Beruflich Dokumente
Kultur Dokumente
1/97)
Universiti Teknologi Malaysia
Disahkan oleh
Alamat Tetap:
2442, TAMAN KEMAMAN, PROF. MADYA DR. MAZLINA ESA
24000 KEMAMAN,
TERENGGANU. Nama Penyelia
Signature :
Name of Supervisor : PROF. MADYA DR. MAZLINA ESA
Date : 29th. APRIL 2006
IMPROVED CHARACTERISTICS OF RADIO FREQUENCY
INTERDIGITAL CAPACITOR
APRIL, 2006
ii
Signature :
Name of Author : LIM YUN ROU
Date : 29th APRIL 2006
iii
To My Beloved Family
iv
ACKNOWLEDGEMENTS
For the success of this study, I would like to thank all those who helped with
the work undertaken in this study especially to:-
ABSTRACT
ABSTRAK
TABLE OF CONTENTS
CONTENTS PAGE
TITLE i
ADMISSION ii
DEDICATION iii
ACKNOWLEDGEMENTS iv
ABTRACT v
ABSTRAK vi
LIST OF TABLES x
LIST OF FIGURES xi
CHAPTER I INTRODUCTION 1
CONTENTS PAGE
1.3 Objectives 5
2.4 Summary 22
3.1 Introduction 24
CONTENTS PAGE
4.1 Introduction 38
CHAPTER V CONCLUSION 65
5.1 Conclusion 65
REFERENCES 69
x
LIST OF TABLES
LIST OF FIGURES
4.8 Simulated return loss and insertion loss responses for the
conventional IDC with 22 mils effective finger length (l
= 22 mils) and increasing total fingers from 1X, 2X to
3X simulated from 1GHz to 40 GHz. (a) N = 6. (b) N =
12. (c) N = 18. 48
LIST OF SYMBOLS
A - Area
C - Capacitance
d - Distance
f - Frequency
fo - Self resonance frequency
h - Dielectric thickness
L - Inductance
l - Fingers length
l - Effective fingers length
M - Number of capacitors connected in a network
N - Total capacitor fingers
N - Total combline fingers per finger
Q - Electric charges
R - Resistance
s - Spacing between fingers
s - Spacing between combline fingers
t - Conductor height
w - Finger width
w - Combline fingers width
Xc - Reactance
Hr - Dielectric relative permittivity
xvii
LIST OF ABBREVIATIONS
INTRODUCTION
(a) (b)
(c) (d)
Figure 1.1: Different types of CMOS or BiCMOS capacitor. (a) MOSFET
gate oxide capacitor, (b) MIM Capacitor, (c) Double-poly capacitor, (d) IDC
1.3 Objectives
(a)
(b)
Figure 1.2: IDC configurations (a) basic, (b) six fingers.
Electromagnetic simulations.
Simulations of the conventional and modified interdigital configurations were
performed using Ansoft HFSS electromagnetic simulation software.
Chapter one discusses the objectives and scope of the project and gives a
general introduction to RFICs and functions of on chip capacitors in RFIC. This
chapter also clearly discusses the motivation behind the study and the limitations in
order to achieve the project objectives.
8
The design and analysis approach used in this project are elaborated in
Chapter three. The overall activities of this study are discussed in this chapter.
Besides, this chapter also discusses in detail the capacitor modification flow of the
proposed interdigital capacitor with combline structure and simulation flow of the
designed capacitors using Ansoft HFSS simulation software.
In Chapter four, the final results are presented and analyzed in detail. This
includes the modifications from the basic configuration to various combline
configurations.
The final chapter concludes the thesis. Suggestions for further improvement
are also presented.
CHAPTER II
LITERATURE REVIEW
(a) (b)
Figure 2.1: Basic capacitor. (a)Two thin plate conductors with fixed distance d, plate
area A, and charges +Q on one plate and Q on the other form a parallel-plate
capacitor, (b) Voltage is applied across the plates, the electric field in the dielectric
displaces electric charges [8].
10
HA
C (Eqn. 2.1)
d
where H # dielectric constant, A # area of the conductor and d # distant between two
conductors.
Figure 2.2: System configuration consisting of two parallel plates in the air. Fixed
dimension d=8 cm, a=3 mm, h=1 mm [9].
The capacitor susceptance versus frequency curves in Figure 2.3 have been
obtained considering the fixed dimension as shown in Figure 2.2. Three different
localizations of the capacitor terminals have been examined as follow.
11
Case A: d2 = d and d1 = 0
Case B: d2 = 3d/4 and d1 = d/4
Case C: d2 = d/2 and d1 = d/2
Observing the graphs in Figure 2.3, one verifies, for some frequencies that
the susceptance becomes infinite (the source connected at points A/B faces a short
circuit). Some frequencies, the susceptance goes to zero (the source at A/B face an
open circuit as in the DC regime). It is also evident that frequency bands exist where
S(Z) is negative, revealing that the device may behave as an inductive component.
From the results, it can be summarized as: system losses were neglected; dielectric
permittivity H was taken as a frequency-independent parameter; fringing field effects
at the system edges were neglected; geometry dimensions a and h were assumed to
be very small as compared to the operating wavelength. Removal of these
assumptions would lead to a more accurate analysis. The work has shown that a
simple parallel plate structure may exhibit a quite unexpected frequency behavior
when the operating wavelength is relatively smaller compared to the component size.
Dimension do affects the performance of capacitor at high frequency range.
12
Figure 2.3: Plots of the device susceptance against frequency for three different
localization of the input terminals. (a) d2 = 8 cm, d1 = 0 cm. (b) d2 = 6 cm, d1 = 2 cm.
(c) d2 = 4 cm, d1 = 4 cm [9].
13
In real world usage, no capacitor is perfect, and will always exhibit some
finite amount of ESR [10]. The ESR varies with frequency for a given capacitor, and
is equivalent because its source is from the characteristics of the conducting
electrode structures and in the insulating dielectric structure. All practical capacitors
behave similarly to the series RLC network shown in Figure 2.4. For the purpose of
modeling, the ESR is represented as a single series parasitic element in the RLC
network [11].
Figure 2.4: Series RLC network representing behavior of a simple parallel plate
capacitor at high frequency [11].
Frequency Impedance
Low Capacitive
High Inductive
At low frequencies, the capacitors impedance looks just like what everyone would
expect from the specified capacitor value. At self resonance, the capacitive and
inductive impedances cancel each other out leaving only a resistive component. The
resistive component is also referred to as ESR. Above self- resonance, the inductive
reactance takes over as it grows much larger than the capacitive reactance and ESR.
The self resonance is given by the formulation [11]:
1
fo (Eqn. 2.2)
2S LC
XC
Q (Eqn. 2.3)
ESR
where |XC| is the absolute value of the reactance in ohms and is given by [11]:
15
1
XC (Eqn. 2.4)
2S f C
Figure 2.5: Conventional interdigital capacitor layout, (a) Cross section view. (b)
Top view.
Gary D. Alley was the founder of interdigital capacitor and proposed the
equivalent circuit at year 1970 as shown in Figure 2.6 [12]. He analyzed the
frequency response of interdigital capacitors, which leads to an optimal design, and
given along with an expression for the static gap capacitance. The capacitor Q factor
is given in terms of its geometry which consists of a planar interdigital conductor
deposited on the surface of a relatively high dielectric constant substrate.
Capacitance values ranging from 0.1 to 10 pF at L band with measured Qs in excess
of 400 are realizable using 2-rail line and space widths on a 99.5-percent alumina
substrate with a dielectric constant of 10.3 [12]. For a maximum capacitance density
with this configuration, the finger width and space width must be equal.
17
Figure 2.6: Interdigital capacitor in Gary D. Alley study and its low frequency
equivalent circuit [12].
Figure 2.7: Coplanar interdigital capacitor. (a) Layout, (b) Circuit model [13].
18
Figure 2.8: Series capacitance of coplanar interdigital capacitor with finger length
varying from 100 m to 350 m [13].
The structure in Figure 2.9 was analyzed by matched at both ends, and
excited by a 50 1 voltage source of unit amplitude. The substrate has thickness h =
1.27 mm and conductivity g = 0.01 S/m, while the ground plane and the metallic
traces forming the device have thickness t = 34 m and conductivity g = 5.8u107
S/m. The magnitude of the scattering parameters in the frequency range 06 GHz,
obtained employing the FDTD technique, is given in Figure 2.10. Comparing these
curves, it appears that only a small high-frequency circuital degradation takes place
in the structure having zig-zag fingers. This happens because zig-zag discontinuities
enhance emission phenomena in terms of surface and volume waves. A capacitance
0.49 pF, for the conventional IDC structure, and a capacitance 0.54 pF, for that
having zig-zag fingers, have been found. Hence, obtaining an increment of about
10% of the parameter. This confirms that better performances can be obtained using
IDC structures having zig-zag fingers.
(a) (b)
Figure 2.11: Equivalent circuit for the IDC structure with zig-zag fingers. (a) Quasi
static network, (b) nth high frequency network [14].
2 S 21
C imaginary (Eqn. 2.5)
(1 S11 S 21 ) (1 S11 S 21 ) Z
2 S 21
R real (Eqn. 2.6)
(1 S11 S 21 ) (1 S11 S 21 ) Z
1 S11 2 S 21
C1 imaginary
(Eqn. 2.7)
(1 S11 S 21 ) Z
From Equations 2.5 to 2.7 [15], the capacitance and Q factor of the IDC can
be subsequently obtained using the S-parameters extracted from the model by
applying the following formulation:
22
Xc 1
Q(Z ) (Eqn. 2.8)
R ZC R
2.4 Summary
It has been observed that IDC is one of the passive components being widely
used in RF circuit designs. Thus it is important to have an IDC which is high quality
and high rangebility to ensure that the same capacitor can be used across different
applications and design purposes. It can be summarized that there are several
23
METHODOLOGY
3.1 Introduction
This chapter discusses step by step flow in this project. At the end of this
chapter, a complete set of flowchart of the project is given.
Figure 3.1 is the flow chart showing the overall project activities. This
project begins with the literature review on fundamentals of capacitor, capacitor
behaviors at high frequency range and study of radio frequency integrated circuits
(RFICs). With the understanding of capacitor theory, scope of literature review is
then focused to the previous research done in interdigital capacitors. It is important to
understand the motivation behind interdigital capacitors, how does it works, what is
the goal and limitation in interdigital capacitor design, critical parameters that affect
the performance and what are the techniques to improve and optimize the
performance.
25
By applying the knowledge and theory gained from the literature studies, this
project is then proceed with the investigation of possibilities to improve the
performance and characteristics of conventional interdigital capacitors. Combline
structure IDC layout modification was proposed to improve the capacitance density
and quality factor (Q) of the capacitor. Simulation of the conventional and proposed
modified interdigital capacitors is then performed using Ansoft HFSS
electromagnetic simulation software.
This section explored the effect of capacitor dimension and placement to the
capacitance value and Q factor. Subsequently, IDC modifications with combline
structure are presented. This chapter also fully discussed the step by step design and
simulation flow of the conventional and proposed interdigital capacitor with
combline structure using Ansoft HFSS electromagnetic simulation software.
combline structures are intended to add more coupling between the two microstrip
lines, and subsequently increase the total capacitance of the capacitor.
Figure 3.3: Planar IDC with combline structure and side feeder.
28
When capacitors are connected in parallel, the total capacitance is the sum of
the individual capacitors' capacitances. If two or more capacitors are connected in
parallel, the overall effect is that of a single equivalent capacitor having the sum total
of the plate areas of the individual capacitors. The total capacitance of capacitors
connected in parallel in Figure 3.4 can be calculated by the formulation:
At high frequency range, capacitors experience parasitic ESL and ESR. When
N number of capacitors with equal capacitance value connected in parallel as shown
in Figure 3.5, it gives the equivalent capacitance with lower effective ESL and ESR.
The effective values are as follow.
29
ESL
ESLeq (Eqn. 3.2)
N
ESR
ESReq (Eqn. 3.3)
N
C eq C u N (Eqn. 3.4)
where ESLeq = Total ESL in parallel connection, ESL = ESL value for each
individual capacitor, ESReq = Total ESR in parallel connection, ESR = ESR value for
each individual capacitor, Ceq = Total capacitance in parallel connection, C =
Capacitance value for each individual capacitor and N = number of capacitors
connected in the network
Equivalent to
Figure 3.5: Capacitor model operating at high frequency range and the simplified
model [17].
To prove that the circuits are equivalent, two models of impedance versus
frequency are computed. The models are parallel and series equivalent circuits,
respectively. The relationships are shown in Figure 3.6. It is clearly seen that both
curves correlated very well.
30
(a) (b)
Figure 3.6: Relationships of impedance versus frequency, (a) parallel and (b) series
equivalent circuit.
(a) (b)
(c)
Figure 3.7: Comparison of different capacitor impedance profiles. (a) One 10 nF versus
two 5 nF capacitors. (b) One 20 nF versus two 10 nF capacitors. (c) One 20 nF versus
four 5 nF capacitors.
a factor of 2 respectively. However, the additional fingers might create more mutual
coupling between fingers end which may increase the fringing effect. Thus, it is
important to simulate and validate these proposed configurations to ensure the
fringing effect will not affect the overall performance of the capacitor.
C1 C2 C3 C4 C5
C1
C2
C3
+ -
C4
C5
conventional IDC simulation is used as the control case while the combline IDC
simulation is used as the experimental case and compared to the control case. The
analysis was done on three-dimensional (3D) basis in order to check the behavior of
both IDCs under high frequency operating condition. This section focuses on the
description of model dimension, material, experimental setting and other critical
criteria of simulation and analysis.
The study is divided into two sections to investigate the effects of finger
length and total capacitor width to the characteristics of the capacitor. Some
assumptions made in the study are tabulated in Tables 3.1 to 3.4. Simulations were
34
then carried out by sweeping the frequency from 1 GHz to 40 GHz. The simulation
results in terms of S, Y and Z parameters were obtained.
Table 3.1: Settings and parameter assumptions in HFSS simulation for experiment 1.
Parameter Values
Total fingers, N 6
(b) 68 mils
(b) 66 mils [ 3X ]
Table 3.2: Settings and parameter assumptions in HFSS simulation for experiment 2.
Parameter Values
Total fingers, N 6
(b) 68 mils
(b) 66 mils [ 3X ]
(a) 3
(c) 14
36
Table 3.3: Settings and parameter assumptions in HFSS simulation for experiment 3.
Parameter Values
(b) 12 [ 2X ]
(c) 18 [ 3X ]
Table 3.4: Settings and parameter assumptions in HFSS simulation for experiment 4.
Parameter Values
(b) 12 [ 2X ]
(c) 18 [ 3X ]
4.1 Introduction
This chapter presents the simulation results extracted from HFSS for both the
conventional and combline IDC configurations with the four experimental settings
stated in Chapter 3. The computation of capacitance values and Q factor of the IDCs
from the simulation results are also presented. The results presented are then
discussed in depth.
This section presents the S- and Z-parameters extracted from HFSS for both
conventional and combline IDCs. These values are used in the later part in the
studies to mathematically compute the capacitance and Q factor based on
formulations presented in Chapter 3. Figure 4.1 shows the simulation models built
in HFSS for experiment 1 as mentioned in Chapter 3 with the assumptions and
settings stated in Table 3.1. Two 50 ohm lumped ports are assigned to the input and
39
output of the models for impedance matching purposes. These models are then
simulated by sweeping the frequency from 1 GHz to 40 GHz with 0.1 GHz
incremental step. Simulation results are shown in Figure 4.2 in the form of return
loss (S11) and insertion loss (S21) responses. The S22 and S12 values are assumed to be
identical to the S11 and S21 measured values since a capacitor is a symmetrical 2-port
device.
(a) (b)
(c) (d)
Figure 4.1: Simulation models in HFSS for conventional IDC with 6 fingers (N=6)
and various effective fingers length, l. (a) Simulation model with 45 view angle.
(b) l = 22 mils [1X], (b) l = 66 mils [3X] and (c) l = 110 mils [5X].
40
Legend:
S11
S21
(a)
Legend:
S11
S21
(b)
Legend:
S11
S21
(c)
Figure 4.2: Simulated return and insertion loss responses for the conventional IDC
with N=6 and increasing finger length from 1X, 3X to 5X simulated from 1 GHz to
40 GHz. (a) l = 22 mils, (b) l = 66 mils and (c) l = 110 mils.
41
From Figure 4.2, it is clearly observed that the slopes of S11 and S21 profiles
change more frequently as the finger length increases from 1X, 3X to 5X. It can be
seen that a dip of S11 = -27 dB occurred at 24 GHz in Figure 4.2(a). Two peaks of
-13 dB and -7 dB appeared at 13 GHz and 38 GHz, respectively. The S21 profile is
seen to be decreasing as the frequency increases. The two profiles cross each other
at 28 GHz when both are of -15 dB. In Figure 4.2(b), the IDC with finger length 3X
longer than the previous model showed that both S11 and S21 have more peaks and
dips, indicating more resonances within the RF range.
The same trend is observed in Figure 4.2(c), where the finger length is 5X
longer than the first model. More resonances appeared. Thus, it can be concluded
that with the same total number of fingers, as fingers length increase, more resonant
present in the same frequency range. This indicates that the capacitor has beneficial
decoupling effects at the self resonant frequencies. As the fingers length increase, it
can be widely used for various applications which operate at different operating
frequencies.
100
10
|Im[Zin/50]|
1
0 5 10 15 20 25 30 35 40
0.1
0.01
Freq (GHz)
(a)
10
1
0 5 10 15 20 25 30 35 40
|Im[Zin/50]|
0.1
0.01
Freq (GHz)
(b)
10
1
0 5 10 15 20 25 30 35 40
|Im[Zin/50]|
0.1
0.01
0.001
Freq (GHz)
(c)
Figure 4.3: Normalized impedance profile for conventional IDC with N = 6 and
increasing effective fingers length from 1X, 3X to 5X simulated from 1 GHz to 40
GHz. (a) l = 22 mils, (b) l = 66 mils and (c) l = 110 mils.
43
The same experiment has been done by simulating the modified IDC with
combline structures added to the capacitor fingers. The purpose of adding the sub-
fingers is to increase the coupling effect between the fingers and subsequently
increase the capacitance. The simulation models built in HFSS for experiment 2 as
mentioned in Chapter 3 with the assumptions and settings stated in Table 3.2 are
shown in Figure 4.4. The same approach is used to match the impedance where two
50 ohm lumped ports are assigned at the input and output sections of the model. The
extracted S11 and S21 are shown in Figure 4.5.
(a) (b)
(c) (d)
Figure 4.4: Simulation models in HFSS for combline IDC with 6 fingers (N = 6) and
various effective fingers length, l. (a) Simulation model with 45 view angle. (b) l
= 22 mils [1X], (b) l = 66 mils [3X] and (c) l = 110 mils [5X].
44
Legend:
S11
S21
(a)
Legend:
S11
S21
(b)
Legend:
S11
S21
(c)
Figure 4.5: Simulated return and insertion loss responses for the combline IDC with
N=6 and increasing finger length from 1X, 3X to 5X simulated from 1 GHz to 40
GHz. (a) l = 22 mils, (b) l = 66 mils and (c) l = 110 mils.
45
From Figure 4.5 it is clearly shown that the impact of increasing finger
lengths to S11 and S21 for the combline IDC is similar as the conventional IDC.
Increasing the capacitor finger length results in more resonances appears in both S11
and S21. Besides, looking from another perspective by comparing the S11 and S21
responses of conventional and combline IDC with the same number of fingers and
same finger length, it is noted that the combline IDC has more resonances compared
to conventional IDC. This behaviour is the expected since combline IDC has
additional sub-fingers that caused more coupling between the fingers end. By
considering both graphs in Figures 4.2(a) and 4.5(a) as examples, S11 for
conventional IDC has one dip while S11 for combline IDC has 2 dips in the same
frequency range. This indicates that the combline IDC which has the same attributes
with the conventional IDC can be used in a wider frequency range for more
applications which operate in different operating frequencies. The same trend can be
observed for Figures 4.2(b) and 4.5(b), and Figures 4.2(c) and 4.5(c).
10
1
0 5 10 15 20 25 30 35 40
|Im[Zin/50]| 0.1
0.01
Freq (GHz)
(a)
10
1
0 5 10 15 20 25 30 35 40
|Im[Zin/50]|
0.1
0.01
Freq (GHz)
(b)
10
1
0 5 10 15 20 25 30 35 40
|Im[Zin/50]|
0.1
0.01
Freq (GHz)
(c)
Figure 4.6: Normalized impedance profile for the combline IDC with N = 6 and
increasing finger length from 1X, 3X to 5X simulated from 1 GHz to 40 GHz. (a) l
= 22 mils, (b) l = 66 mils and (c) l = 110 mils.
47
The next part of the study is to investigate the impact of increasing the
capacitor width by adding total finger number to the capacitor performance. The
following discussion focuses on this topic. Three conventional IDC models with the
same attributes, but maintaining total finger number, are built and simulated using
HFSS. The assumptions are stated in Table 3.3. The models are shown in Figure
4.7 while the extracted S11 and S21 for each case are illustrated in Figure 4.8.
(a) (b)
(c) (d)
Figure 4.7: Simulation models in HFSS for conventional IDC with 22 mils effective
fingers length (l = 22 mils) and various finger numbers, N. (a) Simulation model
with 45 view angle., (b) N = 6 [1X], (c) N = 12 [ 2X] and (c) N = 18 [3X].
48
Legend:
S11
S21
(a)
Legend:
S11
S21
(b)
Legend:
S11
S21
(c)
Figure 4.8: Simulated return loss and insertion loss responses for the conventional
IDC with 22 mils effective finger length (l = 22 mils) and increasing total fingers
from 1X, 2X to 3X simulated from 1GHz to 40 GHz. (a) N = 6. (b) N = 12. (c) N =
18.
49
It can be observed from Figure 4.8 that more resonances occurred as the
width of capacitor increases from 1X, 2X to 3X of the first model. This indicates
that capacitor width did affect the performance of the capacitor. Thus, capacitor
width is one of the factors that can be modified to improve its characteristics.
However, the amplitude delta of increasing the finger length is higher and more
significant compared to finger numbers increment. Despite this, another observation
from Figure 4.8 is that as the capacitor width increases, the first resonance occurred
at a lower frequency. For example, the first resonance observed in Figure 4.8(a) is at
24 GHz while the first resonance observed in Figure 4.8(b) is at 12 GHz and the
subsequent resonances are at 25 GHz and 35 GHz.
The extracted Z11 from HFSS are then normalized and the normalized
impedance profile for conventional IDC with different finger numbers are plotted in
Figure 4.9. The normalized impedance shows similar trend where more resonances
can be observed as the capacitor width increased.
50
100
10
|Im[Zin/50]|
1
0 5 10 15 20 25 30 35 40
0.1
0.01
Freq (GHz)
(a)
10
1
|Im[Zin/50]|
0 10 20 30 40
0.1
0.01
Freq (GHz)
(b)
10
1
0 5 10 15 20 25 30 35 40
|Im[Zin/50]
0.1
0.01
freq (GHz)
(c)
Figure 4.9: Normalized impedance profile for conventional IDC with 22 mils
effective fingers length (l = 22 mils) and increasing total fingers from 1X, 2X to 3X
simulated from 1GHz to 40 GHz. (a) N = 6. (b) N = 12. (c) N = 18.
51
Three combline IDC models with the same attributes but with varying total
finger number are built and simulated using HFSS. The assumptions are stated in the
Table 3.4. The models are shown in Figure 4.10 while the extracted S11 and S21 for
each case are illustrated in Figure 4.11. The normalized impedance profiles are also
illustrated in Figure 4.12.
(a) (b)
(c) (d)
Figure 4.10: Simulation models in HFSS for combline IDC with 22 mils effective
fingers length (l = 22 mils) and various finger numbers, N. (a) Simulation model
with 45 view angle, (b) N = 6, (c) N = 12 and (d) N = 18.
52
Legend:
S11
S21
(a)
Legend:
S11
S21
(b)
Legend:
S11
S21
(c)
Figure 4.11: Simulated return loss and insertion loss responses from 1 GHz to 40
GHz for combline IDC with 22 mils effective finger length (l = 22 mils) and
increasing total fingers. (a) N = 6. (b) N = 12. (c) N = 18.
53
10
1
0 5 10 15 20 25 30 35 40
|Im[Zin/50]|
0.1
0.01
Freq (GHz)
(a)
10
1
0 5 10 15 20 25 30 35 40
|Im[Zin/50]|
0.1
0.01
0.001
Freq (GHz)
(b)
10
1
0 5 10 15 20 25 30 35 40
|Im[Zin/50]|
0.1
0.01
Freq (GHz)
(c)
Figure 4.12: Normalized impedance profile for combline IDC with 22 mils effective
fingers length (l = 22 mils) and various finger numbers, N, (a) N = 6, (b) N = 12 and
(c) N = 18.
54
From Figures 4.11 and 4.12, it can be concluded that the insertion loss, return
loss and impedance profile of combline IDC have the same trend as the conventional
IDC when the capacitor width increases. Nevertheless, the resonances for the
combline IDC is appear more frequent compared to conventional IDC within the
same frequency range. These results are as expected and are attributed by the
presence of additional sub-fingers in the combline IDC cause more coupling
effects compared to the conventional IDC. Thus, impedance varies more as the
frequency increase and thus causes more reflections as a signal passes through the
capacitor. Hence, the return loss and insertion loss vary and resonate more as the
frequency increases.
From the S- and Z-parameters obtained in Section 4.2, the values are used to
calculate the capacitance and Q factor of the capacitors for performance and
characteristics analysis. This section illustrates the calculated capacitance and Q
factor for each capacitor in the four experiments discussed in Section 4.2. The
capacitances of the capacitor with respect to frequency are calculated based on
Equation 2.5, while the Q factor with respect to frequency are calculated based on
Equation 2.8. The computed relationships are presented in Figures 4.13, 4.15 and
4.16 as the capacitance versus frequency comparison plots for both the conventional
and combline IDCs with various finger length within 1 GHz to 40 GHz frequency
range; and Figures 4.14, 4.16 and 4.18 are the Q factor versus frequency comparison
plots for both the conventional and combline IDCs with various finger length within
1 GHz to 40 GHz frequency range.
55
1E-11
9E-12
8E-12
C apacitance (F) 7E-12
6E-12
Conventional_IDC
5E-12
Combline_IDC
4E-12
3E-12
2E-12
1E-12
0
1 4 7 10 13 16 19 22 25 28 31 34 37 40
Freq (GHz)
Figure 4.13: Capacitance versus frequency trend for both conventional and combline
IDCs with 22 mils effective finger length (l = 22 mils) and 6 fingers (N = 6).
Quality Factor for Conventional and Combline IDC with 24 mils finger and N=6
900
800
700
Q u a l ity F a c to r (Q )
600
500 Conventional IDC
400 Combline IDC
300
200
100
0
1.00E+00 1.00E+01 1.00E+02
Freq (GHz)
Figure 4.14: Q factor versus frequency trend for both the conventional and combline
IDCs with 22 mils effective finger length (l = 22 mils) and 6 fingers (N = 6).
56
In Figure 4.13, it is observed that the conventional IDC has the first
resonance frequency at 14 GHz with 6.69 pF capacitance; and the second resonance
is at 36 GHz with 1.09 pF. At these resonances, the Q factor is 0.8083 and 0.4755,
respectively. With the same attributes, combline IDC has more resonances within
1 GHz to 40 GHz frequency range. Its first resonance occurs at 8 GHz with 9.52 pF
capacitance. The next resonance is at 21 GHz with 6.5 pF capacitance while the third
resonance is at 31 GHz with 3.7 pF capacitance. The corresponding Q factors are
1.0394, 0.105 and 0.0683, respectively. It can be concluded that the combline IDC
has higher capacitance and more resonances compared to conventional IDC in the
same frequency range. Besides, the first resonance frequency for combline IDC is
generally lower compared to conventional IDC. However from Figure 4.14, it is
shown that the Q factor for combline IDC is always lower compared to conventional
IDC at all frequencies. This is due to the higher resistive loss caused by the
additional sub-fingers of the combline structures. Generally, Q factor decrease
exponentially as frequency increase for both conventional and combline IDC.
Table 4.1 summarizes the capacitance values and Q factors for both the
conventional and combline IDCs at the respective resonance frequencies obtained
from Figures 4.13 and 4.14.
Table 4.1: Summary of Capacitance and Q factor for both the conventional and
combline IDCs with 22 mils effective finger length and 6 fingers (N = 6).
Capacitance Vs Frequency for Both Conventional and Combline IDC with 66 mils finger and N=6
6.00E-12
5.00E-12
4.00E-12
C ap a citace (F )
C_conventional_IDC
3.00E-12
C_combline_IDC
2.00E-12
1.00E-12
0.00E+00
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39
Freq (GHz)
Figure 4.15: Capacitance versus frequency trend for both the conventional and
combline IDCs with 66 mils effective finger length (l = 66 mils) and 6 fingers
(N = 6).
Quality Factor for Conventional and Combline IDC with 66mils finger and N=6
500
400
Q u a lity F a c to r (Q )
300
Conventional IDC
200
Combline IDC
100
0
1.00E+00 1.00E+01 1.00E+02
-100
Freq (GHz)
Figure 4.16: Q factor versus frequency trend for both the conventional and
combline IDCs with 66 mils effective finger length (l = 66 mils) and 6 fingers
(N = 6).
58
Figure 4.15 shows the capacitance versus frequency for both the conventional
and combline IDCs with 66 mils effective finger length and 6 fingers. Figure 4.16
shows the corresponding Q factor with respect to frequencies ranges from 1 GHz to
40 GHz. Both plots showed the same trend as the comparison plot for the
conventional and combline IDCs with 22 mils effective finger length. Overall,
combline IDC exhibits higher capacitance and more resonances with a lower first
resonance frequency compared to the conventional IDC. Q factor for combline IDC
is lower across the frequency range compared to conventional IDC. For conventional
IDC with 66 mils effective finger length, four main resonance frequencies are at 6
GHz, 14 GHz, 20 GHz and 30 GHz with corresponding 3.82 pF, 4.21 pF, 1 pF and
2.04 pF capacitances. The Q factors at these frequencies are 3.7637, 0.3523, 0.963
and 0.3783, respectively. As mentioned previously, Q factor decrease exponentially
as frequency increase. Combline IDC in this case has six resonance points, which
are at 4 GHz with 5.38 pF capacitance, 9 GHz with 1.01 pF capacitance, 13 GHz
with 0.8 pF capacitance, 20 GHz with 4 pF capacitance, 30 GHz with 0.56 pF
capacitance and lastly at 36 GHz with 1.37 pF capacitance. Out of the six resonance
points, the highest Q factor is obtained at the first resonant frequency with Q factor
of 4.077. Q decreases exponentially across the frequency range.
Table 4.2 summarizes the capacitance values and Q factors for both the
conventional and combline IDCs at their respective resonance frequencies obtained
from Figures 4.15 and 4.16.
59
Table 4.2: Summary of Capacitance and Q factor for both the conventional and
combline IDCs with 66 mils effective finger length and 6 fingers (N = 6).
As the effective finger length increases from 66 mils to 110 mils, Figure 4.17
shows its capacitance versus frequency of the conventional and combline IDC while
the Q factor is presented in Figure 4.18. From Figure 4.17, it is clearly shown that
there are more resonances occurred for IDC having 110 mils fingers compared to the
previous IDC with 66 mils fingers. Q factor for this experiment setting is lower
compared to the IDC having 66 mils fingers. In addition, comparison between the
conventional and combline IDCs having 110 mils fingers showed that the combline
IDC possesses lower Q factor. The capacitance and Q factor at resonance frequency
observed in both Figures 4.17 and 4.18 are summarized in Table 4.3.
60
Capacitance Vs Freq for Both Conventional and Combline IDC with 110mils finger and =6
6.00E-12
5.00E-12
Conventional_IDC
3.00E-12
Combline_IDC
2.00E-12
1.00E-12
0.00E+00
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39
Freq (GHz)
Figure 4.17: Capacitance versus frequency trend for both conventional and combline
IDC with 110 mils effective finger length (l = 110 mils) and 6 fingers (N = 6).
Quality Factor for Conventional and Combline IDC with 110mils finger and
N=6
300
250
Q u ality F acto r (Q )
200
Conventional IDC
150
Combline IDC
100
50
0
1.00E+00 1.00E+01 1.00E+02
Freq (GHz)
Figure 4.18: Quality factor (Q) versus frequency trend for both conventional and
combline IDC with 110 mils effective finger length (l = 110 mils) and 6 fingers
(N = 6).
61
Table 4.3: Summary of Capacitance and Q factor for both conventional and
combline IDC with 110 mils effective finger length and 6 fingers (N = 6).
900
Legend:
800
700 l = 22 mils
l = 66 mils
Q u a l i ty F a c to r (Q )
600
300
200
100
0
1.00E+00 1.00E+01 1.00E+02
Freq (GHz)
Figure 4.19: Comparison of Q factor for conventional IDC with various finger
lengths.
700
Legend:
600
l = 22 mils
500
l = 66 mils
400
Q u a lity F a cto r (Q )
l = 110 mils
300
200
100
0
1.00E+00 1.00E+01 1.00E+02
-100
Freq (GHz)
Figure 4.20: Comparison of Q factor for combline IDC with various finger lengths.
The profiles in Figure 4.20 have the same trend for the combline IDC where
Q dropped not only when frequency increases, but also when the effective finger
length is increased. It is observed that the increment in effective finger length from
22 mils to 66 mils has yield the Q dropped from Q22mils = 580 to Q66mils = 290, which
is equivalent to 50% drop. On the other hand, the increment in effective finger
length from 66 mils to 220 mils has yield the Q dropped from Q66mils = 290 to Q110mils
= 150, which is equivalent to approximately 48% drop.
63
The same comparison can be made between the Q factor of the conventional
IDC with various total finger numbers as well as the combline IDC with various total
finger numbers. The purpose of this comparison is to clearly illustrate the effect of
increasing the capacitor width to its Q factor. The results are shown in Figures 4.21
and 4.22 for the conventional IDC and combline IDCs, respectively.
The maximum Q factor for the conventional IDC with 6 fingers and 22 mils
effective finger length can be obtained at 1 GHz with Q780. This value decreases
exponentially as frequency goes up and then it became close to zero when the
frequency goes higher than 10 GHz. By maintaining the same finger length, another
six fingers are added to the capacitor and thus cause the width to be 2X of the
original model. From the graph, it shows that the Q dropped close to 600, which is
approximately 23% drop. This experiment proceeded by adding another 6 more
fingers to this model and now the capacitor has a total of 18 fingers. The Q factor
again dropped to approximately 400, which is approximately 33% lower than the
second model.
y g
900
Legend:
800
700 N=6
600
N = 12
Quality Factor (Q)
500
N = 18
400
300
200
100
0
1 10 100
-100
Fre q (GHz)
Figure 4.21: Comparison of Q factor for the conventional IDC with various total
finger number, N.
64
700
Legend:
600
N=6
500
N = 12
400
Quality Factor (Q)
N = 18
300
200
100
0
1 10 100
-100
Freq (GHz)
Figure 4.22: Comparison of Q factor for the combline IDC with various total finger
number, N.
Figure 4.22 shows the comparison of Q factor for the combline IDC with
various finger numbers. For the combline IDC with 22 mils effective finger length
and 6 fingers, the highest Q can be obtained at 1 GHz where Q580. This is much
lower compared to the conventional IDC having the same attributes. As the
capacitor width doubled up, Q at 1 GHz dropped to 350 and it continued to drop
when the capacitor width increased to 3X of the original model. The Q factor when
the capacitor has 18 fingers is approximately 180 at 1 GHz. The Q has dropped 40%
as the capacitor width doubled and dropped 69% as the capacitor width becomes 3X
of the original width.
CHAPTER V
CONCLUSION
5.1 Conclusion
Microstrip IDCs are key passive lumped elements, which are widely used in
microwave integrated circuits. The electrical characteristics of IDCs are of interest
over a frequency band, and this has become more crucial as the frequency of most
applications is increasing from Megahertz to Gigahertz range. This thesis presented
an overview of underlying principles of IDC, applications of IDC, and limitations of
the existing IDC. An experimental investigation through simulations on two
attributes of IDC, which are the finger length and capacitor width have been
performed. Besides, a new layout configuration with combline structures has been
explored and simulated.
Comparing conventional and combline IDC with the same total fingers,
combline IDC demonstrates higher capacitance. However the first self
66
resonance frequency for the combline IDC is lower than that of the
conventional IDC.
Q factor for combline IDC is always lower than conventional IDC. This is
due to higher resistive loss seen in combline IDC. Q decreases exponentially
with frequency increase. Higher resistive loss is observed in higher
frequency range.
There are some areas that can be further improved and investigated in order
to obtain a combline IDC with better performance. Since the proposed combline
capacitor managed to achieve a sufficiently high Q due to high resistive substrate
loss, more studies can be done concentrating in low lost substrate material to reduce
resistive loss and increase Q factor. This is important because a capacitor with high
capacitance but low Q is generally not beneficial as most design require high Q
capacitors.
67
This project can also be enhanced with a study to obtain the accurate
electrical model to represent the proposed IDC, so that it can be used in simulations
to accurately represent the behavior of the IDC especially at high frequency region.
Modeling of the proposed IDC is beyond the scope of this thesis.
Studies should be done to look into different capacitor layouts that are able to
produce higher capacitances while having high quality factor. Research can be done
in exploring effect of stack-up (such as coplanar or stripline) to the behavior of IDC.
Figures 5.1 and 5.2 show two different configurations which can be considered.
Figure 5.1: Planar IDC with combline structure and center feeder.
68
Figure 5.2: Coplanar IDC with combline structure and center feeder.
69
REFERENCES
[1] C.-J. Chao, S.-C. Wong, C.-H. Kao, M.-J. Chen, L.-Y. Leu, K.-Y. Chiu,
Characterization and Modeling of On-Chip Spiral Inductors for Si RFICs,
IEEE Transactions on Semiconductor Manufacturing, vol. 15, No. 1,
February 2002, pp. 19-29.
[2] National Semiconductor Corporation, 0.25um CMOS Process Electrical
Design Rules and Characterization Data, May 2001.
[3] TSMC, CMOS RF/ Mixed-Signal 0.25um process design guide, 2000.
[4] Jeffery A. Babcock, et. Al, Analog Characteristics of Metal-Insulator-Metal
Capacitors using PECVD Nitride Dielectrics, IEEE Electron Device
Letters, vol. 22, No. 5, May 2001, pp. 230-232
[5] http://www.eagleware.com/pdf/apps/0009_AccuracyProofMidcap.pdf
[6] J. Kim, J.-O. Plouchartz, N. Zamdmerl, M. Sheronyl, L.-H. Lul, Y. Tan,
M. Yoon, Keith A. Jenkins, M. Kumarl, A. Ray, L. Wagner, 3-Dimensional
Vertical Parallel Plate Capacitors in an SO1 CMOS Technology for
Integrated RF Circuits VLSl Circuits Digest Symposium, 2003.
[7] S. Jenei, S. Decoutere, Stefaan V. Huylenbroeck, G. Vanhorebeek, B.
Nauwelaers, High Q Inductors and Capacitors on Si substrate, IEEE, 2001,
pp. 64-70.
[8] http://www.alaska.net/~natnkell/simplcap.html
[9] J. A. Brandao Fario, Capacitor Modeling at High Frequencies A
Pedagogigal Example of Application of Distributed Circuit Analysis for
Introductory Electromagnetic or Microwave Courses, IEEE Transactions
on Education, vol. 35, No. 3, August 1992, pp. 214-216.
[10] Johanson Technology Corporation, Q and ESR Explained, November 2004.
[11] http://www.ecircuitcenter.com/Circuits/cmodel1/cmodel1.htm
70