Homework #1
Due on May. 5, 2019
1. (a) When electrons of conduction band in a silicon sample accumulate high enough
energy, some of them may acquire enough energy to be emitted out of the silicon
into the air. Can this also happen to holes?
(b) Similar argument can also be applied to a contact between metal and silicon.
Likewise, some of electrons may acquire enough energy to be emitted out of the
silicon into the metal. Can this also happen to holes?
Solution:
(a) No. To emit a hole at this energy, an electron with energy much lower than the
vacuum level must exist to replace the hole. Since there are no electrons with
energy much lower than the vacuum level, thus hot holes can not be emitted out
of the silicon into the air.
(b) Yes. A hot hole located at the energy far below the valence band edge (i.e. far
below the Fermi level). At this energy range, there are plenty of electrons to
exchange position with the hot hole. Thus, hot holes can be emitted out of the
silicon into the metal.
2. The density of the silicon crystal is 2.328 g/cm3, the lattice constant of the silicon
crystal is 5.43 Å (a) Find the atomic density of the silicon crystal (b) How many
silicon atoms per unit cell are there in the silicon crystal.
Solution:
(a) The atomic density of the silicon crystal is
2.328
6.02 10 23 5 10 22 Si/cm3.
28.09
(b) Silicon atoms per unit cell are:
5 1022 Si/cm3 (5.43 10-8)3 cm3/cell = 8 atoms/cell
3. The following resistance measurement were made on a silicon bar:
Voltage applied (V) 90 (mV) 200 (mV) 1 (V)
Current measured (I) 50 (A) 100 (A) 250 (A)
Find the corresponding resistance. What do you conclude about the dependence of
the resistance of this silicon bar on the applied voltage? Explain why ?
Solution:
Voltage applied (V) 90 mV 200 mV 1V
Current measured (I) 50 A 100 A 250 A
Resistance (V/I) 1.8 k 2 k 4 k
The resistance goes up as the impressed voltage is increased. Note that the ohm’s
law is another statement of the mobility relation vdrift E . And the linear vdrift
versus E relation holds for the small electric fields (or low bias level) only. Beyond
E = 2 104 V/cm, vdrift levels off due to the velocity saturation and the resistance of
the silicon bar increased.
4. Argue why the carrier mobility is meaningless if the carrier is moving in vacuum.
Solution:
In semiconductor, the carrier undergo many scatterings and the velocity of
the carrier is the average velocity of final velocity for each scattering. As a result,
qE qE
v v0 t t E where is the mobility. In vacuum, the carrier
m m
moves without sufferring any scattering and the velocity and position of the carrier
qE
is governing by the mechanical laws such as v v0 t . Since there is no
m
scattering, thus the mobility is meaningless.
5. For an abrupt n+-p silicon diode, the n+ doping is 1 1020 cm-3 and the p-type doping
is 3 1016 cm-3.
(a) Draw the energy band diagram at zero bias. Indicate x = 0 as the boundary where
the doping change from n+ to p. Also indicate where the Fermi level is with
respect to the midgap.
(b) Write the equation and calculate the built-in potential.
(c) Write the equation and calculate the depletion width.
(d) Will the built-in potential increase or decrease if the temperature goes up?
Explain why.
Solution:
(a)
kT p 3 1016
For Ei E f ln p 0.0259 ln 0.376(eV ) ,
10
2 ni 1.5 10
kT nn 1 10 20
E f Ei ln 0.0259 ln 0.586(eV )
10
2 ni 1.5 10
x= 0 N+
P
Ec
0.376eV
Ei
Ef
Ev
(b)
The equation of the built-in potential:
Eg
bi 0.38 0.93(eV ) .
2q
(c)
The equation of the depletion width:
Wd
2 si bi
2 11.8 8.854 10 14 0.946
2.03 10 5 (cm)
qN a 1.6 10 19 3 1016
(d) The ni will increase with the temperature exponentially, i.e.
3
2kT 2 E
3
ni 2 2 mn* m*p 4 exp g .
h 2kT
Since the built-in potential follows the equation as
kT N a N d
bi ln ,
q ni 2
thus the value of the built-in potential will decrease if the temperature goes higher.
6. The figure below shows the total charge per unit area in the P-type Si as a function
of Vg for an MOS capacitor at 300K.
(a) What is the oxide thickness?
(b) What is the doping concentration in Si?
(c) Find the potential drop in oxide (ox) when Vg – Vfb = -1 (V).
(d) Find the band bending in Si when Vg – Vfb = 0.5 (V).
Solution:
(a) At Vg - Vfb = -1 (V), the MOS capacitor is in accumulation.
4 10 7 ox 13
tox ox 7 3.45 10 7 8.62 10 7 (cm)
Qs
Cox
ox 1 tox 4 10 4 10
tox = 8.62 (nm)
(b) At threshold,
2 si qN a 2B 2(1 1012 )(1.6 1019 ) N a 2B
Vth V fb 2B or 1 2B
Cox 4 10 7
(1 2B ) 2 (1 2B ) 2
1 2B 2 10 9 N a 2B Na 18
1018
2 10 2B 2 2B
kT N a or Na
where 2B 2 n 2B 0.12 log 10
q ni 1 10
(1 2B ) 2
Na 1018
2 2B
Na
2B 0.12 log 10
1 10
Solve the above two equations by iteration, we have Na = 3.1 1016 cm-3 .
(c) At Vg - Vfb = -1 (V), the MOS capacitor is in accumulation and si 0.
Since Vg V fb ox si , thus 1 ox 0 i.e. ox = -1 (V)
(d) when Vg – Vfb = 0.5 (V), the MOS capacitor is in depletion.
In the depletion region, we have the following relation
2 si qN aSi
Vg V fb si 2
Cox
si qN a si qN a
si 2
2
(Vg V fb )
2Cox 2Cox
si qN a (1 1012 )(1.6 1019 )(3.1 1016 )
0.0155 (V)
2
2Cox 2 ( 4 10 7 ) 2
si 0.0155 0.0155 0.5 si = 0.352 (V).
7. Consider a n+poly-gate MOS capacitor of area 10-4 m2 at room temperature. The
silicon substrate is uniformly doped with ND = 1 1016 cm-3, the oxide thickness
tox = 10 nm, and Qf = 1011 q/cm2. (Use ni = 1 1010 cm-3 and 2.3kT/q = 60 mV)
(a) Calculate the flatband voltage Vfb.
(b) Find the threshold voltage Vth.
(c) Sketch the high frequency C-V curve for this capacitor, indicating values for
Cmax, Cmin, Cfb, Vth, and Vfb on the sketch.
(d) Draw the energy band diagram of this MOS capacitor at Vg = Vth, indicating
the amount of band bending (i.e. the potential drop) in silicon (qSi) and across
the oxide (qox).
Solution:
ox 3.45 1013
Cox 7
3.45 10 7 (F/cm2)
tox 10 10
(a) m si
Eg ND 1 1016
m si
2
0.06 log
n si 0.55 0.06 log 1 1010 si 0.19
i
Qf 1.6 1019 1011
0.046 (V)
Cox 3.45 10 7
Qf
V fb m si si ( si 0.19) 0.046 0.24 (V)
Cox
2 si qN D 2B
(b) This is a PMOSFET and Vth V fb 2B
Cox
kT ni 1 1010
2B 2 n 0.12 log 0.72 (V)
16
q ND 1 10
2 1 1012 1.6 1019 1 1016 0.72
Vth 0.24 0.72 0.24 0.72 0.14
3.45 10 7
Vth 1.1 (V)
(c) V fb 0.24 (V) , Vth 1.1 (V),
Cmax Cox 3.45 10 7 (F/cm2)
Cox A 3.45 107 104 10 8 3.45 10 19 (F)
Cmax
1 1 1 1 Wd max
Cmin Cox Csi Cox si
2 Si 2B ( 2)(1 10 12 )(0.72)
Wd max 19 16
3 10 5 (cm)
qN D (1.6 10 )(1 10 )
1 1012
si
Csi 5
3.3 10 8 (F/cm2)
Wd max 3 10
1 1 1 1 1
7
Cmin 3 108 (F/cm2)
Cmin Cox Csi 3.45 10 3.3 108
3 108 1012 3 1020
i.e. Cmin (F)
1 1 1 1 L
D
C fb Cox C si Cox si
kT 1 10 12
LD 19 16
0.026 4 10 6 (cm)
qN d q 1.6 10 1 10
1 1012
si
Csi 6
2.5 10 7 (F/cm2)
LD 4 10
1 1 1 1 1 C fb 1.45 10 7 (F/cm2)
7
C fb Cox C si 3.45 10 2.5 10 7
i.e. C fb 1.45 10 7 10 12 1.45 10 19 (F)
C
Cmax=3.45E-7(F/cm2), C’max=3.45E-19(F/cm2)
Cfb=1.45E-7(F/cm2), C’fb=1.45E-19(F/cm2)
Cmin=3E-8(F/cm2), C’min=3E-20(F/cm2)
Vg
-1.1V -0.24V
(d) At threshold, Vth V fb ox si with V fb 0.24 (V) , Vth 1.1 (V), and si
= 2B= - 0.72 (V). 1.1 0.24 ox 0.72 = -0.14 (V).
qφOx = 0.14 eV
10nm
Efm Ec
qφSi = 0.72 eV
Ev EfSi
0.36eV
Ec Ei
Ev
300nm
8. Consider the MOS capacitor C-V curve below. The area of the capacitor is 104 µm2.
(a) Is the substrate doping N-type or P-type. What is the doping concentration?
(b) What is the thickness of the oxide in the MOS capacitor?
(c) What is the value of the capacitance at position C on the C-V curve shown above?
(d) At location B on the C-V curve, what is the silicon band bending Si?
Solutions
Cox 50 10 12
Cox 4 8
5 10 7 (F/cm2)
A 10 10
(a) The substrate doping is P-type since the threshold voltage Vth is larger than Vfb.
2 si qN a 2B
Vth V fb 2B
Cox
2(1 1012 )(1.6 1019 ) N a 2B
0.5 ( 1) 2B
5 10 7
1.5 2B 1.13 10 9 N a 2B
(1.5 2B ) 2 (1.5 2B ) 2
Na 18
1018
1.13 10 2B
2
1.28 2B
kT N a or Na
where 2B 2 n 2B 0.12 log 10
q ni 1 10
(1.5 2B ) 2
Na 1018
1.28 2B
Na
2B 0.12 log 10
1 10
Solve the above two equations by iteration, we have Na = 3.1 1017 cm-3 .
ox 3.45 1013 3.45 1013
(b) Cox 5 10 7 (F/cm2) tox = 6.9 (nm)
tox Cox 5 10 7
(c) At position C, MOSC is at the threshold and has a minimum capacitance
1 1 1 1 Wd max
Cmin Cox Csi Cox si
3.1 1017
2B 0.12 log 10
0.9 (V)
1 10
2 Si 2B ( 2)(1 10 12 )(0.9)
Wd max 19 17
6 10 6 (cm)
qN a (1.6 10 )(3.1 10 )
si1 1012
Csi 6
1.67 10 7 (F/cm2)
Wd max 6 10
1 1 1 1 1
7
7
Cmin 1.25 107 (F/cm2)
Cmin Cox Csi 5 10 1.67 10
1.25 107 104 108 1.25 1011
i.e. Cmin (F) = 12.5 (pF)
(d) At point B, Vg = 0 (V). From the figure, Vfb = -1 (V)
In the depletion region, we have the following relation
2 si qN aSi
Vg V fb si 2
Cox
si qN a si qN a
si 2
2
(Vg V fb )
2Cox 2Cox
si qN a (1 1012 )(1.6 1019 )(3.1 1017 )
0.0992 (V)
2
2Cox 2 (5 10 7 ) 2
si 0.0992 0.0992 [0 ( 1)] si = 0.538 (V).
9. The figure below shows the output characteristics of a long channel MOSFET at
300K. The channel length L = 2 (m) and the electrical oxide thickness tox = 100
(Å). The VDsat for these two curves are 1 (V) and 3 (V). Assume the body effect
coefficient m = 1.2.
(a) What is the threshold voltage VTh?
(b) By using the universal curve of the mobility in the lecture note, find the effective
electron surface mobility at VGS = 4 (V).
(c) From the value of IDsat on the curve and the results of part (a) and (b), find the
channel width W of this MOSFET.
(d) Find the channel length L below which the velocity saturation effect is
significant.
Solutions:
VGS VTh
(a) VDsat VTh VGS mVDsat
m
From the ID-VDS curve for VGS = 4 (V), we have VDsat = 3 (V). VTh = 4 – (1.2)(3)
VTh = 0.4 (V).
From the ID-VDS curve for VGS = 1.6 (V), we have VDsat = 1 (V). VTh = 1.6 – (1.2)(1)
VTh = 0.4 (V).
VTh = 0.4 (V)
VTh VFB 2B VGS VTh VGS VTh VFB 2B
(b) Eeff . For a n+poly gate
3tox 6tox 6tox
Eg Eg
MOST, VFB 2B B 2B B 0.13 (V)
2 2
Eeff VGS VTh 0.2 4 0.4 0.13
8
7.6 105 (V/cm) = 0.76 (MV/cm)
6tox 6(100 10 )
From the universal mobility graph,
we have n 300 (cm2/Vs).
(c) From the ID-VDS curve for VGS = 4 (V), IDsat = 2.8 (mA).
1W (V V ) 2 2mLI Dsat
I Dsat eff Cox GS Th W
2 L m eff Cox (VGS VTh ) 2
ox 3.45 1013
Cox 8
3.45 10 7 (F/cm2)
tox 100 10
2 1.2 ( 2 104 ) (2.8 103 )
W 7 2
1 10 3 (cm), or W = 10 (m).
300 (3.45 10 )( 4 0.4)
(d) If EsatL (VGS - VTh)/m, then the velocity saturation effect is significant.
V VTh
Thus, L GS , where Esat 2vsat
mE sat eff
Since the mobility μeff is gate dependent, thus we split the solution into two gate
biase cases, i.e. VGS = 4 V and VGS = 1.6 V.
For VGS = 4 V, we have μeff = 300 cm2/Vs, thus
2vsat 2 (1 107 )
Esat 6.7 104 (V/cm)
eff 300
VGS VTH 4 0.4
L 4.5 105 (cm) ,or L 0.45 (μm)
mEsat 1.2 (6.7 10 )
4
For VGS = 1.6 V, let’s find the corresponding mobility first.
VGS VTH 0.2 1.6 0.4 0.13
Eeff 3.6 105 (MV/cm) 0.36 (MV/cm)
6tox 6 (100 108 )
From the universal mobility graph, we have n 475 (cm 2 / Vs ) .
2vsat 2 (1 107 )
Esat 4.2 10 4 (V / cm)
eff 475
VGS VTH 1.6 0.4
L 2.38 10 5 (cm) ,or L 0.238( m)
mE sat 1.2 (4.2 10 )
4
10. You are working with two novel semiconductors. Material “A” has a saturation
velocity of 1 107 (cm/s) and a mobility of 500 (cm2/Vs). Material “B” has a
saturation velocity of 1 106 (cm/s) and a mobility of 2000 (cm2/Vs).
(a) What is the saturation field for the two materials Esat?
(b) Suppose the long-channel IDsat for a MOSFET made with material “A” is
400µA, what is the long channel IDsat for a MOSFET made with material “B”?
(c) What are the IDsat’s for short-channel 0.2 (µm) MOSFETs fabricated with the
two materials given a supply voltage of 2.5V?
(d) Based on your results in (b) and (c) above, comment on the relative
importance of mobility and saturation velocity in long and short channel
MOSFETs, EXPLAIN. Consider the effect of the regions of operation, i.e.,
linear vs. saturation.
Solutions:
2vsat , A (2)(1 107 )
(a) Esat , A 4 104 (V/cm).
eff , A 500
2vsat , B (2)(1 106 )
Esat , B 1 103 (V/cm).
eff , B 2000
(b) For long channel devices, IDsat eff.
eff , B 2000
I Dsat , B I Dsat , A 400 1600 (A) = 1.6 (mA)
eff , A 500
I Dsat ,long
(c) Note that I Dsat
V
1 Dsat
Esat L
I Dsat ,long , A 400
Thus I Dsat. A 97 (A)
VDsat 2.5
1 1
E sat , A L 4 104 0.2 10 4
I Dsat ,long , B 1600
I Dsat. B 13 (A)
VDsat 2.5
1 1
E sat , B L 1 103 0.2 10 4
(d) Mobility is important in long-channel MOSFETs, since the MOSFET operates
outside the velocity saturation regime. On the other hand, short-channel MOSFETs
operate in the velocity saturation region, so the saturation velocity is often more
important that the mobility, at least from a drive-current consideration in saturation.
In the linear regime, short-channel MOSFETs still depend more heavily on mobility,
since the lateral field is small.
11. Let y be the coordinate from source to drain. Assume the body effect coefficient m
= 1, i.e. Qn(y) = Cox(VG VTn – Vc(y)) for VDS < VDsat. And neglect the velocity
saturation, i.e. v = nEC. For a MOST operated in linear region with drain current
ID, show that the potential VC(y) and the field EC(y) along the channel can be
expressed as:
2I D y
(a) VC ( y ) (VG VTn ) (VG VTn ) 2
W nCox
ID
(b) E C
( nWCox ) (VG VTn ) 2 2 I DW n Cox y
2
Solutions:
(a)
I D WQn v
dVC y ' dVc y '
WCox VG VTh VC y ' eff (E C )
dy ' dy '
VC y
W eff Cox VG VTh VC y ' dVC y '
y
0
I D dy '
0
V 2 y
I D y W eff Cox VG VTh VC y C
2
Solve the equation of VC(y)
2I D y
VC y VG VTh VG VTh
2
( VC = 0 when y = 0 )
W eff Cox
(b)
dVC y
EC
dy
2I D
1 W eff Cox
2 2I D y
VG VTh
2
W eff Cox
ID
CoxW VG VTh 2 I DW eff Cox y
2 2
eff
12. An ideal N-channel MOSFET maintained at T = 300K is characterized by the
following by the following parameters: W = 50 m, L = 5 m, Tox = 0.05 m, NA =
1015 cm-3, N+ poly-Si gate, n = 800 cm2/Vs (which is assumed to be independent
of Vg). Ignore the bulk charge effect (i.e. m = 0). Determine:.
(a) VTh.
(b) IDsat if VGS = 2V
dI D
(c) if VGS = 2V and VDS = 0.
dV DS
(d) dI D if VGS = 2V and VDS = 2V.
dVGS
Solutions: (We use ox = 3.9 0 =3.45 10-13 F/cm, Si = 11.9 0 =1.0 10-12 F/cm, ni
= 1 1010 cm-3 , and 2.3kT/q = 60 mV here)
kT N
(a) B 2.3 log A 0.3 (V)
q ni
ox
Cox 6.9 10 8 F / cm 2
tox
Eg
V FB si ( si B ) 0.85 (V)
2
2 si qN A 2B
VTh VFB ox si V fb 2B 0.049 (V)
Cox
1W
(b) I Dsat n Cox VGS VTh 2 1.16 (mA)
2 L
(c) Since VDS 0(V ) VDsat VGS VTh 2.05 (V), thus the FET is operated in
linear region.
W 1 2
ID n Cox VGS VTh VDS VDS
L 2
I D W
gd n Cox VGS VTh VDS 1.13 (mS)
VDS L
(d) Since VDS 2(V ) VDsat VGS VTh 2.05 (V), thus the FET is operated in
linear region.
W 1 2
ID n Cox VGS VTh VDS VDS
L 2
I D W
gd n CoxVDS 1.10 (mS)
VGS L
13. The output characteristics of a NMOST and a strained NMOST with Leff = 67 nm
and tox = 32 Å are shown below. Estimate the velocity of the carriers and the electric
field near both the source end and the drain end of the channel when the device is
just saturated at VGS – VT = 1.0 V. (Note that this is an experimental problem, and
you should not assume any values such as vsat that is not given. Use m = 1 for
simplicity)
Solutions:
(a) For the normal device,
We found from the output characteristics that Vdsat = 0.4 V and Idsat = 132 μA
(i) Near the drain side
1 1 m 1 1 1
From
Vdsat 0.4 Vgs Vth E sat L 1 Esat 0.067
1
We have E sat 9.95V / m
1.5 0.067
i.e. Esat 9.95 10 V / cm
4
From
3.45 10 13
I dsat 132 10 6 WC ox V gs Vth Vdsat v sat 0.28 10 4 0.6 v sat
3.2 10 7
We have vsat 7.29 10 cm / s
6
(ii) Near the source side
I dsat 132 10 6 WCox Vgs Vth mVc 0 v source
From
0.28 10 4 33.45
13
10
.210 7
1 v source
We have vsource 4.37 10 cm / s
6
The source electric field can then be found from the following relation.
2vsat 2 7.29 10 6
eff 147cm 2 / V s
E sat 9.95 10 4
eff E source 147 E source
vsource 4.37 10 6 cm / s
E source E source
1 1
E sat 9.95 10 4
We have Esource 4.24 10 V / cm
4
(b) For the strained silicon devices,
We found from the output characteristics that Vdsat = 0.42 V and Idsat = 190 μA
(i) Near the drain side
1 1 m 1 1 1
From
Vdsat 0.42 Vgs Vth E sat L 1 Esat 0.067
1
We have E sat 10.81V / m
1.38 0.067
Esat 1.08 105 V / cm
From
3.45 10 13
I dsat 190 10 6
WC ox Vgs Vth Vdsat v sat 4
0.28 10 0.58 v sat
3.2 10 7
We have vsat 1.08 10 cm / s
7
(ii) Near the source side
I dsat 190 10 6 WCox Vgs Vth mVc 0 v source
From
0.28 10 4 33.45
13
10
.210 7
1 v source
We have vsource 6.29 10 cm / s
6
The source electric field can then be found from the following relation.
2vsat 2 1.08 107
eff 200cm 2 / V s
Esat 1.08 105
eff Esource 200 Esource
vsource 6.29 106 cm / s
Esource Esource
1 1
Esat 1.08 105
We have Esource 4.43 104 V / cm
14. An NMOSFET has thinner Tox at the center of the channel and thicker Tox near the
source and drain. This could be approximately expressed as Tox = Ay2 + B. Assume
that Vth is independent of y. Assume m = 1.
(a) Derive an expression for ID.
(b) Derive an expression for VDsat.
(c) Does the assumption of nearly constant VTh suggest a large or small Wd?
Solutions:
dVc
(a) I D WQn v n WCox VGS VTh Vc n
dy
ox
I D W 2
VGS VTh Vc n dVc
Ay B dy
L
V DS
I D Ay
2
W ox VGS VTh Vc n dVc
2
B dy
L 0
2
L
V
A 2 1 DS
I D y 3 By W n ox VGS VTh Vc Vc2
3 L 2 0
2
A 1 2
I D L L2 B W n ox VGS VTh VDS VDS
12 2
W ox 1 2
ID n
A 2 VGS VTh V DS VDS
L L B 2
12
W 1 2
ID n Cox VGS VTh VDS VDS
L 2
ox ox
where Cox
toxeff A 2
L B
12
dI D
(b) Let V DS V Dsat 0 VGS VTh VDsat 0 VDsat VGS VTh .
dVDS
2 si qN A 2B
(c) VTh VFB ox si V fb 2B
Cox
C Si t
VTh V fb 2B 2 2B V fb 1 6 ox 2B
Cox Wd
For a constant threshold assumption, we have to minimize the effect of the variation of
oxide thickness. From the above expression, this suggest a large Wd .