Page 362 - Electrical Properties of Materials

344 Optoelectronics

small quantity of In in GaN leads to localized deep levels that produce rel-
atively stable excitons; and as their constituents move in the strong fields the
recombination is rapid. It has been estimated that the diffusion length of minor-
ity carriers (before recombination) is 50–60 nm. Thus radiative recombination
would occur before a non radiative encounter at a threading dislocation, even
–2
10
with dislocation densities of 10 cm .
As we have said several times, the variable quality, internal strains and piezo
electric fields mask the correlation between bandgap and emitted light. An
interesting summary based on experimental results of numerous InGaN LEDs
∗
∗ In the book edited by B. Gil listed in has been made by P. K. O’Donnell who gives the following equation in terms
further reading Appendix VIII. of x, the molecular fraction of In in InGaN (i.e. In x Ga 1–x N) for the peak of the
emitted light E p (in eV).

E p =3.41 – 4.3x

There is a spread of about 10% in these coefficients, and it applies only for x
between 0–0.5.
In summary, we can say that considerable progress continues in many
†
† For example, recent work [H. Hira- features of LEDs, including substrates of slightly larger GaN “mirror fin-
yama et al., Appl. Phys. Express 1, ish” slices (still grown on sapphire). Skilled provision of buffer layers reduces
051101 (2008)] with deep ultraviolet di- spreading dislocations. Selection of inclined planes for growth has improved
odes has shown that if threading dislo-
cations in the buffer layer are reduced, performance and yield in some devices. But the main goal of room light-
efficiency is improved. The deep UV re- ing is still some way off. Blue LEDs coated with composite phosphors have
–1
gion around 230–380 nm (over 5 eV) produced white light, fairly well matched to sunlight, at 15 lumens W ,giv-
is potentially important as a steriliz- ing about 20 lumens per diode. The snag is that they are still expensive and
ing antibiotic agent for surgical instru-
ments, food-processing machinery, and around 100 are needed to match a 100 W tungsten filament bulb. However, we
water supplies, which usually cannot be routinely use LED torches and lanterns, battery powered. So if we were pre-
flooded with sterilizing liquids. It is get- pared to bulk buy white diodes and rewire our lighting system to 5 V d.c.,
ting close to soft X-rays and could be we could have energy-efficient room lights now. But if this were seriously
useful for imaging the large molecules
that are important for life. The diodes proposed we think the voters would rebel at the capital cost, even although
used were made of an Al x Ga 1-x Nal- they can no longer get their 100 W bulbs from Woolworths. So room lighting
loy as the energy gap is beyond GaN is still a challenge. Maybe InGaN LEDs will have to compete with OLEDs
(see Table 8.3), with an x value of about
0.87. They were grown on sapphire sub- (Organic LEDs) for efficient replacement of our old fashioned lighting, see
strates by low-pressure MOCVD, with Appendix I.
a buffer layer. This is a very difficult Finally, a word about the environment. Electricity is mainly produced by
device region; initially, Hirayama et al. fossil fuel which causes the undesirable CO 2 emission. Over one fifth of the
obtained a diode output of 5 μWwith electricity consumption in developed countries is due to lighting and the pro-
a quantum efficiency of 0.001%, but by
reducing the quantum well thickness to portion is even higher in developing countries. If the use of LED lamps will
only 1.3 nm and introducing an AlN halve that consumption that, would save, it has been calculated, 300 Megatons
buffer layer to reduce dislocations from of CO 2 emission in the US alone. Will that happen? As I said, I feel quite
9
9
–2
about 3.2 × 10 cm –2 to 1.8×10 cm ,
they obtained an output of 150 μW with certain that white LEDs will come in due course. High power white LEDs for
a quantum efficiency of 0.2%. So the practical lighting systems have already been developed by several compan-
device is still of low efficiency but better, ies (for example, Cree, Nicia, and Seoul Semiconductor). Global trials took
a striking advance.
place in 2012 in 12 major cities (including Kolkata, London, Rio de Janeiro,
Sydney, Toronto, and New York) and showed up to 85% energy saving. On
a lesser but a more local scale, trials took place in 2010 at Imperial College,
London, which suggested a lifetime of 7 years (3–4 times that of conventional
lighting) with a payback period of 4 years for the fittings. LED lighting is
coming.