1

Integrated Microelectronic Devices: Physics and Modeling

J. A. del Alamo
Pearson, 2017

Problem Solutions

Chapter 8: The Si Surface and the

Metal-Oxide-Semiconductor Structure
updated: August 25, 2017
2

Problem 8.2
Under zero bias, a MOS structure that has a flatband voltage of zero volts is, of course, at flatband! Hence,
the energy band diagram is:

The work function needed to produce this condition is:

WM = χs + qφn

with

Nc 2.9 × 1019
qφn = kT ln = 0.026 × ln = 0.15 eV
ND 1017

Hence:

WM = 4.04 + 0.15 = 4.19 eV

 3

Problem 8.3
We first need:

#ox 3.45 × 10−13

Cox = = = 3.8 × 10−7 F/cm2
xox 9 × 10−7

NA 3 × 1017
WS = χ + Eg + kT ln = 4.04 + 1.12 + 0.026 × ln = 5.04 eV
NV 3.1 × 1019

1
φbi = (WS − WM ) = 5.04 − 4.04 = 1 V
q

1 ! 1 !
γ= 2#s qNA = 2 × 1.1 × 10−12 × 1.6 × 10−19 × 3 × 1017 = 0.86 V 1/2
Cox 3.8 × 10 −7

VF B = −φbi = −1 V

kT NA 3 × 1017
φsT = 2 ln = 2 × 0.026 × ln = 0.9 V
q ni 1.0 × 1010

! √
VT = VF B + φsT + γ φsth = −1 + 0.9 + 0.86 × 0.9 = 0.72 V

This means that:

• At VG = −3 V : VG < VF B , the MOS structure is in accumulation

• At VG = 0 V : VF B < VG < VT , the MOS structure is in depletion
• At VG = 0.3 V : VF B < VG < VT , the MOS structure is in depletion
• At VG = 3 V : VG > VT , the MOS structure is in inversion

a) For VG = −3 V , φs is:

φs (VG = −3 V ) $ 0

For VG = 0 V , we first compute the extent of the depletion region:

" #
#s 4φbi 1.1 × 10−12 4×1
xd = ( 1 + 2 − 1) = ( 1+ − 1) = 4.4 × 10−6 cm = 44 nm
Cox γ 3.8 × 10 −7 0.862

Then, φs is:
4

qNA x2d 1.6 × 10−19 × 3 × 1017 × (4.4 × 10−6 )2

φs = = = 0.42 V
2#s 2 × 1.1 × 10−12

We repeat for VG = 0.3 V :

" #
#s φbi + VG 1.1 × 10−12 1 + 0.3
xd = ( 1+4 − 1) = ( 1+4 − 1) = 5.3 × 10−6 cm = 53 nm
Cox γ2 3.8 × 10−7 0.862

Then, φs is:

qNA x2d 1.6 × 10−19 × 3 × 1017 × (5.3 × 10−6 )2

φs = = = 0.61 V
2#s 2 × 1.1 × 10−12

For VG = 3 V , φs is:

φs $ φsT = 0.9 V

b) For VG = −3 V :

Qe = 0

Qh = Qs = Cox (VF B − VG ) = 3.8 × 10−7 (−1 + 3) = 7.6 × 10−7 C/cm2

For VG = 0 V :

Qe $ 0

Qh = Qs = −qNA xd = −1.6 × 10−19 × 3 × 1017 × 4.4 × 10−6 = −2.1 × 10−7 C/cm2

We repeat for VG = 0.3 V :

Qe $ 0

Qh = Qs = 1.6 × 10−19 × 3 × 1017 × 5.3 × 10−6 = −2.5 × 10−7 C/cm2

We can refine the Qe calculation by using the subthreshold regime formula:

kT q(V − VT )
Qe $ − CsT exp
q (1 + C
C )kT
sT
ox
 5

In this, CsT is:

γ 0.86
CsT $ Cox ! = 3.8 × 10−7 √ = 1.7 × 10−7 F/cm2
2 2φf 2 0.9

Then:

(0.3 − 0.72)
Qe $ −0.026 × 1.7 × 10−7 exp = −6.3 × 10−14 C/cm2
(1 + 1.7×10 )0.026
−7
3.8×10 −7

Then, the estimate of Qs above is still good.

For VG = 3 V :

Qe = −Cox (VG − VT ) = −3.8 × 10−7 (3 − 0.72) = −8.7 × 10−7 C/cm2

! !
Qh = Qdmax = − 2#s qNA φsT = − 2 × 1.05 × 10−12 × 1.6 × 10−19 × 3 × 1017 × 0.9 = −3.0×10−7 C/cm2

Qs = Qe + Qh = −8.7 × 10−7 − 3.0 × 10−7 = −1.2 × 10−6 C/cm2

c) For VG = −3 V :

Qs 7.6 × 10−7
Eox = − =− = −2.2 × 106 V /cm
#ox 3.45 × 10−13

Similarly, for VG = 0 V :

2.1 × 10−7
Eox = = 6.1 × 105 V /cm
3.45 × 10−13

For VG = 0.3 V :

2.5 × 10−7
Eox = = 7.3 × 105 V /cm
3.45 × 10−13

For VG = 3 V :

1.2 × 10−6
Eox = = 3.5 × 106 V /cm
3.45 × 10−13

d) For VG = −3 V , there is no depletion region.

For VG = 0 V and 0.3 V , it was calculated above: xd = 44nm and 53 nm, respectively.
6

For VG = 3 V ,

Qdmax 3.0 × 10−7

xdmax = − = = 6.3 × 10−6 cm = 63 nm
qNA 1.6 × 10−19 × 3 × 1017

e) For VG = −3 V :

1 1
CLF = Cox 2kT
= 3.8 × 10−7 2×0.026
= 3.7 × 10−7 F/cm2
1+ q(VF B −V )
1+ −1+3

This is a negligible difference from Cox , as expected.

For VG = 0 V :

Cox 3.8 × 10−7

CLF = $ = $ = 1.5 × 10−7 F/cm2
1 + 4 V −V
γ2
FB
1 + 4 0+1
0.862

Similarly, for VG = 0.3 V :

3.8 × 10−7
CLF = $ = 1.3 × 10−7 F/cm2
0.3+1
1 + 4 0.862

For VG = 3 V :

1 1
CLF = Cox 2kT
= 3.8 × 10−7 2×0.026 = 3.7 × 10−7 F/cm2
1+ q(V −VT )
1+ 3−0.72

A very small difference from Cox , as expected.

f ) For VG = −3, 0 and 0.3 V , CHF = CLF .
For VG = 3 V :

1 1
CHF = 1 1 = 1 1 = 1.2 × 10−7 F/cm2
Cox
+ CsT 3.8×10−7
+ 1.7×10−7

g) The threshold voltage was obtained above: VT = 0.72 V .

h) When biased in accumulation, Qs = Qa . Then:

Qa (breakdown) = −#ox Eox (breakdown) = 3.45 × 10−13 × 4 × 106 = 1.4 × 10−6 C/cm2

i) Under inversion conditions:

Qs = Qi + Qd = Qi + Qdmax = −#ox Eox

 7

Then:

Qi = −#ox Eox − Qdmax

At breakdown:

Qi (breakdown) = −#ox Eox (breakdown) − Qdmax

Qdmax was obtained above. Then:

Qi (breakdown) = −1.4 × 10−6 + 3.0 × 10−7 = −1.1 × 10−6 C/cm2

8

Problem 8.4

• Flatband voltage, VF B : ↑
Reason: WM ↑⇒ φbi ↓⇒ VF B ↑ (more positive)
• Surface potential at zero bias, φs (V = 0): ↓
Reason: φbi ↓⇒ φox ↓ and φs ↓ at zero bias
• Surface potential at threshold, φsth: no effect
Reason: φsth is only a function of the doping level in the semiconductor
• High-frequency capacitance in inversion, CHF (inv): no effect
Reason:

1
CHF = 1 1
Cox + Csth

Csth is only a function of the doping level in the semiconductor. Hence CHF is independent of WM .
• NA to maintain VT unaffected: ↓
Reason: VF B ↑⇒ VT ↑. To get VT ↓, we need to reduce the doping level in the semiconductor.
 9

Problem 8.6
Ia) With a p+ filling, the band diagram along the indicated cross section looks like:

Since the Fermi level in the p+ filling is right at the valence band edge, the substrate on both sides right
next to the trench is in accumulation. Inside the trench, there are two depletion regions on both sides.
However, since the doping level is high, their extent is negligible (they have been exaggerated in the figure).
Ib) With the substrate in accumulation, the low-frequency capacitance is basically that due to the oxide
sidewall. There are two sidewalls in series, hence:

1 #ox 1 3.45 × 10−13

C$ = = 2.2 × 10−8 F/cm2
2 xox 2 80 × 10−7

Ic) The built-in potential of each of the substrate-poly MOS structures is, to first order:

kT Nv 3.1 × 1019
φbi $ ln = 0.026 ln = 0.114 V
q NA 2 × 1016

To first order, this field drops across the oxide liner. Hence, the electric field in the liner is:

φbi 0.114
Eox $ = = 1.4 × 104 V /cm
xox 80 × 10−7
10

IIa) With a n+ filling, the band diagram along the indicated cross section looks like:

Since the Fermi level in the n+ filling is right at the conduction band edge, the substrate on both sides
right next to the trench is in depletion. Inside the trench, the sidewalls of the filling are in depletion too.
However, since the doping level is high, their extent is negligible (they have been exaggerated in the figure).
IIb) This is a simple MOS structure as studied in class. The built-in potential is:

Eg kT Nv 3.1 × 1019
φbi = −VF B $ − ln = 1.1 − 0.026 ln = 0.97 V
q q NA 2 × 1017

The capacitance of one of the MOS structures is:

Cox
C$ $
1 + 4 −VγF2 B

with γ given by:

1 ! xox ! 80 × 10−7 !
γ= 2#s qNA = 2#s qNA = 2 × 1.05 × 10−12 × 1.6 × 10−19 × 2 × 1017 = 6.0 V 1/2
Cox #ox 3.45 × 10−13

Then, the capacitance per sidewall is:

4.3 × 10−8
C$ $ = 4.1 × 10−8 F/cm2
0.97
1 + 4 6.02

Since there are two liners in series, the overall capacitance is half of this:

C = 2.0 × 10−8 F/cm2

 11

IIc) The electric field across the oxide in a standard MOS structure in depletion at zero bias is given by:

" #
qNA xox −VF B 1.6 × 10−19 × 2 × 1017 × 80 × 10−7 0.97
Eox = [ 1 + 4 2 −1] = [ 1 + 4 2 −1] = 1.2×105 V /cm
#s γ 1.05 × 10 −12 6.0
12

Problem 8.8
a) A negative current with the indicated sign deposits negative charge on the gate. Hence, the MOS
structure is driven into accumulation. It will never reach inversion.

b) This sign of current will eventually get the structure into inversion. Let us compute the charge in the
depletion region at threshold, Qdmax :

!
Qdmax = − 2#s qNA φsth

for this we need:

kT NA 6 × 1017
φsth = 2 ln = 2 × −0.026 ln = 0.93 V
q ni 1.0 × 1010

Then:
!
Qdmax = − 2 × 1.05 × 10−12 × 1.6 × 1019 × 6 × 1017 times0.93 = −4.3 × 10−7 C/cm2

The amount of charge that needs to be supplied by the current source is:

∆Q = A(Qdmax − Qd ) = 10 × (10−4 )2 (−4.3 × 10−7 + 3.4 × 10−7 ) = −9 × 10−15 C

From this current source, this is going to take:

∆Q 9 × 10−15
∆t = = = 9 × 10−8 s = 90 ns
I 100 × 10−9

However, this is much shorter than τg . Hence, electrons don’t have time to be generated. It will then take
about τg ∼ 1 µs for the inversion layer to appear!

c) This current is 100 times smaller than that of case b. Hence it will take 100 times longer for the required
amount of charge to be delivered to the structure, that is, about 9 µ s. This is now sufficiently longer thatn
τg . Hence, the electrons have ample time to be generated and it will take about 9 µ s for the inversion
layer to appear.
 13

Problem 8.10
a) The flatband voltage is given by:

kT Nv kT Nv
VF B = −φbi = WM − WS = χs − (χs + Eg − ln ) = −Eg + ln
q NA q NA

or

3.1 × 1019
VF B = −1.12 + 0.026 ln = −0.97 V
1017

b) At threshold, xd $ xdmax which is given by:

"
2#s φsth
xmax =
qNA

For this, we need φsth :

kT NA 1017
φsth = 2 ln = 2 × 0.026 × ln = 0.83 V
q ni 1.1 × 1010

Then:
#
2 × 1.05 × 10−12 × 0.83
xmax = = 0.1 µm
1.6 × 10−19 × 1017

c) From Gauss law:

Qs (V = Vth ) Qdmax qNA xdmax

Eox (V = Vth ) = − =− =
#ox #ox #ox

Then:

1.6 × 10−19 × 1017 × 0.1 × 10−4

Eox (V = Vth ) = = 4.6 × 105 V /cm
3.45 × 10−13

d) Using Gauss’ law again:

Qs Qi + Qdmax
Eox = − =−
#ox #ox

Solving for Qi :

Qi = −#ox Eox − Qdmax = −3.5 × 10−13 × 106 + 1.6 × 10−7 = −1.9 × 10−7 C/cm2
14

e) In accumulation, Gauss’ law is:

Qs Qa
Eox = − =−
#ox #ox

Then:

Qa = −#ox Eox = 3.5 × 10−13 × 106 = 3.5 × 10−7 C/cm2

 15

Problem 8.11
a) For t < 0, the MOS is in inversion and

1
C= 1 1
Cox + Csth

At t = 0, the device is switched further into inversion. Since the required additional electron concentration
cannot be generated immediately, the depletion region has to widen past xdmax , hence the capacitance
drops. Given enough time, the electrons are generated and xd collapses back to xdmax and C goes back
to the value that is had for t = 0− .

b) At t = 0−, we have a typical calculation of high-frequency inversion capacitance:

1 1 1
= +
C(t = 0− ) Cox Csth

Cox is

#ox 3.5 × 10−13

Cox = = = 3.5 × 10−7 F/cm2
xox 10 × 10−7

Csth is:
" #
qNA #s 1.6 × 10−19 × 1017 × 10−12
Csth = = = 9.8 × 10−8 F/cm2
2φsth 2 × 0.83

with

kT NA 1017
φsth = 2 ln = 2 × 0.026 × ln = 0.83 V
q ni 1.1 × 1010

All together:

1
C(t = 0− ) = 1 1 = 7.7 × 10−8 F/cm2
3.5×10−7
+ 9.8×10−8
16

c) From t = 0− to t = 0+ , the inversion charge Qi does not change. The potential build up associated
with this is:

φinv = V1 − Vth = 2 − 0.33 = 1.67 V

This implies that at t = 0+ , the potential build-up associated with the depletion layer is

φdepl (t = 0+ ) = V2 − φinv = 5 − 1.67 = 3.33 V

This demands a depletion region equal to:

" #
+ #s φdepl (t = 0+ ) − VF B 1.05 × 10−12 3.33 + 0.97
xd (t = 0 ) = [ 1+4 −1] = [ 1+4 −1] = 2.2×10−5 V /cm
Cox γ2 3.5 × 10−7 0.512

Then:

#s 1.05 × 10−12
Cs (t = 0+ ) = +
= = 4.8 × 10−8 F/cm2
xd (t = 0 ) 2.2 × 10−5

and

1 1
C(t = 0+ ) = 1 1 = 1 1 = 4.2 × 10−8 F/cm2
Cox
+ Cs (t=0+ ) 3.5×10−7
+ 4.8×10−8

d) For t → ∞, the capacitance should recover to the value it had at t = 0− :

C(t → ∞) = C(t = 0− ) = 7.7 × 10−8 F/cm2

 17

Problem 8.14
From problem 8.13, we know that this structure at t = 0 is in depletion. The sign of the current generator
that is turned on at t = 0 is such that the gate is made positive with respect to the body and the source.
Hence, this structure, as time goes on, gets biased deeper into depletion and eventually goes into inversion.
In an initial phase, then, as t increases, the semiconductor goes further into depletion. The depletion
region widens. This pushes holes out of the body through the body contact. In this phase, there is no
inversion layer and then IS = 0. Hence IB = −IG = −10 nA, independent of time.
As time goes on, the structure reaches threshold and eventually inversion. When this happens, an inversion
layer is formed and the extent of the depletion region maxes out. From threshold on, all the extra charge on
the semiconductor goes into the inversion layer through the source contact. The depletion region thickness
doesn’t change anymore. Hence in this regime, IS = −IG = −10 nA and IB = 0.
So, what we have left to do is to estimate the time that it takes from this structure to reach threshold.
This is easy. At threshold the surface potential is:

kT NA 1017
φsth $ 2φf = 2 ln = 2 × 0.026 ln 10 = 0.84 V
q ni 10

The depletion charge at threshold is:

!
Qdmax $ −AqNA xdmax = −A 2#s qNA φsth
!
= −10 × 10−8 2 × 1.1 × 10−12 × 1.6 × 10−19 × 1017 × 0.84 = −1.7 × 10−14 C

From the previous problem, we can compute the charge at t = 0:

Qd $ −AqNA xd = −10 × 10−8 × 1.6 × 10−19 × 107 × 8.7 × 10−6 = −1.4 × 10−14 C

The difference of this charge has to be supplied by the current generator. This takes a time:

|Qdmax − Qd | 1.7 × 10−14 − 1.4 × 10−14

t= = = 0.3 µs
IG 10 × 10−9

Hence, the current through the terminals as a function of time looks like: