[Laser] LED used as APD

[Kaj Wiik]

Hi!

2011/12/27 steve kavanagh <skavanagh73 at yahoo.ca>:
> I came across some more mentions of using LEDs as APDs
...

> http://digital-library.theiet.org/getabs/servlet/GetabsServlet?prog=normal&id=ELLEAK000014000017000553000001&idtype=cvips&gifs=yes&ref=no
>
> (I found the abstract in many places...the paper itself doesn't appear to be
> on line)

I have a copy (two pages, PDF) if someone is interested.

Here are some excerpts:

"Description of l.e.d.s investigated: Three Burrus-type l.e.d.s,
from two different batches of material, were examined. The
behaviour was substantially the same in all three diodes except
that the reverse-breakdown voltage was quite different for the
two material batches. All of the l.e.d.s were AlGaAs double-
heterostructure diodes of the type described in Reference 8.
Two of the l.e.d.s were made from material manufactured
by Laser Diode Laboratories. For these diodes, reverse
breakdown occurred at 10.4 V. The emission occurred at
0.800 um and the output power at 150 mA drive was of the
order of 4 mW.
The third l.e.d. was made from material fabricated by
A. G. Dentai of Bell Laboratories. The emission is at about
0.81 um, and the output power at 150 mA drive is of the
order of 8 mW. Reverse breakdown occurs at 20.4 V. The
structure and composition of this diode is shown in Fig. 2."

Its behaviour as a detector is shown in Fig. 1 for illumination
from a nominally identical l.e.d. source.
Fig. 1 shows the detected signal current against reverse-bias
voltage for five values of incident optical signal power ranging
from 2uW to about 200 uW. The behaviour is substantially the
same over this range of incident power; however, a small
decrease in gain at higher input levels is evident. From Fig. 1,
absolute sensitivity (external quantum efficiency) can be
inferred by comparison with the signal detected on a
calibrated EG&G SGD-100 p-i-n detector. The quantum
efficiency of this detector at 0.81 nm is 0.60, corresponding to
0.39 A/W. The horizontal lines that intersect the curves at
about 14 V reverse bias show the response of this p-i-n
detector. The area of this detector is substantially larger than
that of the l.e.d., so we can be sure that it is 'seeing' all of the
light available to the l.e.d. (If it in fact sees more light, the
quantum efficiency of the l.e.d. is even higher than reported
here.) From Fig. 1 we see a maximum external quantum
efficiency of 5.7 A/W. At this wavelength this amounts to
8.8 electrons per photon.

To determine whether the observed effect was related to
the fact that the wavelength of the received light was very
close to the band edge of the photodetector, the experiment
was repeated using a HeNe laser (6328Å) as the source. No
significant difference in behaviour was observed. (This rules
out the Franz-Keldish effect as a significant contributor.)

The dark current of the l.e.d. increased with increasing back
bias. This resulted in an optimum signal/noise ratio (s.n.r.)
(assuming other noise sources negligible) at a gain value of
about one-half of the maximum gain achievable. At the point
of optimum s.n.r. the external quantum efficiency for the
20 V l.e.d. was about 2.5 A/W—about 6 times higher than
for a typical p-i-n detector."

"Conclusions: Avalanche gain has been demonstrated in double
heterostructure AlGaAs l.e.d.s. In one case, optimum s.n.r. was
achieved with an external quantum efficiency of 2.5 A/W and
a dark current of 6 nA.
No attempt has been made to optimise device parameters
for use as detectors. The devices investigated were 'off-the-
shelf l.e.d.s. With particular attention to detection properties
significant improvement may be possible."

Over 8 electrons per photon, not bad! Also the SNR optimum is well
below breakdown, which makes life a bit easier...

Cheers,
Kaj OH6EH