Download Module 4: Light Emitting Diodes

Module 4: Light Emitting Diodes
4.1. Introduction
The light emitting diode (LED) has a multitude of applications in the electro-optics
field. With the recent interest in organic LEDs, the field is burgeoning. Infrared
devices are used in conjunction with spectrally matched plots--transistors in optoisolation couplers, hand-held remote controllers for video and audio equipment and
fiber optic sensing techniques. Visible spectrum applications include simple status
indicators on stereos, CD players, and scientific instrumentation. Here, we will
discuss their role in displays.
The light emitting diode is one of the simplest opto-electric devices. Compared to its
laser diode counterpart, the fabrication of the LED is amazingly simple since it does
not require an optical cavity. However, this simple fabrication comes with a trade-off
including low optical output, broad and incoherent spectra and slower device
response. For displays, this often suffices; however, a narrower spectral output is
desired for high resolution displays to improve color purity. Laser diodes are not
developed enough to date to be used in displays. This section is devoted to LED
technology and its future in displays.
4.2. History
In 1907, the first light emitting devices were discovered when yellow light was
produced by passing a current through a silicon carbide detector. What subsequently
transpired and led to today's commercial LED being displaced can be traced to the
observation of light emission from p-n junctions in semiconductors in the 1960's, and
to the fabrication of visible injection lasers in gallium arsenide phosphide (GaAsP).
Following this discovery, several groups began the development of GaAsP materials
for display applications.
During the same time frame, development of GaAsP LEDs were also being pursued
by IBM and Bell Laboratories. The work utilized liquid phase epitaxial techniques
instead of the vapor phase epitaxial used in the production of GaAsP devices. They
were successful in developing both red and green light emitters from this technique.
In the 1980's, the red emitting GaAlAs became important commercially. In the case of
green emitting devices, GaAsP has grown by both the liquid and vapor epitaxy
techniques.
A key discovery in 1966 at Bell Laboratories led to the enhanced performance greenemitting devices. The addition of nitrogen to GaAsP was the key. In 1971, the
nitrogen doping technique, in conjunction with the vapor epitaxy method, led to high
performance red, yellow and green devices.
Through the 1970's and early 1980's, the performance of LEDs dramatically
improved. And the cost of LEDs progressively decreased. In the late 1960's and early
1970's, vacuum fluorescent displays dominated the instrument market and the LEDs
were poised to take over.
In the 1970's, the calculator market boomed and established a commercial enterprise
for LEDs. During the timeframe 1973-1975, the LED shared the calculator market
with the vacuum fluorescent displays. Ultimately, the LED dominated the market.
When the LED calculator decreased because of the influx of the lower power LCD, it
was replaced with the growing watch market. This market was also lost to LCDs
shortly thereafter. Today, conventional LEDs are heavily utilized in instrumentation
and because of their power requirement, are not used in portable devices. The
1990's saw a dramatic increase in interest in polymer LEDs, that promised much
easier processing and perhaps an avenue to high resolution displays.
4.3. Operation of Basic LED Technology
The operation of an LED is based on the recombination of electrons and holes in a
semi-conductor. A single p-n junction is the principle semiconductor device. A
schematic of an LED device is shown below. For a forward bias junction, light is
emitted when minority-carrier injection and electron hole recombination takes place.
The recombination process can either be radiative or non-radiative as depicted in the
diagram.
A radiative recombination most often occurs through bound states associated with
impurity atoms, resulting in a photon energy, Ep given by
where Eg is the semiconductor energy gap and EI is the binding energy of the
impurity levels involved. The energy of the emitted photon is inversely proportional to
the wavelength given by
shallow levels EI<<Eg so
. This is if the recombination occurs through
.
Therefore, the color of the emitted light (i.e. its spectral power distribution) can be
controlled by using the appropriate combinations of semi-conducting materials and
impurities.
Examples of semiconductor energy gaps are shown below in the table. The peak
emission wavelength is also tabulated ensuring
.
The most common types of LED materials are formulated from III - V compounds,
because they can be used to fabricate p-n junctions and their output is in the visible
range. The table below shows two and three element compounds, sometimes
referred to as binary or ternary, respectively. The ternary compounds have a tunable
quality in that they can be tuned by adjusting the composition x of that alloy.
Material
Energy Gap,
Peak Emission
System
eV, at 300°K
Wavelength, Å
GaAs
GaP
AlAs
InP
GaAs1-xPx
1.43 direct
2.26 indirect
2.16 indirect
1.35 direct
1.43-2.03 direct
8670
5485
5740
9180
8670-6105
Xc=0.49
AlxGa1-xAs
2.03-2.26 indirect
1.43-1.98 direct
6105-5485
8670-6525
Xc=0.43
GaxIn1-xP
1.98-2.14 indirect
1.35-2.18 direct
6525-5790
9180-5685
Xc=0.62
2.18-2.26 indirect
5685-5485
The underpinning objective of LED technology is to maximize the height output by
increasing the probability for radiative recombination and decreasing the probability
for non-radiative recombination. Non-radiative recombination can occur through
many paths, such as the deep trap state, anger recombination or photon emission.
4.4. Direct - Indirect Transition
In semi-conducting materials, the conduction band and valence band are separated
by an energy gap. The energy difference between the valence and conduction bands
is a function of momentum of the electrons.
In direct semiconductors (e.g. GaAs or InP), the lowest energy conduction band state
occurs at the same momentum or k value as the highest energy valence band state.
In an indirect semiconductor, such as GaP, AlAs, the band extrema occur at different
k values. Because of conservation of momentum, the change in momentum of an
electron and hole involved in radiative recombination must equal the photon
momentum. A photon has negligible momentum so that the momentum of an electron
hole involved in electron-hole recombination must be essentially unchanged.
Therefore, on the energy vs. momentum illustration below
the conservation of momentum can readily be accommodated by small deviations
from the vertical transitions.
For indirect semiconductors, the electron must make a diagonal transition requiring a
substantial change in momentum. This can be accomplished by the scattering of an
electron from


a lattice vibration
impurity atom that permits the conservation of energy and momentum.
There is generally a low probability that this will occur, since an extra participant must
be involved in the recombination process. Consequently, non-radiative recombination
usually dominates in indirect semiconductors.
For the indirect case shown below
the energy change DE defines the photon energy and momentum. According to the
basic equation
where DE is the loss of energy equal to the bandgap, is the wavelength, p is the
photon momentum and h is Plank's constant, but conservation of momentum
additionally requires that the much greater electron momentum on the order of h/za
can be accounted for. Consider a lattice dimension a~ 10 -10 m and  ~ 10-6 m, it is
obvious that both conservation criteria can be met with the participation of a photon.
The two consequences of this result are that the indirect transition is inefficient (must
transfer momentum and hence thermal energy to the lattice) and less likely to occur
than the direct transition (requirement that all three particles simultaneously meet the
energy and momentum requirement). Indirect bandgaps typically have long diffusion
lengths and recombination times, which produce good transistors but ineffective
LEDs.
The importance of the difference between direct and indirect bandgap
semiconductors becomes obvious when LEDs are fabricated using different
compositions of a ternary allow system in which the energy gap changes from direct
to indirect at a certain transition. The ternary example systems include



GaAs1-xPx
GaAs1-xAs
In1-xGaxP
The band structure of GaAs1-xPx, a commercial grade LED, is shown below.
For x = 0, GaAsP is a solid comprised of GaAs and GaP. This compound exhibits a
direct transition since the direct minimum in the conduction band is of lower energy
than the indirect minimum. Since Eg ~ 1.4 eV, it emits in the near IR (~900nm).
If x is increased by the addition of phosphorous, the energy bandgap increases and
the nature of band structure is modified. Notice upon increasing x, the direct bandgap
minimum is much more sensitive than the indirect minimum.
At x = 0.49, the energy minimum is equivalent for both the direct and indirect cases.
After x = 0.49, the indirect minimum becomes of lower energy than the direct and
therefore, the physics of radiative recombination changes since these electrons now
collect the indirect minimum and have a low probability for radiative transitions.
The GaAs1-xPx system is well established, but can only produce wavelengths defined
by the range of energy gap widths, down to green. Blue LEDs require higher
bandgap materials, such as



sic technology is well developed for high temperature semiconductor
applications, but on the downside, it has an indirect bandgap, so its emission
efficiency is poor
GaN (also Fn/Al GaN alloys) is a direct bandgap material producing successful
blue and blue-green devices
II - VI compounds such as ZnS and ZnSe possess direct bandgap transitions
1.5 - 3.6 eV, offering a full spectrum LED with a single material.
4.5. Nitrogen Doping
It has been established that high efficiency LEDs can be fabricated from indirect
materials such as GaP and GaAs1-xPx by doping the crystal with nitrogen. Consider
GaAs1-xPxdirect devices can be fabricated for x<0.49 and for x>0.49 when the indirect
transition occurs, the efficiency falls off by 2 - 3 orders of magnitude.
With the addition of nitrogen, the quantum efficiency can be significantly improved by
more than two orders of magnitude. The figures below demonstrate this phenomena.
The nitrogen doping process "softens" the nature of the direct-indirect transition.
Higher performance devices are possible for alloy compositions. In the figure
presented below, a plot of brightness as a function of peak emission wavelength (and
energy) for GaAs1-xAs with and without nitrogen is presented.
Devices can be fabricated from the green through the red spectral regime The
constant LED performance from red through green is explained by the CIE relative
luminosity function which increases with peak emission energy (and alloy
composition) in the same region as the device efficiency is decreasing. The product
of the CIE curve and the efficiency curve results in the luminous performance as
depicted. The emission energy which is obtained in GaAs1-xPx devices doped with
nitrogen is a function of both nitrogen concentration and alloy composition.
Nitrogen is basically a diffract type of impurity than is commonly used in
semiconductor devices. It has been coined as the isoelectronic impurity since it
comes from column V in the periodic table of elements and therefore has five valence
electrons, as do arsenic and phosphorous that it replaces in the crystal.
As a consequence, nitrogen, unlike a donor or acceptor, introduces no charge
carriers in the crystal lattice. Nitrogen does provide a strong radiative recombination
center, particularly for direct bandgap materials. The reason for this is the unique way
in which carriers abound at the nitrogen site.
In the case of normal donors and acceptors, charge carriers are bound to shallow
impurities by weak Coulombic forces, which are proportional to 1/r and which result in
a large spread in position in any direction. From the uncertainty principle, the
uncertainty in position Dr, is inversely proportional to DK
Consequently, since  r, is large for a weak Coulomb binding, this implies that  K is
small and that the momentum of a carrier bound to a shallow impurity is closely
localized in momentum space in the vicinity of the impurity.
Since nitrogen is isoelectric and has no net charge, Coulombic binding is not
operative. Instead, charge carriers are bound at nitrogen centers by a much shorter
range attraction which results from a combination of the difference in the
electronegativity between the nitrogen atom and the V atom that it replaces, and the
hydrostatic deformation of the lattice around the nitrogen site. In this situation,  r is
smaller for the isoelectronic trap binding and  K is larger. The result is shown below.
The shaded region indicates that the electron is bound to the nitrogen impurity that is
widely distributed in space. The electron density is highest near the x minimum, but
there is also significant probability that the electron can be located anywhere
between G and x with a high probability of being at K = 0. The probability of finding
an electron at K = 0 is related to the separation between the direct and indirect
minima EgG - Eg(x) and that this probability increases as the separation between the
direct and indirect minima decreases. The fact that an electron bound to a nitrogen
impurity has a high probability of being at K = 0 permits the nitrogen doped indirect
energy gap crystal to act as if it were a direct energy gap crystal with vertical
transitions.
The momentum conservation rule can therefore be obeyed even in an indirect
crystal, with the result that nitrogen doping results in a high radiative recombination
efficiency for these materials. This has been effective in spanning the spectral range
between red and green.
Since the probability of an electron being at K = 0 decreases as the separation
between the energy minima increases, the radiative recombination probability and
hence quantum efficiency is expected to decrease as you move from GaAs toward
GaP. This is not what happens. With the addition of nitrogen, the quantum efficiency
is high (compared to these without nitrogen) and decreases as the crystal
composition is shifted toward GaP due to increasing separation between the direct
and indirect energy bandgap minima.
4.6. The GaAlAs System
The properties of the GaAsxAs1-x alloy system are analogous to the GaAs1-xPx
system. For x = 0, you have GaAs in both cases and for x =1, you have GaP and
AlAs in both cases; therefore, the similarities are not too surprising. These two binary
compounds both have indirect energy gaps with similar energies. The respective
ternaries both have direct-indirect transitions between x = 0.4 and x = 0.5, with
GaAlAs occurring at a slightly lower composition, consistent with the lower energy
gap of AlAs.
The GaAsP system has been important for visible LEDs, especially from the
commercial perspective, because it can be grown from the well established vapor
phase epitaxial technology. The GaAlAs requires liquid phase or OM-VPE growth
technology. GaAlAs has the advantage that the Ga and P atoms are similar in size as
compared to Ga and P. As a result, in GaAsP, the atoms do not fit together
particularly well as do the GaAlAs atoms, and there are fewer strain and dislocation
related defects in GaAlAs than GaAsP. This makes it possible to grow very high
quality devices using the GaAlAs system.
4.7. The GaP:ZnO System
Red emitting GaP devices utilize a deep trap state which is similar in many ways to
the nitrogen state. In this situation, the trap is formed by having Zn and O or nearby
lattice sites in the crystal. As in the case of nitrogen doping, this Zn,O state has a
short range potential in real space, because as is the case with an isoelectronic
molecule or pair with a large uncertainty in K space and a high recombination
efficiency, the reality is a short range potential. Since the Zn,O level has binding
energy of ~0.5 eV, the transition energy is approximately 1.8 eV, which falls in the
red part of the spectrum.
4.8. LED Devices
There are a variety of types of LED displays, the simplest being a discrete LED lamp.
Other examples include bar-of-light, seven segment displays, alpha numeric displays
and matrix addressed LED arrays.
4.9. Discrete Emitter
A discrete emitter is the simplest LED; it is essentially a light bulb or indicator lamp.
Common LED indicator lamps are red, green and yellow. The figure below shows the
primary elements of an LED emitter. It is composed of a lead frame, the
semiconductor chips and the encapsulating epoxy lens.
The epoxy is generally colored with dye in order to enhance the contrast ratio of the
lamp. In some instances, glass particles are dispersed in the epoxy to scatter light in
all directions.
4.10. Bar-of-Light Displays
In applications when more than a single light is needed, the bar-of-light technique is
often used. The bar of light is essentially a row of discrete LEDs connected together
with a common reflection cavity as shown below.
The cavity is most commonly filled with glass dispersed epoxy. The result is that
when the LEDs are turned on, the bar lights up uniformly. Bar-of-light displays are
often utilized as long rectangular indicators, or in conjunction with a legend. There
are many variations of bar-of-light indicators. In certain cases, the devices maintain a
given distance between reflecting cavities to form a linear array of small emitters. In
high resolution forms (> 100 elements), they are well suited for industrial process
control systems as a status of position indicator. These devices can be fabricated out
of multiple colored LEDs, so color changes as the bar is progressively illuminated.
4.11. Seven Segment Displays
The seven segment display is the simplest numeric-type display. In these types of
displays that were used in 1980 vintage calculators and wristwatches, the display can
be fabricated using a single piece of GaAsP shown below.
A single piece of n-type GaAsP is masked photolithographically such that the seven
bar shaped p-type regions can be diffused into it, forming the seven segment display.
Metal is deposited on the p-type regions and on the back side of the display to form
the anode and cathode contacts respectively. The most common type of packaging
for these displays is die attached and wireboarded to a circuit board and then
covered with a plastic magnifying glass.
For larger numeric displays, a hybrid assembly technique is employed. Small LED
chips, the same that are used for discrete lamps, are mounted in a reflecting cavity
such that chips appear as uniformly illuminated bars of light. The schematic diagrams
are shown below.
The above construction technique shows an LED chip. This construction technique is
one in which the LED chip is mounted on a printed circuit board and the light is
reflected from a metalalized surface and scattered by a diffusing plate.
The above figure show an epoxy-filled display mounted on a lead frame.
4.12. Alphanumeric Displays
There are two categories of alphanumeric displays: 1) segmented and 2) dot matrix
displays. These can be fabricated in a variety of fonts as shown below.
The addressing schemes will be presented later in this module.
4.13. Matrix Addressed Displays
One way to create a matrix addressed display is to grow a p-type layer on the n-type
substrate and follow this with the growth of a n-type layer. An isolation diffuser is then
driven through he n-type epitaxial layer, and finally the p-type anode region is
diffused. In order to complete the array, a complex multilevel metalization scheme is
required
The above figure is a planar electrically isolated monolithic LED array structure. The
isolation is accomplished by a deep Zn diffusion through the upper n-type layer of an
n-p-n structure.
Another type of array structure for x-y addressability is accomplished with row
isolation obtained by sawing through the epitaxial layer as shown below.
The column contacts are made by stitch bonding from row to row. This process can
be contrasted with the other structure which eliminates the sawing and wire bonding,
but requires more complicated material growth and water processing technology.
4.14. Optical Device Efficiency
Up to this point, we have reviewed the physics of LEDs and some example structures
for displays in determining the device efficiencies. The emission process can be
broken down into three characteristic steps.
1. excitation
2. recombination
3. extraction
Let us begin by discussing excitation with reference to the following three figures.
Photons that are created by minority electron recombination on the p-type side of the
junction are more likely to be successfully emitted from the surface of the device for
devices shown above with an opaque and transparent substrate. For a total LED
current, I, made up of three components, electron, hole and space charge region
recombination components identified as In, Ip and Ir, respectively, the electron
injection efficiency providing the excitation is
All of the processes described above equally apply to both electrons and holes. The
electron mobility, mp, on the other hand, is greater than that of a hole, m p, since
where Nd and Na are defined as n-type donor and p-type acceptor densities,
respectively. A greater gn is attainable for a given doping ratio than hole injection
efficiency, gp. Consequently, LEDs are usually p-n+ diodes constructed with the player at the surface as shown above.
Some of the recombinations undergone by the excess electron distribution, n, in
the p-type region will lead to radiation of the desired photon, but others will not. This
is due to the existence of doping and various impurity levels in the bandgap. The
recombination rate, R, can be expressed in terms of radiative and nonradiative rates,
Rr and Rnr, as
where
tr and tnr are the minority carrier lifetimes associated with the radiative and
nonradiative recombination process and t is the effective lifetime.
The radiative efficiency is defined as
and the internal quantum efficiency ni is
From the above figures, it is obvious that the photons generated on either side of the
junction will traverse through substantial amounts of semiconductor material to be
reabsorbed. In fact, the photon energy may be ideally suited to reabsorption if it
exceeds the semiconductor direct bandgap. It is obvious why GaAs is opaque and
GaP transparent to photons from Ga(As:P) junctions. The greater efficiency is
expected from the transparent substrate with reflecting contact.
The photon must strike the LED surface at an angle less than the critical angle for
total internal reflection, qc, given by the expression
where next and nLED are the refractive indices of the external environment and the
LED, respectively. The critical angle loss attributed to air (next = 1) can be reduced by
encapsulating the device in an epoxy lens cap to increase next > 1 and the angle of
incident at the air interface.
Even for angles less than qc, there is Fresnel loss, with transmission ratio
The total external quantum efficiency is then a fraction of the photon emitted, given
by the expression
where a is the average absorption coefficient n0 is the LED volume and A is the
emitting area.
For practical display applications, the radiation wavelength with respect to spectral
response of the human eye should be considered. The GaP LED that emits green,
for example, is less efficient than the GaAsP LED that emits red; however, the eye
compensates for this deficiency since it is more sensitive to green.
Relatively new heterojunction LEDs offer two mechanisms to improve LED
efficiencies. The electron injection efficiency can be enhanced, but, in addition,
absorption losses through the wider bandgap n-type layer are essentially eliminated
by photons emitted by recombination in the lower 2.0 eV bandgap p-type region. A
GaAlAs heterojunction LED is shown below along with its corresponding energy band
diagram .
4.15. Driving Schemes and Interfacing
In circuit diagram applications, the LED is often modeled as a regular diode, but with
much greater forward voltage, Vp. For displays, one usually seeks maximum
brightness from the material, heavy conducting and VP approaches the contact
potential. Consider the example of moving from GaAs to Gap, the value of VF moves
form 1.5 to 2.0 volts. The dependence of VF on temperature can also be thought of
as regular diodes, but radiant power and wavelength also changes.
Single discrete LEDs are usually driven by logic gates, perhaps as status indicators.
Two examples are shown below.
The gate output will not be able to source or sink sufficient current for visibility and an
amplifier will be required as shown below.
Bar-to-light displays are utilized to indicate signal level (e.g. audio volume or strain)
or with a modification of the position indicator for fine tuning. Operational amplifiers or
voltage comparators used to decode an analog signal into a bar graph or positional
indicator display are shown below.
Matrix addressed LED arrays can be used for high density display panels and
conventional row by column strobing which is controlled by a microprocessor
interface. A simple LED matrix is depicted below, where an LED element will be
turned on by applying the proper signal to one x-axis and one y-axis.
Multiple LEDs are commonly packaged together in a single integrated device,
organized in one of the standard display fonts, with decoding often included within
the package as shown below.
The 7-segment display is adequate for hexadecimal applications, but the 16-segment
is required for alphanumerics. To reduce pin-out requirements, the LEDs of a single
package are either connected to a common cathode or common cathode
configuration with multiple display digits multiplexed as illustrated below.
4.16. Conclusions
The LED is an important optical source which has significant applications to the
display community. For simple display applications discussed here, the LED
fabrication process is simple and easy to drive. The key drawback of LED technology
is the broadband emission, making it difficult to design a broad color gamut.