ECE 4813

Semiconductor Device and Material

Characterization
Dr. Alan Doolittle
School of Electrical and Computer Engineering
Georgia Institute of Technology

As with all of these lecture slides, I am indebted to Dr. Dieter Schroder from Arizona State
University for his generous contributions and freely given resources. Most of (>80%) the
figures/slides in this lecture came from Dieter. Some of these figures are copyrighted and can be
found within the class text, Semiconductor Device and Materials Characterization. Every serious
microelectronics student should have a copy of this book! ECE 4813 Dr. Alan Doolittle
 Doping Profiling
Secondary Ion Mass Spectrometry
Spreading Resistance
Capacitance – Voltage
Threshold Voltage

ECE 4813 Dr. Alan Doolittle

 Secondary Ion Mass Spectrometry
 SIMS is the most common doping profiling technique
 Incident ions knock out atoms and ions from the substrate
 The mass of these ions is analyzed

106 1020

Cs, O Mass 105 1019

Density (cm-3)
Analyzer

Ion Counts
104 1018

103 1017
Si B
102 1016

101 1015

1000 200 400 600

1014
0 0.2 0.4 0.6 0.8 1
Time (s) Depth (µm)

ECE 4813 Dr. Alan Doolittle

 Secondary Ion Mass Spectrometry
 The use of Cs or O results in a modification of the surface
work function (low work function for Cs and high for O)
producing either + or – ions.

 Charging of the sample can effect

both the sputter yield (changes
the acceleration energy) and the
Cs, O Mass
focus into the mass analyzer
Analyzer
 Can be offset by an electron flood gun
for –ions and a +H source (rarely used)
for +ions
Si B

ECE 4813 Dr. Alan Doolittle

 Secondary Ion Mass Spectrometry
 Mass Spectra can be obtained
 Depth Profiling can be obtained
 Time of Flight Versions Dominate technology now

ECE 4813 Dr. Alan Doolittle

 Secondary Ion Mass Spectrometry
 Mass Spectra can be obtained
Oxygen

S1-O
5
10 S2-O
Count Rate (per sec)

S3-O
Mg impurities Cl impurities
104
Oxygen

103

102
10 15 20 25 30 35 40
Mass (amu)
ZnO
ECE 4813 Dr. Alan Doolittle
Secondary Ion Mass Spectrometry

ECE 4813 Dr. Alan Doolittle

 Secondary Ion Mass Spectrometry
 While rarely used in simple SIMS analysis, energy spectrums
for sputtered ions can tell you a great deal about the sample
including in-situ resistivity analysis
 Care must be taken since SIMS often measures charge to
mass ratio. Time of Flight SIMS analyzers correct this
problem.

A
Concentration (cm-3)

Ga
Ga-12V
Ga+12V
Mg

0 20 40 60 80 100
Sputter Time (min)
ECE 4813 Dr. Alan Doolittle
 Spreading Resistance
 The wafer is bevelled extending the layer thickness
 Compare R to calibrated standards; Rsp dominates

R = V / I = 2Rp + 2Rc + 2Rsp ≈ 2Rsp

∆z = ∆x sin(θ)
Beveled Surface
Original Surface I V I
I
~ 20 µm
∆x Rp
Rc

Rsp
Current Spreading

htpp://www.ssm-inc.com

ECE 4813 Dr. Alan Doolittle

 Spreading Resistance
 The spreading resistance of ideal metal-
semiconductor contacts can be calculated
 For flat-bottom probe of diameter d
ρ
Rsp = d
2d
 For hemispherical probe of radius r
ρ
Rsp =
2πr r
 For a “real” probe
ρ
Rsp = k For ρ = 1 Ω-cm,
2πr
r = 1 µm,
k must be experimentally determined Rsp = 1000-2000 Ω

ECE 4813 Dr. Alan Doolittle

 Spreading Resistance
 Spreading resistance is measured and
converted to resistivity and NA or ND profiles

1021 105
1020
ND (cm-3)

1019 104

Rsp (Ω)
1018
1017
103
1016
1015
1014 102
0 0.02 0.04 0.06 0.08 0.1
Depth (µm)

ECE 4813 Dr. Alan Doolittle

 Capacitance-Voltage
 Capacitance-voltage measurements used for
 Doping profiling
 Flatband voltage, oxide charge etc. in MOS devices
 Capacitance is a measured charge responding to a time varying voltage

RL
vo = vi
p RL + 1 /(G + jωC )
n G C ≈ (G + jωC )RLv i for small RL
vi vi

RL In phase:
RL vo
v o ≈ GRLv i ~ G
Out of phase:
v o ≈ ωCRLv i ~ C

Actual capacitance meters use ac current amps with phase-sensitive

detectors to measure “in phase” and “out of phase” components.

ECE 4813 Dr. Alan Doolittle

 Capacitance-Voltage
 Need a device with a space-charge region
 Schottky diode, pn junction, MOS-C, MOSFET
 The dc bias, V, determines W
 The ac bias, vi, measures the capacitance

Qs dQ/dV
Qm + -
+ -
+ - NA Qs = qN AW
-
+
Q Q/V
V
W ~ Vbi − V
W dW
∴ Qs ~ Vbi − V
vi
V
dQm dQs
C= =−
dV dV

ECE 4813 Dr. Alan Doolittle

 Capacitance - Voltage
Qs
Qm + -
Qs = qA ∫ (p − n + N − N )dx ≈ −qA ∫ N Adx
W W
-
V ++ - NA + −

+ - D A
0 0

W dW Depletion region approximation

for a p-type substrate
W
d dW
C = qA
dV ∫ N Adx = qAN A (W ) dV
( neglecting dN A dV )
0

K sε o A K sε o A dW K ε A dC
C= ; W = ; = − s 2o
W C dV C dV

C3 2 K sε o A
N A (W ) = − = W =
qK sε o A2 dC dV qK sε o A2 d (1 C 2 ) dV
;
C

Y-axis X-axis

ECE 4813 Dr. Alan Doolittle

 Capacitance - Voltage
1 10-10

Capacitance (F)
 What to use?
8 10-11
C3
N A (W ) = − 6 10-11
qK sε o A2 dC dV Slope = dC/dV
4 10-11 C
 Or
2 10-11
2 V
N A (W ) =
qK sε o A2 d (1 C 2 ) dV
0
0 2 4 6 8 10
Voltage (V)
 For both
2 1021

K sε o A

1/C2 (F-2)
W =
C 1 1021
Slope =
d(1/C2)/dV
The 1/C2
– V curve
has less curvature! 0
0 2 4 6 8 10
Voltage (V)

ECE 4813 Dr. Alan Doolittle

 Capacitance - Voltage
 C - V curves are always nonlinear
 1/C2 - V curves clearly show carrier or doping
density non-uniformities

Doping Density (cm-3)

600 6 1020 1 1016
Capacitance (pF)

1/C2 8 1015

1/C2 (F-2)
400 Nonlinear 4 1020 p p+
6 1015
(Nonuniform Doping)
Linear 4 1015
200 2 1020
(Uniform Doping)
C 2 1015
0 0 0
-1 1 3 5 7 9 0 0.5 1 1.5 2
Voltage (V) Depth (µm)

ECE 4813 Dr. Alan Doolittle

 Capacitance - Voltage
 C - V curves can determine channel depths in
compound semiconductors

ECE 4813 Dr. Alan Doolittle

 MOS Capacitance - Voltage
 Oxides (insulators) in series with the junctions
create an additional fixed capacitor.

Coxide = εA/tox
Depletion Oxide
K sε o A K sε o A
Layer

Substrate C= ; W=
W C

2K S ε o NA 2K S ε o ND
xn = (Vbi − V A ) and xp = (Vbi − V A )
q N D (N A + N D ) q N A (N A + N D )

2K S ε o (N A + N D )
W = x p + xn = (Vbi − V A )
q NAND

W ~ Vbi − V
ECE 4813 Dr. Alan Doolittle
MOS Capacitance

ECE 4813 Dr. Alan Doolittle

 MOS Capacitance
 Oxides (insulators) in series with the junctions create
an additional fixed capacitor.
 Capacitance approaches the Cox in accumulation.
 More MOS in chapter 8…
1 1 
W = K S ε O A + 
 C COX 
CoxCS
Accumulation C= Accumulation
Cox + CS
1 1

0.8 CFB/Cox 0.8 CFB/Cox

•

•
C/Cox
C/Cox

0.6 Clf 0.6 Clf

0.4 0.4
Chf Chf
0.2 0.2
p-Substrate Cdd Cdd n-Substrate
0 0
-5 -3 -1 1 3 5 -5 -3 -1 1 3 5
Gate Voltage (V) Gate Voltage (V)
ECE 4813 Dr. Alan Doolittle
 Max-Min MOS Capacitance
 Measure the maximum accumulation Capacitance
(high frequency measurement) and the minimum
capacitance in strong inversion
2 K S ε OφSurface ,inv
W@ inversion =
C C qN A
C = ox S
Cox + CS
4φF C22φf 4φF 2
Cinversion
Accumulation NA = ≈
qK S ε O A2  C 
2
qK S ε O A2  C 
2

1 − 2φf  1 − inversion 

1
 Cox   Cox 
0.8
 
CFB/Cox
•
C/Cox

0.6 Clf  Clf = Low frequency capacitance where the

measurement frequency is small enough such that
0.4 the carriers can respond to the stimulus. Sweep
rate of bias is also slow.
0.2
C2φf Chf
 Chf = High frequency capacitance where the
p-Substrate Cdd carriers cannot respond to the stimulus. Sweep rate
0 of bias is also slow.
-5 -3 -1 1 3 5  Cdd = When the bias sweep rate is faster than
generation rate such that inversion cannot take
Gate Voltage (V) place. Typically also measured with a high
frequency stimulus.
ECE 4813 Dr. Alan Doolittle
 Built-In Voltage Determination from
Capacitance-Voltage
 The intercept of the C-V curve can determine the VBI
 In practice, care should be exercised as in practice the ohmic-
contacts (particularly the “back contact”) can lead to errors in
the determination of VBI
 Accounting for the majority
carrier tail in the depletion region 2 1021
(introduces a kT/q factor – this

1/C2 (F-2)
effect is ignored in the depletion
region approximation) The 1 1021
voltage intercept is, Slope =
kT d(1/C2)/dV
Vi = −VBI +
q
 For a p-n junction, 0
0 2 4 6 8 10
kT  N A N D 
VBI = ln 
q  ni2  Voltage (V)
 For a Schottky diode,
kT  N C  kT
Vi = −φB + ln +
q  N D  q

ECE 4813 Dr. Alan Doolittle

 Contactless C - V
 Contact C-V measurements
 pn junctions
d
 Evaporated metal Schottky diodes
 Mercury Schottky diodes
W
 MOS capacitors
 Can also be implemented contactless
 Compressed air escapes through porous
disc; air cushion forms between electrode
and semiconductor surface Pressurized
Air
Bellows
Cair CS   Ceramic
C= ; W = K sε o A 1 − 1  Air Filter
Cair + CS  C Cair  d ≈ 1 µm
C3 2
NA = − =
qK sε o A2dC / dV qK sε o A2d (1/ C 2 ) / dV

http://www.semitest.com

ECE 4813 Dr. Alan Doolittle

 Contactless C -V
 Contact diameter ~
1 mm
 Need calibrated
standard wafer
 Can be used on Ω-cm
product wafers
 Gives doping
profiles
 Used mainly by
150 mm wafer
wafer
manufacturers

ECE 4813 Dr. Alan Doolittle

 Contactless C -V

 Hg probe
 Liquid metal
contacts
 Some “contact”
Photo from MDC Corporation
occurs so some
concern for
contamination (in
production) exists.
 Insulating
substrates can be
used (not shown)
 Two series
capacitors, one being
substantially larger
than the other
Photo from SSM
Corporation

ECE 4813 Dr. Alan Doolittle

 Electrochemical C -V

 Simultaneously
performs CV
analysis while
electrochemically
etching the
semiconductor
 Holes needed for
etching
 P-type is easy
 N-type needs light to
generate holes

 Works best with

direct bandgap
semiconductors but
is used with Si

ECE 4813 Dr. Alan Doolittle

 What Is Measured ?
Debye Length≡ a measure of the
distance over which a charge
kTK sε o imbalance is neutralized by majority
LD = carriers (under steady state conditions).
q 2 ( p + n)
The Debye length sets the spatial limit
(resolution) of an electrically measured
 The previous equations profile.
indicate the doping NA NA
profile is measured NA NA
 The entities that respond p p
p p
to the ac voltage are the
majority carriers, not the
0 x 0 W x
dopant atoms
NA NA
 Detailed modeling has
shown that the majority NA NA
carrier profile is p p
measured p p

0 x 0 W x
ECE 4813 Dr. Alan Doolittle
 Debye, Thomas-Fermi or Other Limit?
 You will hear physicists often using the “Thomas-Fermi” length as a
resolution limit instead of the Debye length – why?
 Both are a measure of the distance over which a charge imbalance is neutralized
by majority carriers (under steady state conditions). The Debye length sets the
spatial limit (resolution) of an electrically measured profile.
 Debye Length valid for non-generate semiconductors at any temperature

kTK sε o
LD =
q 2 ( p + n)

 Thomas-Fermi Screening Length is valid for degenerate semiconductors and metals

and is strictly valid only at low temperatures (but is more generally applied at all
temperatures) 1
 π  6 πK sε o  2
LTF =  
 3( p + n )  q 2 m*

 Quantum Confined Length: When an electron (hole) is confined by a potential well to

form a 2D sheet with planar doping density N2D, its spatial extent is described by a
“wavefunction” that has limited width. In this special and common case, the resolution
limit is described by: 1
7  4 K sε o  2  3
LQCS = 2  
5  9q 2 m* [N 2 D ] 

ECE 4813 Dr. Alan Doolittle

 What Is Measured ?
 C-V and VT profiling methods determine the carrier density,
not necessarily the doping density
 For uniformly doped material: p = NA, n = ND
 For non-uniformly doped material: p ≠ NA, n ≠ ND

NA
NA
p
p
0 x
W.C. Johnson and P.T. Panousis, “The Influence of Debye Length on the C-V Measurement
Doping Profiles,” IEEE Trans. Electron Dev. ED-18, 965-973, Oct. 1973.

 Important Limitation: When the contact area becomes

comparable to the depletion width, a simple parallel plate
capacitor model cannot be used. A 3D solution is needed.
ECE 4813 Dr. Alan Doolittle
 Series Resistance
Capacitance Meter Error
200

Cs rs = 100 -10,000 Ω
Cs

Cs, Cp (pF)
150
Gactual CactualGp Cp
100 100 Ω
Rs
1,000 Ω Cp
rs 50 rs = 10,000 Ω

(a) (b) (c) f = 100 kHz

0
10-7 10-6 10-5 10-4 10-3
G (S)
Device As Seen By C Meter

Cactual   G 
2

Cp = ; Cs = Cactual 1 +  actual  
(1 + rs Gactual ) + (ωrs Cactual )
2 2
  ωCactual  
 Cp, Cs, Gp and Rs are all capacitance meter measured values.
 Series connection is preferred if series resistance is important!
 Never trust a Capacitance measurement with a quality factor (Q=ωC/G) < 5.
ECE 4813 Dr. Alan Doolittle
 “Deep” Concerns
 When an acceptor or defect energy is deep in the bandgap a concern
as to whether the carriers can adequately respond to the ac stimulus
is warranted.

 If for a p-type material with (for example) a deep acceptor at EA eV

above the valance band,
EC
 EA 
 
 kT 
e 1
τ emission = >
σ p vthermal NV ωmeasurement
EA

EV
σ p ≈ 10 −15 cm 2 vthermal ≈ 105 cm / sec NV ≈ 1019 cm −3
 So for 0.16 eV deep acceptor (In in Si or Mg in GaN), τemission~ 0.5 uS
(~5 x [1/ω] for a 1MHz signal). τemission increases by a factor of ~10
when EA=0.22 eV.

 In such cases, the capacitance does not accurately reflect p or NA

ECE 4813 Dr. Alan Doolittle
 Series Resistance
 If more than 2 parameters are needed, then more than one frequency will be
required.
ω22CS 2 − ω12CS1
Cactual =
ω22 − ω12
 There are limits to the 2-frequency technique and cautions should be
exercised to insure measurements are valid. The reader is encouraged to
examine Agilent application note 4294-3 : “Evaluation of MOS Capacitor Oxide
C-V Characteristics Using the Agilent 4294A”, section 7.

 MAIN POINT: LEAKAGE CURRENT CAN LOWER APPARENT CAPACITANCE

AND IS BIAS DEPENDENT – HUGE ERRORS!
 Thin dielectrics
 Semiconductors with inverted surfaces

ECE 4813 Dr. Alan Doolittle

 MOSFET Threshold Voltage
 Based on I-V not C-V so the technique scales better.
 The threshold voltage, VT, dependence on substrate bias
can be used to determine the doping profile under the gate
VG
S VD=0.1V 2qK sε oN A (2φF + VSB )
VT = VFB + 2φF +
Cox
n n
VSB > 0 for n − MOSFET
p
dVT 2qK sε oN A
= =m
VB d 2φF + VSB Cox
2
Cox m2
N A (W ) =
2qK sε o
2K sε o (2φF + VSB )
W =
qN A

ECE 4813 Dr. Alan Doolittle

 MOSFET Threshold Voltage
 Measure VT as a function of back bias VB
 Assume a value for 2φF , say 0.6 V; plot VT vs. (2φF +VB)1/2
 Find NA from the slope,
 Use this NA to then find a new 2φF and replot VT vs. (2φF +VB)1/2
 One or two iterations are sufficient
 Find density NA and depth W and plot the profile
Doping Density (cm )
-3

17
1 10
16
8 10 C-V V
T
16
6 10
16
4 10 SRP
SUPREM
16
2 10

0
0 0.1 0.2 0.3 0.4 0.5
Depth (µm)

D. Feldbaumer and D.K. Schroder, “MOSFET Doping Profiling,”

IEEE Trans. Electron Dev. 38, 135-140, Jan. 1991.

ECE 4813 Dr. Alan Doolittle

 Profiling Limits
Schottky:
 There are two limits
 Close to surface NA
 Junction breakdown
W
102 2K sε oVbi
Si, T=300K W = ≈ 5 − 6LD
Distance (µm)

101 qN A
WBD
100
MOS-C:
10-1
3LD
10-2 NA
10-3
1014 1015 1016 1017 1018 1019 1020 3LD
NA (cm-3)
kTK sε o 410
LD = = cm
q 2N A NA
for Si at T = 300 K
ECE 4813 Dr. Alan Doolittle
 Review Questions
 What is secondary ion mass spectrometry?
 Name a disadvantage of spreading resistance
profiling.
 How is the capacitance measured?
 Why is 1/C2 - V preferred over C – V?
 What is important in contactless C – V?
 What is measured in most profiling techniques,
i.e., doping density or majority carrier density?
 What is the Debye length?
 What does series resistance do?
 How does the threshold voltage technique work?
 What determines the profiling limits?

ECE 4813 Dr. Alan Doolittle