11.1. Consider a photodetector operating in the neighborhood of 60 GHz.

How
many cycles of green light are there in a single cycle of 60 GHz? Can a
photodetector be used to follow the oscillations of the electromagnetic field
associated with this light?

For the RF signal,

1 1
TRF = = = 17 ps
vRF 60 × 109

For the optical signal, the wavelength is about 500 nm, from Figure 11.1.

The period is

1 λ 500 × 10−9
Top = = = = 0.0017 ps
vop c 3 × 108

17
The RF period is = 10000 times longer, so 10,000 cycles of
0.0017

optical light fit into one cycle of the RF. Therefore even the fastest

detectors can only measure a time average intensity compared to the

oscillation of light.

11.2. Light with wavelength λ=700 nm is incident on a sample of GaAs.

a) Where in the spectrum does this radiation lie?

From Figure 11.1, this is at the red edge of visible light.

 b) At what depth is the incident flux (neglecting Fresnel loss) reduced to
10% of its value at the surface? 그리고 1%로 감소되는 깊이는?

To find the absorption coefficient from Figure 11.4, we must convert this

wavelength into energy. Using the Golden Rule:

1.24 1.24
E(eV) = = = 1.77eV
λ ( μm) 0.700

From Figure 11.4, α=2.1×104 cm-1.

The decay in intensity is given by

 FL (x) = FL (0)e −α x or

FL ( x) − ( 2.1×104 cm −1 ) x
= 0.1 = e , ln (10) =2.3
FL (0)
ln( 0.1) −4
x10% = − −1 = 1.1× 10 cm = 1.1μm
2.1 × 10 cm
4

For 1% remaining flux,

ln ( 0.01)
x1% = − = 2.2 × 10−4 cm = 2.2 μ m
2.1 × 104 cm −1
c) The color is changed to orange. Now how deep does the light penetrate
(to the 10% level)?

From Figure 11.1, orange light has a wavelength of about 600 nm, or a

photon energy of 1.24/0.6=2.07eV.

From Figure 11.4, at this energy the absorption coefficient of GaAs

increases to 5.3×104cm-1, and

ln (0.1)
x10% = − −1 = 0.43μm
5.3 × 10 cm
4

11.3. a) Calculate the Fresnel reflection at normal incidence for light going from
air to glass (n=1.5).

(1− 1.5)20.52
R= = = 0.04 or 4% is reflected.
(1+ 1.5)2 2.52
b) Explain why you can see into a store window and see your reflection at
the same time, but at night looking out your window from a lighted room you can
only see your reflection?”

When you look into a store window, you see 4% of the sunlight reflected

from you and off the surface of the glass, but you also see the light from
 the merchandise being transmitted through the window. To your eye both

images are superimposed.

At night, there is very little light from the far side of the window, so your

4% reflection of the room light is the only thing you can see.

11.4. Show that Equation (11.8) follows from Equation (11.6).

Equation (11.6) is:

1 dJn (x) Δn
+ GL (x) − p = 0
q dn τn

But we know

GL ( x) = α (1 − R ) FL ( x) = α (1 − R ) FLi e −α x

and we also know

dn
Jn = qDn
dx

Using n=np,

dJ n d 2 np
= qDn
dx dx 2

Therefore

1⎛ d 2 np ⎞ Δn p
⎟⎟ + α (1 − R ) FLi e −
−α x
⎜ qDn =0
q ⎜⎝ dx 2 ⎠ τn

1⎡ ⎛ d 2 n p 0 d 2 Δn p ⎞⎤ Δn p
⎟⎟ ⎥ + α (1 − R ) FLi e −
−α x
⎢ qDn ⎜ + =0
q ⎢⎣ ⎜ dx 2 dx 2 ⎥ τ
⎝ ⎠⎦ n

But since np0 is a constant, this reduces to

 1⎡ d 2 Δn p ⎤ Δn p
⎢ qDn ⎥ + α (1 − R ) F e −α x
− =0
q ⎣⎢ dx 2 ⎦⎥
Li
τn

which is Equation (11.9.)

11.5. For the circuit shown in Figure P11.1

a) Plot the I-Va characteristic for the diode with photocurrent IL=100μA, 200μA,
and 300μA. Let IDark=I0=10-14 A and the ideality factor n=1.

RL=10 kΩ

Va

b) On your graph, also plot the load line for Va =+5V and Va=-5 V

c) Find the current flowing through the circuit for each load line and plot it
against IL. Recalling that the photocurrent IL is proportional to the intensity of the
light, under which bias regime should one operate photodiodes if one wants the
output current to be proportional to intensity?

To produce a linear variation in output current with incident light, reverse

(or zero) bias should be used.

11.6. A photodiode is made of Al0.1Ga0.9As. The refractive index at λ=800 nm is

3.65. If the junction depth is 0.3 μm, and the junction width is 1.5 μm, find the
quantum efficiency ηQ and the responsivity Rph. Assume the light is incident from
air.

(1 − n ) = (1 − 3.65 )
2 2
(−2.65) 2
The Fresnel reflection is R= = = 0.32
(1 + n ) (1 + 3.65)2
2
4.652

To find the absorption coefficient, we use

 1.24 1.24
E (eV ) = = = 1.55eV . From Figure 11.4, we find the
λ ( μ m) 0.8μ m

absorption coefficient is α=9000 cm-1. After the surface layer, the flux

remaining is

FL ( xn ) = FL (0.3μ m) = (1 − R ) FLi e−α xn = (0.68)( FLi )e

( )(
− 9×103 cm −1 0.3×10−4 cm )

= 0.52 FLi

The amount of light remaining past the depletion region is

FL ( w) = FL ( xn )e−α w

so the amount absorbed in the depletion region is

( ) (1.5×10 ) ⎤
FL (0.3μ m) ⎡1 − e
−4

= 0.52 FLi (1 − 0.26 ) = 0.38 FLi

− 9×103

⎣⎢ ⎦⎥

If every photon absorbed produces current, then

ηQ = (1 − 0.32 ) ⎡⎣e−α x (1 − e−α w ) ⎤⎦ = (0.68) [ 0.38] = 0.26

n

The responsivity is

q qλ (1.6 × 10−19 )( 0.8 × 10−6 )

R ph = ηQ = ηQ = ( 0.26 ) = 0.17
hν hc ( 6.62 × 10−34 )( 3 × 108 )

11.7. For the following solar cell whose I-V characteristics are shown in Figure
P11.2, find Isc, Voc, and η. The incident power is 15 mW.
 From the graph we find ISC = -4mA, VOC = 0.95V, Im=3.5 mA, and

Vm=0.83 V. The fill factor is

Im Vm 3.5 × 0.83
FF = = = 76%
IscVoc 4.0 × 0.95

and the power conversion efficiency is

I V
η = FF sc oc = ( 0.76 )
( 4 ×10 ) ( 0.95) = 0.19 = 19%
−3

PLi 15 ×10−3

11.8. Which absorbs more of the total solar spectrum, GaAs or Ge?

From Figure 11.8, we see that Ge has a small bandgap, and thus

can absorb all wavelengths below about 1.8μm, whereas GaAs

has bandgap such that is can only absorb wavelengths smaller

than about 0.9 μm, so GaAs is transparent to a large fraction of

the solar spectrum.

11.9. If a photon of wavelength at the solar spectrum peak of λ=0.5μm (green) is
absorbed by Si, the electron and hole have excess energy as shown on the energy
band diagram of Figure P11.3. If both carriers scatter down (or up) to the band
edges, what percentage of the absorbed energy is lost as heat?

EC

EV

From the Golden Rule, the total energy absorbed from the photon is

1.24 1.24
E (eV ) = = = 2.48eV
λ ( μ m) 0.5

If both the electron and hole relax to the band edges, the energy different

between them now is just 1.12 eV. Therefore the total energy given up as

heat is

Eph-Eg=2.48-1.12=1.36eV, or more than half.

11.10. If the doping in the p region of an n+p photodiode is decreased, one would
expect the diffusion length in a solar cell to increase. Verify (or contradict) this by

repeating Example 11.4 with N A' =1016 cm-3. For a factor of 10 change in doping,

what was the change in ηQ?

The parameter that is affected is the diffusion length Ln, which is

increased to 400μm (Figure 3.23). We now have

 (10−2 μm −1 )(400 μm)(1 − 0.2)e −10
−2

α Ln (1 − R )e− αx n
⋅ 0.4

ηQ (λ = 1μm) = = = 64%
(αLn + 1) (10 −2 μm−1 )(400μm) + 1
and

αLn (1− R)e −α x (1μm )(400 μm)(1− 0.2)e

n
−1 −1⋅0.4

ηQ (λ = 0.5μm) = = = 53%
(αLn + 1) (1μm−1 )(400μm) + 1

We see that the lighter doping made a significant change in the efficiency

at infrared but made little difference at green.

11.11. If a GaAs photodiode has a junction depth of 0.2 μm, and if light absorbed
in the surface layer is considered lost to surface recombination, what is the total
fractional loss in photons in the surface layer? Let the photon energy be 1.4 eV,
and repeat for Eph=1.8 eV. Where on the spectrum are these two energies?

At Eph=1.4 eV, α=4000 cm-1 (From Figure 11.4), and at Eph=1.8 eV,

α=30,000 cm-1

The light remaining after crossing the surface layer at 1.4 eV is

FL ( xn ) −( 4000) (0.2×10 −4 )
= e −α xn = e = 0.92 , so 8% is lost.
FL (0)

At 1.8 eV,

FL ( xn ) − ( 3×104 )( 0.2×10−4 )
= e −α xn = e = 0.55 is kept or 45% is lost.
FL (0)

Figure 11.1 shows that 1.4 eV is in the near infrared, and 1.8 eV is in the

red.

11.12. a) Explain why the PIN diode would break down first at the corners if
there were no guard ring.
The field lines are concentrated at the corners. Field lines must start and end on

charges, but if the number of positive and negative charges have to be equal,

the field lines have to be closer (higher field strength) in the corners.

b) Draw the energy band diagram for a PIN diode under high reverse bias,
and indicate the breakdown mechanism.

(Figure appears in part (c).) The breakdown mechanism is tunneling.

c) Draw the energy band diagram for a p+-p—-i-n junction under the same
reverse bias. Explain why this structure will break down at higher voltages.
 The p- region drops some of the voltage, so that the field in the intrinsic

region is reduced.

11.13. a) What should the concentration x of arsenic be in a GaAsxP1-x LED

designed to emit band to band at λ=670 nm (Figure 11.13)?

Using the Golden Rule, we find the energy corresponding to this wavelength:

1.24 1.24
E( eV) = = = 1.85eV
λ ( μm) 0.67

From Figure 11.13, we find that this

band-gap results from an As concentration

of 0.65.
b) If you also take into account the fact that the electrons are concentrated a little
above EC and the holes are concentrated a little below EV, how much does that
change the band-gap you would choose?

The peaks of the electron and holes distributions are about 1/2 kT away from

the band edges. If this adds a total of kT=0.026 eV to the emission energy, that

is still a small change and we can neglect this effect.

11.14. Recall that the peak of the electron distribution with energy is about 1/2 kT
above EC, and the peak of the holes distribution is about 1/2 kT below EV If each
distribution is approximated as having an overall width of kT, estimate the spectral
width of the emission. Assume the material emits at 1.3 μm. Hint: to find Δλ, use
E=hc/λ, and take the derivative dE/dλ to obtain an expression for Δλ in terms of
ΔE.

From the figure, we could estimate the range of possible energy

emissions is about 2kT=2x0.026 eV=ΔE. Taking the derivative as

suggested, we have
−1
(6.6 × 10 J ⋅ s)(3 × 108 m / s)Δ λ = 0.052eV
1 −34
ΔE = hc Δλ = −
λ (1.3 × 10 m)
2 −6 2

Δλ =
(0.052eV ⋅1.6 × 10 −19
J / eV )
= 71nm
−13
1.17 × 10 J/m
11.15 a) Explain why nitrogen is not a donor in GaP

Nitrogen is in Column V of the periodic table, and so is phosphorus.

Therefore if an N atom replaces a P atom in the lattice, there are no extra

electrons in the binding and thus no donor.

b) Explain why the N has no long range forces in GaP

Since the nitrogen is not a donor, it will not be ionized in the lattice.

Since it is electrically neutral it will not attract electrons, (unless the

electron gets so close as to fall into the N well).

11.16. Optical fiber manufacturers battled the OH- ion (resulting from water) for
years. These ions, when incorporated into the glass, produce a strong absorption at
1.4 μm (see Figure 11.16). They have finally managed to nearly eliminate it. Is
there a similar absorption in the earth’s atmosphere?
A look at the solar spectrum in Figure 11.8

shows almost no light at λ=1.4 μm reaches the earth, although there is

radiation at that wavelength outside the earth’s atmosphere. Thus water

vapor is a significant source of absorption in the atmosphere, as well.

11.17. What semiconductor materials can be used to produce emission at 1 eV?
Of these, are any compatible with readily available substrates (e.g. GaAs or InP)?

From Figure 11.17,

considering only direct gap materials, the possibilities are in the following

systems: InGaAs, InAsP, InGaAsP, InSbP, InAlSb. Of these the InGaAsP

quaternary compounds can be lattice-matched in InP substrates.

11.18. Find the frequency difference between the Fabry-Perot resonances of an

edge-emitting laser diode chip in which the effective index the mode sees is 3.5,
the wavelength is 900 nm, and the chip length is 100 μm. If the gain curve is 50 nm
wide, how many Fabry Perot resonances are there in this range for this diode?

We can follow the procedure in Example 11.7, but it is quicker to take the

−1
derivative: dq = 2 ndΔλ and recognize that the change in q from one mode to
λ2

the next is Δq=1. Then we have

−1
dq = 1 = 2(3.5)(100 × 10−6 m)Δ λ
(900 × 10 m)
−9 2

or
 Δλ =
(900 × 10 ) = 1.16nm
−9 2

where we have neglected the minus sign

2(3.5)(100 × 10 ) −6

because we are interested in the separation, not the direction. There are

50/1.16≈43 peaks under the gain curve.

11.19. A diode begins to lase when the gain in the cavity exceeds the losses. One
source of loss is the partially reflective mirrors at either end of the cavity. Some
percentage of the light power is lost each time the light strikes one of the mirrors.
How would the power-current curve of a laser be changed if coatings are added to
the facets to increase the reflectivity?

We would expect it takes less current for the gain to exceed the loss since the

loss is less. Thus, the threshold would be reduced.

11.20. Lasers are often characterized with an L-I-V curve, or one that plots light,
current, and voltage, like the one in Figure P-11.4 for a VCSEL. The L-I (power,
or light, vs. current) curve uses the left axis, and the V-I curve uses the right axis.
For the laser shown in the figure, what is the ratio of the optical power emitted to
the electrical power dissipated in the device, at an operating current of 10 mA?
What happens to the rest of the power?

At 10 mA, the diode emits about 5.4 mW of optical power. It consumes

P=IV=(10mA)×(3V)=30mW. Thus the “wall-plug efficiency” is 5.4/30=18%.

The rest of the power is converted to heat.

11.21. Comment on the probability of absorption (zero, low, medium, high) by a
photon of λ=600 nm (red) by the following materials:

Solution: a photon of 600 nm has a photon energy of 1.24/0.6 μm = 2.07 eV, so

if the semiconductor has a band gap larger than 2.07 eV, this photon cannot be

absorbed at all. For those whose band gap is correct, we look for indirect vs.

direct gap materials. Indirect materials have a low probability of absorption;

direct materials have a high probability.

Si: low: Si has an indirect gap, and Eg=1.12 eV<2.07 eV.

Ge: high: indirect, the band gap is small enough. However there is a direct

transition at hν = 0.8eV

GaAs: high: direct and band

gap is small enough

InAs: high, direct and band gap

is small enough
SiC: zero: band gap is too large,

GaN: zero: direct but band gap

is too high(3.5 eV)

CdS: zero: direct but band gap

too high.(2.5 eV)