Problem Set # 10

ECEN 3320 Fall 2013

Semiconductor Devices
November 18, 2013 – Due December 2, 2013

1. An n-channel MOSFET has W/L = 10 and oxide thickness xox =30 nm. Assume the carrier
mobility in the channel is constant at µn =500 cm2 /Vs.
(a) Calculate the required inversion layer carrier concentration, Qn /q, in order to obtain
channel resistance of 500 Ω.
Solution: The drain current, ID is given by
W
ID = µn Cox [VGS − VT ]VDS
L
in the linear regime and
" #
W V2
ID = µn Cox [VGS − VT ]VDS − DS
L 2

in the quadratic regime. The inversion layer charge, Qn (= Qinv ) is given by

Qn = Cox [VGS − VT ].

The channel conductance is given by

∂ID
gd = |V
∂VDS GS
and evidently is a linear conductance when ID is expressed in the linear approxima-
tion such that the linear conductance is given by
W W
gd = µn Cox [VGS − VT ] = µn Qn
L L
The channel resistance , rd = 1/gd , can be expressed as
L
rd =
µn Qn W
and the charge, Qn as
L
Qn = .
µn rd W

1
 For a 500 Ω channel, the Qn then is
1
Qn = = 4 × 10−7 C/cm2 = 2.5 × 1012 charges/cm2 .
10 × 500 × 500
(b) Calculate the gate voltage above the threshold, VG −VT , required to produce the inversion
layer charge, Qn /q, found in (a).
Solution: We can write that
Qn Qn xox 4 × 10−7 3 × 10−6
VGS − VT = = = = 3.48 V
Cox ox 3.9 × 8.85 × 10−14
2. An n-channel enhancement-mode MOSFET is biased as shown below. The threshold voltage
was found to be VT =1V. Using the long-channel MOSFET theory, calculate and plot ID vs
VDS for

Figure 1: An n-channel enhancement mode MOSFET.

(a) VGD = 0,
Solution: Here we have that VGS = VDS . We cannot have quadratic behavior (VDS
cannot be less than VGS − VT ). We have
(
ID =
ID0
  VDS ≤ VT
qVDS
ID0 exp kT VDS > VT

(b) VGD = VT /2 and

Solution: Here we have VGS = VDS + VT /2. Again, we can have no quadratic behavior.
We have
(
ID =
ID0
  VDS ≤ VT /2
qVDS
ID0 exp kT VDS > VT /2

(c) VGD = 2VT .

2
 Solution: Here we have that VGS = VDS + 2VT . We will have quadratic behavior. VD S
is always less than VGS − VT . We still have cut-off when VGS ≤ VT , that is, when
VDS ≤ −VT . Knowing that
" #
W V2
ID = µn Cox (VGS − VT )VDS − DS when VDS ≥ −VT
L 2

and plugging in for VGS in terms of VDS and VT , we find

(
ID0 VDS ≤ −VT
ID = W
2
VDS
µn Cox L [(VT VDS + 2 ] VDS > −VT

Figure 2: An n-channel enhancement mode MOSFET I-V curves. VT = 1 was assumed for all of the
numerical calculations.

3. Suppose that the I − V curve for an n-channel MOSFET has saturation drain current IDsat =
2 × 10−4 A, the drain voltage at the onset of saturation VDSsat = 4 V and threshold voltage
VT = 0.8 V.

(a) Find the gate voltage VGS at the onset of saturation

Solution: We have that
W h 2
i
ID = µn Cox (VGS − VT )VDS − VDS /2 when VDS ≤ VGS − VT .
L
We then know that

VDS,sat = VGS,sat − VT

so that

VGS,sat = VDS,sat + VT = 4.8 V

W µn Cox
(b) Find the value of the conduction parameter, 2L .

3
 Solution: Above saturation, we can write that
W 2
ID,sat = µn Cox VDS,sat
2L
or that
W ID,sat
µn Cox = 2 = 1.25 × 10−5 mhos/V
2L VDS,sat
(c) Determine ID when VGS = 2V and VDS = 2V.
Solution: This is just plugging in to the expression for saturated (VDS ≥ VGS − VT )
current
W h 2
i
ID = µn Cox (VGS − VT )VDS − VDS /2
L
W
= µn Cox (VGS − VT )2 /2
L
= 1.25 × 10−5 × 1.22 = 1.8 × 10−5 amps
4. Start with the inversion layer charge density equation, Qn = −Cox [VG − VT − V (y)].
Solution: We have that
∂n
Jn = Dn q + qnµn E = Jdif f + Jdrif t .
∂y
The Qn (y) we are using is just qn(y) × A × L where A is the channel cross sectional area
and L is the channel length in the above notation. We can then write that
∂Qn
J n × A × L = Dn + Qn µn E.
∂y
(a) Derive an expression for diffusion current in the channel.
Solution: Starting from the definition of the diffusion current above, we find
∂Qn ∂E(y)
Jdif f = Dn = −Dn Cox
∂y ∂y
(b) Derive an expression for drift current in the channel.
Solution: Starting from the definition of the drift current above, we find
∂E(y)
Jdrif t = Qn µn E = −µn Cox [VG − VT − V (y)]
∂y
(c) Derive an expression for the ratio of diffusion and drift currents.
Solution: Taking the ratio of drift to diffusion currents, we find
Jdrif t µn
= [VG − VT − V (y)].
Jdif f Dn
kT
Using the Einstein relation that Dn = q µn , we find that
Jdrif t (VG − VT − V (y))
= .
Jdif f kT /q
The drift current is greater than the drift by essentially the amount that the driving
voltage is greater than the thermal noise voltage kT /q.