Problem 1

Analyze a p-type silicon MOS capacitor characterized by the following substrate

doping concentrations:
NA = 1015 cm−3 and NA = 1018 cm−3.
Assume a constant surface potential of ψs = 0.5 V applies to both
configurations.

[20 Marks]

(a) Utilizing the depletion approximation, derive a closed-form analytical

expression for the electrostatic potential profile ψ(x) within the
semiconductor starting from Poisson’s equation. Explicitly list all
assumed conditions. [4
Marks]

(b) Apply a suitable numerical technique to the exact Poisson equation to

determine the internal potential distribution. State the relevant
boundary conditions clearly. [4 Marks]

(c) Graph the potential profiles derived from Parts (a) and (b) for both
doping levels on a single set of axes. The plots must span from the
semiconductor-oxide interface deep into the bulk. [4
Marks]

(d) Assess the deviation between the numerical and analytical methods by
graphing the error as a function of depth. [4
Marks]

(e) Drawing upon your findings, provide commentary on: [4 Marks]

(i) The overall validity of the depletion approximation.

(ii) How variations in substrate doping influence the spatial potential profile.

Solution
(a) Analytical Derivation using the Depletion Approximation
To derive the potential profile ψ(x), we start with the 1D Poisson’s equation in
the semiconduc-tor:

d2ψ
dx2 = −
ρ(x)
1
ϵs
(1)

Assumed Conditions for the Depletion Approximation:

Abrupt Depletion Edge: The transition between the fully depleted region and the
neutral bulk is instantaneous at a depth x = W .
Complete Depletion of Mobile Carriers: Within the depletion region (0 ≤ x ≤ W ),
the concentration of holes and electrons is negligible compared to the ionized
acceptor concentration (p ≈ 0, n ≈ 0).
Charge Density: The space charge is comprised entirely of static, ionized acceptor
atoms. Therefore, ρ(x) = −qNA for 0 ≤ x ≤ W , and ρ(x) = 0 for x > W .
Derivation:
Substitute the charge density into Poisson’s equation for the region 0 ≤ x ≤ W :

d2ψ dx2 =
qNA ϵs

(2)

Integrate once with respect to x to find the electric field, applying the boundary
condition that the electric field is zero at the depletion edge ( dψ = 0 at x = W ):
dψ = qNA (x − W ) (3)
dx ϵs
Integrate a second time to find the potential profile, applying the boundary
condition that the potential in the neutral bulk is zero (ψ(W ) = 0):
ψ(x) = qNA (x − W )2 + C (4)
2ϵs
Since ψ(W ) = 0, the constant C = 0.
The closed-form analytical expression for the potential profile is:
ψ(x) = qNA (W − x)2 for 0 ≤ x ≤ W (5)
2
2ϵs
ψ(x) = 0 for x>W (6)

Numerical Technique for the Exact Poisson Equation

To find the internal potential distribution exactly, we must account for both fixed
ionized dopants and mobile charge carriers. The exact Poisson equation for a p-
type semiconductor is:
d2ψ q
dx2 = −ϵ (p − n − NA) (7)

2
Assuming Boltzmann statistics, the carrier concentrations are p = N e−βψ and n
= n eβψ,

where q

A NA

β = kT . Substituting these yields the non-linear Poisson equation:

Numerical Technique:
d2ψ dx2
= qNA
ϵs

−e

3
−βψ +
βψ

2
A

(8)

The Finite Difference Method (FDM) combined with the Newton-Raphson

iteration is the standard approach to solve this non-linear boundary value
problem.
Discretize the spatial domain x into N nodes from x = 0 to x = L (where L ≫ W ).
Approximate the second derivative using central finite differences: d2ψi ≈
ψi+1−2ψi+ψi−1 .

Formulate a system of non-linear algebraic equations.

dx2
∆x2

Use the Newton-Raphson method to iteratively update the potential guess at each
node until the residual error falls below a defined tolerance.
Boundary Conditions:
At the oxide-semiconductor interface (x = 0): Dirichlet boundary condition. ψ(0) =
ψs = 0.5 V.
Deep in the bulk (x = L): Dirichlet and Neumann boundary conditions. ψ(L) = 0 V
and

dψ
dx x=L

4
= 0.

(b) Potential Profile Plots

Given a fixed surface potential of ψs = 0.5 V, the analytical depletion widths are:
I
W d= 2εsψs
. qNA

For the lighter doping (NA = 1015 cm−3), Wd = 0.804 µm. For the heavier
doping (NA = 1018 cm−3), Wd = 25.4 nm.
The analytical boundary conditions dictate that ψ = 0.5 V at x = 0, and ψ = 0 at
x = Wd.

Figure 1: Plot of the Analytical Approximation

5
 Figure 2: Plot of the Numerical Exact Solution

Figure 3: Superimposed Comparison: Analytical vs. Numerical Methods

Figure 4: Absolute error margin between the numerical calculation and the
depletion ap-proximation over depth.

(c) Quantitative Error Analysis

The discrepancy between the simplified analytical approach and the exact
numerical solution can be represented as:
e(x) = ψexact(x) − ψana(x).

6
Here, ψexact(x) is the numerical result of the non-linear Poisson equation,
while ψana(x) is the closed-form depletion approximation. The absolute
magnitude of the error is:
|e(x)| = |ψexact(x) − ψana(x)|,

and the relative percentage error is defined as:

%Error = |ψexact(x) − ψana(x)| 100.

|ψexact(x)| ×

By calculating this error continuously with respect to depth x, we observe

a key pattern. Because the analytical model enforces an abrupt, non-
physical boundary where the field drops instantly to zero, whereas the exact
solution correctly models the gradual screening effect of majority carriers,
the≈deviation peaks precisely near the theoretical depletion edge (x Wd). In
this transition tail, the numerical solution retains a small finite potential
that the analytical model ignores.

(d) Observations and Comments

(i) Viability of the Depletion Approximation To verify the approximation
analyti-
/
cally, it is helpful to benchmark ψ against the bulk Fermi potential ϕ = ln
)
.
kT NA
s B
q ni
NA Profile Bulk Potential (ϕB) Strong Inversion Threshold
(2ϕB) 1015 cm−3 ≈ 0.288 V ≈ 0.576 V
10 cm
18 −3
≈ 0.467 V ≈ 0.934 V

For both doping levels, the given ψs = 0.5 V satisfies the condition ϕB < ψs
< 2ϕB. Con-sequently, the device resides in the depletion/weak-inversion
regime and has not yet breached strong inversion. This justifies the use of
the depletion approximation: the concentration of mobile minority electrons
is still negligible compared to the static ionized acceptor density. The
numerical curves closely track the analytical parabola, diverging only
slightly near Wd.
Furthermore, the ratio of the depletion boundary to the extrinsic Debye length
yields:
/
Wd = 2ψs ≈ 6.2,
LD VT
which is independent of doping NA. Since Wd is significantly larger than LD,
the transition region where the approximation fails represents only a minor
fraction of the total space-charge width, proving the approximation is
structurally sound for both scenarios.
7
(ii) Influence of Substrate Doping on Potential Curvature

• In the lightly doped instance (1015 cm−3), the space-charge width is

comparatively
≈ large (Wd 0.804 µm). The spatial potential drops off very
gradually, resulting in a wide, sweeping parabolic curve.

8
• For the heavily doped counterpart (1018 cm−3), the depletion width
shrinks dramati-cally
≈ (Wd 25.4 nm). The same 0.5 V drop is forced to
happen over a microscopic distance, creating a highly compressed,
√
steep potential profile. This also triggers a
significantly stronger surface electric field, which scales as Es ∝ NA.

• While the lightly doped device exhibits a broader, smoother tail region
≈
near x Wd, the heavily doped device forces the numerical and analytical
curves to overlap almost identically due to the vastly reduced Debye
smoothing length.

9
Problem 2
The Bipolar Junction Transistor (BJT) can be modeled mathematically for
large-signal be-haviors using the Ebers-Moll framework.

[10
Marks]

(a) Evaluate whether the Ebers-Moll equations hold universally across all
standard oper-ating regimes of a BJT. Provide an explanation of how
this model accounts for physics in the cut-off, forward-active,
saturation, and reverse-active states. [5
Marks]

(b) Construct an Ebers-Moll BJT model within a circuit simulation

environment (like LTspice) utilizing suitable parameters. Produce and
analyze the following graphs:

(i) Input and output curves for the Common-Base (CB) setup.
(ii) Input and output curves for the Common-Emitter (CE) setup.

[5 Marks]

Solution
(a) Applicability of the Ebers-Moll Model Across BJT Regimes
The Ebers-Moll (EM) representation serves as a robust large-signal
equivalent circuit, map-ping BJT behavior to a pair of interacting p-n
junction diodes. It remains mathematically valid across all four operating
quadrants (cut-off, forward-active, saturation, and reverse-active). However,
it is fundamentally a static model; it ignores dynamic high-frequency charge
storage, the Early effect (base-width modulation), and high-level injection
phenom-ena.

1. Cut-off Region: Both the emitter-base (EB) and collector-base (CB)

junctions face reverse bias. Diode currents IF and IR approach zero,
rendering the transistor an open switch.

2. Forward-Active Region: The EB junction is forward-biased while the

CB junction is reverse-biased. Forward injection IF drives the device (IR
≈ 0), resulting in the classic amplification behavior IC ≈ βIB.
3. Saturation Region: Both junctions are forward-biased. Here, forward
and reverse mechanisms (IF , αF and IR, αR) heavily influence the current
flow, dropping VCE near zero and acting like a closed switch.

1
0
4. Reverse-Active Region: The EB junction is reverse-biased and the CB
junction is forward-biased. The transistor operates backward. Due to
physical asymmetries, the
reverse gain is terrible (αR ≪ αF ), so this mode is generally avoided in design.

1
1
 Figure 5: Equivalent Schematic of the Ebers-Moll Model (PNP Variant)

By executing Kirchhoff’s Current Law (KCL) at the emitter node (Node 1)

of the PNP model:
IE + αRIR − IF = 0 ⇒ IE = IF − αRIR.
Likewise, applying KCL at the collector node (Node 2):
IC + IR = α F IF ⇒ IC = αF IF − IR.

The fundamental diode currents obey the standard Shockley diode equations:
( ) ( )
IF = IE0 eVBE/(ηVT ) − 1 , IR = IC0 eVBC/(ηVT ) − 1 ,

where IE0 and IC0 denote the respective reverse saturation currents, αF and αR
represent the common-base current gains, η is the emission ideality factor,
and VT = kT/q.
Substituting these definitions yields the universal Ebers-Moll equations:
( ) ( )
IE = IE0 eVBE/(ηVT ) − 1 − αRIC0 eVBC/(ηVT ) − 1 , IC
= αF IE0 (eVBE/(ηVT ) − 1 −
) IC0 e(VBC/(ηVT ) − 1 . )
These unified equations gracefully collapse to model specific regions:

Cut-off Setting With VBE < 0 and VBC < 0, the exponential terms vanish. We find
IF ≈ −IE0 and IR ≈ −IC0. The device only passes negligible reverse leakage.

Forward-Active Setting With VBE > 0 and VBC < 0, the reverse current collapses
to
IR ≈ −IC0. The collector current simplifies to:

IC ≈ αF IE0 eV BE /(ηVT ).

Saturation Setting With VBE > 0 and VBC > 0, the exponential terms for both
junctions dominate, indicating heavy carrier injection from both sides,
pushing the device into deep saturation.

Reverse-Active Setting With VBE < 0 and VBC > 0, the collector physically
injects carriers while the emitter collects them, governed purely by the

1
2
inefficient αR parameter.

1
3
Summary By unifying the operation of the device under one mathematical
framework, the Ebers-Moll equations seamlessly transition between
operational states, proving highly effective for foundational DC circuit
evaluation and SPICE simulation.

(b) Ebers-Moll Modeling via LTspice Simulation

An equivalent Ebers-Moll NPN model was constructed in LTspice, relying on
discrete ideal diodes and voltage-controlled/current-controlled sources. The
following global variables were defined:
Model Property Configured
Value Forward CB current gain, αF
0.99
Reverse CB current gain, αR 0.5
Diode saturation current, IS 1 ×
10−14 A
Ideality multiplier, η 1
Thermal Voltage, VT (at 300 K) 25.85
mV Ambient Temperature 300 K
Silicon Bandgap, Eg 1.11 eV

(i) Common-Base (CB) Input Characteristics

A DC domain sweep was performed on VEB while anchoring VCB at distinct steps:

.dc VEB 0 0.9 0.01 VCB list 1 5 9

VEB traverses from 0 to 0.9 V in 0.01 V increments, with VCB locked at 1 V, 5 V, and
9 V.

• The resulting input curve perfectly mimics a traditional forward-biased p-n

junction.

• Emitter current IE remains microscopic until surpassing the built-in cut-

in threshold, after which it scales exponentially.

• Demonstrates the characteristically low input impedance inherent to CB

topologies.

1
4
Figure 6: Schematic of the CB Ebers-Moll
Equiva-lent Circuit

1
5
 Figure 7: Simulated Input Response for CB Topology

(ii) Common-Base (CB) Output Characteristics

.dc VCB -0.75 10 0.005 VEB 0.6 0.7 0.05
VCB sweeps from −0.75 V to 10 V, while VEB remains fixed at 0.60 V, 0.65 V, and
0.70 V.

• The collector current IC scales steeply for low voltages before

stabilizing into a nearly flat plateau (the active region).

• Higher VEB drive biases yield more severe emitter injection, shifting the
flat curves progressively upward.

• Extending VCB into the negative quadrant forcefully forward-biases the

collector, driv-ing the device into hard saturation.

• The flat, horizontal nature of the active region profiles highlights an

exceptionally high output resistance.

Figure 8: Test Bench for CB Output Curves

1
6
 Figure 9: Simulated Output Response for CB Topology

(iii)Common-Emitter (CE) Input Characteristics

.dc VBE 0 1.0 0.01 VCE list 0 2 5

VBE varies from 0 to 1.0 V in 0.01 V increments, while VCE is stepped through
0 V, 2 V, and 5 V.

• Similar to the CB trace, the base current profile reflects a standard diode
curve.
• IB holds at essentially zero until the junction cut-in voltage (≈0.6 to 0.7 V) is
overcome.
• Stepping VCE creates only minor deviations in the curves because the
primary current driver is the heavily forward-biased EB interface.

• Exhibits a moderate input resistance, distinctively higher than that of

the CB config-uration.

Figure 10: Schematic of the CE Ebers-Moll

Equiv-alent Circuit

1
7
 Figure 11: Simulated Input Response for CE Topology

(iv) Common-Emitter (CE) Output Characteristics

.dc VCE 0 10 0.02 VBE 0.6 0.7 0.05
VBE is stepped from 0.60 V to 0.70 V while VCE is swept from 0 to 10 V to obtain the
output characteristics.
• At nominal VCE values near zero, the base-collector junction flips to
forward bias, immediately plunging the transistor into saturation
(characterized by the initial vertical surge in IC).
• Pushing past the saturation threshold smoothly transitions the device
into the active regime, where IC flattens out, largely ignoring further
VCE increases.
• Ramping up base drive shifts the entire saturation-active curve
vertically due to beta multiplication.
• The collector current remains nearly constant in the active region,
indicating the high output resistance predicted by the ideal Ebers–Moll
model.

1
8
Figure 12: Test Bench for CE Output Curves

1
9
 Figure 13: Simulated Output Response for CE Topology

Conclusion
The LTspice simulations reliably replicate the physical mechanics predicted
by the Ebers-Moll equations. Both the Common-Base and Common-Emitter
architectures showcase dis-tinct regional states perfectly captured by the
simplified dual-diode math. These generated plots confirm that
manipulating standard p-n junction biases is mathematically sufficient to
model both saturation switching and active large-signal amplification in a
BJT.

2
0