22-Elec-B10 Electro-Optical Engineering · May 2015
Question 3 of 6: GaAs LED Output and a Fabry–Perot Laser Diode
Nivaar worked solution (AI-drafted; not reviewed by a licensed engineer)
Notes on this paper
Paper format. National Exams, May 2015 — 07-Elec-B10
Electro-Optical Engineering. Three hours, closed book; one 8.5 in ×
11 in double-sided sheet of hand-written notes and one approved Casio or
Sharp calculator are permitted. Six questions are printed; any five constitute a
complete paper and all questions are of equal value (20 marks each).
All six are solved here, because the set is a study resource
rather than a timed sitting. The constants used throughout are those printed on
page 1 of the paper:
$\varepsilon_0 = 8.854\times10^{-12}$ F/m,
$\mu_0 = 4\pi\times10^{-7}$ H/m,
$c = 2.998\times10^{8}$ m/s,
$q = 1.602\times10^{-19}$ C,
$h = 6.626\times10^{-34}$ J·s,
$k = 1.381\times10^{-23}$ J/K, and the semiconductor data
(GaAs: $n = 3.63$, $E_g = 1.41$ eV; Si: $E_g = 1.11$ eV;
InGaAsP: $n = 3.5$).
Reference texts. The Engineers Canada syllabus for this
examination code draws on the standard fibre-optic and photonics texts, and the
solutions below cite them by chapter:
G. Keiser, Optical Fiber Communications, 5th ed., McGraw-Hill
— the primary reference for fibre dispersion, sources, detectors,
receivers and link budgets.
B. E. A. Saleh and M. C. Teich, Fundamentals of Photonics, 3rd ed.,
Wiley — semiconductor sources, laser rate equations and photodetector
noise.
J. Wilson and J. F. B. Hawkes, Optoelectronics: An Introduction,
3rd ed., Prentice Hall — device-level treatment of LEDs, laser diodes and
photodiodes.
J. M. Senior, Optical Fiber Communications: Principles and Practice,
3rd ed., Pearson — rise-time budgets and receiver front-end comparison.
Question 3: GaAs LED Output and a Fabry–Perot Laser Diode (20 marks)
Given. Two GaAs emitters — an LED biased at 15 mA
at room temperature, and a Fabry–Perot laser diode driven ten times above
its threshold — with the data collected below. The exam constant sheet
supplies $E_g = 1.41$ eV and $n = 3.63$ for GaAs.
Given data
Quantity
Symbol
Value
(a) LED junction temperature
$T$
27 °C = 300 K
(a) LED bias current
$I$
15 mA
(a) External quantum efficiency
$\eta$
15 %
(b) Cavity length × stripe width
$L \times w$
200 µm × 5 µm
(b) Emission wavelength
$\lambda$
850 nm
(b) Threshold / operating current
$I_{th}$ / $I$
2 mA / 20 mA
(b) Internal (material) loss
$\alpha$
25 cm$^{-1}$
(b) Internal quantum efficiency
$\eta_i$
80 %
Find. For the LED, the peak wavelength, the spectral width
and the emitted optical power; for the laser, the threshold gain and photon
lifetime, the photon density per unit area, the power generated inside the
cavity and the power leaving one facet.
Emission spectrum of the GaAs LED. Spontaneous recombination across a 1.41 eV gap fixes the peak at 879 nm, and the thermal spread of carriers in the bands sets the linewidth.
Approach. For the LED, the band gap fixes the photon energy
and hence the wavelength, the carrier thermal energy fixes the linewidth, and
the external quantum efficiency converts injected electrons into emitted
photons. For the laser, the round-trip condition — gain equals internal
loss plus mirror loss — fixes the threshold gain, from which the photon
lifetime, the steady-state photon density and the output power all follow.
Peak wavelength of the LED. Spontaneous recombination in
GaAs releases a photon of about one band gap, so
$$\lambda = \frac{hc}{E_g}
= \frac{(6.626\times10^{-34})(2.998\times10^{8})}{1.41\times1.602\times10^{-19}}
= 8.794\times10^{-7}\ \text{m}
= \boxed{\;879\ \text{nm}\;}$$
This sits in the first fibre window, consistent with the 850 nm systems of
the other questions on this paper.
Spectral width from the carrier thermal spread. The
electrons and holes that recombine are distributed over roughly $kT$ of energy
in each band, and the standard result for a bulk LED line is a full width at
half maximum of about $1.8\,kT$ in energy:
$$\Delta E \approx 1.8\,kT = 1.8(1.381\times10^{-23})(300)
= 7.46\times10^{-21}\ \text{J} = 0.0465\ \text{eV}.$$
Converting an energy width to a wavelength width with
$\Delta\lambda = \lambda^{2}\Delta E/(hc)$,
$$\Delta\lambda = \frac{(879.4\times10^{-9})^{2}(7.46\times10^{-21})}
{1.986\times10^{-25}}
= 2.90\times10^{-8}\ \text{m} = \boxed{\;29\ \text{nm}\;}$$
which is the familiar few-tens-of-nanometres linewidth of a near-infrared LED,
and roughly a hundred times broader than a laser diode of the same material.
Optical power emitted by the LED. The external quantum
efficiency is the number of photons escaping per injected electron, so
$$P = \eta\,\frac{I}{q}\,h\nu = \eta\,\frac{I}{q}\,E_g
= \eta\,I\,\frac{E_g}{q}
= (0.15)(15\times10^{-3})(1.41\ \text{V})
= \boxed{\;3.17\ \text{mW}\;}$$
The electronic charge cancels neatly when the gap is written in electronvolts:
the power in milliwatts is simply efficiency times current in milliamps times
gap in volts. No emitting area is quoted, so "intensity" is reported
here as the total radiant flux; if the emitting facet area were known the
irradiance would be this power divided by that area.
Facet reflectivity of the laser. A cleaved GaAs–air
interface reflects by Fresnel's normal-incidence formula,
$$R = \left(\frac{n-1}{n+1}\right)^{2}
= \left(\frac{3.63-1}{3.63+1}\right)^{2} = 0.323 .$$
No coatings are mentioned, so both facets share this value — about a third
of the incident power returned, which is enough to sustain oscillation only
because the semiconductor gain per pass is very large.
Threshold gain from the round-trip condition. Steady
oscillation requires that one round trip return the field unchanged, so the
modal gain must make up the distributed material loss and the loss out of the
two mirrors:
$$g_{th} = \alpha + \frac{1}{L}\ln\frac{1}{R}
= 25 + \frac{1}{0.02\ \text{cm}}\ln\frac{1}{0.323}
= 25 + 56.6
= \boxed{\;81.6\ \text{cm}^{-1}\;}$$
The mirror term dominates: for a cavity only 200 µm long, letting
light escape is a bigger loss than absorbing it, which is exactly the design
intent.
Photon lifetime. The photon lifetime is the time constant
with which the cavity would empty if the gain were switched off, and it is set
by the total loss the photon meets while travelling at the group velocity
$v_g = c/n$:
$$\tau_p = \frac{1}{v_g\,g_{th}}
= \frac{1}{(8.259\times10^{9}\ \text{cm/s})(81.6\ \text{cm}^{-1})}
= \boxed{\;1.48\ \text{ps}\;}$$
A picosecond-scale photon lifetime against a nanosecond-scale carrier lifetime
is what puts the relaxation-oscillation resonance of such a laser in the
gigahertz range (see Question 6(d)).
Steady-state photon density per unit area. Above threshold
every extra injected carrier that recombines radiatively feeds the lasing mode,
and in the steady state that supply balances the cavity loss:
$$\frac{dS}{dt} = \frac{\eta_i (I-I_{th})}{qV} - \frac{S}{\tau_p} = 0
\quad\Longrightarrow\quad
S\,d = \frac{\eta_i (I-I_{th})\,\tau_p}{q\,A},$$
where $A = L\,w = (0.02)(5\times10^{-4}) = 1.0\times10^{-5}$ cm$^2$ is the
cavity footprint. Substituting,
$$S\,d = \frac{(0.8)(18\times10^{-3})(1.485\times10^{-12})}
{(1.602\times10^{-19})(1.0\times10^{-5})}
= \boxed{\;1.33\times10^{10}\ \text{photons/cm}^{2}\;}$$
The active-layer thickness is not given, so the sheet density (photons per unit
area) is the meaningful quantity; dividing by a typical 0.1 µm
quantum-well-plus-barrier thickness would give a volume density of order
$10^{15}$ cm$^{-3}$.
Optical power generated inside the cavity. The same carrier
flux, weighted by the photon energy
$h\nu = hc/\lambda = 1.459$ eV, gives
$$P_{\text{gen}} = \eta_i (I-I_{th})\,\frac{h\nu}{q}
= (0.8)(18\times10^{-3})(1.459\ \text{V})
= \boxed{\;21.0\ \text{mW}\;}$$
This is the power stimulated into the mode; it is not all available outside,
because some of it is absorbed by the material loss on the way.
Power leaving one facet. The generated power divides
between the two loss channels in proportion to their coefficients, so the
fraction escaping through the mirrors is
$$\frac{\alpha_m}{\alpha_m+\alpha} = \frac{56.56}{81.56} = 0.693 ,$$
giving a total emitted power of $(21.0)(0.693) = 14.57$ mW. The two facets
are identical, so each radiates half:
$$P_{\text{facet}} = \boxed{\;7.28\ \text{mW}\;}$$
Equivalently the external differential quantum efficiency is
$\eta_d = \eta_i\,\alpha_m/(\alpha_m+\alpha) = 0.555$, i.e. a slope
efficiency of $\eta_d\,h\nu/q = 0.81$ W/A shared between the two ends.
Fabry-Perot cavity of the GaAs laser. Round-trip oscillation requires the modal gain to balance the internal loss plus the loss through the two cleaved facets; the escaping fraction is what appears as useful output.