Integrated Microelectronic Devices: Physics and Modeling

J. A. del Alamo
Pearson, 2017

Problem Solutions
c Pearson 2018

Chapter 3: Carrier Generation and Recombination

updated: January 10, 2018

INSTRUCTORS:
Please do not distribute these solutions to students in elec-
tronic form!
5. Problem 3.1
We have a p-Si with

We are to assume that dominant trap state in this ample is located at and concentration of traps

. Furthermore, we are told and

Since we are given excess hole and electron conentrations, the situation is clearly in outside thermal
equilibrium. We assume that we are at room temperature and that we are in the non-degenerate
regime

(i) band-to-band optical process generation and recombination

events [pg. 98]
First, we need to calculate what equilibrum electron and hole concentrations would be at thermal
equilibrium

= , . Since , =

Then, the optical generation rate is

The optical recombination rate is
We can see that the recombination rate far exceeds the generation rate. This makes sense since we
have excess carriers.

(ii) Auger process by hot electrons and hot holes [pg.99]

hot electrons
Generation from hot electrons can be found by isolating the generation term in (3.34)

Recombination can also be found from (3.34),

hot holes
Similarly, we only need to look at (3.35) and figure out which terms belong to generation from
hot holes
Generation should be the one associated with since it should only be dependent on
how many holes there are currently. Note that the and the term came from substituting for
so we only have to worry about one constant, .
Recombination on the otherhand is dependent on how many holes are available to absorb the
energy from the recombination and also how many electron and holes there are to recombine.
 Therefore, it is the term.
Generation:

Recombination:
Again, we see that the recombination rate exceeds the generation rate

(iii) Trap assisted processes [pg. 99]

We need the value of , the concentration of occupied traps to solve for the different rates. We
can do so using (3.41) [pg. 100]

is about 3 orders of magnitudes smaller than , an expected result since we have a p-type
silicon so the Fermi level is below the mid-bandgap
rate of electron capture

rate of electron emission

rate of hole capture

rate of hole emission

Solutions courtesy of Ethan Lee, MIT, Sept. 2017

Problem 3.2
a) Right before the radiation is turned off, the excess carrier concentration is:

n0 (t = 0− ) = p0 (t = 0− ) = gl τ = 1020 × 2 × 10−5 = 2 × 1015 cm−3

Since n0  NA , this is a low-level injection situation. Hence,

n(t = 0− ) ' n0 (t = 0− ) = 2 × 1015 cm−3

and

p(t = 0− ) ' NA = 1017 cm−3

The net recombination rate is:

n0 (t = 0− ) 2 × 1015
U (t = 0− ) = = = 1020 cm−3 · s−1
τ 2 × 10−5

We could have written this result right away since in steady state, U = gl = 1020 cm−3 · s−1 .
b) Since the carrier lifetime in this sample is 20 µs, just after 100 ns after the light is turned off the excess
carrier concentration has not had any time to change in an appreciable way. Hence,

n0 (t = 100 ns) = p0 (t = 100 ns) ' n0 (t = 0− ) = 2 × 1015 cm−3

The carrier concentrations are:

n(t = 100 ns) ' n0 (t = 0−) = 2 × 1015 cm−3

and

p(t = 100 ns) ' NA = 1017 cm−3

The net recombination rate is also as in part a):

n0 (t = 100 ns) 2 × 1015

U (t = 100 ns) = = = 1020 cm−3 · s−1
τ 2 × 10−5

c) 50 µs after the light is turned off, the excess carrier concentration is:

/2×10−5
n0 (t = 50 µs) = gl τ e−t/τ = 2 × 1015 e−5×10 = 1.6 × 1014 cm−3
−5

The carrier concentrations therefore are:

 n(t = 50 µs) ' n0 (t = 50µs) = 1.6 × 1014 cm−3

and

p(t = 50 µs) ' NA = 1017 cm−3

The net recombination rate is:

n0 (t = 50 µs) 1.6 × 1014

U (t = 50 µs) = = = 8.2 × 1018 cm−3 · s−1
τ 2 × 10−5

d) After steady-state is reached, the sample is in thermal equilibrium. Hence n0 (t → ∞) = 0 and U (t →

∞) = 0. The carrier concentrations therefore are:

n2i 1.1 × 1020

n(t → ∞) = = = 1.1 × 103 cm−3
NA 1017

and

p(t → ∞) = NA = 1017 cm−3

Problem 3.4
a) First, for a sample with this doping level, no ' 1018 cm−3 and po ' 102 cm−3 .
We can now easily estimate the recombination rate in thermal equilibrium due to the radiative process.

Go,rad = Ro,rad = rrad no po = 2 × 10−15 × 1018 × 102 = 2 × 105 cm−3 · s−1

The generation and recombination rates due to the two Auger processes are also easy to estimate:

Go,eeh = Ro,eeh = reeh n2o po = 1.8 × 10−31 × 1036 × 102 = 1.8 × 107 cm−3 · s−1

Go,ehh = Ro,ehh = rehh no p2o = 9.5 × 10−32 × 1018 × 104 = 9.8 × 10−10 cm−3 · s−1

The Auger process involving hot electrons is vastly more likely than that involving hot holes because this
is n-type material.
We can estimate the rate of generation and recombination due to the trap assisted process if we can
estimate the lifetime due to this process. We can do this because we know the lifetime of the sample and
we can estimate the lifetime due to the other processes. Of all other processes, only the Auger process
involving hot electrons matters. The lifetime associated with this is:

1 1
τeeh = = = 5.6 × 10−6 s
reeh n2o 1.8 × 10−31 × 1036

The trap assisted lifetime is therefore:

1 1
τtr = 1 1 = 1 1 = 1.2 × 10−6 s
τ
− τAuger 10−6
− 5.6×10−6

In an n-type sample τtr ' τho . From this, we can workout the rate of hole emission and hole capture (but
nothing more):

po 102
ro,hc = ro,he = = = 8.3 × 107 cm−3 · s−1
τho 1.2 × 10−6

The rates of electron emission and electron capture are orders of magnitude above this figure. We cannot
estimate it since we do not know τeo .
b) The carrier lifetime associated with the radiated process is given by:

1 1
τrad = = = 5 × 10−4 s
rrad no 2 × 10−15 × 1018

With n0 = p0 = 1016 cm−3 , we are in a low-level injection condition. Hence, the net recombination rate
due to the radiative process is given by:
 n0 1016
U = = = 2 × 1019 cm−3 · s−1
τrad 5 × 10−4

We computed above the carrier lifetime due to the hot-electron-assisted Auger process. The net recombi-
nation rate due to this process is then:

n0 1016
Ueeh = = = 1.8 × 1021 cm−3 · s−1
τeeh 5.6 × 10−6

The carrier lifetime due to the hot-hole-assisted Auger proces is:

1 1
τehh = = = 1.1 × 1027 s
rehh p2o 9.5 × 10−32 × 104

The net recombination rate due to this process is:

n0 1016
Uehh = = = 9.5 × 10−12 cm−3 · s−1
τehh 1.1 × 1027

which, of course, is extremely rare since there are very few holes here to assist in this process.
Finally, the net recombination rate due to the trap-assisted process is:

n0 1016
Utr = = = 8.3 × 1021 cm−3 · s−1
τtr 1.2 × 10−6

This is the most likely of all processes because it is characterized by the smallest lifetime.
Problem 3.5
(a) ( )
( ) ( )

( ) ( )

And in general, we have:

( ) ( )( )
(( ) ) ( )( )

( )∑

( ) (( ) )

( ) ∑

( )

( )

In the case of , ( )

In the case of

( ) ( )( )∑( )

( )
( )[ ]

(b) For simplicity, let the constant ( ) , then

( ) ∑

And n’(mT) converges to

 ( ) ∑

( )
So, , and the stead state is established after about
pulses. In the case of , stead state is obtained after the first pulse, as expected
from Figu. 3.24 in the notes.

Solutions courtesy of Wenjie Lu, MIT, Sept. 2013

Problem 3.9
The maximum lifetime that can be obtained on a semiconductor occurs when the sample is completely
free of traps. In this case, the lifetime is set by the internal recombination mechanisms. For an n-type
sample, the electron-assisted Auger process is dominant. For this smaple, the lifetime is given by:

1 1
τeeh = = = 5.6 s
reeh n2o 1.8 × 10−31 × 1030

This is very long, so we should also check the lifetime determined by the radiative process:

1 1
τrad = = = 0.5 s
rrad no 2 × 10−15 × 1015

This is bigger, so the Auger lifetime dominates and the maximum lifetime that is to be expected is:

τ ' τeeh = 0.46 s

Problem 3.11
a) The differential equation governing this system is:

d n0 n0
= Gext − (1)
dt τ

The homogeneous solution is:

n0h (t) = Ae−t/τ (2)

In this case Gext (t) = g0 + g1 sin ωt. Therefore we try a particular solution of the form:

n0p (t) = B sin ωt + C cos ωt + D (3)

We plug this into the differential equation above and get:

1
Bω cos ωt − Cω sin ωt = g0 + g1 sin ωt − (B sin ωt + C cos ωt + D) (4)
τ

We now identify terms and solve for the unknowns:

g1 τ
B = (5)
1 + ω2 τ 2
ωτ
C = −g1 τ (6)
1 + ω2 τ 2
D = go τ (7)

We match initial condition n0 (t = 0) = n0h (t = 0) + n0p (t = 0) = 0 and get:

ωτ
A = g1 τ − go τ (8)
1 + ω2 τ 2

All together, then, for t ≥ 0:

ωτ g1 τ
n0 (t) = (g1 τ − go τ )e−t/τ + (sin ωt − ωτ cos ωt) + go τ (9)
1 + ω2 τ 2 1 + ω2 τ 2

This can be rewritten as:

ωτ g1 τ
n0 (t) = (g1 τ 2 2
− go τ )e−t/τ + √ sin(ωt − φ) + go τ (10)
1+ω τ 1 + ω2 τ 2

with:

tan φ = ωτ (11)
Following are graphs of the function with g0 = 1, g1 = 5, ω = 3 and τ equal to 1, 3 and 5, respectively.
The solution has the form of a sinusoidal signal that rides on top of a rising exponential.
b) The solution achieves steady state when the initial transient has died off. This requires that t  τ . In
this case, the solution is:

g1 τ
n0 (t  τ ) ' go τ + √ sin(ωt − φ) (12)
1 + ω2 τ 2

c) For the solution to be considered quasi-static, we need it to achieve steady state and that the phase
shift φ is very small. This requires that t  τ and ωτ  1. In this case, the solution becomes:

n0 (t  τ ) ' (go + g1 sin ωt)τ (13)

This is a situation in which the frequency of the light intensity waveform is low enough in the scale of the
inverse of the carrier lifetime for the carrier concentration to closely track the illumination.
 Problem 3.15

For n-Si at room temperature with ND = 1018 cm-3, τ = 1.0 µs, according to Fig 3.17.

(a) As , so at ( )
Since ( ) , it is LLI, therefore
( )
( ) ( )
And the net recombination rate at t = T,
( )

(b) The excess electron concentration after 500 ns after the light pulse is
( ) ( )

Therefore
( )
( )
( )

Solutions courtesy of Wenjie Lu, MIT, Sept. 2013

 Problem 3.16
We need to calculate the minimum carrier generation rate required to drive Si into high-level
injection, i.e. . We want to do this with a HeNe laser . We are told the
sample as a thickness of . Since we are at high-level injection,

We cannot start with the carrier dynamics equation,

since the term is only valid in the low-level injection condition. We must find what the

equivalent term would be for high-level injection.

We face a problem here because we

don't know the trap concentrations so we can't calculate and .

However, we can see that for high-level injection Auger recombination processes will dominate due
to the cubic dependence on the excess carrier concentration. Then, we have

We want to shine the HeNe laser such that this excess carrier concentration is maintained. This
happens when . We thus have

Since each photon generates a pair of electron and hole, essentially equals the density of photons
that need to be absorbed per unit volume per second to maintain this excess carrier level.
With a laser , each photon has energy

The sample is 0.1 so photon flux is

Then, the power density is simply photon flux times the energy of each photon:

Solutions courtesy of Ethan Lee, MIT, Sept. 2017