微波退火技術應用於新穎式透明非晶態銦鎵鋅氧薄膜電晶體之研究

國立交通大學

光電系統研究所

碩士論文

微波退火技術應用於新穎式透明非晶態銦鎵鋅氧薄膜電晶體

之研究

Study of Microwave annealing Technology on Novel

Transparent Amorphous Indium Gallium Zinc Oxide Thin

Film Transistors

研 究 生 : 羅婉柔 Yuan-Jou Lo

指導教授 : 劉柏村 教授 Prof. Po-Tsun Liu

林建中 教授 Prof. Chien-Chung Lin

微波退火技術應用於新穎式透明非晶態銦鎵鋅氧薄膜電晶體之研究

Study of Microwave Annealing Technology on Novel Transparent

Amorphous Indium Gallium Zinc Oxide Thin Film Transistors

研 究 生 : 羅婉柔 Yuan-Jou Lo

指導教授 : 劉柏村 教授 Prof. Po-Tsun Liu

林建中 教授 Prof. Chien-Chung Lin

國立交通大學

光電系統研究所

碩士論文

A Thesis

Submitted to Institute of Photonic System College of Photonics

National Chiao Tung University in partial Fulfillment of the Requirements

for the Degree of Master

in

Photonic System

October 2012

Hsinchu, Taiwan, Republic of China

微波退火技術應用於新穎式透明非晶態銦鎵鋅氧薄膜電晶體

之研究

研究生: 羅婉柔

指導教授: 劉 柏 村 教授

林 建 中 教授

國 立 交 通 大 學

光電系統研究所

摘 要

非晶態氧化半導體(Amorphous Oxide Semiconductors: AOSs) 具有高載子遷移

率而且可低溫沉積、可撓曲、透明性以及均勻度佳等特點已受到廣泛的重視。目

前被研究的氧化半導體有 ITO、IZO、TiO2

、ZnO、In2

O

、Ga2

O

、IGO、a-IGZO…

等。其中以透明非晶態氧化銦鎵鋅(Amorphous InGaZnO: a-IGZO)薄膜當作主動層

(Active layer)的薄膜電晶體(Thin Film Transistors: TFTs)，具有載子遷移率與可靠

度比傳統氫化非晶矽薄膜電晶體(a-Si:H TFT)高、均勻性優於低溫複晶矽薄膜電晶

體(Low Temperature Polycrystalline Silicon TFT: LTPS TFT) 而且可低溫製程，因

此 a-IGZO 薄膜電晶體具有取代氫化非晶矽薄膜電晶體與低溫複晶矽薄膜電晶體

來製作主動矩陣有機發光顯示器(Active Matrix Organic Light Emitting Display:

AMOLED)的潛力。但是 a-IGZO 存在著一些本質上的缺點，例如對光以及環境中

的水、氧很敏感進而影響元件的穩定性。然而，a-IGZO TFT 長時間操作的可靠度

也是需要探討。本論文主要研究 a-IGZO 經由微波退火後的穩定性分析、元件長

時間操作的可靠度分析與紫外光影響元件穩定性分析等三個主要方向來進行探

討，以期得到改善元件穩定性的方法。從實驗結果我發現，當增加微波退火時間

和能量，可以提升元件穩定性與可靠度。以這個特性，最後我提出新穎式微波退

火技術以改善傳統爐管退火長時間以及高溫的特性。

Study of Microwave Annealing Technology on Novel

Transparent Amorphous Indium Gallium Zinc Oxide Thin

Film Transistors

Student : Yuan-Jou Lo

Advisors : Prof. Po-Tsun Liu

Prof. Chien-Chung Lin

College of Photonics & Institute of Photonic System

National Chiao Tung University, Hsinchu, Taiwan

Abstract

Amorphous oxide semiconductors (AOSs) are attracted much attention due to

high mobility, low temperature deposition, flexible,transmission, and uniformity.It

has been investigated of AOSs, such as ITO, IZO, TiO

, ZnO, In

O

, Ga

O

, IGO,

a-IGZO,etc. Especially,the thin film transistors (TFTs) with a-IGZO thin film as

active layer perform higher mobility and better reliability than conventional

hydrogenated amorphous silicon TFT (a-Si: H TFT). In addition, the uniformity of

a-IGZO TFT is also superior to Low Temperature Polycrystalline Silicon TFT

(LTPS TFT). Therefore, the a-IGZO TFTs have the potential to replace a-Si: H TFT

and LTPS TFT forming Active Matrix Organic Light Emitting Display (AMOLED).

However, a-IGZO there are some inherent defect, such as sensitive to water and

oxygen in ambient and light illumination thereby affect the device stability. The

device reliability under GBS also has to be considered.In this thesis, we studied the

interaction between microwave annealing and furnace annealing a-IGZO thin film

transistors.Time dependence threshold voltage variation reliability are investigated.

Photo-reaction of a-IGZO TFT under UV-illumination is also investigated. From

the experimental results, increasing microwave annealing time and power can

improve the device stability and reliability under GBS. Finally we successfully

investigated that microwave annealing can get better electrical characteristics,

stability and reliability than furnace annealing.

誌謝

首先要感謝我的指導教授劉柏村教授為我們建立了一個資源豐富的研

究環境，讓我的碩士班生涯裡受益良多！無論在求學和生活中，老師都給予了

我最無私的指導與關懷。除了設備完善讓我們可以有世界級的研究成果，高水

準的的師資與研究團隊可以隨時與世界頂尖研究成果爭雄，成為更全方位的人

才，可以在未來為校爭光，回饋培育我們的社會。

在實驗的過程中首先要非常感謝鄧立峯與傅治翔學長，一直帶領我們

學弟妹，也不時給我們關心並給予指點並有情有義的帶著我們畢業。還要感謝

我的兩位夥伴王薇雅與張智翔，與我在無塵室內的量測機台共度無數時光。在

漫漫長夜也有破曉時分，擁有得來不易的令人欣慰的成果時，喜悅可以彼此共

享。同時也感謝實驗室中一起成長的好夥伴們揚順、光廷、震碩、立煒學長以

及亦君、俊彥、沁卉等這兩年以來的互相扶持與鼓勵，陪伴我度過這每一刻研

究生涯。還有豐榮、玫諍與韋勳等的加入也使實驗室注入了新的活力。一起相

約打球、聚餐或出遊玩樂等無數歡樂時光也將是珍貴而不可取代的美好回憶。

最後，我要感謝我的父母家人，你們的養育之恩、栽培之情與不離不

棄的陪伴，比任何言語都要深刻而有力的支持著我，讓我能夠繼續堅定地朝著

理想而前進。我會用一切的努力來報答，並用這篇論文獻給我最愛的家人。

Chinese Abstract………..………...I

English Abstract………....…..II

Acknowledgements…………...……….….III

Contents………...IV

Figure Captions...……….……....……..……....VI

Table Captions...VIII

Contents

Chapter 1 Introduction

1.1.

General Background

1.2.

Amorphous Oxide Semiconductors

1.2.1. Introduction to Amorphous Oxide Semiconductor………….……….4

1.2.2. The Carrier Transport Mechanism of AOSs……….………...7

1.2.3. Amorphous IGZO………..……...9

1.3. Microwave Annealing Process

1.3.1. Interaction of Microwaves with Matter...14

1.3.2. Microwave Processing of Materials……….………….……...…..……18

1.4. Thesis Organization

Chapter 2 Experiment Procedures

2.1. Experiment Procedures

2.3. Basic Microwave Heating System

2.4. Parameter Extraction Method

2.5. Measurement of Reliability on a-IGZO TFTs………..….

Chapter 3 Results and Discussion

3-1. The effect of different MWA time on a-IGZO TFT

3.1.1. a-IGZO TFT with SiO2 gate insulator………...31

3.1.2. a-IGZO TFT with SiNx gate insulator……….…...33

3-2. The effect of different MWA Power on a-IGZO TFT

3.2.1. a-IGZO TFT with SiO2 gate insulator……….…….…...35

3.2.2. a-IGZO TFT with SiNx gate insulator……….……….…...37

3-3. The comparison of MWA and Furnace annealing on a-IGZO TFT

3.3.1. a-IGZO TFT with SiO2 gate insulator……….…..39

3.3.2. a-IGZO TFT with SiNx gate insulator………..….41

3-4. Material Analysis Results

Chapter 4 Experimental Results and Discussion

4-1. Conclusions

Figure Caption

Fig. 1-1 The basic operation of the pixel array: After the scan line was selected,

the data line applied a voltage to charge the storage capacitor…………....

Fig. 1-2 Cross section of a LCD’s subpixel………..………...

Fig. 1-3 Schematic orbital drawing of electron pathway (conduction band bottom)

in

conventional

silicon-based

semiconductor

and

ionic

oxide

semiconductor………...………...

Fig. 1-4 Transmittance of a-IGZO film in visible light region...

Fig. 1-5 The carrier concentration as a function of O

pressure during the deposition

in a- InGaZnO

and a-In

Zn

O

...

Fig.

1-6

(a)

The microwave heating system, (b)

the setup in the microwave

chamber.………...

Fig. 2-1 The cross-section of a-IGZO TFTs...

Fig. 2-2 Schematic DC sputtering system………...

Fig.

3-1

The

electrical

characteristics

of

different

MWA

time

for

SiO

gate

insulator. …..………..…………..……..

Fig.

3-2

The

PBIS

and

NBIS

results

for

different

MWA

time

for

SiO

gate

insulator..………...

Fig.

3-3

The

electrical

characteristics

of

different

MWA

time

for

SiN

gate

insulator………...

Fig.

3-4

The

PBIS

and

NBIS

results

for

different

MWA

time

for

SiN

gate

insulator………...

Fig.

3-5

The

electrical

characteristics

of

different

MWA

power

for

SiO

gate

insulator………..…………....

Fig.

3-6

The

PBIS

and

NBIS

results

for

different

MWA

power

for

SiO

gate

insulator………...

Fig.

3-7

The

electrical

characteristics

of

different

MWA

power

for

SiN

gate

Fig.

3-8

The

PBIS

and

NBIS

results

for

different

MWA

power

for

SiN

gate

insulator………..………...

Fig. 3-9 Electrical characteristics comparison with MWA and furnace annealing for

SiO

gate insulator………...

Fig. 3-10 The PBIS and NBIS comparison of MWA and furnace annealing for SiO

gate insulator………...

Fig. 3-11 Electrical characteristics comparison with MWA and furnace annealing

for SiN

gate insulator……….

Fig. 3-12 The PBIS and NBIS comparison of MWA and furnace annealing for SiN

gate insulator………...

Fig. 3-13 The X-ray diffraction results for different MWA time, MWA power and

furnace annealing…...……….……....

Fig. 3-14 The scanning electron microscope results for different MWA time, MWA

power and furnace annealing………...……...

Fig. 3-15 The X-ray photoelectron spectroscopy results for different MWA time,

MWA power and furnace annealing……….…..

Fig. 3-16 The X-ray photoelectron spectroscopy results of 2P_100s and furnace

Table Caption

Table 1-1 The comparison of a-Si TFT, poly-Si TFT and TAOSs TFT………

Table 2-1 Experiment flow chart………..……

Table 2-2 The experiment flow of devices under light illumination………....

Table 3-1 The data of MWA different time for SiO

gate insulator………...

Table 3-2 The data of different MWA time for SiN

gate insulator………...

Table 3-3 The data of MWA different power for SiO

gate insulator………….….

Table 3-4 The data of different MWA power for SiN

gate insulator………….….

Table 3-5 The data of comparison with MWA and furnace annealing for SiO

gate

insulator………...…...

Table 3-6 The data of comparison with MWA and furnace annealing for SiN

gate

insulator………...

Table 3-7 The data of optical band-gap for different MWA time, MWA power and

furnace annealing………..………...………...

Chapter 1 Introduction

1.1 General background

Display is the widely used productions. We can find a lot of applications in our daily life.

Therefore, display technologies become more and more important in recent years. Display

industry can be considered to start from cathode ray tubes and transfer to TFTs. In this case,

we need the brief introduction about TFT-LCD. The basic structure of a TFT-LCD panel, as

shown in Fig. 1-1 [1], may be thought of as two glass substrates sandwiching a layer of liquid

crystal. The upper glass substrate is fitted with a color filter, while the back glass substrate has

transistors fabricated on it. Furthermore, the black matrix, an opaque polymer, which is

deposited on the upper glass structure, blocks the environment light to illuminate the front of

TFT and incidentally avoids the incorrect color mixture by back light through color filter. [2]

As shown in Fig. 1-2 [2] we can see the cross-section of TFT-LCD’s subpixel. However,

nothing shields the light from back illumination, so the photocurrent still takes place in LCD

operation. Transistors can be classified into the depletion-mode and the enhancement-mode,

based on whether drain current flows through TFT when no voltage is applied to gate

electrode. For the depletion-mode TFT, when the device is on, drain current flows through the

device when no gate voltage is applied. For the enhancement-mode TFT, when the device is

data line applied a voltage to charge the storage capacitor, which the driving signals addressed

the sub-pixel and therefore the liquid crystal twisted in order to correspond the gray level that

data line provided.

After the scan line was selected, charging a voltage to open the switch, the data line

applied a voltage to charge the storage capacitor, which the driving signals addressed the

sub-pixel and therefore the liquid crystal twisted in order to correspond the gray level that

data line provided. But at the same time TFTs leak current from the data line or the storage

capacitor, the gray level is not precise anymore, especially deep black level—the most

sensitive to human eyes. On the other hand the contrast performance (white level brightness over black level brightness) usually depends on black level, so contrast degrades by TFT’s leakage current at off-state. What is more, many drawbacks of image quality, say, flicker and

vertical crosstalk, occur at the condition that TFTs having large off-current.

Fig. 1-2 Cross section of a LCD’s subpixel [2]

Recently, a new electronics is emerging for applications because we need the

performance more efficient. So far, organic molecule semiconductors have been almost

exclusively examined for such applications but their performance and chemical instability are

not sufficient for practical applications. According to above statements, we can find that TFTs

play important roles in the flat panel display industry. For different applications, the methods

of improving TFT become more and more important. In this case, we have five requirements

for TFT. First, we want higher refresh rates, fast pixel switching speeds, high on/off current

1.2 Amorphous Oxide Semiconductors

1.2.1 Introduction to Amorphous Oxide Semiconductor

There’re a lot of materials for TFTs’ active layers. For example, we can see there are

three kinds of TFT which are much wider used in flat display industry. These three are a-Si

TFT, poly-Si TFT and transparent amorphous oxide semiconductor (TAOS) TFT. As we show

in the Table 1-1 [5] we can see the comparison of different kind of TFTs. A-Si TFT has good

stability and uniformity, but the performance isn’t good enough. So it is usually choose a-Si

TFT for large size display. [6] Poly-Si TFT has good stability and performance, but its

uniformity isn’t good enough. So it is usually choose poly-Si TFT for small size display. [6]

Transparent amorphous oxide semiconductor TFT has not only good stability and uniformity

but also good performance. So we chose TAOSs TFT as our active layer in this study.

According to the above statements, we can find that TAOSs TFT not only have good

performance but also have other five advantages. TAOS TFT has high mobility, good

uniformity and high optical transparency. It can deposit at room temperature and ease of

fabrication. It also has low off current, large allowance in the choice of gate insulator and

compatible with the present flat display panel industry.

Research on amorphous semiconductors started in 1950s to seek appropriate materials.

The largest impact on electronics is the discovery of hydrogenated amorphous silicon (a-Si:H)

by Spear and LeComber in 1975. This is the first material which can control carrier concentration by impurity doping as in crystalline and it opened a new frontier called ‘Giant Micro-electronics’ which means electronics based on circuits fabricated on a large area

substrate. [7] The structure of inverted-staggered type is the most common one used in

TFT-LCD products nowadays. [8] The fabrication process in TFT manufacture industry needs

many thin film deposition techniques. Take the inverted-staggered bi-layer a-Si:H TFT for

instance, the gate insulator (GI) and semiconductor layers can be deposited in a in-situ step

(one-pump-down process). That in-situ step means “processes without breaking vacuum

between two processes.” By using this “in-situ step” method, a clean interface, which is very

important for high-performance device characteristics, can be easily obtained. [9] Besides, the

characteristics of the interface between each layer can be easily and well controlled by only

distribution, and dangling bond passivation. The well step of fabrication can guarantee the

good ohmic contact properties during the source/drain and semiconductor layer regions. In

addition, the major advantages of this material are its good electronic properties such as high

photoconductivity and the ability to achieve low cost fabrication at low temperatures. [10]

Transparent amorphous oxide semiconductors (TAOS) are good candidates for high

performance thin film transistor devices during TFT-LCD application. It’s the materials for

high performance, low processing temperature electronic devices with high mobility >> 10

cm2 (Vs)-1 and high transmittance about 75% in visible light portion of the electromagnetic

spectrum and even when their films are deposited at room temperature. [11] However, the

TAOS materials have uncontrollable carriers generated from oxygen vacancy. Therefore, it is

important to design and explore a suitable material having both properties of large mobility

1.2.2 The Carrier Transport Mechanism of AOSs

The mobility of a-Si:H (~1 cm2(Vs)-1) is much smaller than that of single crystalline Si

(~200 cm2(Vs)-1) due to the intrinsic chemical bonding nature. The average carrier paths in

covalent semiconductors, such as a-Si:H, consist of strongly directive sp3 orbital. The bond

angle fluctuation significantly alters the electronic levels, causing high density of deep

tail-states, as shown in Fig. 1-3. [12]

Fig. 1-3 Schematic orbital drawing of electron pathway (conduction band bottom) in conventional silicon-base semiconductor and ionic oxide semiconductor. [12]

In contrast, transport oxides constituting of heavy post transition metal cat-ions with the

(n-1)d10ns0 electron configuration, where n≧5, are the transparent AOS (TAOS) candidates

having large mobility comparable to those of the corresponding crystals. [13] The electron

pathway in oxide semiconductor is primarily composed of spatially spread ns orbital with an

orbital is possible. The degree of overlap of the ns orbital is insensitive to the distorted

metal-oxygen-metal bonding. This feature shows why the Hall mobility of AOSs is similar to

the corresponding crystalline phase, even under the room temperature deposition of

1.2.3 Amorphous In-Ga-Zn-O TFTs

Among the TAOS materials, amorphous indium gallium zinc oxide (a-IGZO) is one of

the most glaring candidates serving as semiconductor layer in thin film transistor (TFT). [14]

However there are still many critical issues existed in a-IGZO TFT, especially for the easy

absorption and desorption reaction of the oxygen atom with the surrounding atmosphere. As

the oxygen species are absorbed from the ambient atmosphere, they can capture electrons in

the conducting channel and form a depletion region beneath back channel layer. By following

the equation of O g e O  s   _  ₂ ) (

2 _{, the resultant buildup of absorbed negative space}

charges O2-(s) easily repells conduction electrons and positively shifts Vth of a-IGZO TFT. [14]

Whereas the desorption of oxygen atoms in a-IGZO back channel will result in the left shifts

of Vth. This random reaction of absorption and desorption happening also leads to some

uniformity problems. Some researching groups have applied passivation layer method to

shield the back channel from the contact with ambient air. But the sequel processes would

strongly affects the original properties of a-IGZO film. Therefore, the fundamental method to

release the issue of environmental influence should be the improvement of the film quality of

itself.

TAOSs have attracted keen attention since the high performance thin-film transistors can

by obtained by using the amorphous In-Ga-Zn-O (a-IGZO) thin films for the semiconductor

TFT performance is also confirmed by using the sputter deposition [15], which demonstrates

the possibility of the large-area applications. The dependence of the TFT characteristics on the

metal composition is investigated in detail by a novel combinational approach, since the

multi-metal AOSs can take any ratios of the composition.

The a-IGZO film has electrons as majority carriers, which is mainly affected by the

oxygen vacancies and oxygen interstitials during deposition processes. [16] The ion bonding

structure makes the a-IGZO TFT exhibit high field-effect carrier mobility even in the

amorphous phase. [17] Even if the a-IGZO TFT owns many superior characteristics, the

sensitivity to atmosphere is a extremely critical issue for the a-IGZO TFT application. [18]

The environment-dependent metastability was attributed to oxygen adsorption/desorption

reactions to the backchannel of the a-IGZO TFT device. The random reactions between the

ambient air and the a-IGZO backchannel layer can not only change the oxygen vacancies in

the a-IGZO film but result in a threshold voltage shift with days going by, and even device

uniformity problems. [19] In addition to isolate the a-IGZO layer from exposing to the

atmosphere, the electrical stability and uniformity of the a-IGZO film can be improved by the

optimization of the chemical stoichiometry or adjusting oxygen content inherently.

For the In2O3-Ga2O3-ZnO ternary system, the incorporation of cat-ions with large ionic

valance such as Ga3+ and Al3+ to high conductive oxides such as In2O3 and ZnO is effective to

such as a-ITO and c-ZnO have high density carrier density, hence is difficult to control the

device characteristics. Besides, amorphous In-Ga-Zn-O (a-IGZO) is transparent throughout

the visible spectrum. The transmittance is greater than 80 percent from 400 nm to 850 nm as

shown in Fig. 1-4. [21] Because only In3+ meets the electron configuration criterion

(n-1)d10ns0 (n≧5) of heavy post transition metal cat-ion for ionic AOS (IAOS) among the

three cations. The mobility is primary determined by the fraction of In2O3 content and the

highest value of ~40 cm2(Vs)-1 is obtained for the samples containing the maximum In2O3

fraction. The large ionic valence ions such as Ga3+ combine with high conductive oxides such

as In2O3 and ZnO to control the carrier concentration effectively because of the strong

metal-oxygen bonds. [22] In other words, Ga3+ suppresses carrier generation via oxygen

vacancy formation because Ga ion forms stronger bond with oxygen than Zn and In ion. [23]

Therefore, the InGaZnO4 composition was chosen as the AOS for channel layer of the

transparent TFT. A critical issue of semiconductor materials for TFT applications is

Wavelength (nm) 300 400 500 600 700 % T 50 60 70 80 90 100

Fig. 1-4 Transmittance of a-IGZO film in visible light region. [21]

This is particularly important for AOSs because electron carriers can be easily generated.

It is vital to choose materials which control carrier concentration at low levels (~1014 cm-3) to

achieve low off current (I off) and large on-off current ratio (Ion/I off). [24] In practice, the

effect of binary amorphous materials in the In2O3-ZnO system is employed in commercial

flexible transparent conductive films by depositing on plastic sheet. Thus, the effect of partial

oxygen pressure was studied on the carrier concentration in a-InGaZnO4 and a-In2Zn3O6 and

the results are shown in Fig. 1-5. [25] The carrier concentration in the a-InGaZnO4 is

distinctively reduced to below 1013 cm-3 by increasing PO2 to 8 Pa, on the other hand, it

remains at 1018 cm-3 in the a-In2Zn3O6 deposited under the same condition. It is evident that

incorporation of Ga3+ is supposed to attract the oxygen ions tightly due to its high ionic

around the chemical composition of InGaZnO4 is not sensitive to the variation in the

composition, as shown in Fig. 1-5. [26] Thus, the InGaZnO4 system has better electrical

properties than the In2Zn3O6 system. [27] In addition, the a-IGZO TFTs’ processes are similar

to that of a-Si based TFT. It means the existing production lines can be used. In this case, we

can save a lot of money from buying new equipment.

Fig. 1-5 The carrier concentration as a function of O2 pressure during the deposition in a-

1.3 Microwave Annealing

1.3.1 Interaction of Microwaves with Matter

Conventional heating usually involves the use of a furnace, which heats the walls of the

furnaces by convection or conduction. The core of the sample takes much longer to achieve

the target temperature. Microwave heating is able to heat the target compounds without

heating the entire furnace, which saves time and energy. It is also able to heat sufficiently thin

objects throughout using volumetric heating, rather than through the outer surface. Different

materials convert microwave radiation to differing amounts of heat. The selectivity of

different materials allows the object to be heated at differing speeds as well. [34] [38]

Microwave heating is the perspective techniques, which heat the volume of the wafer,

not just its surface. It used very loosely for electromagnetic radiation in millimeter and radio

frequency spectrum. Microwave processing is quite the same in thermal processing of ceramic

materials. In 1990, Buchta used a microwave generator operating at 2.45GHz and a power

about 1500W to heat 5-inch wafers 125mm to about 1000oC over period of a few seconds.

Recently Thompson and his group used a resonant chamber with a magnetron source. Dr.

Bykov used a 30 GHz gyration device and a resonant processing chamber. They only

demonstrated that microwave can activate at high temperature. In 2007, Jeff M. Kowalski

report that microwave can activate of the heavily doped implanted layers in the range of

situation, and can keep doped less diffusion. The conventional heating is only limited by thermal diffusivity and surface temperature. So, the deep of surface can’t receive the heating energy. But microwave generate heat directly inside the exposed material as result of

molecular motion. Energy is transferred throughout the entire wafer. [29] Therefore, the

effective of activated is more than conventional heating. Electromagnetic (EM) radiation is a

very crucial form of energy available to mankind. It consists of electric and magnetic fields

that fluctuate sinusoidally in planes perpendicular to each other and propagate at the speed of

light. EM radiation does not need a medium to in which to travel. [30] The dual nature of EM

radiation is evident through its wave-like behavior in the case of interference and diffraction

and its particle-like behavior in the case of phenomena like the photoelectric effect. The

quanta of EM radiation are termed as photons. The frequency ν and the wavelength λ are inversely proportional to each other, related by ν = c/λ, where c is the speed of light in vacuum. The energy E of the EM radiation depends linearly on the frequency ν, given by E = h ν where h is the Planck’s constant. [31]

The electromagnetic spectrum is classified into regions of increasing frequencies (or

equivalently, energies): radio waves, microwaves, infrared, visible light, ultraviolet, X-rays,

and gamma rays. Microwaves are generally taken to have frequencies from 300 MHz to 300

GHz which correspond to wavelengths of 1 m down to 1 mm, respectively. Microwaves have

RADAR, electronic warfare, radiation therapy, non-destructive testing of materials, etc. [32]

The interaction of microwaves with materials takes place through the two components of

the microwave radiation: the electric field E and the magnetic field H. The response of a

material when exposed to an electromagnetic radiation may be understood through the

dielectric constant ε of the material. The dielectric constant, also known as the permittivity of

the material, describes the ability of the material to be polarized in the applied electric field.

To understand the dielectric response to sinusoidal fields such as the microwaves, complex

permittivity ε* is employed: ε* = ε’ + i ε”. The real part of the dielectric constant is a measure

of the penetration of microwave energy in the material; while, the imaginary part indicates the

ability of the material to store the energy. The dielectric properties vary with temperature and

frequency. [33]

In any given material, various entities such as the free electrons, valence electrons, ions,

molecular dipoles, and interfacial charges respond to the applied electric and magnetic field.

The sinusoidal fields cause the charged species to polarize and vibrate. Different charged

species all have different natural frequencies of vibration. The conversion to heat occurs

because of the lag of the response of the material to the applied electromagnetic field. In the

heating of dielectric materials, it is assumed that the magnetic field does not contribute to

microwave absorption and the heating occurs entirely due to the electric field.

a. Electronic polarization: When an atom is subjected to an external electric field,

displacement of the electron cloud with respect to the nucleus gives rise to formation of

a dipole. Valence electrons shift much more easily than the tightly bound core electrons.

Covalent crystals have large dielectric constants owing to the displacement of the valence electrons. Thus, materials like silicon (εr’=11.9) and germanium (εr’=11.9) have high real components of the dielectric constant; hence, microwaves easily

penetrate these materials.

b. Dipole polarization: Under the application of an external electric field, polar molecules

orient themselves with the field. The lag associated with this response and the

inter-molecular collisions lead to dielectric heating. In some materials, the polarization

can be retained due to the need for thermal activation for molecular rotation, which gives rise to the formation of “electrets”.

c. Ionic or atomic polarization: Relative displacement of the positive and negative ions or

atoms within molecules and crystal structures from their equilibrium lattice sites gives

rise to ionic polarization.

d. Interfacial polarization: This involves the accumulation of free charges at interfaces

located within the material: grain boundaries, phase boundaries and defect regions.

Under the application of an electric field, the mobile charges are displaced and

1.3.2 Microwave Processing of Materials

Microwaves generate rapidly changing electric fields and will generally heat any

material containing mobile electric charges, such as polar molecules in a solvent or

conducting ions in a solid. Polar solvents are heated as their component molecules are forced

to rotate with the field and lose energy in collisions. Microwave heating a material depends to a great extent on its ‘dissipation’ factor, which is the ratio of dielectric loss or ‘loss’ factor to dielectric constant of the material. The dielectric constant is a measure of the ability of the

material to retard microwave energy as it passes through; the loss factor is a measure of the ability of the material to dissipate the energy. In other words, ‘loss’ factor represents the amount of input microwave energy that is lost in the material by being dissipated as heat. [35]

Therefore, a material with high loss factor is easily heated by microwave energy. In fact, ionic

conduction and dipolar rotation are the two important mechanisms of microwave energy loss

(i.e. energy dissipation in the material).

Microwaves are reflected from the surface and therefore do not heat metals. Metals in

general have high conductivity and are classed as conductors. Conductors are often used as

conduits (waveguide) for microwaves. Materials which are transparent to microwaves are

classed as insulators. Insulators are often used in microwave ovens to support the material to

be heated. Materials which are excellent absorbers of microwave energy are easily heated and

Non-homogeneous material (in terms of dielectric property) may not heat uniformly, that

is, some parts of the materials heat faster than others. This phenomenon is often referred to as

thermal runaway. This condition can be minimized by keeping the sample in mixing or

fluidized condition. Volumetric heating is the key characteristic of microwave processing. In

conventional heating, the thermal energy is transferred to the material from the outside to the

inside, creating a temperature gradient. Small penetration depth of infrared (less than 0.1 mm)

leads to energy deposition being limited to the surface layers. [37] Microwave heating

overcomes this through absorption of the microwave energy throughout the volume of the

material. Since the surface loses energy by radiation, the core of the material is usually hotter

and the temperature profile is the inverse of that seen in conventional heating. Volumetric

heating has the advantage of uniform and rapid processing of materials leading to an

increased throughput. Rapid heating in semiconductors provides the advantage of minimal

diffusion of various species into the substrate.

Poorly absorbing materials (those with small values of ε”) can be hard to heat using microwaves. One common solution to this is the use of microwave susceptors to provide

hybrid heating. [38] Microwave processing can also be employed for selective heating of

1.4 Thesis Organization

This thesis is divided into four chapters. The main purpose of my thesis is to develop an new process method to improve the TFTs’ characteristics. In my thesis, I use a-IGZO as my active layer for increasing the performance of device. Then developed new processing method: MWA for enhancing TAOS and transistor’s characteristics. We will discuss the intrinsic electrical characteristics, stability and reliability in the following pages.

In chapter 1, the brief overview of flat display panel industry, operations of the TFT-LCD, the carrier transport mechanism of AOSs, amorphous In-Ga-Zn-O TFTs and microwave annealing method are introduced.

In chapter 2, the experiment procedures are introduced. The sputtering system and microwave heating system are also described. The measurement and extraction of electrical parameters are also described. The measurement of reliability on a-IGZO TFTs is described.

In chapter 3, the intrinsic electrical characteristics of a-IGZO TFTs with different MWA

time and power with two kind of gate insulator (SiO2 and SiNx) were discussed. Then we will

discuss the comparison of MWA annealing and furnace annealing. Then we will show the results of material analysis with UV-visible, X-ray diffraction, scanning electron microscope and X-ray photoelectron spectroscopy.

Chapter 2 Experiment Procedures

2.1 Experiment Procedures

Table 2.1 is the experiment flow path in my experiment. First of all, we fabricate the TFT

device we want. Then we do different treatment to our device. The special part is we will do

microwave annealing to our device after they are fabricated. The treatment can be classified to

two ways, one is electrical analysis and the other is material analysis.

The TFT devices were chosen as a bottom-gated passivation-free inverted staggered

structure and fabricated on a glass substrate. First, a 100-nm-thick Mo layer was formed as a

gate electrode in a dc sputtering system and a 150-nm-thick silicon nitride (SiNx) was

subsequently deposited on the patterned gate electrode by plasma-enhanced chemical vapor

deposition (PECVD). The active channel layer of a 50-nm thick IGZO layer was formed by

dc sputtering with a power of 100 W at room temperature in argon (Ar) ambiance with flow

rate 10 SCCM (SCCM denotes cubic centimeter per minute at STP) with target of In:Ga:Zn:O

= 1:1:1:4 at.%. The sputtering was carried out at a working pressure of 3 x 10-3 torr and the

base pressure was below 5 x 10-6 torr. Then, a 100-nm-thick indium tin oxide (ITO) was

formed serving as source/drain electrodes by RF sputtering system and all the layers were

defined by shadow mask. The channel width and length of a-IGZO TFTs were varied from 200 to 1000 μm. As we see at Fig. 2-1. Sequentially, all samples were microwave annealed at

was denoted as microwave power 600~700 W for annealing time 100s. The MWA processing

time is defined as the period when the microwave power was turned on. In our experiment,

the 5.8GHz microwave source has been employed for annealing process. A sample with

conventional thermal annealing process was also fabricated at 450 °C for 1 h in a furnace with

N2 gas flow rate of 10L/hr under atmosphere pressure for comparison. All electrical and

reliability measurements were carried out by using the semiconductor parameter analyzer,

Keithley 4200. For the X-ray photoelectron spectroscopy measurement, X-ray diffraction,

scanning electron microscope, 50-nm a-IGZO thin films with different microwave annealing

conditions were deposited separately on n-type Si wafer. And measure transmittance with

Table 2-1 Experiment flow chart

2.2 Sputtering systems

The experimental methods of the fabrication of the a-IGZO TFT are described. Besides,

the principle of sputtering system including the RF sputtering and the DC sputtering is

described. In this chapter, we will introduce two sputtering systems. One is RF sputtering and

the other is DC sputtering.

RF sputtering can be applied to the deposition of both insulating conducting materials.

The substrate is located above the target so that the sputtered atoms can be deposited on to the

substrate. A RF power supply generates plasma at the frequency of 13.56 MHz. A RF

sputtering system is shown in Fig. 2-2. [39]

DC sputtering has the advantage of higher deposition rate and is less expensive than RF

sputtering. A DC sputtering is shown in Fig. 2-3 [40], the substrate is located above the target

and acts as the anode. DC sputtering is commonly applied to deposit conductive materials.

The plasma creates ions which are accelerated towards the target by a negative DC bias on the

target. The ions bombard the target surface and dislodge the target atoms, which then deposit

onto the substrate. The sputtering is performed in vacuum, typically between 1 mTorr and 50

mTorr. A lower chamber pressure increases the mean free path, which is the distance between

Fig. 2-2 Schematic RF sputtering system [39]

2.3 Basic Microwave Heating System

The microwave heating system is made up of four basic components: power supply,

magnetron, applicator for the heating if the target material and waveguide for transporting

microwaves from the generator to the applicator. Fig. 2-4(a) shows a simplistic diagram of the

microwave heating system.

The microwave system mainly contains: [38]

a. Microwave power supply: supplies high voltage power for the microwave source. Supply

includes internal alarms to prevent damage the microwave source, i.e. over volt alarm or

over temperature alarm.

b. Microwave source: generates the microwave energy required for processing. Internal

interlocks prevent overheating, i.e. water flow switch or over temperature sensor.

c. Isolator: eliminates excessive microwave energy form the process chamber to prevent

damage to the microwave source.

d. Coupler: a port to measure forward microwave energy going into the process chamber.

e. Waveguide: delivers generated microwave energy into the process chamber.

f. Process chamber: an octagonal prism and vessel designed to isolate wafers from the

atmosphere while gases and microwave as specified by the recipe are applied to the wafers. The chamber’s geometry promoted a uniform microwave energy field. As the Fig. 2-4 (b). [38]

When the process starts, the loading stage under the process chamber sealed. And then,

the stage rotates slowly for increase the uniformity of the microwave absorption. After ten

minutes N2 gas pre-purge, the microwave power supplies turns on. The susceptors above and

below the wafer can prevent particles from the environment during process. The addition of

filler wafers (bare silicon) above and below the process wafer can prevent plasma generation.

[39]

Fig. 2-4 (a) Microwave heating system.

Fig. 2-4(b) The setup in the microwave chamber. The distance between adjacent slots was 1 cm only. [38]

2.4 Parameter Extraction Method

The device electrical properties were measured by a Keithley 4200 IV analyzer in a

light-isolated probe station at room temperature. In IDS-VGS measurement, the typical

drain-to-source bias was swept from VGS=-20 V to VGS=30 V. In this session, we describe the

methods of typical parameters extraction such as threshold voltage (Vth), subthreshold swing

(SS) and field effect mobility (

μ

Threshold voltage (Vth) was defined from the gate to source voltage at which carrier

conduction happens in TFT channel. Vth is related to the gate insulator thickness and the flat

band voltage. Plenty of methods are available to determine Vth which is one of the most

important parameters of semiconductor devices. This thesis adopts the constant drain

current method, which is, the voltage at a specific drain current NID is taken as Vth, that is, Vth

= VG (NID) where Vth is threshold voltage and NID stands for normalized drain current.

Constant current method is adopted in most studies of TFTs. It provides a Vth close to that

obtained by the complex linear extrapolation method. Generally, the threshold current NID =

ID/(W/L) is specified at 1 nA in linear region and at 10 nA in saturation region; W and L

represent for TFT channel length and width, respectively. [41]

Subthreshold swing (SS, V / dec.) is a typical parameter to describe the control ability of

gate toward channel which is the speed of turning the device on and off. It is defined as the

magnitude. SS is related to the process, and is irrelevant to device dimensions. SS can be

lessened by substrate bias since it is affected by the total trap density including interfacial trap

density and bulk density. In this study, SS was defined as one-half of the gate voltage required

to decrease the threshold current by two orders of magnitude (from 10-8A to 10-10A). [41] The

threshold current was specified to be the drain current when the gate voltage is equal to Vth.

Typically,

μ

V). [41] The TFT transfer I-V characteristics can be expressed as

_] 2 1 ) [( C μ = ID FE ox VG Vth VD VD2 L W _{  (2-1)} Where,

COX is the gate oxide capacitance per unit area,

W is channel width,

L is channel length,

Vth is the threshold voltage.

If VD is much smaller than VG – Vth (i.e. VD << VG – Vth) and VG > Vth, the drain current

can be approximated as:

D th G V V V L W ) ( C μ = ID FE ox  (2-2)

The gm is defined as:

D V L W ox FE m =μ C g (2-3) Thus, m g WV C L μ_FE  (2-4)

2.5 Measurement of reliability on a-IGZO TFTs

In this session, we will introduce two ways of reliability’s measurement. One is gate DC bias stress condition and the other is light illumination’s measurement.

The DC gate bias stress condition was set to VG = +37.5 V for Positive Gate Bias stress

(PGBS) and VG = -37.5 V for Negative Gate Bias Stress (NGBS). Converted into electric field

(E) is 2.5 MV/cm, while source and drain electrodes are connected to ground from 0s to

2000s. The sample was stressed at room temperature (25℃).

The devices was place in the dark environment (in the black box) and under room

pressure and temperature. Table 2-2 shows the experiment flow of devices under light

illumination. The a-IGZO TFTs with different annealing conditions were measured under

different wavelength light which ranged from 900 nm (visible light) to 300 nm (UV light).

The light source was a halogen optic lamp from OSRAM Inc. at 150 W generating light

intensity about 63315 lx. We find the devices fabricated by the standard manufacturing

processes should be placed in the box for few minutes even few hours to get stable electrical

performances. And light_1_on means the specific wavelength illumination on device under

measuring.

Chapter 3 Results and Discussion

3.1 The effect of different MWA time on a-IGZO TFT

3.1.1 a-IGZO TFT with SiO

gate insulator

First, we discuss the intrinsic characteristics of SiO2 gate insulator. The drain-to-source

voltage (VDS) was fixed at 10 V. The gate-to-source voltage (VGS) was varied from -20 V to

30 V. We fix the MWA power to 1P (600~700W) and change the MWA annealing time for

100s / 300s / 600s. Fig. 3-1 shows the results of SiO2 gate insulator with different MWA time.

We can see the line of 100s is obviously different from the other two. Table 3-1 gives us more

specific results about different MWA time affect to the a-IGZO TFT with SiO2 gate insulator.

We can find that for annealing time 100s to 300s have great increases in mobility and decrease

the amounts of SS / Vth. For the condition of MWA time 600s, we can find that it has almost

the same mobility, Vth and SS value as the condition of MWA time 300s.

Table 3-1 The data of MWA different time for SiO2 gate insulator

Second, we will discuss the reliability of different MWA time with SiO2 gate insulator.

Fig.3-2 shows Vth variation of a-IGZO TFTs after being gate bias stressed with electrical field

of 2.5 MV/cm for 2000s in atmosphere. Vth shifted in the direction of positive voltages under

positive gate bias stress (PGBS). The shift amounts decrease with higher MWA time. On the

other hand, we can find that negative gate bias stress (NGBS) didn’t shift a lot. We find the

shift amounts of Vth don’t have obvious difference through NGBS.

Second(s) 0 500 1000 1500 2000 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 1P_100s 1P_300s 1P_600s

3.1.2 a-IGZO TFT with SiN

gate insulator

First, we discuss the intrinsic characteristics of SiNx gate insulator. The VDS was fixed at

10 V. The gate-to-source voltage (VGS) was varied from -20 V to 30 V. We fixed the MWA

power to 1P (600~700W) and changed the MWA annealing time for 100s / 300s / 600s. Fig.

3-3 shows the results of SiNx gate insulator with different MWA time. We can see the line of

1P_100s is obviously different from the other two. Table 3-2 gives us more specific results for

different MWA time the a-IGZO TFT with SiNx gate insulator. We can find that for annealing

time 100s to 600s have great increases in mobility and decrease the SS / Vth. For the condition

of MWA time 600s, we can find it has the best results than the other two.

Second, we will discuss the reliability of different MWA time with SiNx gate insulator.

Fig.3-4 shows Vth variation of a-IGZO TFTs after being gate bias stressed with electrical field

of 2.5 MV/cm for 2000s in atmosphere. Vth shifted in the direction of positive voltages under

positive gate bias stress (PGBS). The shift amounts decrease with higher MWA time. On the

other hand, we can find that negative gate bias stress (NGBS) didn’t shift a lot. We find the

shift amounts of Vth don’t have obvious difference through NGBS.

Second(s) 0 500 1000 1500 2000 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 1P_100s 1P_300s 1P_600s

3.2 The effect of different MWA Power on a-IGZO TFT

3.2.1 a-IGZO TFT with SiO

gate insulator

First, we discuss the intrinsic characteristics of SiO2 gate insulator. VDS was fixed at 10 V.

VGS was varied from -20 V to 30 V. We fixed the MWA time to 100s and changed the MWA

annealing time for 1P (600~700W) and 2P (1200~1400W). Fig. 3-5 shows the results of SiO2

gate insulator with different MWA time. Table 3-3 gives us more specific results for different

MWA power the a-IGZO TFT with SiO2 gate insulator. We can find that for annealing power

1P (600~700W) to 2P (1200~1400W) have great increases in mobility and decrease the SS /

Vth.

Second, we will discuss the reliability of different MWA time with SiO2 gate insulator.

Fig.3-6 shows Vth variation of a-IGZO TFTs after being gate bias stressed with electrical field

of 2.5 MV/cm for 2000s in atmosphere. Vth shifted in the direction of positive voltages under

positive gate bias stress (PGBS). The shift amounts decrease with higher MWA power. On the other hand, we can find that negative gate bias stress (NGBS) didn’t shift a lot. We find the

shift amounts of Vth don’t have obvious difference through NGBS.

Second(s) 0 500 1000 1500 2000 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 1P_100s 2P_100s

3.2.2 a-IGZO TFT with SiN

gate insulator

First, we discuss the intrinsic characteristics of SiNx gate insulator. VDS was fixed at 10 V.

VGS was varied from -20 V to 30 V. We fixed the MWA time to 100s and changed the MWA

annealing time for 1P (600~700W) and 2P (1200~1400W). Fig. 3-7 shows the results of SiNx

gate insulator with different MWA time. Table 3-3 gives us more specific results about

different MWA power the a-IGZO TFT with SiNx gate insulator. We can find that for

annealing power 1P (600~700W) and 2P (1200~1400W) has great increases in mobility and

decrease the SS / Vth.

Second, we will discuss the reliability of different MWA time with SiNx gate insulator.

Fig.3-8 shows Vth variation of a-IGZO TFTs after being gate bias stressed with electrical field

of 2.5 MV/cm for 2000s in atmosphere. Vth shifted in the direction of positive voltages under

positive gate bias stress (PGBS). The shift amounts decrease with higher MWA time. On the

other hand, we can find that negative gate bias stress (NGBS) didn’t shift a lot. We find the

shift amounts of Vth don’t have obvious difference through NGBS.

Second(s) 0 500 1000 1500 2000 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 1P_100s 2P_100s

3.3 The Comparison of MWA and Furnace Annealing on a-IGZO

TFT

3.3.1 a-IGZO TFT with SiO

gate insulator

First, we discuss the intrinsic characteristics of SiO2 gate insulator. VDS was fixed at 10 V.

VGS was varied from -20 V to 30 V. We chose the best two conditions: 1P_300s and 1P_600s,

comparing with furnace annealing. Fig. 3-9 shows the results of SiO2 gate insulator with

different MWA conditions and furnace annealing. Table 3-5 gives us more specific results of

different MWA conditions and furnace annealing affect to the a-IGZO TFT with SiO2 gate

insulator. We can find that for the condition of MWA time 300s and 600s have the similar

results, comparing with furnace annealing, we can find that MWA has great increases in

mobility and decrease the SS / Vth.

Fig. 3-9 Electrical characteristic comparison with MWA and furnace annealing for SiO2 gate

Table 3-5 The data of comparison with MWA and furnace annealing for SiO2 gate insulator

Second, we will discuss the reliability of different MWA time with SiO2 gate insulator.

Fig.3-10 shows Vth variation of a-IGZO TFTs after being gate bias stressed with electrical

field of 2.5 MV/cm for 2000s in atmosphere. Vth shifted in the direction of positive voltages

under positive gate bias stress (PGBS). The shift amounts decrease with higher MWA time

and almost the same as furnace annealing for the condition of MWA time 600s. On the other hand, we can find that negative gate bias stress (NGBS) didn’t shift a lot. We find the shift

amounts of Vth don’t have obvious difference through NGBS.

Second(s) 0 500 1000 1500 2000 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 1P_300s 1P_600s furnace

3.3.2 a-IGZO TFT with SiN

gate insulator

First, we discuss the intrinsic characteristics of SiNx gate insulator. VDS was fixed at 10 V.

VGS was varied from -20 V to 30 V. We chose the best two conditions: 1P_ 600s and 2P_100s,

comparing with furnace annealing. Fig. 3-11 shows the results of SiNx gate insulator with

different MWA conditions and furnace annealing. Table 3-6 gives us more specific results

about different MWA conditions and furnace annealing affect to the a-IGZO TFT with SiNx

gate insulator. We can find that for annealing time comparing with furnace annealing, MWA

condition for 1P_600s have great increases in mobility and decrease the SS / Vth.

Fig. 3-11 Electrical characteristic comparison with MWA and furnace annealing for SiNx gate

Table 3-6 The data of comparison with MWA and furnace annealing for SiNx gate insulator

Second, we will discuss the reliability of different MWA time with SiNx gate insulator.

Fig.3-12 shows Vth variation of a-IGZO TFTs after being gate bias stressed with electrical

field of 2.5 MV/cm for 2000s in atmosphere. Vth shifted in the direction of positive voltages

under positive gate bias stress (PGBS). The shift amounts decrease for MWA condition of

2P_100s and almost the same as furnace annealing. On the other hand, we can find that

negative gate bias stress (NGBS) didn’t shift a lot. We find the shift amounts of Vth don’t have

obvious difference through NGBS.

Second(s) 0 500 1000 1500 2000 -1 0 1 2 3 4 5 6 1P_600s 2P_100s furnace

Fig. 3-12 The PBIS and NBIS comparison of MWA and furnace annealing for SiNx gate

3.4 Material Analysis Results

In this chapter, we will introduce four different material analysis results. First is

UV-visible, second is X-ray diffraction, third is scanning electron microscope and the last is

X-ray photoelectron spectroscopy.

First, a-IGZO TFTs with different annealing treatment were measured under different

wavelength light which ranged from 900 nm (visible light) to 300 nm (UV light). The

decrease of the light transmittance in UV light region can be attributed to the decrease of

optical energy band-gap (Eg). The optical band-gap extract from different wavelength’s

transmittance of each condition was collected and compared in Table 3-7. It was clearly observed that the film didn’t change their band-gap a lot from different MWA time and power. Comparing with furnace annealing, it’s still not change their band-gap a lot. Which means that after post-annealing the transmittance didn’t have obvious difference.

Table 3-7 The data of optical band-gap for different MWA time / power and furnace annealing

Second, X-ray diffraction (XRD) is used to observe the active layer pattern shapes and

continue time is 1 second. Fig. 3-13 shows the results of a-IGZO TFTs with difference

annealing conditions for 50nm by XRD. We can find that the XRD results for different MWA time and power didn’t change films structure. Comparing with furnace annealing, MWA still remain amorphous phase. So we can find that different annealing treatment didn’t change the films structure.

Fig. 3-13 The X-ray diffraction results for different MWA time / power and furnace annealing

Third, scanning electron microscope (SEM) images indicate the formations of films

surface structure. Fig 3-14 shows the results of SEM for different annealing conditions. We can find that different MWA time and power didn’t change the surface of films. Comparing with furnace, it’s still have obvious difference between MWA. So we can find that different

.

Fig. 3-14 The scanning electron microscope results for different MWA time / power and furnace annealing

The last, I will show the X-ray photoelectron spectroscopy (XPS) results. The results of XPS analysis on O1s spectrum in each a-IGZO thin film was showed in Fig. 3-15. Fig. 3-15 shows the XPS analysis results of O1s spectrum for the a-IGZO thin film with microwave annealing treatments. Two components of O1s peaks could be fitted by Gaussian Lorentzian deconvolution, which centered at 530.6 and 531.4 eV, respectively. The lower binding energy centered at 530.6 eV, denoted as peak A, originated from the lattice oxygen ions with neighboring metal atoms. [42] The higher binding energy peak at 531.4 eV, denoted as peak B, corresponds to O2_ ion at an oxygen-deficient region in the matrix of the a-IGZO film. [43] According to the results of XPS, peak A increased and peak B decreased while the microwave annealing duration increased, as shown in Fig. 3-15. Also, the XPS spectrum of a-IGZO film with high-power microwave annealing was composed of a high intensity of peak A and low intensity of peak B. This figure revealed a high content of lattice oxygen ions in the microwave-annealed a-IGZO films with few oxygen-deficient regions as the microwave

assisted the oxygen ions binding with metal atoms and suppressed the formation of oxygen-deficient region in the a-IGZO films when increasing both the microwave power and annealing duration.

Fig. 3-15(b) shows the XPS O1s spectrum of a-IGZO thin film with 2P microwave

annealing for 100 s and furnace annealing at 450oC for 1 hour for comparison. The O1s peak

of a-IGZO film with microwave annealing composed of a higher intensity of peak A and a

lower intensity of peak B than those of the 450oC furnace-annealed one. These material

analysis results were consistent with the electrical improvement in device performance and reliability of a-IGZO TFT, since microwave annealing facilitated the formation of lattice oxygen and eliminated the defects originating from oxygen deficiency. Energy transfer to the a-IGZO TFT was even more effective by the microwave annealing process than that of the conventional furnace annealing.

Chapter 4 Conclusions

4.1 Conclusions

According to the statements of previous chapter, we can have the following conclusions. After 1P_300s MWA treatment, it was observed that the a-IGZO TFTs represented a lower threshold voltage (it changes from 3.53V to 3.13V), a lower sub-threshold swing (it changes

from 1.25 V/dec. to 0.62 V/dec.), and a higher mobility (it changes from 5.33 cm2/V-s to 5.81

cm2/V-s) for SiO2 gate insulator. After 1P_600s MWA treatment, it was observed that the

a-IGZO TFTs represented a lower threshold voltage (it changes from 1.12V to 1.02V), a lower sub-threshold swing (it changes from 0.41 V/dec. to 0.30 V/dec.) and a higher mobility

(it changes from 11.7 cm2/V-s to 13.4 cm2/V-s) for SiNx gate insulator. The stability also get

great improve by MWA. We can find that the shift amounts of Vth were decrease with

increasing MWA time and power for both two different G.I. under the PBIS.

Through material analysis of UV-visible, XRD and SEM, we can find that the optical band-gap, phase structure, grain size and the ratio of atom composition of a-IGZO thin films were not damaged by different kind of post-treatments. But when see the results of XPS, we can find that the oxygen defects were repaired and the oxygen bonding was become stronger by increasing MWA time and power.

We can find that MWA can have the performance as good as (even better than) furnace annealing for low thermal budget and short working time. This method has great potential for applying in high-resolution display products and flexible electronics.

In summary, this work has demonstrated the feasibility of high performance and reliable a-IGZO TFTs with microwave annealing process. Microwave annealing with low thermal budget can reduce the manufacturing process period and improve electrical characteristics of

a-IGZO TFTs, due to the effective absorption of microwave energy by the a-IGZO active layer. This selective heating also potentially avoided the damage to materials neighboring the a-IGZO channel layer in the TFT device structure during thermal processes. With optimum microwave annealing around 1200W for 100s in this work, electrical performance and reliability of a-IGZO TFT are more significantly promoted than with furnace annealing at

450oC for 1h. Results of this study significantly contribute to microwave annealing