLEDs will be able to deliver. estimates establish a time frame of five to ten years be-
Due to the way in which they generate light, OLEDs fore OLEDs will become an important incoherent light
will provide a very different light source to existing source.

10.4 General Light-Source Survey

Table 10.2 gives a summary of the main types of light
sources and selected electrical and light technical prop-
erties as discussed in this chapter.

Table 10.2 Electrical and technical data of the basic types of light sources (state of affairs 2003)
Light source Electrical input Luminous Luminous Color rendering
power (W) flux (lm) efficacy (lm/W) quality
Incandescent 10–1000 80 –15000 8 –15 Excellent
Halogen 20–2000 300–60000 15–30 Excellent
Low-pressure Hg discharge 7–150 350–15000 50–100 Good
High-pressure Hg discharge 50–1000 2000–60000 40–60 Good
Metal halide discharge 20–2000 1600–24000 80–120 Good to excellent
Low-pressure Na discharge 20–200 2000–40000 100–200 Poor
High-pressure Na discharge 40–1000 1600–14000 40–140 Moderate to good
Sulphur microwave discharge up to 5000 up to 450000 80–90 (system) Good
White dichromatic inorganic LED 1–5 20 –150 20–30 Good
White trichromatic inorganic LED 1 20 –25 20–25 Excellent
Organic LED (at 1000 cd/m2 ) 15 mW (per cm2 ) 0.3 lm (per cm2 ) 30 Excellent

References

10.1 J. R. Coaton, A. M. Marsden: Lamps and Lighting 10.4 G. Derra, E. Fischer, H. Mönch: UHP-Lampen:
(Wiley, New York 1997) Lichtquellen extrem hoher Leuchtdichte, Phys. Bl.
10.2 W. Elenbaas: Light Sources (Crane, Russek, New 54(9), 817 (1998)
York 1972) 10.5 Y. A. Ono: Electroluminescent Devices (World Scien-
10.3 M. Born, T. Jüstel: Umweltfreundliche Lichtquellen, tific, Singapore 1995)
Physik J. 2(2), 43 (2003) In German 10.6 S. Nakamura, S. Pearton, G. Fasol: The Blue Laser
Diode (Springer, Berlin, Heidelberg 1997)
582 Part C Coherent and Incoherent Light Sources

10.7 A. Zukauskas, M. S. Shur, R. Caska: Introduction to 10.11 R. Müller-Mach, G. O. Müller, M. Krames, T. Trottier:
Part C 10

Solid-State Lighting (Wiley, New York 2002) p. 122 High-power phosphor-converted light-emitting
and references therein diodes based on III-nitrides, IEEE J. Sel. Top. Quan-
10.8 S. M. Sze: Physics of Semiconductor Devices (Wiley, tum Electron. 8, 339 (2002)
New York 1981) 10.12 M. Yamada, T. Naitou, K. Izuno, H. Tamaki, Y. Mu-
10.9 M. R. Krames, M. Ochiai-Holcomb, G. E. Höfler, razaki, M. Kameshima, T. Mukai: Red-enhanced
C. Carter-Coman, E. I. Chen, I.-H. Tan, P. Gril- white-light-emitting diode using a new red phos-
lot, N. F. Gardner, H. C. Chui, J.-W. Huang, phor, Jpn. J. Appl. Phys. 42, L20 (2003)
S. A. Stockman, F. A. Kish, M. G. Craford, T. S. Tan, 10.13 R. Müller-Mach, G. O. Müller: White light emitting
C. P. Kocot, M. Hueschen, J. Posselt, B. Loh, diodes for illumination, SPIE Proc. 3938, 30 (2000)
G. Sasser, D. Collins: High-power truncated- 10.14 C. W. Tang, S. A. Van Slyke: Organic electrolumines-
inverted-pyramid (Alx Ga1−x )0.5 In0.5 P/GaP light emit- cent diodes, Appl. Phys. Lett. 51, 913 (1987)
ting diodes exhibiting > 50 external quan- 10.15 J. H. Borroughes, D. D. C. Bradley, A. R. Brown,
tum efficiency, Appl. Phys. Lett. 75, 2365 R. N. Marks, K. MacKay, R. H. Friend, P. L. Burns,
(1999) A. B. Holmes: Light-emitting diodes based on con-
10.10 R. Müller-Mach, G. O. Müller, T. Jüstel, P. J. Schmidt: jugated polymers, Nature 347, 539 (1990)
US Patent 2003/0006702