Process Optimization 4H-SiC SiO2 dielectric interface.

DEGREE PROJECT, IN NANOTECHNOLOGY , SECOND LEVEL
STOCKHOLM, SWEDEN 2015
Process optimization for the
4H-SiC/SiO2 interface.
CHARILAOS-KIMONAS KARALAS
KTH ROYAL INSTITUTE OF TECHNOLOGY
SCHOOL OF INFORMATION TECHNOLOGY

i
Acknowledgements
First of all, I would like to express my gratitude to Professor Anders Hallèn for giving me chance of
working with him. His collected behaviour and his advices were of great help and have importance in achieving
my goals.
I would also like to thank my supervisor Sethu Saveda Suvanam for always being there when I need help
and also for the encouragement he gave me throughout thesis work.
Furthermore, I would like to thank Kostakis for all his help from the very first moment that I arrived
Sweden. Also, I would like to thank my friends Milad and Öner for all their support during my studies, without
forgetting of course the help from Linnea and Sb.
A great thank goes to a special person, Matilda, for being patient and tolerating me and my anxiety all this
time.
Moreover I would like to thank my family and my friends from Greece, Giannis, Martha, Evdokia,
Menelaos, Kostas and Maria. Without their help and support, at some intense periods in my life, I might never
have been here. And last but not least, I thank my Master Christos for making me believe in myself again and my
old mentor, Professor Xanthos N. Maintas.
Finally this thesis is dedicated to all the oppressed of this world, with the expression of the desire and hope
that any kind of exploitation in this society will be extinct one day.

iii
Abstract
This thesis aims to optimize the process for the 4H-SiC/SiO2 interface formations. The experiments are
made on metal-oxide-semiconductor (MOS) structures, where the semiconductor is an n-type epitaxially
grown 4H-SiC thin film. The oxide is fabricated either with thermal oxidation, or by using plasma-enhanced
chemical vapour deposition (PeCVD), utilising two different tools, Precision 5000 Mark II (P5000) and
Plasmalab 80Plus system (Pekka). The deposition temperature is varied for the thermally grown oxide, while
power, pressure and gas ratio of N2O/SiH4 is investigated for the PeCVD method. Also the post deposition
annealing (PDA) temperature is studied for both techniques. The oxide formation and PDA is done in N2O
ambient in order to study the effect of nitrogen passivation of the traps that exist at the interface of 4H-
SiC/SiO2. After the dielectric formation the structures are electrically and structurally characterized. The
electrical characterization is done by capacitance-voltage (CV) and current-voltage (IV) measurements while
the structural characterization is done with atomic force microscopy (AFM). The density of interface traps
(Dits) is extracted using the Terman method from CV data. It is observed that the flatband voltage drops
almost to zero when the samples are annealed in nitrogen rich ambient, resulting in a more electrically
uniform oxide. Also, Dits can also be reduced by nitrogen treatment when the oxide is deposited by the
PeCVD technique. However, it appears that the Terman method cannot determine the amount of traps along
the entire bandgap and it is clear that a large amount of Dits are still present closer to the conduction band.
Finally, it is found that there is a larger spread in the data extracted from the samples deposited by P5000 in
comparison to Pekka, indicating that Pekka is a more reliable tool for oxide deposition in SiC substrate.

v
Contents
CHAPTER 1. INTRODUCTION..................................................................................................... 1
CHAPTER 2. MATERIAL PROPERTIES AND MOS THEORY................................................. 3
2.1 MATERIAL PROPERTIES.................................................................................................................. 3
2.1.1 Silicon Carbide properties ................................................................................................... 3
2.1.2 Silicon Dioxide properties.................................................................................................... 4
2.1.3 Traps in SiC/SiO2 interface.................................................................................................. 4
2.1.4 Passivation of the interface traps......................................................................................... 5
2.2 MOS THEORY................................................................................................................................ 6
CHAPTER 3. FABRICATION PROCESS...................................................................................... 9
3.1 PRE-DEPOSITION SURFACE CLEANING PROCESS ............................................................................ 9
3.2 OXIDE DEPOSITION........................................................................................................................ 9
3.2.1 PeCVD ................................................................................................................................. 9
3.2.1.1 P5000.........................................................................................................................................10
3.2.1.2 Pekka.........................................................................................................................................11
3.2.2 Thermal Growth................................................................................................................. 11
3.3 POST OXIDATION/DEPOSITION ANNEALING .................................................................................. 12
3.4 GATE METALLIZATION................................................................................................................. 12
3.4.1 Lithography........................................................................................................................ 12
3.4.2 Aluminium etching ............................................................................................................. 13
3.5 FABRICATION PARAMETERS OF THE SAMPLES ............................................................................. 13
CHAPTER 4. CHARACTERIZATION................................................................................................. 15
4.1 ELECTRICAL MEASUREMENTS..................................................................................................... 15
4.1.1 Probestation 3 semi-automatic........................................................................................... 15
4.1.2 CV ...................................................................................................................................... 15
4.1.2.1 Defects in the oxide...................................................................................................................17
4.1.2.2 Terman Method .........................................................................................................................18
4.1.3 IV........................................................................................................................................ 19
4.2 AFM............................................................................................................................................ 19
4.2.1 AFM system........................................................................................................................ 19
4.2.2 Tapping Mode .................................................................................................................... 20
CHAPTER 5. RESULTS................................................................................................................ 23
5.1 P5000.......................................................................................................................................... 23
5.1.1 Influence of Pressure.......................................................................................................... 23
5.1.2 Influence of Power ............................................................................................................. 24
5.1.3 Influence of Gas ratio......................................................................................................... 25
5.1.4 Influence of Annealing Temperature.................................................................................. 26
5.2 PEKKA ......................................................................................................................................... 27
5.2.1 Influence of Pressure.......................................................................................................... 28
5.2.2 Influence of Gas ratio......................................................................................................... 29
5.2.3 Influence of Annealing Temperature.................................................................................. 30
5.3 THERMAL GROWTH ..................................................................................................................... 32
5.3.1 Influence of Growth Temperature...................................................................................... 32

vi
5.3.2 Influence of Annealing ....................................................................................................... 33
CHAPTER 6. DISCUSSION.......................................................................................................... 35
CHAPTER 7. CONCLUSIONS ..................................................................................................... 39
APPENDIX..................................................................................................................................... 41
REFERENCES............................................................................................................................... 43

1
Chapter 1. Introduction
By considering the main criteria for high power electronic systems, such as power density, efficiency,
and reliability, one can see the limitation of silicon based devices. For example, the maximum temperature for
operation of Si based devices is 150-175 o
C, which make them unsuitable for harsh environmental conditions
and high power densities [1]. Due to this fact, researchers head for new alternative material systems for high
power and high temperature applications. Silicon carbide of the 4H polytype (4H-SiC), which is one of the
three SiC polytypes that are used in power electronics, has very suitable properties for these applications, such
as wide band gap (3.2 eV), high thermal conductivity (4.9 W/cmK), high saturated carrier velocity (2×107
cm/s) and high breakdown electric field (2.2 MV/cm) [1,2]. With these excellent material properties, 4H-SiC
has become a leading material for future electronic power devices. Another advantage is that the dielectric
material silicon dioxide (SiO2) is the native oxide of SiC and this can be thermally grown [2,3]. Dielectric
materials are needed to passivate the semiconductor surface and also to form gate oxide in field effect devices.
Although promising properties of SiC have been extensively studied and verified, there are still many
challenges waiting to be overcome in order to manufacture reliable devices based on SiC. One of these
challenges is the high density of interface traps between SiC and SiO2 [4]. Such traps can capture electrons or
holes and are very difficult to control in the process. They cause degradation and problems with long term
stability of devices and also lower the mobility of charge carriers underneath the gate in MOS devices.
The main method today to reduce the negative effect of such traps is to passivate them by introducing
nitrogen at the interface. The nitrogen atoms form chemical bonds with defects, thus passivating the electrical
activity at these bonds so they no longer can trap and de-trap charges. During the past two decades different
techniques have developed for adding nitrogen, for instance oxide deposition, or thermal oxidation in N2O
ambient and/or post deposition/oxidation annealing in N2O ambient [4,5,6].
In this thesis two different ways of oxide fabrication and the effect of annealing in N2O ambient will
be studied. The first of them is plasma-enhanced chemical vapour deposition (PECVD) by using two different
instruments, Precision 5000 Mark II (P5000) and Plasmalab 80Plus system (Pekka). The other way to
fabricate the oxide layer is by thermal growth. By varying the different parameters of the fabrication, such as
power, pressure and the gas ratio of N2O/SiH4 for PeCVD, and also the post-deposition/growth annealing
temperature for both methods, we hope to optimize the oxide formation process for 4H-SiC device fabrication.
The samples used for the study are MOS capacitors with n-type 4H-SiC as semiconductor, SiO2 is the
dielectric and a circular aluminium gate forms the metal contact. The characterization techniques that are used
are capacitance-voltage (CV) and current-voltage (IV) for electrical characterization, atomic force microscopy
(AFM) for the structural characterization.
The study shows that it is possible to reduce the interface traps by forming the SiO2 in nitrogen
containing ambient. Post deposition annealing in nitrogen ambient results in a more homogeneous oxide in
terms of electrical behaviour. Finally, when comparing the two setups for SiO2 deposition, Pekka appeared to
be more reliable.

3
Chapter 2. Material properties and MOS theory
In this chapter the properties of SiC and SiO2 will be discussed and also the SiC/SiO2 interface will be
examined. Finally, there will be a short review of MOS theory to define the parameters used for analysis.
2.1 Material properties
2.1.1 Silicon Carbide properties
Silicon carbide (SiC) is composed of stacked layers of closed-packed planes of silicon and carbon
atoms. SiC exists in 300 different crystal structures, which are called polytypes [7], depending on the stacking
sequence of the layers. Among them only three polytypes are presently of interest for the semiconductors
industry due to their remarkable electric properties. The first is 3C-SiC, where C stands for cubic, which is
also called β-SiC. The other two are 4H-SiC and 6H-SiC, where H stands for hexagonal lattice. The hexagonal
lattice polytypes are referred to as α-SiC. The stacking sequence of Si-C bi-planes of these different polytypes
is illustrated in Figure 2.1.
Figure 2. 1 The three different types that are used in the electronics industry [7]
A comparison of the various material properties of the three different polytypes is given in Table 2.1.
In this investigation the 4H-SiC polytype is used.
Table 2.1 Electrical properties of 3C-SiC, 4H-SiC and 6H-SiC [8]
3C-SiC 4H-SiC 6H-SiC
Band Gap (eV) 2.4 3.2 3
Critical Electric Field (MV/cm) 2 2.2 2.5
εr 9.7 10 10
λ (W/cmK) 5 4.9 5
μn (cm2
/Vs) 750-1000 800-1000 370-500
μp (cm2
/Vs) 40 115 90

4
2.1.2 Silicon Dioxide properties
Silicon dioxide (SiO2) is the natural oxide of Si and SiC, which makes it a preferable candidate for
gate dielectric and passivation. Silicon dioxide has a relative dielectric constant of 3.9, a band gap of 9 eV,
and an offset of 2.7 eV in comparison to the conduction band of 4H-SiC. The critical electric field and thermal
conductivity of SiO2 are 10 MV/cm and 0.015 W/cmK, respectively [9]. There are different techniques to
fabricate the SiO2 dielectric, either with thermal growth or by deposition. Deposition can be done in different
ways, such as chemical vapour deposition (CVD), plasma enhanced chemical vapour deposition (PeCVD) and
atomic layer deposition (ALD) [9].
2.1.3 Traps in SiC/SiO2 interface
The channel mobility for 4H-SiC is found to be lower by two to three orders of magnitude than the
bulk mobility because of the existence of traps along the interface between SiC and SiO2 [6,10]. There are two
main reasons for these traps to be formed, a) interfacial carbon clusters and b) near interfacial oxide defects
[11,12]. The behaviour of the interfacial carbon clusters vary depending on the size of the cluster. The energy
spectrum for electrons of small sized clusters is similar to the electronic energies of wide band gap amorphous
hydrogenated carbon (a-C:H). Figure 2.2a shows the electronic energies around the bandgap of a-C:H. The
filled states to the right in Fig. 2.2a, which are π-bonds of sp2
-hybridised carbon atom, are distributed up to 1.4
eV above the valence band of 4H-SiC. As long as those states are filled, they are neutral. Though it is possible
that electrons transit from these states to the Fermi level of the semiconductor, resulting in a positive charge
and, in the case of a p-type semiconductor, this phenomenon will affect its mobility. On the other hand, the
empty states of the antibonding π*
levels appear to be more than 1 eV higher than the upper conduction band
of 4H-SiC, which has the broadest band gap of all the SiC polytypes, so they do not seem to affect the
performance of the device [11,12].
Figure 2.2.b shows the energy spectrum of graphite. As the size of the cluster increases, it starts to
behave more like graphite with a continuous electronic spectrum, which can be charged positively if it traps a
hole, or negatively if it traps an electron in p-type and n-type SiC, respectively. In other words, depending on
the relative position of the Fermi level of SiC and the neutrality level of graphite, which coincide with the
Fermi level of the cluster, found at 2.5 eV above the valence band of SiC, electrons may be emitted or be
trapped[11,12].
Figure 2.2.c illustrates the energy spectrum of the oxide defects, who will affect the performance of the
n-type MOS of 4H-SiC [12]. The near interfacial oxide defects, which may be oxygen vacancies or carbon
interstitial, have an energy level are around 2.8 eV below the conduction band of SiO2 as can be seen from the
figure.

5
Figure 2. 2. The energy distributions of different sources of interface traps: a) from a wide energy gap a-C:H, b) from a graphite
like cluster and c) from oxide defects. Zero energy level is placed at the conduction band of SiO2 and the band edges of Si are given
as reference [11].
2.1.4 Passivation of the interface traps
Afanasev et al. found in the early 90’s that the density of interface traps (Dit) in SiC/SiO2 is in the
order of 1013
eV-1
cm-2
close to the conduction band [10], which is several orders of magnitude higher than at
a well-passivated Si/SiO2 interface. One of the common methods to passivate the interface traps is to
introduce nitrogen (N) [10,13,14]. The most probable explanation of why N can actually passivate the traps at
the SiC/SiO2 interface is that N can replace the interstitial carbon, or can be placed in a vacancy. Also N can
react with the carbon clusters and form CN molecules, which later diffuse into the oxide away from the
interface [13].
There are different ways of inserting N into a MOS structure. Firstly, the oxide can be thermally grown
in nitric oxide (NO) or nitrous oxide (N2O) ambient [10,13,14]. It has been reported that Dit decreased to
2×1011
eV-1
cm-2
for NO oxidation at 1175 o
C and to 4×1011
and 1×1012
eV-1
cm-2
for N2O oxidation at 1300
o
C and 1410 o
C, respectively [6]. Another way is, after the fabrication of the dielectric, to anneal the samples
in NO, N2O and N2 ambient [10,13,14]. For NO post oxidation annealing (POA) it has been reported that the
Dit value dropped to 1-2×1012
eV-1
cm-2
and 3-5×1011
eV-1
cm-2
for annealing temperatures 1175 o
C and 1410
o
C, respectively, while the channel mobility increase to 30-40 cm2
V-1
s-1
. For N2O POA at 1175 o
C the Dit
reduced to 3×1011
eV-1
cm-2
and the mobility rose at 49 cm2
V-1
s-1
. On the other hand, when oxide was
deposited and annealed in N2O ambient at 1150 o
C there are references that measured the Dit at 4-8×1011
eV-1
cm-2
and the channel mobility at 24-40 cm2
V-1
s-1
. Finally, for N2 POA the Dit was found 4×1011
eV-1
cm-2
[6]. All the above Dits were measured at 0.2 eV below the conduction band of 4H-SiC.

6
2.2 MOS theory
One of the most important electronic components, which are used in integrated circuits, is the metal-
oxide-semiconductor (MOS). The MOS structure consists of the following: a semiconductor as the substrate
(or body), doped p- or n-type. On top of the substrate is an oxide deposited, or grown, depending on the
oxidation method and on top of this oxide a metal is placed which function as the gate contact. An ideal MOS
capacitor is at flatband condition when there is no applied voltage at the gate. The energy band diagram is the
following.
Figure 2. 3. Energy band diagram of the flatband condition (VG=0) for an Ideal MOS capacitor for a) nMOS b) pMOS [15]
In Figure 2.3 EV, EC, Ef, Eg and Ei are the valence band edge, conduction band edge, Fermi level,
energy band gap and the intrinsic level (Ei= Eg/2) of the semiconductor, respectively. Using the elementary
charge q, the electron affinity expressed in electron volts is qχ of the semiconductor and qχi is the electron
affinity of the oxide. The energy differences ΨBn, ΨΒp, Φn, Φp are the fermi potentials with respect to midgap
and the band edges [16]. The quantities Φm are the metal work function and ΦB the potential difference
between the metal fermi level and the conduction band of the dielectric. The top level is the vacuum level
used as a reference level for free electrons.
As can be seen from the figure, under the conditions of flat bands the difference between the metal
work function and the semiconductor work function (Φms) is zero:
𝛷 𝑚𝑠 = 𝛷 𝑚 − (𝑥 +
𝛦 𝑔
2𝑞
+ 𝛹 𝛣 𝑝
) = 𝛷 𝑚 − (𝜒 +
𝛦 𝑔
𝑞
− 𝛷𝑝) = 0 𝑓𝑜𝑟 𝑝 − 𝑡𝑦𝑝𝑒 (2.1a)
𝛷 𝑚𝑠 = 𝛷 𝑚 − (𝑥 +
𝛦 𝑔
2𝑞
− 𝛹 𝛣 𝑛
) = 𝛷 𝑚 − (𝜒 + 𝛷𝑛) = 0 𝑓𝑜𝑟 𝑛 − 𝑡𝑦𝑝𝑒 (2.1b)

7
In real cases, however, Φms is seldom negligible and must be taken into account. In addition, there are
always in the real case extra charges in the oxide and at the interface that needs to be taken into account, and
will result in band bending of the structure.
In case the gate voltage is not zero, three different situations are distinguished: accumulation, depletion
and inversion. Figure 2.4 illustrates the cross section of the MOS capacitor and the band diagrams for these
states for a p-type semiconductor.
Figure 2. 4. The charge distribution in the top figure and band diagram for a p-type MOS capacitor at the bottom figure, for a)
accumulation b) depletion and c) inversion is shown. The Fermi level in the semiconductor is denoted by a dashed line [15, 17]
a) Accumulation
When negative voltage is applied at the gate for p-type substrate, the band edges of the semiconductor
are bent upwards which results in the accumulation of holes at the interface of the semiconductor and
the oxide.
b) Depletion
When a positive voltage is applied, lower than the threshold voltage which will be defined below, the
bands are bending downwards, while the intrinsic level is still above the Fermi level. This results in
the appearance of depletion region with width Wdep which is given by:
𝑊𝑑𝑒𝑝 = √
2 𝜀 𝑠 𝛷 𝑠
𝑞𝑁 𝐴
(2.2)
where εs is the electrical permittivity of the semiconductor, qΦs is the band bending (Φs is called
surface potential) and NA is the doping concentration. The charge in the depletion region is equal to –
qNAWdep.

8
c) Inversion
The voltage at which the inversion begins is called threshold voltage and is given by:
𝑉𝑡 = 𝑉𝑓𝑏 + 𝛷 𝑠𝑡 +
√𝑞𝑁 𝐴2𝜀 𝑠 𝛷 𝑠𝑡
𝐶 𝑜𝑥
(2.3)
As seen for an ideal MOS capacitor (Fig. 2.3), the flatband voltage Vfb=0, but in a real case some
voltage needs to be applied to reach flat band conditions. When the voltage, which is applied, is higher
than threshold, the bands are bending more until the intrinsic level goes below the Fermi level, which
results in that the p-type substrate is inverted to n-type, i.e. there will be a higher concentration of
electrons than holes at the interface. A practical criterion for when the semiconductor surface is
sufficiently n-type to allow an electron channel to appear is given by the strong inversion. The
condition for strong inversion is when the surface potential is as high above the Fermi level as it was
below in the unbiased case, i.e. for
𝛷 𝑠𝑡 = 2𝛷 𝐵 = 2 (
𝑘𝑇
𝑞
) 𝑙𝑛 (
𝑁 𝐴
𝑛 𝑖
) (2.4)
Here, Cox is the oxide capacitance, k is the Boltzmann constant, T the temperature and ni the intrinsic
carrier concentration [16,18]. The n-type substrate behaves the opposite way.

9
Chapter 3. Fabrication Process
This chapter describes the fabrication of the MOS structures step by step. In particular, the different
techniques and equipment used for dielectric deposition, or growth, are explained. The chapter also includes a
table of the sample names and their preparation schemes.
3.1 Pre-deposition Surface Cleaning process
The wafer that has been used is n-type epitaxial 4H-SiC grown on 4o
off-axis purchased from SiC Crystal
AG with 3×1015
cm-3
nitrogen doping. The wafer has been diced into square pieces with area of 1 cm2
. The
first step of the fabrication is the surface cleaning, which has been done in two stages, in order to remove the
impurities like organic and metal particles [19]. The first stage of the cleaning is a standard 7up/IMEC
process:
 5 minutes in 3 l HSO4: 1 l H2O2 solution
 5 minutes in de-ionised water (DI)
 100 seconds IMEC clean in 6 l DI H2O: 60 ml isopropanol : 60 ml HF (50%) solution
 rinse again in DI water
Afterwards 150 nm of Nickel (Ni) has been deposited at the wafer back side using metal evaporation followed
by rapid thermal annealing (RTA) at 950 o
C (for the samples where either the oxide has been thermally
grown, or they have been annealed in the furnace, the nickel deposition has been done after the
oxidation/annealing) . Then a weak RCA1 cleaning process consisting of:
 100 ml NH4: 2 l DI H2O: 400 ml H2O2 solution at 60 ºC for 10 minutes was used
 rinse in DI water for 5 minutes
 kept 30 seconds in HF (1:50)
 rinse in DI water for 60 seconds
3.2 Oxide Deposition
The cleaning procedure has been followed by the formation of the dielectric. The samples have been
divided in three batches of 10, 13 and 3 samples. In the first two batches the oxide has been deposited by two
versions of the plasma enhanced chemical vapour deposition (PeCVD) method, the first by using the Applied
Materials P5000 processing tool (10 samples) and the second batch by using Plasmalab 80Plus (Pekka)
processing tool (13 samples). The oxide of the third batch has been thermally grown (3 samples).
3.2.1 PeCVD
As mentioned before, one of the oxide formation method is plasma enhanced chemical vapour
deposition (PeCVD). Plasma is the fourth state of matter and it can be described as a gas containing positive
ions and negative electrons. The main advantage of this method is that the energy required for initiating the
oxide formation, is provided by an RF source, which accelerates the positive ions towards the surface where
they deposit their kinetic energy and facilitate the chemical processes leading to oxidation [19,20,21]. Thus
the deposition can take place at lower temperatures than plain chemical vapour deposition (CVD). Silane

10
(SiH4) and nitrous oxide (N2O) are the precursors that have been used for the formation of the oxide and the
reaction is the following [21].
𝑆𝑖𝐻4 (𝑔𝑎𝑠) + 4 𝑁2 𝑂 (𝑔𝑎𝑠) → 𝑆𝑖𝑂2 (𝑠𝑜𝑙𝑖𝑑) + 2 𝐻2 𝑂 (𝑔𝑎𝑠) + 4 𝑁2 (𝑔𝑎𝑠)
The configuration of PeCVD process is illustrated in Figure 3.1.
Figure 3. 1. PeCVD process for the formation of SiO2 [21]
As it can be seen from Figure 3.1, the wafer is positioned at the bottom electrode, where heat can be
provided. SiH4 and N2O are inserted into the chamber through the inlets at the top. Between the two electrodes
a high voltage is applied, typically a few hundred volts, thus plasma is created. The plasma particles collide
with the reactant gases and produce ions and radical reactive species. These products are adsorbed and they
react with the surface of the wafer, building up a SiO2 layer. The byproducts are desorbed from the surface
and driven away though the outlets [19,20,21]. The typical parameters that one can use to control the process
are the RF power applied to maintain the plasma, wafer heating, gas ratios, and gas inlet pressure. In this
experiment the wafer temperature is kept constant at 400 and 300 °C for P5000 and Pekka, respectively.
3.2.1.1 P5000
Precision 5000 Mark II (P5000) is a dry etching and PeCVD system with four process chambers. It is
manufactured by Applied Materials Inc. [22,23]. Chamber A is used for SiO2 and SiN etching, chamber B is
for Al, TiW, TiN and Al (500nm)/TiW (100nm) etching, chamber C is for poly-Si and c-Si etching and
chamber D is for SiO2 and SiN deposition. For this experiment the deposition was done in chamber D at 400
o
C while the pressure and the RF power varied from 1.5 Torr to 6 Torr and from 50 W to 300 W, respectively.
Lastly, the gas ratios of N2O/SiH4 that were used are 600/30 sccm, 800/60 sccm, 600/60 sccm and 1200/60
sccm. The unit “sccm” means standard cubic centimeters per minute.

11
3.2.1.2 Pekka
The 80Plus is PeCVD system, used for deposition of dielectric thin films, is manufactured by
Plasmalab [24]. The deposition temperature was 300 o
C, the pressure ranged from 0.2 Torr to 0.8 Torr, the
power from 10 W to 20 W and the different gas ratios of N2O/SiH4 were 425/710 sccm, 500/53 sccm, 700/75
sccm and 650/142 sccm.
3.2.2 Thermal Growth
Thermal growth is used for the third batch and it was done in N2O ambient. The process proceeds in
several steps. First the N2O dissociates according to the following two reactions [25].
2𝑁2 𝑂 → 2𝑁𝑂 + 𝑁2
2𝑁2 𝑂 → 2 𝑁2 + 𝑂2
The main contribution to the oxidation is from O2, which follows the Deal-Grove model [26].
𝑆𝑖𝐶 + 𝑂 → 𝐶𝑂 + 𝑆𝑖
𝐶𝑂 + 𝑂 → 𝐶𝑂2
𝑆𝑖 + 2𝑂 → 𝑆𝑖𝑂2
On the other hand, NO is either removing excess carbons [13], or form SiOxNy (x<2, 0<y<2) as it can be seen
from the reaction below:
Si + xNO → SiO 𝑥N 𝑦 +
𝑥 − 𝑦
2
𝑁2
N2 is inert and it diffuses out of the dielectric [25].
The oxide of two of the three samples in batch 3 has been grown at 1250 °
C for 15 hours and, as for
the third sample, it has been grown at 1150 o
C for 12 hours. The oxidation of the samples took place in the
furnace which is shown in Figure 3.2.

12
Figure 3. 2. Thermco furnace either for thermal oxidation or for annealing.
Figure 3.3 schematically illustrates the process.
Figure 3. 3. Schematic diagram of the dielectric fabrication and the back contact formation
3.3 Post oxidation/deposition annealing
After the oxide fabrication, several samples have been annealed, either with rapid thermal annealing
by using Mattson 100 RTP System for 300 sec at temperatures of 850, 1000 and 1150 °C, or in the furnace
for 3 hours at 1150 °C.
3.4 Gate metallization
3.4.1 Lithography
For the formation of the gate contact, a layer of 150 nm thick Aluminium is deposited on the dielectric
by metal evaporation. Then the samples are coated with positive photoresist spun at 4000 rpm for 60 sec to
form a 1.4 µm thick resist (see Fig. 3.4). They are soft baked at 90 °
C for 90 sec. to harden the resist and then
aligned with a mask, defining the areas of the top circular gate contacts with different diameters of 50-500 μm

13
and then expose to UV-light for 7 seconds. In order to remove the exposed photoresist, the samples are
developed in CD26 solvent for 40 seconds. Finally, the samples are hard baked at 110 °
C for 20 min in an
oven to harden the photoresist that is still on the samples, covering the areas where the Al should be
preserved.
3.4.2 Aluminium etching
The top aluminium contact is dry etched in Applied Material P500 processing tool. It is done by using
the gases BCl3 and Cl2 with ratio 40/30 sccm for 40 sec. Finally, the remaining photoresist is striped by
dipping the samples in acetone. The gate formation process is schematically illustrated in Figure 3.3.
Figure 3. 4. Schematic diagram of the gate formation
3.5 Fabrication Parameters of the samples
In Tables 3.1-3 all the parameters, including the thickness of the oxides for the different fabrication
processes, are tabulated.

14
Table 3.1 P5000 samples
P5000
Sample
number
Deposition
Temperature
(C)
Pressure
(Torr)
Power
(W)
Spacing
(mils)
SiH4
(sccm)
N2O
(sccm)
PDA
(C)
Annealing
Time
(sec)
Thickness
(nm)
1 400 2.8 100 450 60 1200 1150 300 47
2 400 2.8 100 450 60 1200 N/A - 44
3 400 2.8 100 450 60 1200 1000 300 46
4 400 4.5 300 525 30 600 N/A - 72
5 400 4.5 300 525 30 600 1150 300 68
6 400 6 100 450 60 1200 1150 300 68
7 400 1.5 300 450 60 1200 1150 300 45
8 400 1.5 50 450 60 1200 1150 300 44
9 400 2.8 100 450 60 800 1150 300 45
10 400 2.8 100 450 60 600 1150 300 45
Table 3.2 Pekka samples
Pekka
Sample
number
Deposition
Temperature
(C)
Pressure
(Torr)
Power
(W)
SiH4
(sccm)
N2O
(sccm)
PDA
(C)
Annealing
Time
(sec)
Thickness
(nm)
11 300 0.8 20 710 425 1150 10800 46
12 300 0.2 10 53 500 1150 300 80
13 300 0.4 10 75 700 1150 300 44
14 300 0.2 10 142 650 N/A - 47
15 300 0.2 10 142 650 1150 300 45
16 300 0.5 20 710 425 1150 300 50
17 300 0.5 20 710 425 1150 10800 51
18 300 0.8 20 710 425 1150 300 46
19 300 0.8 20 710 425 N/A - 52
20 300 0.8 20 710 425 1150 300 45
21 300 0.8 20 710 425 1000 300 47
22 300 0.8 20 710 425 850 300 47
23 300 0.5 20 710 425 N/A - 47
Table 3.3 Thermally Grown samples
Thermal
Growth
Sample
number
Deposition
Temperature
(C)
Furnace
Time
(hours)
POA
(C)
Annealing
Time
(sec)
Thickness
(nm)
24 1150 12 1150 300 27
25 1250 15 N/A - 36
26 1250 15 1150 300 34

15
Chapter 4. Characterization
In this chapter the experimental setup and the different types of measurements for the characterization
and analysis of the samples will be discussed.
4.1 Electrical Measurements
4.1.1 Probestation 3 semi-automatic
The electrical measurements are accomplished by using Probe Station 3 semi-automatic in KTH
electrum manufactured by Cascade Microtech. The tool is shown in Figure 4.1.
Figure 4. 1. Probestation Semi-automatic for CV and IV measurements.
The samples are placed on the chuck and are fixed by turning on the vacuum pump. The microscope is
connected to the computer on the left and, by rotating the handles of one of the four probes, which are used for
the electrical contact, the probe is placed onto the gate of the MOS structure. During the measurements, the
light is turned off. From the control unit to the right we choose either the CV or the IV configuration for our
measurements.
4.1.2 CV
The configuration for the CV measurements is the following. Two different voltage sources were
superimposed on the device terminals, one AC voltage measurement signal, whose magnitude (50 mV) and
frequency (200 kHz) are kept constant, and one DC bias voltage whose magnitude is changed in steps with
time. By integrating the current from the AC signal over time, the gate charge Qg is derived. Then the

16
capacitance is calculated from Qg and the DC voltage [27]. The voltage is applied at the bottom of the MOS
structure (Nickel side), so the whole back side will be at the same potential. With a probe tip on the upper side
individual MOS structures can be measured. The voltage interval is mostly from -5 V to 5 V, although for
several samples who’s Vfb are close to + 5 V, other start and stop voltages is used. The full interval is still 10
V each time. The voltage is swept from low to high bias in steps of 0.1 V/s. In this experiment, a high
frequency (200 kHz) is used for measuring the CV curves. In this case, the minority carries and the interface
traps do not have time to response to the AC signal [28]. A typical CV curve, where the three different states
of the MOS structure are shown, is illustrated in Figure 4.2.
Figure 4. 2. High frequency capacitance vs voltage curve, where the inversion, depletion and accumulation region are shown for
Sample 3.
A MOS structure with an n-type substrate under voltage bias can be seen as a system of two capacitors
in series connection - the oxide capacitor and the depleted semiconductor capacitor. With a positive bias the
majority carriers, electrons, are accumulated in the channel at the semiconductor-dielectric interface, and the
total capacitance is equal to the oxide capacitance. When the gate voltage is smaller than the flatband voltage,
a depletion region is formed, which acts like a capacitor connected in series with the oxide capacitor, thus the
total capacitance is decreased. Depending on the frequency there are two different responses from the MOS
structure on negative bias. At low frequency (typically less 100 Hz) the minority carriers have time to respond
to the AC signal and they form a capacitor in parallel with the depletion region capacitor, which results in
increasing the total capacitance until the total capacitance again reaches the oxide capacitance value. At high
frequency, the minority carriers are not generated and transported fast enough to give a response to the AC
signal and the total capacitance reaches its minimum value when the inversion is achieved and the depletion
region has its maximum width [28,29].
In this experiment two parameters will be extracted from the measurements. Firstly, the flatband
voltage Vfb, which can be found from the horizontal shift of the CV curve in respect to the theoretical one, as
illustrated in Figure 4.3. Secondly, the density of interface traps, Dit, which will be extracted with the Terman
method from the measured CV curves. The Terman method will be discussed in Section 4.1.2.2.

17
Figure 4. 3. The shift of the CV curve for sample 21 according to the theoretical curve shows the Vfb. The theoretical curve has no
extra charges at the interface or in the oxide, in addition to equal work functions for the semiconductor and contact metal, resulting
in Vfb=0 V.
4.1.2.1 Defects in the oxide
In reality there are always charges in the oxide and at the interface. These charges are often classified
in four different types and this makes the MOS structure deviate from the ideal behaviour, as described in
Section 2.2. A committee, which was set up by the Electronics Division of the Electrochemical Society and
the IEEE in January 1978, decided on a terminology for the different oxide charges [30], which are displayed
in Figure 4.4.
Figure 4. 4. A cross section of a MOS structure, which shows the different types and location of the charges inside the dielectric.
1) Mobile ionic charges (Qm) are the result of ionic impurities, such as Na+
, Li+
and K+
. Negative ions
and heavy metals are also added to this type of charge for oxidation at temperature over 500 o
C.

18
2) Oxide trapped charge (Qot): Holes or electrons can be trapped due to ionizing radiation, avalanche
injection in the bulk of the oxide, or similar processes. Depending of the sign of these charges there is
the analogous shift in the positive or negative direction of the CV curve with respect to the ideal curve.
3) Fixed oxide charge (Qf): These charges are structural defects created during the oxidation process. The
presence of these defects can be observed in the CV curves as a hysteresis loop when the CV
measurement is swept from accumulation to inversion and vice versa.
4) Interface trapped charge (Qit): These can be positive, or negative charges and they are caused by
broken bonds, structural or oxidation-induced defects and metal impurities at the interface, or near
interface. They can be identified by evaluating the slope of the CV curve in the depletion region.
[28,30,31]
4.1.2.2 Terman Method
One of the methods to determine the Dit is the Terman method using high frequency CV
measurements. Even though the interface traps do not respond to a higher frequency AC signal, they respond
to slow change of the DC voltage by filling, or emptying the traps that are below the conduction band,
according to the relative bending of the Fermi level. This results in stretching of the CV curve with respect to
the ideal, theoretical curve at the depletion, as can be seen in Figure 4.5a [28]. From the comparison of the
two curves, the surface potential, Ψs, versus the gate voltage is extracted and plotted, as in Figure 4.5b. From
the slope of this curve, the Dit can be calculated according to Equation 4.1 [28].
𝐷𝑖𝑡 = (
𝐶 𝑜𝑥
𝑞2
) (
𝑑𝑉𝑔
𝑑𝛹𝑠
− 1) − (
𝐶 𝑑𝑒𝑝
𝑞2
) (4.1)
Figure 4. 5. a) Experimental and theoretical (ideal) high frequency capacitance-voltage curve, showing the non-parallel shift of the
experimental curve. b) Surface potential versus experimental voltage for the extraction of Dit for Sample 6

19
4.1.3 IV
The configuration for current-voltage measurements is almost the same as the capacitance-voltage, but
now only a DC bias is applied, which is increased in steps of 0.2 V/sec. The maximum voltage is estimated
for each sample as the voltage where the electric field of 10 MV/cm is reached, which depends also on the
oxide thickness. A typical IV curve is shown in Figure 4.6. While the voltage increases, so called Fowler-
Nordheim tunnelling current may appear. Then electrons trapped at the defects in the oxide gain enough
energy to jump into the conduction band which results in the Frenkel-Poole emission of electrons and, finally,
when the critical field is reached, a chain of defects creates a conduction path and the breakdown of the oxide
occurs [32].
Figure 4. 6. Current-Voltage curve, in which the different types of currents and the breakdown is indicated for Sample 14
4.2 AFM
4.2.1 AFM system
The surface roughness of the samples is measured using a commercial setup Dimension 3000, which
was manufactured by Veeco/Digital Instruments. The principle of the Atomic Force Microscopy (AFM) is to
drag/push a small needle across a surface and monitor the vertical motion of this needle. To do this, a sharp
tip, which is connected to a cantilever placed in a holder, scans the surface of the sample. A laser beam is
aligned with the backside of the tip and is reflected to a segmented photodiode. The photodiode can then
detect any kind of deformation of the cantilever. Another piezoelectric element is located at the base of the
cantilever, which can translate any deformation into an electrical signal. The atoms of the tip interact with the
atoms of the surface via Van der Waals forces [33]. A schematic view of the AFM is shown in Figure 4.7.

20
Figure 4. 7 A schematic illustration of the AFM system and the image of Sample 20 [34].
4.2.2 Tapping Mode
There are three different operation modes for AFM, contact, non-contact and tapping mode. For this
experiment the latter is used. The principle of tapping mode is that the piezoelectric element makes the
cantilever oscillate close to its resonant frequency and keeps the oscillating amplitude constant. When the tip
reaches a step on the investigated surface the amplitude decreases, but when it reaches a void the amplitude
increases. In the detector a sinusoidal signal arrives, which tracks the changes in the amplitude and translate
them to an image [33].
The purpose of the AFM measurements for this thesis is to measure the surface roughness from the
images that have been recorded by Nanoscope IIIa software and to check if there is any correlation between
the Vfbs or Dits with the roughness root mean square (rms).
The configuration set-up for the measurement is the following: first the laser is aligned from the knobs
at the top of the SPM (scanning probe microscopy) head until the spot is visible in the Dimension Head Filter
Screen and the detected signal is between 4-6 V. Then the photodetector is adjusted from the knobs at the left
side of the SPM head, until the spot goes to the center of the screen. In Figure 4.8 a picture of the tool and of
the SPM head is shown.

21
Figure 4. 8. Commercial AFM system tool a) Computer b) SPM Head.

23
Chapter 5. Results
In the following sections, the influence of the different oxide fabrication parameters, such as pressure
of the gases, power, gas ratio and post deposition annealing (PDA) for the PECVD method and growth
temperature and post oxidation annealing (POA) for the thermally grown oxides, on the electrical
characteristics will be investigated. The error bars presented in this report is a standard deviation taken on 8-
12 different MOS devices on each sample. A summary of all the electrical characteristics for various samples
can be found in Appendix 1.
5.1 P5000
This section elucidates the electrical measurement and surface analysis results obtained for the metal
oxide semiconductor (MOS) structures with oxide deposited using plasma enhanced chemical vapour
deposition (PECVD) in an Applied Mechanic P5000 processing tool. Figure 5.1 a and b shows the capacitance
voltage (CV) and current voltage (IV) curves of three samples (P1, P2 and P9 for the CV and P6, P7 and P10 for
the IV) with different deposition parameters, respectively. From now on the samples grown by P5000 tool will
be named as Px, where x is the sample number. The CV plots are chosen in order to show the spread of the
results, while the IVs show samples that break at low, medium and high electrical field. As it can be seen from
the Figure 5.1, this method is quite unstable since the values of the CV curves spread. In general, the flatband
voltage (Vfb) varies from -10 V to 3.5 V and the density of interface traps (Dits) from 2×1011
eV-1
cm-2
to
1×1012
eV-1
cm-2
. A similar trend is observed for IV curves and an average critical field is found to be 3.5
MV/cm.
Figure 5. 1. a) An example of high frequency capacitance-voltage curves for samples whose oxide has been deposited with P5000 b)
a sample of current density versus electric field curves.
5.1.1 Influence of Pressure
For these investigation two samples P1 and P6 have been utilized. Sample P1 and P6 are deposited at a
pressure of 2 and 6 Torr respectively while the other oxide parameters power and gas ratio (N2O/SiH4) have
been fixed at 100 W and 1200/60 respectively. After oxide depositions both the samples have been annealed
at 1150 O
C for 300 seconds. Figure 5.2 a and b show the flatband voltage and density of interface traps results
for fabricated samples, respectively.

24
Figure 5. 2. a) Effect of pressure on Vfb and sample stability; b) Effect of pressure on Dits and sample stability. Samples fabricated
at 100W deposition power, 1200/60 N2O/SiH4 gas ratio and annealed at 1150 °C for 300 seconds.
Increasing the deposition pressure results in a reduction of the concentration of negative oxide charges
indicated by shift of Vfb from 3.55 V for sample P1 towards to zero (0.8 V) for sample P6 as shown in Figure
5.2.a. Also, the standard deviation of different devices for a process varies from 1.4 V for 2.8 Torr to 0.4 V for
6 Torr, showing that deposition at higher pressure i.e. 6 Torr is more stable. On the other hand, change in Dits
is meagre as a function of deposition pressure as can be seen in Figure 5.2.b. Thus the deposition pressure
does not seem to affect the Dits, which is around 2.7×1011
eV-1
cm-2
; although in this case the stability
decreases significantly at higher pressure (2×1011
eV-1
cm-2
) compared to the lower pressure (0.4×1011
eV-1
cm-2
). Finally, by increasing the pressure, the critical electric field of oxide was found to increase from 3
MV/cm to 5.5 MV/cm.
While the pressure is increased, the surface roughness increases dramatically and the rms values are
0.7 nm and 6.7 nm for P1 and P6, respectively.
5.1.2 Influence of Power
In this section the influence of deposition power is discussed. The deposition power is varied from 50
W for sample P7 and 300 W for sample P8 while the other oxide deposition parameters have been fixed as
following: Pressure of 1.5 Torr, the gas ratio of (N2O/SiH4) was 1200/60 and the PDA temperature was 1150
o
C for 300 seconds. Figure 5.3 a and b shows the Vfb and Dits results obtained for CV curves for the samples
in this study.

25
Figure 5. 3. a) Effect of power on Vfb and b) effect of power on Dits. Samples fabricated at the pressure of 1.5 Torr, 1200/60
N2O/SiH4 gas ratio and annealed at 1150 o
C for 300 seconds.
It can be observed from Figure 5.3.a that the sample with the lowest deposition power has a Vfb of
0.1+0.4 V and Dits of 4×1011
eV-1
cm-2
. While, higher power has a Vfb of -4+0.8 V, indicating an increase in
positive charges. Also, Dits increase by 4 times to 1.5×1012
eV-1
cm-2
and they have larger spread in
comparison to the sample with low deposition power. The critical electric field is found to be 3.5 MV/cm and
4 MV/cm for low and high deposition power, respectively. Thus it can inferred that change in power have a
minimal effect on the dielectric breakdown field.
As a result of increased power, surface roughness varies from 0.9 nm for sample P7 to 1.2 nm for
sample P8.
5.1.3 Influence of Gas ratio
For this investigations gas ratio of N2O/SiH4 have been varied from 20 to 10 while other parameters
have been fixed as follows: pressure of 2.8 Torr, power of 100 W and a PDA at 1150 o
C for 300 seconds.
Figure 5.4 a and b show influence of gas ratios on the flatband voltage and density interface states,
respectively. It can be observed that the gas ratio of 20, sample P1, have a Vfb of 3.5 V and Dits of 2.7×1011
eV-1
cm-2
. While, in case of 13.33 (sample P9) and 10 (sample P10) the Vfbs are -10.6 V and -0.2 V and the
density of traps 2×1012
eV-1
cm-2
and 5.5×1011
eV-1
cm-2
respectively. Also, it can be seen that higher the gas
ratio, less stable is the Vfb, but the stability dramatically increases for the Dits.
As for the maximum electric field that the oxide is able to withstand, the general trend shows that the
critical electric field decreases as the gas ratio increases. For gas ratios of 20, 13.33 and 10 the measured fields
are 3 MV/cm, 2 MV/cm and 1.5 MV/cm, respectively.

26
Figure 5. 4. a) Effect of gas ratio on Vfb; b) effect of gas ratio on Dits. Samples fabricated at 100 W deposition power, pressure of
2.8 Torr and annealed at 1150 o
C for 300 seconds.
Gas ratio does not seem to affect the surface roughness as the rms values are very low, 0.7, 0.6 and 0.8
nm for sample P1, P9 and P10, respectively.
5.1.4 Influence of Annealing Temperature
Two different groups of sample were used to investigate the effects of annealing temperature on Vfb
and Dits. The first group consist of two samples P4 and P5 deposited under the following conditions: 4.5 Torr
pressure, 300 W power and (N2O/SiH4) 20 gas ratio. P4 is not annealed whereas P5 is annealed at 1150 o
C for
300 s. The second group consists of three samples P1, P2 and P3. P3 and P1 are annealed at 1000 o
C and 1150
o
C, respectively and P2 is unannealed reference sample. These samples are deposited with a pressure of 2.8
Torr, power of 100 W and gas ratio (N2O/SiH4) 20.
Figure 5. 5. a) Effect of annealing on Vfb for group of samples P4 and P5 (4.5Torr, 300W, 20 N2O/SiH4) and group of samples P1,
P2 and P3 (2.8 Torr, 100 W, 20 N2O/SiH4); b) effect of annealing on Dits for the previous groups.
Figure 5.5.a shows the effect of annealing temperature on Vfb. It can be seen that in the first group
annealing causes reduction in the concentration of the negative oxide charges but it leads to slightly lower
stability. On the contrary, in the second group, unannealed reference have a negative Vfb of -3.97 V and
annealing at temperature of 1000 o
C shifts Vfb closer to zero. Further increasing the PDA temperature to 1150
o
C shifts Vfb to a positive voltage of 3.55 V.

27
Table 5.1.Vfbs and the standard deviation for group of samples P4 and P5 (4.5 Torr, 300 W, 20 N2O/SiH4) and group of samples P1,
P2 and P3 (2.8 Torr, 100 W,20 N2O/SiH4)
P4 P5 P2 P3 P1
Vfb (V) 2.17 0.62 -3.97 0.64 3.55
Standard Deviation (V) 0.26 0.60 0.89 0.62 1.39
Annealing Temperature (°C) N/A 1150 N/A 1000 1150
Figure 5.5.b shows the effect of PDA temperatures on Dits extracted from the CV curves. For samples
P4 and P5, annealed at 1150 o
C, the annealing process caused Dits to slightly decrease, but made them more
unstable. In the second set of samples, Dits dropped more than one third of the value of the not annealed
sample and made it 4 times more reliable, but between P3, annealed at 1000 o
C and P1, annealed at 1150 o
C
there is no significant difference.
Table 5.2.Dits and the standard deviation for group of samples P4 and P5 (4.5 Torr, 300 W, 20 N2O/SiH4) and group of samples P1,
P2 and P3 (2.8 Torr, 100 W, 20 N2O/SiH4)
P4 P5 P2 P3 P1
Dit (1011
eV-1
cm-2
) 5.69 4.37 8.33 2.51 2.73
Standard Deviation (1011
eV-1
cm-2
) 1.96 2.30 4.15 0.93 0.42
Annealing Temperature (°C) N/A 1150 N/A 1000 1150
For both of the groups annealing leads to an increase of the critical field. Especially for the 1st
group
(P4 and P5) there is a considerable change from 1.5 MV/cm for the unannealed sample to 6 MV/cm for the
sample annealed at 1150 o
C. For 2nd
group (P1, P2 and P3) the effect is less pertinent and varies from 3
MV/cm, 3.5 MV/cm and 4 MV/cm for the unannealed, 1000 o
C and 1150 o
C PDA temperatures, respectively.
The next table shows the rms values of roughness of the above groups. Except sample P4, which is not
annealed and the roughness is 5.7 nm, all the others are around 0.7 nm.
Table 5.3.Roughness rms for group of samples P4 and P5 (4.5 Torr, 300 W, 20 N2O/SiH4) and group of samples P1, P2 and P3 (2.8
Torr, 100 W, 20 N2O/SiH4)
P4 P5 P2 P3 P1
Roughness
(nm) 5.7 0.7 0.6 0.8 0.7
5.2 Pekka
In this section, electrical characterization and surface analysis results of samples with the gate oxides
deposited using Plasmalab 80Plus (Pekka) plasma enhanced chemical vapour deposition (PECVD) system are
presented. Different set of samples are fabricated by varying chamber conditions, pressure, gas ratio and
annealing temperature in order to study the effect on device performance. A summary of all the samples is
tabulated in Appendix 1. From now on the samples fabricated by Pekka tool will be named as Pex, where x is
the number of the sample.

28
Figure 5. 6 a) A sample of high frequency capacitance-voltage curves for samples whose oxide has been deposited with Pekka b) a
sample of current density versus electric field curves
Figure 5.6.a shows the spreading of the CV curves of three different samples Pe14, Pe16 and P22,
deposited using this tool. The Vfb varies from 3 V to 0.5 V and the Dits from 1011
eV-1
cm-2
to 5×1011
eV-1
cm-
2
. Despite the positive results in CV measurements, IV curves, from samples Pe13, Pe15 and Pe20, show unstable
behaviour. Figure 5.6.b show a wide distribution of IV curves with an average critical field around 5 MV/cm.
5.2.1 Influence of Pressure
In order to study the influence of pressure, three groups of samples are organized. In each group, there
are two samples. For the first group the pressure vary from 0.2 Torr (Pe12) to 0.4 Torr (Pe13) For the other two
groups it is varied from 0.5 Torr (Pe17, Pe23) to 0.8 Torr (Pe11, Pe19). The samples in the first group are
deposited under the condition of 10 W power and gas ratio of (N2O/SiH4) of 10. They are annealed at 1150 °C
for 300 sec. The samples in the other two groups are fabricated under the condition of 20 W power and
(N2O/SiH4) 425/710 gas ratio. Samples Pe17 and Pe11 are annealed at 1150 °C for 3 hours, whereas samples
Pe23 and Pe19 are unannealed.
Figure 5. 7. a) Effect of pressure on Vfb for group of samples Pe13 and Pe12 (10W, 10 N2O/SiH4 annealed at 1150 °C for 300
seconds), Pe17 and Pe11 (20W, 425/710 N2O/SiH4 annealed at 1150 °C for 3 hours) and Pe23 and Pe19 (20W, 425/710 N2O/SiH4 and
not annealed); b) effect of pressure of Dits for the same groups.
Figure 5.7.a shows, as the pressure is increased, the Vfbs of all samples approach zero, meaning that
defects inside the oxide decrease as well as the stability of the samples. However, the trend shows an

29
alteration depending on the set of samples. The first two groups, which are annealed, start from a Vfb of 1.34
V for Pe12 and drops to 0.5 V for the rest of the samples. For the last set of samples, which are not annealed, as
pressure increases the Vfb decreases 4 times.
Table 5.4.Vfbs and the standard deviation for group of samples Pe13 and Pe12 (10W, 10 N2O/SiH4 annealed at 1150 °C for 300
not annealed)
Pe12 Pe13 Pe17 Pe11 Pe23 Pe19
Vfb (V) 1.34 0.55 0.61 0.45 4.78 1.46
Standard Deviation (V) 0.68 0.39 0.96 0.36 1.36 0.88
Pressure (Torr) 0.20 0.40 0.50 0.80 0.50 0.80
Figure 5.7.b illustrates change in Dits for the same set of samples. As it can be seen, for the first group
of samples, Pe12 and Pe113, changing the pressure does not have any effect on value of Dits, but only on the
stability, which increase. For the rest of two groups, increasing the pressure results in decreasing Dit values as
can be seen in the second group, where there is an 80% reduction for 0.5 to 0.8 Torr.
Table 5.5.Dits and the standard deviation for group of samples Pe13 and Pe12 (10W, 10 N2O/SiH4 annealed at 1150 °C for 300
not annealed)
Dit (1011
eV-1
cm -2
) 2.99 3.40 7.54 1.61 6.37 5.25
Standard Deviation (1011
eV-1
cm -2
) 2.17 0.86 4.63 0.89 4.29 4.52
Pressure (Torr) 0.20 0.40 0.50 0.80 0.50 0.80
The critical field decreases from 5 MV/cm for sample Pe12 to 2 MV/cm for sample Pe13, while for
sample Pe11 is 7 MV/cm and it slightly decreases to 6 MV/cm for sample Pe17.
In the next table (Table 5.6) the roughness of the previous sample groups are shown.
Table 5.6.Roughness rms for group of samples Pe13 and Pe12 (10W, 10 N2O/SiH4 annealed at 1150 °C for 300 seconds), Pe17 and Pe11
(20W, 425/710 N2O/SiH4 annealed at 1150 °C for 3 hours) and Pe23 and Pe19 (20W, 425/710 N2O/SiH4 and not annealed)
Roughness (nm) 2.9 1.2 0.9 0.8 1.1 2.7
5.2.2 Influence of Gas ratio
In order to study the influence of gas ratio, two samples are deposited, Pe12 and Pe15. The samples are
deposited under the condition of 10 W of power, pressure of 0.2 Torr and a PDA of 1150 °C for 300 s. The
gas ratios of N2O/SiH4 are varied from 500/53 to 650/142 for this study.

30
Figure 5. 8. a) Effect of gas ratio on Vfb; b) effect of pressure of Dits. Samples fabricated at 20W deposition power, 0.2 Torr
pressure and annealed at 1150 °C for 300 seconds.
Figure 5.8 a shows the change in Vfb for two different gas ratios. The Vfb and their standard deviation
are basically unchanged while the gas ratio decreases (500/53 to 650/142) as the values are 1.40 V and 1.34 V
for Vfb and 0.65 V to 0.68 V for the standard deviation. Also the Dits (Figure 5.8.b) do not change, since for
the gas ratio of 500/53 is 2.99×1011
eV-1
cm-2
and 2.26×1011
eV-1
cm-2
for gas ratio of 650/142. But the standard
deviation decreases from 2.17×1011
eV-1
cm-2
to 6.68×1010
eV-1
cm-2
, respectively.
The maximum electric field the oxide can tolerate increases for lower gas ratio from 5 to 7.5 MV/cm.
As in P5000, the gas ratio does not change the roughness of the surface. The rms values are 2.9 for
sample Pe12 and 2.4 for sample Pe15.
5.2.3 Influence of Annealing Temperature
In order to study influence of annealing temperature, two groups of samples, are selected. In the first
group there are four samples, Pe22, Pe21, Pe20 and Pe18. The oxide is deposited under the condition of 20 W
power, 0.8 Torr pressure and (N2O/SiH4) 425/710 gas ratio and the annealing temperatures are 850 o
C, 1000
o
C and 1150 o
C, for sample Pe22, Pe21 and Pe20, respectively. The same parameters of power, pressure and gas
ratio are used for sample Pe18, but the PDA is performed in two steps. First, one third of the required oxide is
deposited, in this case 15 nm and PDA at 1150 o
C for 300 s. Second, the remaining oxide is deposited, in this
case 35 nm followed by PDA at 1150 o
C for 60 s. The second group of samples, Pe14 and Pe15, are fabricated
under the condition of 10 W power, (N2O/SiH4) 650/142 gas ratio and 0.2 Torr. Sample Pe14 is un-annealed,
whereas Pe15 is annealed at 1150 o
C.

31
Figure 5. 9 a) Effect of annealing on Vfb; b) effect of annealing on Dit for the group of samples Pe22, Pe21, Pe20 and Pe18 (20 W, 0.8
Torr, 425/710 (N2O/SiH4)) and the group of samples Pe14 and Pe15 (10 W, 0.2 Torr, 650/142 (N2O/SiH4))
Figure 5.9.a shows the change in Vfb versus PDA temperature. Sample Pe22 (annealed at 850 o
C) and
sample Pe21 (annealed at 1000 o
C) exhibit similar behaviour in terms of Vfb, which is around 0.5 V and also
standard deviation which is 0.4 V. As the annealing temperature goes to 1150 o
C, as in the case of sample Pe20,
there is an increase in the amount of the positive oxide charges which leads Vfb to drop to -4.53 V and a
decreased stability by 2 times. On the other hand, in sample Pe18, it has been observed that the positive oxide
charges are reduced. Flatband voltage for this sample is -1.15 V and the standard deviation is reduced by half
and reach the value of 0.5 V. Flatband voltage shows a slight decrease as the sample is annealed from Pe14,
which is not annealed and having a Vfb of 1.48 V, to Pe15 which is annealed at 1150 o
C and having the Vfb of
1.40 V. In terms of stability, the standard deviation change is also very small, 0.62 V for Pe14 and 0.65 V for
Pe15.
Figure 5.9.b shows the change in Dits versus annealing temperature. The measured Dits for sample Pe22
is 7.94×1010
eV-1
cm-2
and its standard deviation is 6.71×1010
eV-1
cm-2
, while for sample Pe21, the Dit is
2.04×1011
eV-1
cm-2
and the standard deviation is 1.53×1011
eV-1
cm-2
. Sample Pe20 has Dit of 1.16×1012
eV-1
cm-
2
with standard deviation of 5.32×1011
eV-1
cm-2
and Dit of sample Pe18 is 3.87×1011
eV-1
cm-2
with the standard
deviation of 1.66×1011
eV-1
cm-2
. Therefore, as seen by these results, Dit is directly proportional to the PDA
temperature, but for the last sample, Pe18, which was exposed to a different annealing procedure, the value is
decreased. The same trend also applies to the standard deviation, which means that the stability of the samples
decreases for an increase in the annealing temperature while for sample Pe18, it becomes more stable than
sample Pe20. Sample Pe14 and Pe15 have Dit of 4.20+1.50×1011
eV-1
cm-2
and 2.26+0.67×1011
eV-1
cm-2
,
respectively. The results show that annealing the sample causes a decrease in Dits, but the stability increases.
For the first set of samples, increase in PDA temperature leads to a drop in the critical electric fields,
since the values are 7 MV/cm for Pe20 and Pe18, 2.5 MV/cm for Pe21 and 1.5 MV/cm for Pe22. For the second set
of samples, annealing the sample decreases dramatically the critical electric field from 7.5 MV/cm to 2
MV/cm for Pe15 and Pe14, respectively.
Annealing appears to increase the roughness of the samples. For the first group the rms value goes
from 0.5 nm for the unannealed sample, Pe14, to 2.5 nm for sample Pe15, which was annealed at 1150 o
C. The
same trend is followed by the second group too. The values are 0.7 nm, 1.2 nm, 1.4 nm and 2.4 nm for the
sample Pe22, Pe21, Pe20 and Pe18, respectively.

32
5.3 Thermal Growth
The last process that has been examined is thermal growth process. There are only three samples, T24,
T25 and T26, which have thermally grown oxide. In general these oxides are more stable with Vfb around 1 V
and Dit around 1.5×1011
eV-1
cm-2
. The oxide breakdown occurred around 3 MV/cm. Figure 5.10 a and b
shows CV and the IV curves for the three different samples.
The rms values of the roughness are 0.8 nm, 1.2 nm and 3.2 nm for the samples T24, T25 and T26.
Sample T26 is the not annealed sample of the thermally grown ones.
Figure 5. 10. a) A sample of high frequency capacitance-voltage curves for samples whose oxide has been grown thermally b) a
sample of current density versus electric field curves
5.3.1 Influence of Growth Temperature
Figure 5. 11. a) Effect of grown temperature on Vfb; b) effect of grown temperature on Dits.
In this section the effect of growth temperature on the thermally grown oxide samples is shown. Two
temperatures, 1150 o
C and 1250 o
C, are defined, for sample T24 and T26, respectively. Figure 5.12 a shows that
Vfb decreases from 1.19 V for sample deposited at 1150 o
C to 0.81 V for 1250 o
C grown sample, as well as the
decrease in the standard deviation from 0.51 V to 0.47 V. Figure 5.12 b shows the change in Dit with respect
to the growth temperature. The Dit for the sample grown at 1150 o
C is 1.99×1011
eV-1
cm-2
and it increases

33
slightly to 2.1×1011
eV-1
cm-2
for the sample grown at 1250 o
C. Standard deviation also decreases from
1.92×1011
eV-1
cm-2
to 1.28×1011
eV-1
cm-2
, respectively, showing that the stability of the samples improved by
increasing growth temperature.
5.3.2 Influence of Annealing
Figure 5. 12. a) Effect of annealing temperature on Vfb; b) effect of annealing temperature on Dits
This section investigates the influence of PDA temperature for two samples; T25 with no annealing
(N/A) and T26 annealed at 1150o
C. Figure 5.11 a, shows that Vfb decreases after annealing as well as the
stability of the sample. In case of unannealed sample (T25), the measured Vfb is 2.00 V with a standard
deviation of 0.35 V. In the case of the annealed sample (T26), Vfb is 0.81 V and the standard deviation is 0.47
V. The trend is opposite for Dit (Figure 5.11 b). For the unannealed sample (T25), the measured Dit and
standard deviation are 1.23×1011
eV-1
cm-2
and 4.53×1010
eV-1
cm-2
, respectively and for the annealed sample
(T26) Dit is 2.1×1011
eV-1
cm-2
and standard deviation is 1.28×1011
eV-1
cm-2
.

35
Chapter 6. Discussion
In this chapter a summary of the important results from the experiments will be given and some
general trends will be discussed. However, first it should be stated that there is often a considerable spread in
the values measured on each sample, as seen by the relatively large error bars (standard deviation). Several
samples also deviate from any trends, which makes the analysis difficult. These inconsistencies may be related
to the small size of the samples and that each sample is cleaned and handled separately, which could lead to a
spread in data.
It is, nevertheless, observed that for all methods, annealing the samples generally results in a Vfb closer
to zero. Tables 6.1-3 show all the Vfbs values and the corresponding annealing temperatures for the two
PeCVD tools and the thermal growth.
Table 6.1. Annealing temperatures versus Vfb for P5000 samples.
P5000 Flatband Voltage (V)
Annealing
Temperature
(°C) P1 P2 P3 P4 P5 P6 P7 P8 P9 P10
1150 3.50 - - - 0.62 0.79 -4.07 0.10 -10.61 -0.18
1000 - - 0.64 - - - - - - -
N/A - -3.97 2.17 - - - - - -
Table 6.2. Annealing temperatures versus Vfb for Pekka samples.
Pekka Flatband Voltage (V)
Annealing
Temperature
(°C) Pe11 Pe12 Pe13 Pe14 Pe15 Pe16 Pe17 Pe18 Pe19 Pe20 Pe21 Pe22 Pe23
1150 0.45 1.34 0.55 - 1.40 -1.03 0.61 -1.15 - -4.53 - - -
1000 - - - - - - - - - - 0.54 - -
850 - - - - - - - - - - - 0.49 -
N/A - - - 1.48 - - - - 1.46 - - - 4.78
Table 6.3. Annealing temperature versus Vfb for thermally grown oxide samples.
Thermally grown oxide Flatband Voltage (V)
Annealing Temperature
(°C) T24 T25 T26
1150 1.19 - 0.80
N/A - 2.00 -
As shown by the tables 6.1-3, annealing leads to a reduction of oxide defects. This effect is perhaps
more clearly seen for samples processed in Pekka, since in P5000 several samples, like P1, P7 and P9, still
have high absolute value of Vfb even though they are annealed. This difference between Pekka and P5000
could possibly be explained by that the power of P500 is one order of magnitude higher than Pekka. Even if it

36
is not possible to directly compare this parameter between the two setups due to different geometries and
designs, it could imply that the ions arrive at the sample surface with higher energy in P5000, which causes
more defects in the oxide structure.
For the PeCVD tools, it is also found that the samples with low Dits are nearly always combined with a
low absolute value of Vfb. Figure 6.1 shows the results for both tools.
Figure 6. 1 Dits versus Vfbs for all a) P5000 and b) Pekka samples.
P5000 sample values are plotted in Figure 6.1a. As it can be seen it starts from a high Dit, in the order of 1012
eV-1
cm-2
and Vfb less than -10 V and it drops linearly to 2×1011
eV-1
cm-2
with Vfb almost zero, or slightly
positive values for a majority of the samples. Figure 6.1b shows values from Pekka samples. Again it begins
from a Dit of the order of 1012
eV-1
cm-2
with Vfb -5 V, it drops to a minimum with Dits of 2×1011
eV-1
cm-2
with Vfb close to zero V, but then Dit values saturates or even increases again at 6×1011
eV-1
cm-2
while Vfb
rises to 5 V. The two graphs in Fig. 6.1 shows that for both tools there is a cluster of samples with Dits less
than 5×1011
eV-1
cm-2
while at the same time the Vfb is less than 1 V. This behaviour indicates that there exist
a rather wide range of process parameters for the PeCVD techniques that gives low Dit values and flatband
voltage around 0 V. A possible explanation is that the nitrogen treatment in PeCVD is an effective method to
passivate the oxide defects since both properties, Dit and Vfb, are minimized.
For the Pekka tool it is, furthermore, found that there is a correlation with Vfbs and Dits with the
surface roughness of the samples. By comparing all the samples that have been annealed at 1150 °C, this trend
is illustrated in Figure 6.2.

37
Figure 6. 2 a) Roughness (RMS) versus Vfb and b) Dit versus the product of roughnessGas ratio for samples which oxide deposited
with Pekka tool and have been annealed at 1150 °C
As can be seen from Figure 6.2a, a lower roughness is combined with Vfb closer to zero. A small value of the
roughness could mean that the oxide is more homogeneous, leading to fewer oxide defects, which leads to low
Vfb. For larger roughness values, the flatband voltage will increase to either more positive or more negative
values.
No clear trend could be seen for the relation between the surface roughness and the density of interface
traps. However, in Figure 6.2b the product of roughness and gas ratio is shown with respect to Dit and here the
trend is clear. The Dit values decrease and saturate at low values for large values of the roughness together
with high levels of nitrogen. The relation in Fig. 6.2b is not easy to understand, but it seems that if the
roughness is large at the same time as the nitrogen gas is enhanced, this will lead to a better uptake of the
nitrogen and lower Dit. Of course, a high uptake of nitrogen could also lead to an increased roughness.
Furthermore, it is observed that by simultaneously increasing the power, pressure and the gas ratio during the
deposition process, the Dits tend to decrease. In Figure 6.3 the density of interface traps of the samples that are
annealed at 1150 °C are compared to the product of power, pressure and gas ratio. The Dit reduction is strong
in the beginning and then tends to level out. Power, pressure and gas ratio can all be correlated with the
energy and amount of nitrogen that is deposited and reach the interface. The more nitrogen at the interface, the
lower value of Dits.

38
Figure 6. 3 Dits versus (PowerPressureGas Ratio) for samples with oxide deposited in Pekka and annealed at 1150 °C
From P5000 the sample that combines lowest Dit (2.82+2.19×1011
eV-1
cm-2
) and low Vfb (0.79+0.42
V) is P6. It is fabricated with the highest pressure (6 Torr), medium power (100 W), high gas ratio
(N2O/SiH4=20) and it is annealed at 1150 °C. As for Pekka, sample Pe22 is the best with three times lower Dit
(7.94+6.71×1010
eV-1
cm-2
) compared to P6 and about half the Vfb (0.48+0.29 V) of the P6. The fabrication
parameters for Pe22 are: pressure 0.8 Torr, power 20 W, which are the highest for this tool, gas ratio of 0.6, but
it is annealed at only 850 °C. On the other hand the samples that have highest Dit and Vfb are P2 and Pe23 from
P5000 and Pekka, (8.33+4.15×1011
eV-1
and -3.96+0.88 V, and 6.37+4.29×1011
eV-1
cm-2
, 4.78+1.36 V,
respectively), respectively. P2 is fabricated with power and pressure of 100 W and 2.8 Torr, while the gas ratio
is (N2O/SiH4) 20 and it is not annealed. The parameters for Pe23 are: power 20 W and pressure 0.5 Torr, with
gas ratio (N2O/SiH4) of 0.6, and it is also not annealed.
Finally, as for the samples with thermally grown oxide, they all appear to have low Dits, in the order of
1-2×1011
eV-1
cm-2
and low Vfb of 1-2 V. Growth temperature appear to have no significant effect to Vfb and
Dit, whereas annealing results in a more uniform oxide, reducing the flatband voltage.

39
Chapter 7. Conclusions
In this thesis, the process for oxide fabrication in order to optimize the interface between 4H-SiC and
SiO2 is presented. Plasma-enhanced chemical vapour deposition (PeCVD), which is done in the Electrum
cleanroom with two different tools, P5000 and Pekka, and thermal oxidation are the two methods that are
investigated. Capacitance-voltage (CV) and current-voltage (IV) are utilised for the electrical characterization,
whereas the structural characterization is done by using atomic force microscopy (AFM).
From our experiments a few main conclusions can be extracted, even if a large spread in the data
makes it difficult to find very clear trends. The main results are the following:
 Annealing in N2O ambient is very efficient to reduce the flatband voltage to nearly zero.
 Nitrogen treatment can also reduce the density of interface traps, Dit, introduced by the PeCVD
technique.
 Terman method is used for extracting the interface trap density (Dit) from CV data. This method
underestimates the real number of traps and it is clear that a large number of Dits are still present.
 P5000 seems to be a more unreliable tool than Pekka for oxide deposition, since there is a larger
spread in the data for samples processed in P5000.
 It should also be of interest to complement these electrical measurements with more extensive
structural characterization of the samples, to see if the cause for the large variation in measured data
could be, for instance, pinholes.
From the investigation, it is clear that nitrogen annealing has a positive effect on the flatband voltage and
the interface trap densities, although it is not possible to reduce the Dit values to levels that are acceptable for
devices. For future work alternative dielectric deposition techniques, as well as other dielectric systems,
should be investigated.

41
Appendix
In the next tables the results from the electrical and structural characterization are summarised for all
samples.
P5000 samples
Sample
number
Flatband
Voltage
Standard
Deviation
Density of
interface traps
Standard
Deviation
Critical
Electric
Field Roughness
(V) (V) (1011
eV-1
cm-2
) (1011
eV-1
cm-2
) (MV/cm) (nm)
1 3.55 1.39 2.73 0.42 3.0 0.7
2 -3.97 0.89 8.33 4.15 3.5 0.6
3 0.64 0.62 2.51 0.93 4.0 0.8
4 2.17 0.26 5.69 1.96 1.5 5.7
5 0.62 0.60 4.37 2.30 6.0 0.7
6 0.79 0.42 2.82 2.19 5.5 6.7
7 -4.07 0.77 15.3 5.62 3.5 0.9
8 0.10 0.43 3.92 1.89 4.0 1.2
9 -10.61 0.81 20.19 6.94 2.0 0.6
10 -0.18 0.21 5.49 4.22 1.5 0.8
Pekka samples
Sample
number
Flatband
Voltage
Standard
Deviation
Density of
interface traps
Standard
Deviation
Critical
Electric
Field Roughness
(V) (V) (1011
eV-1
cm-2
) (1011
eV-1
cm-2
) (MV/cm) (nm)
11 0.45 0.36 1.61 0.89 7.0 0.8
12 1.34 0.68 2.99 2.17 5.0 2.9
13 0.55 0.39 3.40 0.86 2.0 1.2
14 1.48 0.61 4.20 1.49 2.0 0.5
15 1.40 0.65 2.26 0.67 7.5 2.4
16 -1.03 0.96 7.72 5.48 2.5 1.2
17 0.61 0.96 7.54 4.63 6.0 0.9
18 -1.15 0.52 3.87 1.65 7.0 2.4
19 1.46 0.88 5.25 4.52 7.0 2.7
20 -4.53 1.12 11.63 5.32 7.0 1.4
21 0.54 0.43 2.04 1.53 2.5 1.2
22 0.49 0.30 0.79 0.67 1.5 0.7
23 4.78 1.36 6.37 4.29 - 1.1

42
Thermally grown samples
Sample
number
Flatband
Voltage
Standard
Deviation
Density of
interface traps
Standard
Deviation
Critical
Electric
Field Roughness
(V) (V) (1011
eV-1
cm-2
) (1011
eV-1
cm-2
) (MV/cm) (nm)
24 1.19 0.51 1.99 1.92 2.0 0.8
25 2.00 0.35 1.23 0.45 3.5 1.2
26 0.81 0.47 2.10 1.28 4.0 3.2

43
References
[1] Lee, H.-S. (2008). Fabrication and Characterization of Silicon Carbide Power Bipolar Junction
Transistors. Stockholm: KTH, Royal Institute of Technology. PhD Thesis
[2] Afanas, V. V. (1999). Electronic properties of SiO2 / SiC interfaces. Microelectronic Engineering, 48,
241-248.
[3] J. A. Cooper, J. (1997). Advances in SiC MOS Technology. Physica Status Solidi (a), 162(1), 305.
[4] Rozen, J., Dhar, S., Pantelides, S. T., Feldman, L. C., Wang, S., Williams, J. R., et al. (2007). Suppression
of interface state generation upon electron injection in nitrided oxides grown on 4H-SiC. Applied
Physics Letters, 91(15), 153503.
[5] Li, H.-F., Dimitrijev, S., Harrison, H. B., & Sweatman, D. (1997). Interfacial characteristics of N2O and
NO nitrided SiO2 grown on SiC by rapid thermal processing. Applied Physics Letters, 70(15), 2028.
[6] Roccaforte, F., Fiorenza, P., Greco, G., Vivona, M., Lo Nigro, R., Giannazzo, F., et al. (2014). Recent
advances on dielectrics technology for SiC and GaN power devices. Applied Surface Science, 301, 9-
18.
[7] Zetterling, C.-M. (1997). Silicon Dioxide and Aluminum Nitride as Gate Dielectric for High Temperature
and High Power Silicon Carbide MOSFETs. Stockholm: KTH, Royal Institute of Technology. PhD
Thesis
[8] Soo, M. T., Cheong, K. Y., & Noor, A. F. (2010). Advances of SiC-based MOS capacitor hydrogen
sensors for harsh environment applications. Sensors and Actuators B: Chemical, 151(1), 39-55.
[9] Wolborski, M. (2006). Characterization of Dielectric Layers for Passivation of 4H-SiC Devices.
Stockholm: KTH, Royal Institute of Technology. PhD Thesis
[10] Chung, G. Y., Tin, C. C., Williams, J. R., McDonald, K., Di Ventra, M., Pantelides, S. T., et al. (2000).
Effect of nitric oxide annealing on the interface trap densities near the band edges in the 4H polytype
of silicon carbide. Applied Physics Letters, 76(13), 1713.
[11] Afanasev, V. V., Bassler, M., Pensl, G., & Schulz, a. M. (1997). Intrinsic SiC/SiO2 Interface States.
physica status solidi (a), 162, 321.
[12] Afanas ev, V. V., Ciobanu, F., Dimitrijev, S., Pensl, G., & Stesmans, A. (2004). Band alignment and
defect states at SiC/oxide interfaces. Journal of Physics: Condensed Matter, 16(17), S1839-S1856.
[13] Jamet, P., Dimitrijev, S., & Tanner, P. (2001). Effects of nitridation in gate oxides grown on 4H-SiC.
Journal of Applied Physics, 90(10), 5058.
[14] Dimitrijev, S., & Li, H.-f. (1997). Nitridation of silicon-dioxide films grown on 6H silicon carbide.
Electron Device Letters, 18(5), 175-177.
[15] Retrieved from
http://www.iue.tuwien.ac.at/phd/hehenberger/dissap2.html

44
[16] M. Sze, S., & Ng, K. K. (2006). Physics of Semiconductor Devices. New Jersey: JOHN WILEY &
SONS.
[17] Retrieved from http://ecee.colorado.edu/~bart/book/book/chapter6/ch6_2.htm
[18] Hu, C. C. (2010). Modern Semiconductor Devices. Pearson Education Inc.
[19] Plummer, J. D., Deal, M. D., & Griffin, P. B. (2000). Silicon VLSI Technology: Fundamentals, Practice
and Modeling. New Jersey : Prentice Hall Inc.
[20] Masud-Ur-Rashid, M. (2012). Characterization of Dynamic Elastic Modulus and Damping Property of
CNx Coating Material by Experimental Modal Analysis and Finite Element Approach. Stockholm:
KTH, Royal Institute of Technology. Master Thesis
[21] Wosinski, L. (2003). Technology for photonic components in silica / silicon material structure.
Stockholm: KTH, Royal Institute of Technology. PhD Thesis
[22] Retrieved from http://lims.electrumlab.se/WebForms/Equipment/EquipmentView.aspx?toolId=19
[23] O’Brien, S., & Nash, J. (2009). Manual P5000. Semiconductor & Microsystems Fabrication Laboratory.
[24] Retrieved from http://lims.electrumlab.se/WebForms/Equipment/EquipmentView.aspx?toolId=74
[25] Bhat, M., Jia, H. H., & Kwong, D. L. (1995). Growth kinetics of oxides during furnace oxidation of Si
in N2O ambient. Journal of Applied Physics, 78(4), 2767.
[26] Sanjeev Kumar Gupta and Jamil Akhtar (2011). Thermal Oxidation of Silicon Carbide (SiC) –
Experimentally Observed Facts, Silicon Carbide - Materials, Processing and Applications in Electronic
Devices, Dr. Moumita Mukherjee (Ed.), ISBN: 978-953-307-968-4, InTech
[27] C - V Characterization of MOS Capacitors Using the Model 4200-SCS Semiconductor Characterization
System. (2896).
[28] Schroder, D. K. (2005). Oxide and Interface Trapped Charges, Oxide Thickness. In Semiconductor
Material and Device Characterization. New Jersey: John Wiley & Sons, Inc.
[29] Taur, Y., & Ning, T. H. (2009). Fundamentals of Modern : VLSI Devices. CAMBRIDGE University
Press.
[30] Deal, B. E., & Nicollian, E. H. (n.d.). Standardized Terminology for Oxide Charges Associated with
Thermally Oxidized Silicon. Journal of Electrochemical Society, 127(4), 979-981.
[31] Suvanam, S. S. (2011). A comparison of free carrier absorption and capacitance voltage methods for
interface traps measurements. Stockholm: KTH, Royal Institute of Technology.
[32] Usman, M. (2012). Impact of Ionizing Radiation on 4H-SiC Devices by. Stockholm: KTH, Royal
Institute of Technology.
[33] Liu, X. (2014). Surface Forces and Friction Effects of adsorbed layers and surface topography.
Stockholm: KTH, Royal Institute of Technology.

45
[34] Platz, D. (2013). Reconstructing force from harmonic motion. Stockholm: KTH, Royal Institute of
Technology.

TRITA -ICT-EX-2015:18
www.kth.se

Process Optimization 4H-SiC SiO2 dielectric interface.

More Related Content

Similar to Process Optimization 4H-SiC SiO2 dielectric interface. (20)

Process Optimization 4H-SiC SiO2 dielectric interface.