國立成功大學

微電子工程研究所
碩士論文
(初稿)

以磁控濺鍍法沉積鋁鎵錫氧化物薄膜電晶體及其光電元
件之探究

Research of Aluminum Gallium Tin Oxide Thin Films and

their Optoelectrical Devices Fabricated by RF Sputtering

研究生 : 吳東翰 Student: Tung-Han Wu

指導教授 : 陳志方 Advisor: Jone-Fang Chen

Institute of Microelectronic
Department of Electrical Engineering
National Cheng Kung University
Tainan, Taiwan, R.O.C.
Thesis of Master of Science
July 2025

中華民國 一百一十四 年 七月
I
以磁控濺鍍法沉積鋁鎵錫氧化物薄膜電晶體及
其光電元件之探究

吳東翰* 陳志方**

國立成功大學微電子工程研究所

摘要
本論文討論了使用射頻磁控濺鍍法沉積氧化鋁鎵錫薄膜下，使用

不同製程參數下對薄膜特性的影響，並進一步在光電元件中使用這

些薄膜，包含使用在紫外光感測器以及在薄膜電晶體上的應用。

一開始，本研究使用磁控濺鍍沉積氧化鋁鎵錫薄膜，藉由改變二

氧化錫靶材的瓦數以及通入的氧氣量來改變製程結果，並進行光

學、結構和材料的分析。藉由光學分析得知，在可見光的範圍內氧

化鋁鎵錫薄膜具有高透明度，平均穿透率能超過 90%。而藉由原子

力顯微鏡分析，可以得到不同通氧量以及二氧化錫瓦數下製成的氧

化鋁鎵錫薄膜都有均勻的表面，且表面粗糙度軍方根值均小於 2 奈

米。材料分析則利用 X 射線進行，其結果顯示隨著氧通量的增加，

薄膜中的氧空缺也隨之減少。

在第二部分中，同樣利用磁控濺鍍法沉積氧化鋁鎵錫來製作光感
I
測器，並利用不同的通氧量以及不同二氧化錫的瓦數作為製程中的

調變參數。一開始將二氧化錫的瓦數由 20 瓦增加到 50 瓦，可以發

現不論光暗電流皆有增加。光電流由 2.38×10-7 (A)上升到 4.22×10-4

(A)，暗電流則由 1.98×10-7 (A)上升到 3.01×10-4 (A)。而改變通氧量

時則使用了 0% 到 6%，其中鋁鎵氧化物靶材固定 20W，二氧化錫

靶材固定 40W，一共四種不同的製程參數來進行比較。由結果可

知，隨著通氧量的增加，光感測器的光暗電流皆有下降，由光電流

由 4.32×10-5 (A)下降到 1.97×10-9 (A)，暗電流則由 1.44×10-5 (A)下降

到 1.69×10-9 (A)。同時，我們量測了時間內響應，發現不論開或關

都未能在設定的時間內回到照光前的初始狀態。

第三部分中，我們利用磁控濺鍍法來氧化鋁鎵錫薄膜電晶體。一

開始是使用改變二氧化錫的瓦數來減少缺陷以及提升自由載子數

量，並以此來改善薄膜電晶體的特性。接下來再利用改變通入的氧

氣比例來了解通氧量以及電晶體電性之間的關係，發現在通氧量 4%

下能使得開關電流大小有最佳的表現(2.83×103)。再利用改變薄膜電

晶體結構，由原本的一層通道改為兩層，以利提供更多載子來幫助

開啟晶體。結果來說，改善結構後的開關電流比提升為 1.13×104，

臨界電壓為 3.87(V)，而次臨界擺幅大小則為 0.3(V/decade)。數據證

實使用雙通道的結構確實能使薄膜電晶體的特性更進步。

II
 最後，利用 40W/20W 的薄膜電晶體進行照光量測，較大的響應

發生在波長 220 奈米到 260 奈米之間，得到的光拒斥比為

1.39×102。同時，我們量測了時間內響應，發現不論開或關都未能在

設定的時間內達成理想值，呈現和光感測器一樣的結果，兩者都無

法在短時間內正確地完成開關的動作。

關鍵字:氧化鋁鎵錫、紫外光感測器、薄膜電晶體、紫外光電晶體

作者*

指導教授**

III
Research of Aluminum Gallium Tin Oxide Thin Films

and their Optoelectrical Devices Fabricated by RF

Sputtering
Tung-Han Wu* Jone-Fang Chen**
Institute of Microelectronics

National Cheng Kung University, Tainan, Taiwan, R.O.C

Abstract

This research discusses the differences between various process

parameters of the characteristics of aluminum gallium tin oxide
(AlGaSnO) thin films deposited by RF magnetron sputtering. Also,
their applications in ultraviolet(UV) optoelectronic devices, such as UV
photodetectors and thin-film transistors(TFTs) are also mentioned.
In the first part, AlGaSnO thin films were deposited by magnetron
sputtering, with adjusting different co-sputtering power as the process
parameters. Optical, structural and material analysis were conducted.
By optical analysis, the result showed that AlGaSnO thin films
exhibited high transparency in the visible light range, which
transmittance is over 90%. Atomic Force Microscopes (AFM) analysis
revealed that AlGaSnO thin films are smooth. Although changing the
oxygen flow rate and SnO2 power, the root mean square (RMS) surface
roughness values are still below 2nm. Material analysis showed that the
number of oxygen vacancies in the AlGaSnO thin films decreased as
the increase of the oxygen flow rate. This analysis was finished by X-
ray photoelectron spectroscopy(XPS).
In the second part, AlGaSnO UV photodetectors were produced by
IV
magnetron sputtering by varying the oxygen flow rate from 0% to 6%.
According to the data, the current increases as we elevate the SnO2
power. For changing oxygen flow rate, the current decreases which is
because introducing the oxygen can help to eliminate the oxygen
vacancy. During the experiment, AGO power is 20W. To conclude, the
PD in 4% shows the best performance above all the parameters. The
photo and dark current are 4.37×10-6 and 2.46×10-6 respectively. The
rejection ratio is 3.35 and the EQE is 73.6%. At the end, time-resolved
response of AGTO MSM photodetectors is discussed and its result is
shown. Its rising and falling time is over 150 seconds. In other words,
all of the devices cannot react to the light quickly.
For the next part, the characteristics of AGTO thin-film transistors
are discussed. We use RF-sputtering with different parameters including
changing SnO2 power and oxygen flow rate in order to figure out the
relation between the performance of devices and different variables.
Among all the experiment, the AGO power is set as 20W. Increasing
SnO2 power can effectively upgrade the on and off current, and results
in the better performance. Subsequently, we change the oxygen flow
rate from 0% to 6% in order to understand the effect of it. Rising the
oxygen flow rate results in the declining trend of threshold voltage
because introducing the oxygen can get rid of oxygen vacancies. As a
result, it can efficiently improve the performance of transistors.
Combining the results from both of parameters, we find that the best
performance of AGTO thin-film transistor is 40W/20W O2 4%. The
performance of device’s On/Off ratio is 2.83×103, and the threshold
voltage is 4.52(V) Also, the subthreshold swing 0.539(V/dec).
Next, we change the single-layer thin-film transistor to double-layer
in order to improve the performance. Using the new structure, more
carriers will be provided by the 0% layer so that the performance of
V
 transistors will be enhanced. As a result, the On/Off ratio increases from
2.83×103 to 1.13×104. Also, the threshold voltage decreases from
4.52(V) to 3.87(V). After that, we mention about the performance of
visible-blind phototransistors, which is fabricated with 40W/20W and
the rejection ratio is 1.39×102. At the end, time-resolved response of
AGTO MSM photodetectors is discussed and its result is shown. Its
rising and falling time is over 150 seconds. In other words, all of the
devices cannot react to the light quickly.
Keyword: Aluminum Gallium Tin Oxide; UV photodetector; Thin-film
transistor; UV phototransistor

Author*

Advisor**

VI
 Contents
摘要…………………………………………………………………….....I
Abstract……………………………………………...…….................…IV
Content……………………………………………………………...…..VI
Table Caption…………………………………………………………....IX
Figure Captions………………………………………………………….IX
Chapter 1……………………………………………………………….....1
1.1 Background and Motivation………………………………………….1
1.2 Organization of This Thesis…………………………………………..2
Reference…………………………………………………………………3
Chapter 2………………………………………………………………….5
2.1 Theory of Photodetector………………………………………………5
2.1.1 Responsivity of the Photodetector………………………………….6
2.1.2 Photo to dark Current Ratio (PDCR)……………………………….6
2.1.3 Ultraviolet(UV)-to-visible Rejection Ratio…………………………7
2.2 Theory of Thin-Film Transistor……………………………………….7
2.2.1 Threshold Voltage(Vth)…………………………………………….10
2.2.2 Field-Effect Mobility(μ)………………………………………….10
2.2.3 On/off Current Ratio(Ion/Ioff)………………………………….……11
2.2.4 Subthreshold Swing(SS)…………………………………………..11
2.2.5 Interface Trap Density(Nit)…………………………………….…..12
2.3 Experimental Equipment…………………………………………….12
2.3.1 Radio-frequency Sputtering System………………………………12
2.3.2 Plasma-enhanced Chemical Vapor Deposition(PECVD)………….13
2.3.3 Electron Beam Evaporation……………………………………….14
2.3.4 Thermal Evaporation System…………………………………….15
2.3.5 X-ray Diffraction Analysis (XRD)……………………………….16
2.3.6 Atomic Force Microscopes(AFM)……………………………….17
VII
 2.3.7 X-ray Photoelectron Spectroscopy(XPS)…………………….…..20
2.3.8 Measurement Systems……………………………………………20
Reference………………………………………………………………23
Chapter 3………………………………………………………………...23
3.1 Growth of AlGaSnO Thin Film……………………………………...23
3.2 Optical Characteristics……………………………………………...24
3.3 X-ray Diffraction (XRD) Analysis………………………………….27
3.4 Atomic Force Microscopes (AFM) Analysis………………………..29
3.5 X-ray Photoelectron Spectroscopic (XPS) analysis…………………31
Reference………………………………………………………………..34
Chapter 4………………………………………………………………...35
4.1 Motivation...........................................................................................35
4.2 Fabrication of AGTO MSM Photodetectors…………………………36
4.3 Characteristics of AGTO MSM Photodetectors……………………..37
4.3.1 Characteristics of AGTO MSM Photodetectors with changing oxygen
flow rate…………………………………………………………………37
4.3.2 Characteristics of AGTO MSM Photodetectors with changing SnO2
power……………………………………………………………………42
4.3.3 Time-resolved Response of AGTO MSM Photodetector…………46
4.4 Summary…………………………………………………………….47
Reference………………………………………………………………..48
Chapter 5………………………………………………………………...51
5.1 Motivation…………………………………………………………...51
5.2 Fabrication of AGTO Thin-Film Transistors………………………...52
5.3 Characteristics of AGTO Thin-film Transistors……………………..53
5.3.1 Characteristics of AGTO Thin-film Transistors with different SnO2
power……………………………………………………………………53
5.3.2 Characteristics of AGTO Thin-film Transistors with different oxygen
flow rate…………………………………………………………………60
VIII
5.3.3 Double-channel thin-film transistor…………………………….....62
5.4 Elemental Analysis of AGTO Thin-film transistors………………...65
5.4.1 Transmission Electron Microscopy(TEM) Analysis………………66
5.4.2 Energy Dispersive Spectra (EDS) Analysis……………………….67
5.5 Characteristics of AGTO Thin-film Phototransistors……………….68
5.6 Time-Resolved Response of AGTO Thin-film Phototransistors……71
5.7 Summary…………………………………………………………….73
Reference………………………………………………………………..74
Chapter 6………………………………………………………………...79
6.1 Conclusion…………………………………………………………..79
6.2 Future Work…………………………………………………………81
Table Captions
Table 2-1. The dimension of the seven crystal systems…………..…....19
Table 3-1. Bandgap of AGTO thin films………………………………..27
Table 3-2. Average root mean square (RMS) surface roughness values
…………………………………………………………………………29
Table 3-3. The proportion of OII with changing oxygen flow rate….…32
Table 4-1. Characteristics of AGTO MSM PDs with different SnO2
power…………………………………………………………………..47
Table 4-2. Characteristics of AGTO MSM PDs with different oxygen flow
rate………………………………………………………………...…...41
Table 4-3. The average response time of AGTO MSM PDs…………..48
Table 5-1. Characteristics of different SnO2 power AGTO TFT………56
Table 5-2. Characteristics of different oxygen flow rate AGTO TFT….63
Table 5-3. Comparison between Single-layer TFTs and Double-layer
TFTs……………………………………………………………………66
Table 5-4. Rising and falling time of 40W/20W AGTO phototransistor.
…………………………………………………………………………72

IX
 Figure Captions
Fig. 2-1. Three types of transitions under ultraviolet(UV) illumination
………………………………………………………………………..…6
Fig. 2-2. The structure of thin-film transistor……………………………8
Fig. 2-3. Four types of different TFT structures………………………..10
Fig. 2-4. Schematic diagram of RF sputtering system…………………14
Fig 2-5. Schematic diagram of PECVD ………………………………..15
Fig. 2-6 Schematic of X-ray Bragg’s diffraction……………………….18
Fig. 2-7 Seven types of crystal system………………………………...18
Fig. 3-1. Schematic diagram of the AGTO thin film…………………..24
Fig. 3-2. The transmittance spectrum of AGTO thin films with different
SnO2 co-sputtering power……………………………………………..25
Fig. 3-3. The transmittance spectrum of AGTO thin films with different
oxygen flow rate……………………………………………………….26
Fig. 3-4. Tauc plot of AGTO thin films with different SnO2 co-sputtering
power…………………………………………………………………..26
Fig. 3-5. Tauc plot of AGTO thin films with different oxygen flow rate
…………………………………………………………………………27
Fig. 3-6. XRD spectrum of AGTO thin films with different SnO2 power
………………………………………………………..………………..28
Fig. 3-7. XRD spectrum of AGTO thin films with different oxygen flow
rate……………………………………………………………………..29
Fig. 3-8. O1s XPS spectra of 0% 40w/20w AGTO………………….…30
Fig. 3-9. O1s XPS spectra of 4% 40w/20w AGTO………………….…33
Fig. 3-10. O1s XPS spectra of 6% 40w/20w AGTO…………….……..33
Fig. 4-1. (a) Metal mask of interdigitated electrodes and AGTO thin film
on the substrate (b) Schematic of AGTO photodetector……………….37
Fig. 4-2. (a) I-V curve of 20W/20W O2 4% PD (a) I-V curve of 30W/20W

X
O2 4% PD (a) I-V curve of 40W/20W O2 4% PD (a) I-V curve of
50W/20W O2 4% PD……………………………………………….….38
Fig. 4-3. (a) I-V curve of 40W/20W O2 0% PD (b) I-V curve of 40W/20W
O2 4% PD (c) I-V curve of 40W/20W O26% PD………………….....….40
Fig. 4-4. (a) I-V curve of 40W/20W O2 0% PD (b) I-V curve of 40W/20W
O2 4% PD (c) I-V curve of 40W/20W O2 6% PD…………………….....42
Fig. 4-5. (a) Responsivity of 40W/20W O2 0% PD (b) Responsivity of
40W/20W O2 6% PD…………………………………………………...45
Fig. 4-6. Time-resolved response of 40W/20W photodetector…………47
Fig. 5-1. Schematic of AGTO thin-film transistor……………………..53
Fig. 5-2. ID-VG characteristics of 20W/20W 4% AGTO TFT…………..56
Fig. 5-3. ID-VG characteristics of 30W/20W 4% AGTO TFT…………..56
Fig. 5-4. ID-VG characteristics of 40W/20W 4% AGTO TFT………….56
Fig. 5-5. ID-VG characteristics of 50W/20W 4% AGTO TFT.……...….57
Fig. 5-6. ID-VD characteristics of 20W/20W 4% AGTO TFT………….57
Fig. 5-7. ID-VD characteristics of 30W/20W 4% AGTO TFT.…………58
Fig. 5-8. ID-VD characteristics of 40W/20W 4% AGTO TFT.…………58
Fig. 5-9. ID-VD characteristics of 50W/20W 4% AGTO TFT.…………59
Fig. 5-10. ID-VG characteristics of 40W/20W 0% AGTO TFT.………..60
Fig. 5-11. ID-VG characteristics of 40W/20W 6% AGTO TFT.………..61
Fig. 5-12. ID-VD characteristics of 40W/20W 0% AGTO TFT.………..61
Fig. 5-13. ID-VD characteristics of 40W/20W 6% AGTO TFT.………..62
Fig. 5.14 Schematic of AGTO double-layer thin-film transistor……....64
Fig. 5-15. (a) ID-VG characteristics of AGTO Double-Channel thin-film
transistor (b) ID-VD characteristics of AGTO Double-Channel thin-film
transistor…………………………………………………….………....64
Fig. 5-16. The TEM cross section image of 40W/20W AGTO TFT.…..66
Fig. 5-17. Cross section image of 40W/20W AGTO TFT………….….67
Fig. 5-18. Elements in 40W/20W AGTO TFT…………………………67
XI
Fig. 5-19 Cross section image of 40W/20W AGTO TFT……………...68
Fig. 5-20. EDS spectrum of AGTO 40W/20W AGTO TFT…………...68
Fig. 5-21. ID-VG Characteristics of 40W/20W AGTO Phototransistor...70
Fig.5-22. Photo responsivity of 40W/20W AGTO phototransistor at
different applied gate voltage………………………………………….71
Fig.5-23. Energy-band diagram of AGTO phototransistor under
illumination as well as different applied gate voltage: (a) thermal
equilibrium (VG=0V), (b) VG<Vth, and (c) VG ˃Vth……………………72
Fig. 5-25. Time-resolved response of 40W/20W AGTO phototransistor
……………….………………………………………………………...72

XII
 Chapter 1

Introduction

1.1 Background and Motivation

Transparent oxide semiconductors(TOSs) have caused widespread

discussion due to the diverse applications in many fields like light-
emitting diodes(LEDs), active-matrix liquid crystal displays(AMLCDs),
and resistive random-access memory(RRAM), sensors, and so on [1].
There are many candidates for TOS materials, such as indium oxide [2],
gallium oxide [3], tin oxide [4], zinc oxide [5]. Due to their high
conductivity, large bandgap and good optical transparency in the visible
wavelength region, these materials are appropriate for TFTs.
As a transparent semiconductor material with ultra wide-bandgap
semiconductor, Ga2O3 has been used in high-power electronics and
optical devices. Also, the big bandgap allowed Ga2O3 to operate as
photodetectors within the UV energy range, making them prime
candidates for future solar-blind photodetectors [6].
There are many studies focusing on III-oxide alloys, such as Al2O3 and
Ga2O3. On the other hand, GTO was used because of high carrier
concentration in the past. However, it is also a big problem for depletion
[7]. Then, some papers were studying the IGTO as TFTs. In this way, the
mechanical stability and electrical properties could be improved a lot [8].
Unfortunately, there were many reports talking about the instability of this
material [9]. Therefore, in order to improve the stability and decrease the
leakage current, we doped Al for replacing In in IGTO TFT. What’s more,
due to the low cost of aluminum oxide, we can reduce the expense on
producing at the same time.

1
1.2 Organization of This Thesis
This thesis presents a research of AGTO thin-films and its
applications on optoelectronic devices, such as UV photodetectors,
thin-film transistors and UV phototransistors. The thesis is separated
into several parts:
Chapter 2 talks about the overview of related theories and the
instruction of the experimental equipment used to measure the devices.
Chapter 3 puts all the fabrication process of AGTO thin films in detail.
In this part, the thin films are deposited in different process, such as
changing oxygen flow rate and co-sputtering power. Besides, structural,
optical and electrical characteristics results are shown and discussed in
this part.
Chapter 4 is focusing on the AGTO UV PDs. In this part, the PDs in
different SnO2 power and oxygen flow rate will be produced. And its
responsivity, UV-to-visible rejection ratio, photo/dark current ratio and
the time response of UV PDs are displayed in details. Also, its time-
resolved response is discussed.
Chapter 5 aims at AGTO TFTs. In this chapter, AGTO TFTs
fabricated with silicon dioxide(SiO2) as gate insulator are introduced in
and out. For the first part, the results adjusting oxygen flow rate will be
shown, and the one changing co-sputtering power are mentioned behind.
Also. the characteristics of these TFTs, such as threshold voltage, field-
effect mobility, on/off drain current ratio, subthreshold swing and
interface trap density are analyzed deeply. Subsequently, the double-
channel thin-film transistor is introduced, too. At the end, the AGTO
phototransistor’s characteristics is exhibited, and the TEM analysis
results is demonstrated.
Chapter 6 summarizes the whole thesis’s finding and discussion
related to AGTO thin films. Last but not the least, the future work about
2
this thesis are proposed in this part.

Reference

[1] Hosono, Hideo. “Recent Progress in Transparent Oxide

Semiconductors: Materials and Device Application.” Thin Solid
Films, vol. 515, no. 15, May 2007, pp. 6000–6014,
https://doi.org/10.1016/j.tsf.2006.12.125. Accessed 25 May 2022.
[2] F. Gherendi, M. Nistor and N. B. Mandache, "In2O3 Thin Film
Paper Transistors," in Journal of Display Technology, vol. 9, no.
9, pp. 760-763, Sept. 2013, doi: 10.1109/JDT.2013.2278036.
[3] Liang, Huili, et al. “Recent Progress of Deep Ultraviolet
Photodetectors Using Amorphous Gallium Oxide Thin
Films.” Physica Status Solidi (A), vol. 218, no. 1, 14 Oct. 2020, p.
2000339, https://doi.org/10.1002/pssa.202000339. Accessed 29
Oct. 2022.
[4] Bae, Kang-Hwan, et al. “Effects of Hydrogen Plasma Treatment
on the Physical and Chemical Properties of Tin Oxide Thin Films
for Ambipolar Thin-Film Transistor Applications.” Ceramics
International, vol. 48, no. 17, Sept. 2022, pp. 24540–24549,
www.sciencedirect.com/science/article/pii/S0272884222016546?v
ia%3Dihub, https://doi.org/10.1016/j.ceramint.2022.05.097.
Accessed 25 Mar. 2025.
[5] Yang, Jianwen, et al. “Amorphous Nickel Incorporated Tin Oxide
Thin Film Transistors.” Journal of Physics D Applied Physics,
vol. 50, no. 35, 28 June 2017, pp. 355103–355103,
https://doi.org/10.1088/1361-6463/aa7c53. Accessed 25 Mar.
2025.

3
[6] Chen, Xuanhu, et al. “Gallium Oxide-Based Solar-Blind
Ultraviolet Photodetectors.” Semiconductor Science and
Technology, vol. 35, no. 2, 17 Jan. 2020, pp. 023001–023001,
https://doi.org/10.1088/1361-6641/ab6102. Accessed 8 Oct. 2023.
[7] Hu, Gengtao, et al. “Solution Processed Amorphous Gallium-
Incorporated Tin Oxide Thin-Film Transistors.” Japanese Journal
of Applied Physics, vol. 59, no. 5, 13 Apr. 2020, pp. 050906–
050906, https://doi.org/10.35848/1347-4065/ab88c0. Accessed 25
Mar. 2025.
[8] Kang, Jiyeon, et al. “Migration of Indium Ions in Amorphous
Indium–Gallium–Zinc-Oxide Thin Film Transistors.” Applied
Surface Science, vol. 258, no. 8, 1 Dec. 2011, pp. 3509–3512,
www.sciencedirect.com/science/article/pii/S0169433211018630?v
ia%3Dihub, https://doi.org/10.1016/j.apsusc.2011.11.104.
[9] Seo, Seok-Jun, et al. “Improved Negative Bias Illumination
Instability of Sol-Gel Gallium Zinc Tin Oxide Thin Film
Transistors.” Applied Physics Letters, vol. 99, no. 15, 10 Oct.
2011,
pubs.aip.org/aip/apl/article/99/15/152102/341119/Improved-
negative-bias-illumination-instability-of,
https://doi.org/10.1063/1.3646388. Accessed 25 Mar. 2025.

4
 Chapter 2
Relevant Theory and Experimental Equipment
2.1 Theory of Photodetector
Obviously, the mechanism of generating electron-hole pairs from
photon absorption is very important to photodetectors(PDs). When the
photon’s energy is larger than the bandgap of semiconductor material, an
electron can be excited from the valence band to the conduction band. As
a result, the ideal photodetector will be the one which can transfer the
incident radiation into electrical energy in 100 percent. Not to mention,
this can be an increase in current when the device is illuminated [1][2].
There are several types of transitions that take place in the
photodetectors:
(a) Intrinsic band-to-band transition: In the same semiconductor material,
electrons move from the valence band to the conduction band.
(b)Valence band to acceptor level transition: There is an electron
transferring from the valence band to an acceptor level, and the whole
process stays in the semiconductor material. Not to mention, there is an
electron-hole pair generated.
(c) Donor level to conduction band transition: This type of transition is
that an electron moves from a donor level to the conduction band. Same
with the previous one, there is also an electron-hole pair created.
To conclude, the significant thing is that the cut-off wavelength of a
semiconductor can change because of many factors, such as difference in
bandgap or other situations. The variation is considered in compound
semiconductors composed by different material not sharing the same
bandgaps.

5
 Fig. 2-1. Three types of transitions under ultraviolet(UV) illumination

2.1.1 Responsivity
This parameter is important for analyzing photodetectors works well
or not. Normally, the responsivity is used for realizing the
photodetector’s ability to convert incident light to electrical signal. In
other words, we can simplify that the responsivity qualifies the
detector’s performance, including sensitivity and efficiency in detecting.
The formula is expressed by:
𝐼𝑝ℎ𝑜𝑡𝑜 −𝐼𝑑𝑎𝑟𝑘
𝑅= (2 − 1)
𝑃𝑜𝑝𝑡

In this formula, Iphoto is the current under illumination, Idark is the current
in the dark environment, and Popt is the power of incident power.
Obviously, the relation between optical bandgap and responsivity is
close in high degree because the bandgap will influence the absorption
edge directly.
2.1.2 Photo to dark Current Ratio (PDCR)
The photo-to-dark Current Ratio (PDCR) represents the sensitivity of
6
photodetector to light. It is defined by the ratio of the current incident light
to the current without the light, and the formula is expressed by:
𝐼𝑝ℎ𝑜𝑡𝑜
𝑃𝐷𝐶𝑅 = (2 − 2)
𝐼𝑑𝑎𝑟𝑘
2.1.3 Ultraviolet(UV)-to-visible Rejection Ratio
This parameter can represent the ability to distinct the unwanted and
background light signals from the desired one. By detecting the desired
light signal, the rejection ratio can also be used for distinguishing the
specific light from the incident one. It can say that the ratio describes the
efficiency in:
𝑅(𝑈𝑉 𝑟𝑎𝑛𝑔𝑒)
RR = (2 − 3)
𝑅(𝑣𝑖𝑠𝑖𝑏𝑙𝑒 𝑟𝑎𝑛𝑔𝑒)

In addition, this parameter can understand whether the specific device is

suitable for UV detector or UV-to-visible. As definition, UV is a light
range from 200(nm) to 400 (nm). It can be separated onto three parts:
UVA[320(nm) – 400(nm)], UVB[280(nm) -320(nm)], and UVC [200(nm)
– 280(nm)]. According to the formula, the ratio can be large if the material
is quite appropriate for photodetectors.
2.2 Theory of Thin-Film Transistor
Normally, a thin-film transistor (TFT) is a device composed of several
parts: source, channel layer, dielectric layer, drain and gate. The structure
of TFT is shown in Fig. 2-2.
The physical mechanism of TFT is similar to metal oxide
semiconductor field-effect transistor(MOSFET). The difference between
both of them majorly is about the way channel is established. For
MOSFETs, when the suitable voltage is applied to the devices, the path is
generated by inversion. On the other hand, for TFTs, the path is formed
by accumulation. Because the channel path in TFTs comes from
7
accumulation, it is the majority carriers in channel layer that make a TFT
n-type or p-type. Separately, n-type TFT’s majority carriers are electrons
and p-type TFT’s are holes [3].
There are three regions operating the transistors. At the beginning, the
drain voltage(VD) isn’t applied, which causes no current flow appearing.
In this region, the device is cutoff no matter how the gate voltage(VG)
adjusts. Subsequently, when the drain voltage gradually increases but still
smaller than the saturation voltage(Vsat), the transistor is defined as in the
linear region. Continuously, after the drain voltage is larger than the
saturation voltage, the electronic component comes into the saturation
region. To conclude, we can describe the current in linear and saturation
region in the formulas below:
𝑊 1
𝐼𝐷,𝑙𝑖𝑛 = 𝐶𝑜𝑥 𝜇 [(𝑉𝐺 − 𝑉𝑡ℎ )𝑉𝐷 − 𝑉𝐷2 ] ; 𝑉𝐷 ≤ 𝑉𝑠𝑎𝑡 (2 − 4)
𝐿 2

1𝑊
𝐼𝐷,𝑠𝑎𝑡𝑢𝑟𝑎𝑡𝑖𝑜𝑛 = 𝜇𝐶𝑜𝑥 (𝑉𝐺 − 𝑉𝑡ℎ )2 ; 𝑉𝐷 ≥ 𝑉𝑠𝑎𝑡 (2 − 5)
2𝐿
In the formulas, W and L is the width and length of channels respectively,
μ is the field-effect mobility, Cox is the capacitance of insulating layer and
Vth is the threshold voltage.

Fig. 2-2. The structure of thin-film transistor

8
 Normally, four types of TFT structures are widely used, which is
staggered bottom-gate, staggered top-gate, coplanar bottom-gate and
coplanar top-gate are illustrated in Fig. 2-3. With the top-gate structure,
the better ability to control gate and lower parasitic capacitance are two
major merits [4]. However, the dielectric layer requires higher process
temperature, which channel layer has lower tolerance. As a result, we
choose bottom gate structure instead of top-gate structure in order to
avoid the temperature problems.

Fig. 2-3. Four types of different TFT structures

2.2.1 Threshold voltage(Vth)

Threshold voltage means the minimum gate voltage to activate the
transistor and enable the enough current to flow between source and drain
terminals. In other words, we can say that the threshold voltage is the
specific value of gate voltage that makes transistors start to form current
when it comes to n-type TFT. If the gate voltage is under the threshold
voltage, the transistor is inactive, which means there is just a small current
9
between both terminals. Therefore, it can be briefly concluded that this
parameter is critical for evaluating the performance of TFTs. And there
are several ways to get the value, such as transfer characteristic curve,
constant current method and so on. In this thesis, the extrapolation method
is chosen. ID1/2 versus VG figure is plotted and the intersection is the
threshold voltage where ID1/2 equals to 0 [5].

2.2.2 Field-Effect Mobility (μ)

In the semiconductor material, a force will show up after the electric
field applied, and this will trigger the acceleration of carriers. The net
acceleration leads to the drift current. For the strength, the average drift
velocity is proportional to electric field and can be calculated in the
equation below:
𝑉𝑑 = 𝜇𝐸 (2 − 6)
μ is the field-effect mobility and E is the electric field.
Hence, mobility can be regarded as a material’s ability to react with the
electric field. In this study, the field-effect mobility is considered in the
saturation region, and come from transconductance:
𝜕𝐼𝐷 𝑊
𝑔𝑚 = = 𝜇𝐶𝑜𝑥 (𝑉𝐺 − 𝑉𝑡ℎ ) (2 − 7)
𝜕𝑉𝐺 𝐿
𝑔𝑚 𝐿
𝜇𝑒𝑓𝑓 = (2 − 8)
𝑊𝐶𝑜𝑥 (𝑉𝐺 − 𝑉𝑡ℎ )
gm is the transconductance, W and L is the width and length of TFT
respectively, Cox is the capacitance of dielectric layer and VG is the gate
voltage.
2.2.3 On/off Current Ratio (Ion/Ioff)
This ratio can demonstrate the difference between on and off states. It
also demonstrates the switching efficiency and performance of TFTs. The
ratio is defined as the biggest on-state drain current divided by the

10
smallest off-state drain current. With the better on/off current ratio, the
device’s performance is better to transit between conducting and non-
conducting state.
2.2.4 Subthreshold Swing(SS)
Usually, the parameter is used to analyze the switching efficiency of
transistors, especially for the one in the subthreshold region. If the
parameter is low, the switching efficiency is much better. In other words,
the better performance of TFTs is recognized. Several factors play in a
role, including trap density, gate oxide thickness, mobility of carriers and
so on. The subthreshold swing is defined below:
𝜕𝑉𝐺
SS = (2 − 9)
𝜕log(𝐼𝐷 )
where VG is gate voltage and ID is drain current.
2.2.5 Interface Trap Density(Nit)
Interface traps are electronics states that exist in the interface of oxide
layer and semiconductor. Namely, they are the defects which will
influence the performance of devices. It is related to some problems like
increasing leakage currents, variation in threshold voltage and so on [6].

2.3 Experimental Equipment

In this part, all of the experimental equipment, which is used for
fabricating AGTO thin-film transistors, will be introduced thoroughly.
2.3.1 Radio-frequency Sputtering System
This process is one kind of physical vapor deposition(PVD) category,
and is used in the AGTO thin films. At first, many high-energy particles
are generated by glow discharges, and bump the surface of target
materials[7]. During the bumping process, the target material absorbs the
energy from high-energy particles, and squeezing atoms by momentum

11
conversion, which is illustrated in Fig. 2-3. As a result, a great deal of
atoms appear due to the bombardment and are deposited on the substrate
in the sputter chamber. Finally, these atoms accumulate and form AGTO
thin film.

Fig. 2-3. Momentum exchange process

During the whole process, the pressure in the chamber is set at 5х10-6
(Torr) to eliminate the potential problems from the environment. In order
to trigger the plasma, inert gas is necessary. Here, we choose Argon(Ar)
to be introduced into the chamber to achieve the goal because of higher
mass and low inactivity [8].
Fig. 2-4 describes the setup of RF sputtering system which is used in
this thesis. There is a vacuum diffusion pump that can help maintain the
process pressure at constant of 5х10-6(Torr). Also, several gases,
including Argon(Ar), Oxygen(O2) and Nitrogen(N2) are introduced into
the system. Besides, maximum power capacity of the sputtering system
is 500(W).

12
 Fig. 2-4. Schematic diagram of RF sputtering system
Normally, four factors will affect the sputtering yield(S), and they are
(1)Energy of incident ions (2)Target materials (3) Angle of incident ions
(4)Crystal structure of target‘s surface. The sputtering yield is defined as
the ratio of number of sputtered atoms divided by number of bumped ions.
If the ratio is bigger, it can be concluded that there is a better performance
for the system. Except for semiconductor, metal, alloy, and insulator are
appropriate for deposited by sputtering system. Furthermore, with
simultaneous deposition of different targets, much more semiconductor
can be produced by co-sputtering [9].
2.3.2 Plasma-enhanced Chemical Vapor Deposition(PECVD)
Plasma-enhanced Chemical Vapor Deposition(PECVD) is usually
chosen for depositing various thin films. To catalyze chemical reactions,
the plasma is indispensable. Before the process starts, the chamber is
forced to be clean to eliminate the impurities. After that, precursors gases
are introduced into the chamber. The precursor gas is composed of desired
material, and it can be single gas or mixture of many different gases. Then,
the radio frequency(RF) or microwave power source is used to activate
13
plasma. These two sources can ionize the precursor gas molecules and
form the plasma consisting of charged particles. Finally, the radicals are
freed. These high-energy particles provide enough energy for accelerating
reactions. And this makes the precursors to react with others and deposit
in the substrate to become thin films. Nowadays, PECVD is often selected
in manufacturing a big area of thin films [10]. There are several reasons
for choosing PECVD, including lower process temperature, high quality
and efficiency of producing films. Therefore, SiO2 is deposited by
PECVD in this research.

Fig 2-5. Schematic diagram of PECVD

2.3.3 Electron Beam Evaporation
Briefly speaking, this kind of method can be sorted as a classical type
of physical vapor deposition(PVD), which is widely used to deposit thin
films. During the process, the target material will be bumped with electron
beams generated from the tungsten filament, which can be seen as
magnets by applied large voltage. As a result, the filament can also be

14
regarded as the anode. Then, the kinetic energy of electron beams will be
transferred to the thermal energy so that the target material is melted easily.
Subsequently, the target material is vaporized and forms the thin film on
the substrate.
Several advantages of electron beam evaporation are discussed. The
biggest one is that only the source is heated so that we can minimize the
contamination to the surroundings. We can easily control the depositing
rate by manifesting the reasonable power for upgrading the voltage for
filaments. Hence, we can get the higher heating transferring efficiency
and better control ability to deposition rate by this method. What’s more,
most of the materials can be deposited by this method, and we don’t have
to increase the temperature of substrates.
Another advantage is that this method can deposit higher melting point
materials. During the process, when the high voltage is applied, the
tungsten coil will emit electrons, which are guided by electric and
magnetic fields. This will help to bombard the right material in the
crucible. Because of higher energy in the beams, this method can melt
material easily, and the range of evaporation can be greatly controlled.
Finally, the higher impurity thin films can be deposited.
2.3.4 Thermal Evaporation System
By this way, we can deposit the thin films on the substrate. The whole
process takes place in the vacuum chamber, with aluminum pellet is
heated to high temperature so that it can transit from solid to vapor state.
Then, these vaporized material will deposit on the substrate and form the
thin film. During the deposition procedure, the vapor’s source is close to
the target in order to condense the material on the right place. Namely, the
temperature of substrate is carefully controlled to ensure the great quality
of thin film.

15
2.3.5 X-ray Diffraction Analysis (XRD)
To learn the crystal phase or structure of a material, the X-ray
diffractometer is widely chosen. The main principle of this method is
working with the radiation. X-ray radiations are scattering in specific
directions after reacted with materials. This phenomenon comes from
constructive and destructive interference. A scattered waves may have
destructive and constructive at the same time. When atoms of crystal are
set in the periodic way at microscopic level, this situation takes place
easily. In the crystal, each of the atoms are arranged in same interval.
There is an equation, so-called Bragg’s Law is used to describe the
phenomenon.:
2d sin 𝜃 = 𝑛𝜆 (2 − 10)
Where d is the distance between two parallel planes, θis the angle
between the incident ray and the surface of the crystal, n is natural
numbers and λ is the wavelength of rays. Briefly speaking, the
constructive interference happens when the distance is as large as the
positive integer multiplies the wavelength of incident X-ray beam.
Besides, we can verify which kinds of crystal structure it is. There are
many kinds of structures in the Figure 2-7.

16
Fig. 2-6 Schematic of X-ray Bragg’s diffraction

Fig. 2-7 Seven types of crystal system

17
 Table 2-1. The dimension of the seven crystal systems
Crystal system Cell lengths Cell angles
Cubic a=b=c α = β = γ = 90°
Rhombohedral a=b=c α = β = γ ≠ 90°
Monoclinic a≠b≠c α = γ = 90°, β ≠ 90°
Triclinic a≠b≠c α ≠ β ≠ γ ≠ 90°
Hexagonal a=b≠c α = β = 90°, γ = 120°
Tetragonal a=b≠c α = β = γ = 90°
Orthorhombic a≠b ≠ c α = β = γ = 90°

2.3.6 Atomic Force Microscopes(AFM)

This kind of analysis is widely used in various surface analysis, and
can display the three-dimensional images clearly. There are three types
of modes, such as contact mode, non-contact mode and semi-contact
mode. It is conducted by measuring forces between the probe and the
sample. In this study, contact mode is chosen to get the result.
2.3.7 X-ray Photoelectron Spectroscopy(XPS)
X-ray Photoelectron Spectroscopy(XPS) is usually used to know the
material’s properties including elemental composition and so on. With
X-ray, the surface of sample will absorb it and emit secondary electrons.
By measuring the energy of photoelectrons, the elements composed in
thin films is found out because each of element has unique binding
energy.
2.3.8 Measurement Systems
In this study, electrical characteristics are measured by
semiconductor parameter analyzer (model B1500A, Agilent
Technologies). During all the measuring process, the equipment is kept
18
under standard room temperature and atmosphere pressure condition.
In other study, this analyzer is widely utilized in measuring various
kinds of passive and active devices.
To know the characteristics of phototransistors and photodetectors,
the light source is needed. In this research, we chose a 130W Xenon
lamp as light source which can cover the wavelength from 220nm to
500nm. The power of light source is tested by Ophir power and energy
meter.
2.4 Summary
In this chapter, many equipment we will use in the latter chapters are
introduced in detail, including plasma-enhanced chemical vapor
deposition(PECVD), thermal evaporation system, electron beam
evaporation and son on. Also, several parameters used to evaluate the
performance of thin-film transistors are mentioned and its formulas are
shown. For example, subthreshold swing(SS), field-effect mobility and
threshold voltage are introduced. At the end, three main analysis
methods are shown and B1500A which is our main measurement
system is briefly mentioned. For the analysis, they are atomic force
microscopes(AFM), X-ray photoelectron spectroscopy(XPS) and X-ray
diffraction analysis (XRD). Each of them can tell us different situation
in the thin films. For example, we can know the roughness of thin films
by AFM analysis and the proportion of oxygen vacancy can be told by
XPS analysis.

Reference

[1] Kirk S, Giboney. “IEEE Xplore Search Results.” Ieee.org, 2025,

ieeexplore.ieee.org/search/searchresult.jsp?searchWithin=%22Pub
lication%20Number%22:22&searchWithin=%22Volume%22:45
19
 &searchWithin=%22Issue%22:8&searchWithin=%22Start%20Pa
ge%22:1310. Accessed 26 Mar. 2025.
[2] Selamneni, Venkatarao, and Parikshit Sahatiya. “Mixed
Dimensional Transition Metal Dichalcogenides (TMDs) VdW
Heterostructure Based Photodetectors: A Review.” Microelectronic
Engineering, vol. 269, 17 Dec. 2022,p.111926,
www.sciencedirect.com/science/article/abs/pii/S01679317220022
09, https://doi.org/10.1016/j.mee.2022.111926.
[3] Ohta, Hiromichi, and Hideo Hosono. “Transparent Oxide
Optoelectronics.” Materials Today, vol. 7, no. 6, June 2004, pp. 42–
51, https://doi.org/10.1016/s1369-7021(04)00288-3. Accessed 12
June 2022.
[4] Lin, Wen-Kai, et al. “The Influence of Fabrication Process on Top-
Gate Thin-Film Transistors.” Thin Solid Films, vol. 519, no. 15, 26
Jan.2011,pp.5126–5130,
www.sciencedirect.com/science/article/pii/S0040609011002094?v
ia%3Dihub, https://doi.org/10.1016/j.tsf.2011.01.159.
[5] Stankus, Christopher, and Moinuddin Ahmed. “Comparing
Smoothing Techniques for Extracting MOSFET Threshold
Voltage.” Solid-State Electronics, vol. 164, Feb. 2020, p. 107744,
https://doi.org/10.1016/j.sse.2019.107744. Accessed 2 Jan. 2022.
[6] Kimura, Mutsumi, et al. “Extraction of Trap Densities at Front and
Back Interfaces in Thin-Film Transistors.” Japanese Journal of
Applied Physics, vol. 43, no. 1, 13 Jan. 2004, pp. 71–76,
https://doi.org/10.1143/jjap.43.71. Accessed 10 Apr. 2025.
[7] Matsuoka, Morito, and Shun-ichi Tohno. “Radio Frequency
Electric-Mirror Sputtering Deposition.” Journal of Vacuum
Science & Technology a Vacuum Surfaces and Films, vol. 13, no.
5, 1 Sept. 1995, pp. 2420–2426,
20
 pubs.aip.org/avs/jva/article/13/5/2420/456836/Radio-frequency-
electric-mirror-sputtering, https://doi.org/10.1116/1.579483.
Accessed 10 Apr. 2025.
[8] Huai, Ying, et al. “Characteristic Study of an Atmospheric-
Pressure Radio-Frequency Capacitive Argon/Nitrogen Plasma
Discharge.” IEEE Transactions on Plasma Science, vol. 42, no. 6,
June 2014, pp. 1648–1653,
ieeexplore.ieee.org/document/6813668. Accessed 10 Apr. 2025.
[9] ATTA, A A, et al. “Optical Characteristics of Transparent Samarium
Oxide Thin Films Deposited by the Radio-Frequency Sputtering
Technique.” Pramana, vol. 87, no. 5, 7 Oct. 2016, Accessed 10 Apr.
2025.
[10] Beck, R.B., et al. “PECVD Formation of Ultrathin Silicon Nitride
Layers for CMOS Technology.” Vacuum, vol. 70, no. 2-3, 31 Jan.
2003,pp.323–329,
www.sciencedirect.com/science/article/pii/S0042207X02006656?
via%3Dihub, https://doi.org/10.1016/S0042-207X(02)00665-6.

21
 Chapter 3
Characteristics of AlGaSnO Thin Film
3.1 Growth of AlGaSnO Thin Film
At first, the procedure of fabricating AGTO thin film will be
introduced below.
(1) The quartz substrates are cleaned thoroughly. We put them in the
acetone, isopropyl alcohol and deionized water in the ultrasonic cleaner
coming one for another. After cleaning, the substrates are dried by
nitrogen gun in order to eliminate particles.
(2) The AlGaSnO thin film which is 100 nm is sputtered by RF-
magnetron sputtering from SnO2 and AlGaO targets. Several different
process parameters, including changing oxygen flow rate and SnO 2
sputtering power, are conducted while the total gas flow is 50 sccm. Also,
the process pressure remains at 5 mTorr and the power of AlGaO is set as
20W. The distance between substrate holder and the target guns is 9 cm
and the angles are 30°. The holder is rotated with the constant speed if 15
rpm. The schematic diagram of AGTO thin films is illustrated at Figure
3-1.

Fig. 3-1. Schematic diagram of the AGTO thin film

22
 3.2 Optical Characteristics
This part will talk about the transmittance properties of AGTO thin
films. The transmittance spectrums are shown in Fig. 3-2. -Fig. 3-4. As
we can see, the transmittance across the visible light spectrum, which
covers from 400 to 700nm, is over 80%. On the contrary, the
transmittance across the ultraviolet region is much lower. The formula
below can describe the optical bandgap by Tauc relation:
(αhν)𝑛 = 𝐴(ℎ𝜈 − 𝐸𝑔) (3 − 1)
where the absorption coefficient is α, h= 4.136 × 10−15 (eV·s) is
Planck’s constant, ν is the frequency of incident photon, A is a constant
and Eg is the bandgap. For the parameter n, it equals to 2 for direct allowed
transitions and 0.5 for indirect allowed transitions. For the calculation, n
is assumed to be 2. The optical bandgap is found by plotting (αhν)2 versus
hν curve, and calculates the x-axis intercept of the line tangent to the
inflection point. The bandgap result is shown in Table 3-1. The one
changing SnO2 power is illustrated in Fig. 3-2. and another one with
adjusting oxygen flow rate is in Fig. 3-3. No matter how the SnO2 power
increases, the transmittance remains 90%. For increasing oxygen flow
rate, there is no obvious change elevating from 0% to 6%. On the other
hand, the bandgap decreases from 5.22(eV) to 5.02 (eV) when we elevate
the SnO2 power from 20W to 50W.

23
Fig. 3-2. The transmittance spectrum of AGTO thin films with different
SnO2 co-sputtering power

Fig. 3-3. The transmittance spectrum of AGTO thin films with different
oxygen flow rate

24
Fig. 3-4. Tauc plot of AGTO thin films with different SnO2 co-sputtering
power

Fig. 3-5. Tauc plot of AGTO thin films with different oxygen flow rate

25
 Table 3-1. Bandgap of AGTO thin films
Parameter Bandgap(eV)
20W/20W O2 4% 5.22
30W/20W O2 4% 5.14
40W/20W O2 4% 5.09
50W/20W O2 4% 5.02
40W/20W O2 0% 5.00
40W/20W O2 6% 5.03
3.3 X-ray Diffraction (XRD) Analysis
The XRD spectra of AGTO thin films are shown in Fig. 3-6.-Fig. 3-7.
In these figures, no obvious peak is observed in these samples. As a result,
either changing SnO2 power or adjusting oxygen flow rate cannot lead to
crystallite formation [1]. Hence, we can say that all these samples are
amorphous.

Fig. 3-6. XRD spectrum of AGTO thin films with different SnO2 power
26
Fig. 3-7. XRD spectrum of AGTO thin films with different oxygen flow
rate

3.4 Atomic Force Microscopes(AFM) Analysis

Atomic Force Microscopes(AFM) Analysis is used to gain the
information of the surface roughness. For example, some researches
mentioned about the effect of annealing to root mean square surface
roughness values [2]. This analysis is conducted to the AGTO thin films
in this research. Fig. 3-8. are the figures of different parameters’ AFM
images, including changing SnO2 and oxygen flow rate. According to the
results, there is no obvious relation between changing SnO2 power,
27
oxygen flow rate and root mean square (RMS) surface roughness values.
All of the values are below 2nm. The detailed results are listed in Table
3-2.
Table 3-2. Average root mean square (RMS) surface roughness values
Parameter RMS(nm)

40W/20W O2 0% 1.64

40W/20W O2 4% 1.48

40W/20W O2 6% 1.72

20W/20W O2 4% 1.56

30W/20W O2 4% 1.45

50W/20W O2 4% 1.89

60W/20W O2 4% 1.57

28
(a) (b) (c)

(d) (e) (f)

(g)

Fig. 3-8. AFM images of AlGaSnO thin films with different parameters:
(a) 20W/20W 4% (b) 30W/20W 4% (c) 40W/20W 4% (d) 50W/20W
4% (e) 60W/20W 4% (f) 40W/20W 0% (g) 40W/20W 6%

29
3.5 X-ray Photoelectron Spectroscopic (XPS) Analysis
Nowadays, many studies choose this method to figure out the relation
between the oxygen vacancy and other parameters. For instance, some
research works on finding the effect of annealing to the proportion of
oxygen vacancy [3]. In order to complete this analysis, the X-rays are
necessary. The target samples are irradiated with X-rays, which comes
from monochromatic X-ray source. The photons will go through the
samples and react with the material’s atoms, and cease many
photoelectrons emit. After that, we can get the information about the
samples, including the composition of thin films and its electronic
structure by the kinetic energy and intensity of these emitted
photoelectrons. By checking the binding energies of the emitted electrons,
the elements in the material can be found out. The binding energy of
electrons changes with its atomic species and chemical environment.
This part will focus on analyzing the oxygen vacancies. For thin-film
transistors, the oxygen vacancies are very important to the electrical
properties and considered as donor defects in oxide semiconductors. To
understand how the oxygen vacancy changes with different parameters,
the XPS O(1s) spectra are fitted with two Gaussian peaks. Fig. 3-8. to 3-
10. shows the XPS O(1s) spectra of AGTO thin films under different
oxygen flow rate from 0% to 6%. The first peak at 530(eV) is the oxygen
atoms bonded with metal atoms(OI), and the second peak at 531.5(eV) is
the oxygen vacancies(OII). When the oxygen flow rate increases from 0%
to 6%, the proportion of oxygen vacancies [OII/(OI+OII)] decreases from
48.11% to 37.27%. It indicates that the oxygen can help to fill some
oxygen vacancies existing in the semiconductor oxide. The proportion of
oxygen vacancies [OII/(OI+OII)] is listed in Table 3-3.
30
Table 3-3. The proportion of OII with changing oxygen flow rate
Oxygen flow rate OII/Ototal(%)
0% 48.11
4% 46.28
6% 37.11

Fig. 3-8. O1s XPS spectra of 0% 40w/20w AGTO

31
Fig. 3-9. O1s XPS spectra of 4% 40w/20w AGTO

Fig. 3-10. O1s XPS spectra of 6% 40w/20w AGTO

32
Reference
[1] D.M., Priyadarshini, et al. “Effect of Annealing Ambient on SnO2
Thin Film Transistors.” Applied Surface Science, vol. 418, 30 Nov.
2016,pp.414–417,
www.sciencedirect.com/science/article/pii/S0169433216326915?
via%3Dihub, https://doi.org/10.1016/j.apsusc.2016.11.233.
[2] Lounis, Zakia, et al. “Surface Stoichiometry Analysis by AES,
EELS Spectroscopy and AFM Microscopy in UHV Atmosphere of
SnO2 Thin Film.” Journal of Electron Spectroscopy and Related
Phenomena, vol. 226, 23 Apr. 2018, pp. 9–16,
www.sciencedirect.com/science/article/pii/S036820481730289X?
via%3Dihub, https://doi.org/10.1016/j.elspec.2018.04.006.
[3] Cao, Xiaoping, et al. “Structural Characterization of Pd-Doped
SnO2 Thin Films Using XPS.” Surface and Interface Analysis,
vol. 24, no. 9, 16 Sept. 1996, pp. 662–666,
https://doi.org/10.1002/(sici)1096-
9918(19960916)24:9%3C662::aid-sia155%3E3.0.co;2-c.

33
 Chapter 4
The Fabrication and characteristics of AGTO MSM
UV Photodetectors
4.1 Motivation
Nowadays, UV photodetectors(PDs) are studied and gain a great deal
of attention because of its widespread applications in many different
fields, such as military applications, spectroscopy, and optoelectronic
circuits [1][2]. Normally, solar-blind means wavelength from 200nm to
280nm are cut, and this range of wavelength cannot go through Earth’s
atmosphere. As a result, many solar-blind UV PDs are continuously
applied on many different fields, such as biochemical detection,
environment mentoring and military. Also, it is significant that these
PDs possess great stability.
To find the low cutoff wavelength PDs, wide-bandgap
semiconductor materials have been studied for a long time [3]. Then,
Ga2O3 has been investigated over and over due to its high bandgap
(4.9eV to 5.3eV). Also, this material has great responsivity in the UV
spectrum [4][5]. At first, many researchers found the potential of β-
Ga2O3. Unfortunately, it is the high temperature that the process of this
material required [6]. Therefore, researcher changed for other way-
choosing Ga2O3-based alternatives [7]. For example, many different
materials like In2O3 which provides carriers [8][9], and choose Al2O3 as
passivation layer can make photodetector more stable [10]. What’s
more, these UV PDs can be produced by sputtering at room temperature
[11].
In this part, we will focus on the AGTO UV PDs. The MSM structure
is chosen, and concentrate on the effects of adjusting oxygen flow rate
and SnO2 power. Then, we will discuss the results including optical
34
characteristics and electrical properties.

4.2 Fabrication of AGTO MSM Photodetectors

At first, we will talk about the whole procedure of process:
(1) The quartz substrates are thoroughly cleaned by the ultrasonic
oscillator. Acetone, isopropyl alcohol and deionized water are
continuously used. After cleaned, the substrates are dried by nitrogen
gun.
(2) 100 nm-thickness AGTO thin films are deposited on the substrate
by RF co-sputtering with changing different oxygen flow rate. Before
the process, the chamber is pumped, and its pressure is at 5х10-6 (Torr).
During the process, the pressure is set as 5mTorr. For the sputtering, the
SnO2 power is 40 Watt and AGO target power is 20 Watt. The holder
rotates at the constant speed of 15 rpm, and the distance between the
substrates and the holder is 9(cm).
(3) Lastly, the Ni/Au(30/70nm) metal contacts are deposited as top
electrodes above the semiconductor by thermal evaporation with
specific masks. The dimensions of finger structures are illustrated in Fig.
4-1. (a)-(b), which is in the interdigitated structure. The width(W) and
length(L) is 0.1(mm) and 1.2(mm) respectively, and the finger-to-finger
spacing is 0.2(mm).

35
Fig. 4-1. (a) Metal mask of interdigitated electrodes and AGTO thin
film on the substrate (b) Schematic of AGTO photodetector

4.3 Characteristics of AGTO MSM Photodetectors

4.3.1 Characteristics of AGTO MSM Photodetectors with changing
different SnO2 power
In this section, the relation between different SnO2 power and
performance of PDs is discussed. We changed SnO2 power from 20W
to 50W. The wavelength of light source is from 500(nm) to 220 (nm),
and the applied voltage is from -10(V) to 10(V). The dark current for
20W/20W O2 4%, 30W/20W O2 4%, 40W/20W O2 4%, 50W/20W O2
4% is measured at 1.98×10-7(A), 2.00×10-7(A), 2.46×10-6(A),
3.01×10-4(A). And the photocurrent under 300(nm) are 2.38×10-7(A),
7.12×10-7(A), 4.37×10-5(A), and 4.22×10-4(A). According to the data,
there is an increasing trend of currents when we elevate SnO2 power
from 20W to 50W. The I-V characteristics of different SnO2 power
ranging from 20W to 50W are shown in Fig. 4-2. (a)-(d).

36
Fig. 4-2. (a) I-V curve of 20W/20W O2 4% PD

Fig. 4-2. (b) I-V curve of 30W/20W O2 4% PD

37
Fig. 4-2. (c) I-V curve of 40W/20W O2 4% PD

Fig. 4-2. (d) I-V curve of 50W/20W O2 4% PD

38
Fig. 4-3. (a) Responsivity of 20W/20W O2 4% PD

Fig. 4-3. (b) Responsivity of 30W/20W O2 4% PD

39
Fig. 4-3. (c) Responsivity of 40W/20W O2 4% PD

Fig. 4-3. (d) Responsivity of 50W/20W O2 4% PD

40
Table 4-1. Characteristics of AGTO MSM PDs with different SnO2 power
Sputtering Photo Dark Photo/Dark EQE Rejection
power current current current Ratio
(A) (A) Ratio
-7 -7
20W/20W 2.38×10 1.98×10 1.20 4.02% 1.82

30W/20W 7.12×10-7 2.00×10-7 3.56 12.0% 3.33

40W/20W 4.37×10-6 2.46×10-6 1.78 73.6% 3.35

50W/20W 4.22×10-4 3.01×10-4 1.40 N/A N/A

4.3.2 Characteristics of AGTO MSM Photodetectors with changing

different oxygen flow rate
To investigate the characteristics of AGTO MSM PDs, several
samples are fabricated under 0% and 6%. The wavelength of light
source is from 500(nm) to 220 (nm), and the applied voltage is from
-10(V) to 10(V) in Fig. 4-2. (a) - (c). The responding responsivity is
shown in Fig. 4-3. (a)- (c). The results show that there is a decreasing
trend in current from changing oxygen flow rate from 0% to 6%. It can
be explained by the oxygen vacancies. When the oxygen is introduced
into the chamber during the process, the oxygen vacancies in the AGTO
thin film can be filled. Therefore, as we pumped much oxygen into the
chamber, more vacancies can be repaired. This cause the decrease in
number of carriers and results in the lower current.
To be specific, the dark currents for 40W/20W O2 0%, 40W/20W O2
4% and 40W/20W O2 6% are 4.32×10-5 (A), 2.46×10-6(A), 1.69×10-9
(A). Also, the photocurrent under the 240 (nm) light source are
4.32×10-5(A), 4.37×10-6 (A) and 1.97×10-9 (A) for each setting
respectively, which demonstrates a decreasing trend.

41
Fig. 4-4. (a) I-V curve of 40W/20W O2 0% PD

Fig. 4-4. (b) I-V curve of 40W/20W O2 4% PD

42
 Fig. 4-4. (c) I-V curve of 40W/20W O2 6% PD

Fig. 4-5. (a) Responsivity of 40W/20W O2 4% PD

43
 Fig. 4-5. (b) Responsivity of 40W/20W O2 6% PD

Table 4-2. Characteristics of AGTO MSM PDs with different oxygen

flow rate
Oxygen flow Photo Dark Photo/ EQE Rejection
rate current current Dark Ratio
(40W/20W) (A) (A) current
Ratio
0% 4.32×10-5 1.44×10-5 3.000 N/A N/A
4% 4.37×10-6 2.46×10-6 1.776 73.6% 3.35
6% 1.97×10-9 1.69×10-9 1.166 11.8% 1.72

There is a method to calculate the photoconductive gain, we can

analyze the scale by external quantum efficiency (EQE) [12] [13]. This
parameter describes the number of carriers circulating through a

44
photodetector per absorbed photon and per unit time. The formula of
EQE is:
ℎ𝑐 ∆𝐼 1
EQE = × × (4 − 1)
𝑒𝜆 𝑃 𝐺
where h is Plank constant, c is the light velocity, e is the electronic

charge, λis the wavelength of radiation, ∆I is the difference between

photo current and dark current, P is the light power density irradiated
on the PDs and G is the photoconductive gain.
The photo responsivity from 500(nm) to 220(nm) of oxygen flow rate
from 0% to 6% is shown in Fig.4-4. (a)-(d). Separately, the responsivity
of 4% and 6% is 1.56 and 8.41. We can find a clear relation between
oxygen flow rate and responsivity. UV-to-visible rejection ratio (RR) is
defined as the responsivity at 240 (nm) divided by the one at 500 (nm).
The equation is written as follow:
𝑅𝑒𝑠𝑝𝑜𝑛𝑠𝑖𝑣𝑖𝑡𝑦(𝑈𝑉 𝑟𝑎𝑛𝑔𝑒)
RR = (4 − 2)
𝑅𝑒𝑠𝑝𝑜𝑛𝑠𝑖𝑣𝑖𝑡𝑦(500𝑛𝑚)

4.3.3 Time-Resolved Response of AGTO MSM Photodetectors

The time-resolved response of AGTO MSM photodetectors are

discussed in this part. The definition of rising time is the time interval
for the current increasing from 10% to 90% of the peak under
continuous illumination. On the contrary, the falling time means the
time interval for the current decreasing from 90% to 10% of the peak
without illumination [14]. We chose the 240-nm light and the bias
voltage which is 10(V). To measure the time-resolved response, we
choose 40W/20W photodetector, and the result is illustrated in Fig. 4-6.
The device is switched on and off for 5 cycles with the interval of 150
45
seconds. However, the device cannot reach the saturation current and
return to the initial state before illumination. In other words, the device
cannot correctly finish the on and off switching in the short time. Its
average response time are shown in Table 4-3.

Fig. 4-6. Time-resolved response of 40W/20W photodetector

Table 4-3. The average response time of AGTO MSM PDs

Rising time(seconds) Falling time(seconds)
40W/20W ˃150 ˃150

4.4 Summary
To conclude, the AGTO MSM photodetectors are fabricated and the
effect of changing SnO2 power and oxygen flow rate is discussed. When
elevating SnO2 power from 20W to 50W, there are more Sn that can
provide more carriers so that the current can enlarge. The dark current
increases from 1.98×10-7(A) to 3.01×10-4(A) and the photo current
increases from 2.38×10-7(A) to 4.22×10-4(A). For changing oxygen
flow rate, the dark current decreases from 0% as 4.32×10-5 (A) to 6%
as 1.69×10-9 (A). And the photocurrent under 240(nm) decreases from

46
4.32×10-5 (A) to 1.97×10-9 (A). As we can see, the current decreases as
we elevate the oxygen flow rate, which is because introducing the
oxygen can help to get rid of oxygen vacancy so that the current
decreases. When the oxygen flow rate is 4%, the PD shows the best
performance above all the parameters. The photo and dark current are
4.37×10-6 and 2.46×10-6 respectively. The rejection ratio is 3.35 and the
EQE is 73.6%, which is the largest among all the samples. At the end,
time-resolved response of AGTO MSM photodetectors is discussed and
its result is shown. Its rising and falling time is over 150 seconds. In
other words, all of the devices cannot react to the light quickly.

Reference
[1] Fontana, Márcio, et al. “Photosensitivity Characterization of
Nanostructured Tin Oxide Films and Alternative Photodetector
Application.” IEEE Sensors Journal, vol. 11, no. 4, Apr. 2011, pp.
869–874, ieeexplore.ieee.org/document/5582140,
https://doi.org/10.1109/jsen.2010.2068542. Accessed 11 Apr.
2025.
[2] Li, Meiya, et al. “Radio Frequency Sputtered Zinc Oxide Thin
Films with Application to Metal–Semiconductor–Metal
Photodetectors.” Thin Solid Films, vol. 515, no. 18, June 2007,
pp. 7357–7363, https://doi.org/10.1016/j.tsf.2007.03.026.
Accessed 20 May 2022.
[3] Zhou, Changqi, et al. “Ultraviolet Photodetectors Based on Wide
Bandgap Oxide Semiconductor Films.” Chinese Physics B, vol.
28, no. 4, Apr. 2019, p. 048503, https://doi.org/10.1088/1674-
1056/28/4/048503. Accessed 19 May 2022.

47
[4] Qin, Yuan, et al. “Review of Deep Ultraviolet Photodetector
Based on Gallium Oxide.” Chinese Physics B, vol. 28, no. 1, 1
Jan. 2019, pp. 018501–018501, https://doi.org/10.1088/1674-
1056/28/1/018501. Accessed 27 Apr. 2023.
[5] Liang, Huili, et al. “Recent Progress of Deep Ultraviolet
Photodetectors Using Amorphous Gallium Oxide Thin Films.”
Physica Status Solidi (A), vol. 218, no. 1, 14 Oct. 2020, p.
2000339, https://doi.org/10.1002/pssa.202000339.
[6] Zhong, Mianzeng, et al. High-Performance Single Crystalline UV
Photodetectors of β-Ga2O3. Vol. 619, 1 Jan. 2015, pp. 572–575,
https://doi.org/10.1016/j.jallcom.2014.09.070. Accessed 17 June
2023.
[7] Zhang, H., et al. “Trace Amount of Niobium Doped β-Ga2O3
Deep Ultraviolet Photodetector with Enhanced Photo-Response.”
Optik, vol. 243, 2 June 2021, p. 167353,
www.sciencedirect.com/science/article/pii/S0030402621009724?
via%3Dihub, https://doi.org/10.1016/j.ijleo.2021.167353.
[8] Young, Sheng-Joue, and Yi-Hsing Liu. “Ultraviolet
Photodetectors with 2-D Indium-Doped ZnO Nanostructures.”
IEEE Transactions on Electron Devices, 2016, pp. 1–5,
ieeexplore.ieee.org/document/7506319,
https://doi.org/10.1109/ted.2016.2582506. Accessed 11 Apr.
2025.
[9] “Fabrication and Measurement of Cosputtered Indium Gallium
Oxide Ultraviolet...: EBSCOhost.” Ebscohost.com, 2024,
web.p.ebscohost.com/ehost/pdfviewer/pdfviewer?vid=0&sid=b6a
62d75-d935-4335-a75d-3d47c4fbbac2%40redis. Accessed 11
Apr. 2025.

48
[10] Qu, Jiaqi, and Jun Chen. “Near-Infrared Photodetector Based on
MoS2 QDs/GaAs Heterojunction with an Al2O3 Interface
Passivation Layer.” Micro and Nanostructures, Apr. 2022, p.
207231, https://doi.org/10.1016/j.micrna.2022.207231. Accessed
30 Apr. 2022.
[11] Li, Huixin, et al. “Flexible Ultraviolet Photodetector Based ZnO
Film Sputtered on Paper.” Vacuum, vol. 172, 19 Nov. 2019, p.
109089,
www.sciencedirect.com/science/article/pii/S0042207X19323437?
via%3Dihub, https://doi.org/10.1016/j.vacuum.2019.109089.
[12] Katayama, Haruyoshi, et al. “Measurement of Absorption and
External Quantum Efficiency of an InAs/GaSb Type II
Superlattice.” Infrared Physics & Technology, vol. 70, 31 Oct.
2014, pp. 53–57,
www.sciencedirect.com/science/article/pii/S1350449514002369?
via%3Dihub, https://doi.org/10.1016/j.infrared.2014.10.014.
[13] Xing, Meijiao, et al. “High External Quantum Efficiency in
ZnO/Au/Ga2O3 Sandwich–Structured Photodetector.” Applied
Surface Science, vol. 618, May 2023, p. 156705,
https://doi.org/10.1016/j.apsusc.2023.156705. Accessed 2 Apr.
2023.
[14] Kin-Tak Lam, Artde Donald. “Time-Resolved Response
Improvement of Oxygen-Doped A-In-Ga-Sn-O Metal-Semic...:
EBSCOhost.” Ebscohost.com, 2024,
web.p.ebscohost.com/ehost/pdfviewer/pdfviewer?vid=0&sid=fb7
9f90b-1613-4d80-95ad-3d7eb52becee%40redis. Accessed 11
Apr. 2025.

49
 Chapter 5

The Fabrication and Characteristics of AGTO Thin-Film

Transistors

5.1 Motivation
More and more research focus on thin-film transistors(TFTs) [1].
Classically, semiconductor materials are deposited on glass substrate.
And these TFTs have great reputation due to high field-effect mobility,
low-temperature process, low leakage current and great optical
transparency [2] - [4].
Nowadays, transparent oxide semiconductor (TOS) thin-film
transistors are applied in many fields, including active-matrix displays,
such as LCD screens [5]-[8]. What’s more, thin-film transistors can
even be used in active-matrix liquid crystal displays(AMLCDs).
In the past, Gallium oxide(Ga2O3) is investigated deeply due to large
bandgap. Ga2O3 thin-film transistors have many advantages, such as
high field-effect mobility and stability, and so on [9]. These makes
Ga2O3 become a potential material for next generation power electronic
devices and high-frequency electronic devices [10] [11]. On the other
hand, AGO has been extensively discussed due to its larger bandgap
than Ga2O3. The high bandgap of AGO shows more potential on [12].
Unfortunately, as thin-film transistor’s material, AGO cannot provide
enough carriers as the channel layer. Hence, this thesis decides to utilize
SnO2 to deal with this problem. In other words, SnO2 is expected to give
more carriers so that the problem can be solved [13] [14].
This part will focus on AGTO TFTs produced by RF sputtering.
Three different changes are made and discussed in this chapter. That
is –changing SnO2 power, oxygen flow rate and types of insulating
50
layers. The electrical properties are mentioned and compared in the
following articles.

5.2 Fabrication of AGTO Thin-Film Transistors

Several steps are taken to fabricate the devices: At the beginning, 2
cm х 2 cm quartz substrates are cleaned using ultrasonic bath, soaring
the substrates into acetone, isopropyl alcohol and deionized water step
by step. Each time of soaring is set as 5 minutes. After that, a 70-nm
thick aluminum (Al) bottom gate is deposited on the substrate by E-gun
evaporation system, and the 200nm-thick SiO2 dielectric layer followed
by plasma-enhanced chemical vapor deposition (PECVD).
Then, AGTO thin films are sputtered on the SiO2 dielectric layer with
changing SnO2 power from 20W to 60W. All the thickness is 30 nm.
The atomic ratio of AGO is 5:95. During all the process, the pressure
keeps at 5 mTorr. After that, different oxygen flow rate from 0% to 6%,
and the co-sputtering power combination is SnO2/AGO ratio of
40W/20W. Total gas flow rate is 50 sccm.

At the end, 70-nm thick Al is deposited as source(S) and drain(D)

electrodes by E-gun evaporation system. The schematic of AGTO thin-
film transistors is depicted in Fig. 5-1.

51
 Fig. 5-1. Schematic of AGTO thin-film transistor

5.3 Characteristics of AGTO Thin-film Transistors

5.3.1 Characteristics of AGTO Thin-film Transistors with

different SnO2 power

In order to find out the relation between electrical properties and

SnO2 power, AGTO thin-film transistors are fabricated and measured in
the same way. At the beginning, we have tried not to insert any oxygen.
Unfortunately, most of thin-film transistors from changing 20W to 80W
cannot work normally due to excessive oxygen vacancy. Only 40W can
perform in the right way, Therefore, we decide to introduce 4% oxygen
to improve the problem. During the process, the AGO target ‘s power is
fixed at 20 W, the oxygen flow rate is 4% and SnO2 target’s power
adjusts from 20 W to 50 W, which ID-VG characteristics are illustrated
in Fig. 5-2. – Fig. 5-5. All the ID-VD characteristics AGTO TFT is
52
illustrated in Fig. 5-6. – Fig. 5-9. The applied gate voltage(VG) is from
-10 (V) to 15(V) and the drain voltage is fixed at 8(V). According to the
results, there is an obvious trend between electrical properties and SnO2
power. The On/Off ratio of 20W/20W, 30W/20W, 40W/20W, 50W/20W,
TFTs are 6.62 × 102, 1.69× 103, 2.83× 103, 1.06 × 101 respectively. The
subthreshold swings are 0.873(V/decade), 0.455(V/decade),
0.552(V/decade),1.609(V/decade). The threshold voltages are 6.02(V),
4.82(V), 4.53(V), 2.71(V). Obviously, the one with 40W/20W
possesses the best electrical properties, which has the largest On/Off
ratio and the smallest threshold voltage.
As the SnO2 power increases, the threshold voltage shifts to the left.
In AGTO TFTs, the main carriers come from Sn’5s orbit and that why
the content of Sn is important for influencing the current of TFTs. As
the Sn content increases, Ga and Al content decreases. To conclude, we
can say that Sn plays a role in supplying carriers. Therefore, we decide
to change the power on SnO2 in order to increase the content of Sn,
which may help to increase carriers and contribute to the larger On/Off
ratio. As we can see from the result, the ratio really increases from
6.62×102 to 2.83×103 by elevating the power from 20W to 40W and
reduces to 1.06 × 101 at 50W.
However, there is also a problem from increasing the SnO2 power.
According to the data, we can see that the trap density continues to rise
from 5.93×1012 to 7.82×1012 (cm-2) as the power increases from 20W
to 50W. Normally, more carriers will be captured when the trap density
is larger, and will make the thin-film transistors performed badly
because of lower ability for gate voltage to control the channel. This
also reflects on the subthreshold swing. According to the result, the
subthreshold swing shares the same time to reach maximum with the
trap density, which takes place when SnO2’s power is 50W. The
53
 performance of different SnO2 power thin-film transistors are listed in
Table 5-1. As a result, we try to find other way to improve the
performance.

Table 5-1. Characteristics of different SnO2 power AGTO TFT

SnO2 SS(V/dec) μ(cm2V-1s-1) On/Off Vt(V) Dit (cm-2)
/AGO(5%) ratio
20w/20w 0.873 4.97 6.62×102 8.29 5.93×1012

30w/20w 0.552 4.63 1.69×103 4.82 6.23×1012

40w/20w 0.539 7.23 2.83×103 4.53 2.25×1012

50w/20w 1.609 1.58 1.06×101 2.71 7.82×1012

Fig. 5-2. ID-VG characteristics of 20W/20W 4% AGTO TFT

54
Fig. 5-3. ID-VG characteristics of 30W/20W 4% AGTO TFT

Fig. 5-4. ID-VG characteristics of 40W/20W 4% AGTO TFT

55
Fig. 5-5. ID-VG characteristics of 50W/20W 4% AGTO TFT

Fig. 5-6. ID-VD characteristics of 20W/20W 4% AGTO TFT

56
Fig. 5-7. ID-VD characteristics of 30W/20W 4% AGTO TFT

Fig. 5-8. ID-VD characteristics of 40W/20W 4% AGTO TFT

57
 Fig. 5-9. ID-VD characteristics of 50W/20W 4% AGTO TFT

5.3.2 Characteristics of AGTO Thin-film Transistors with

different oxygen flow rate

In order to understand the relation between oxygen flow rate and

the thin-film transistor’s performance, we choose the best transistor
mentioned above to be the sample. In Fig. 5-10. and Fig. 5-11. illustrates
the ID-VG characteristics of TFTs which is fabricated at SnO2 40W and
AGO 20W, and its oxygen flow rate is 0% and 6% respectively. For
elevating the oxygen flow rate, the threshold voltage increases because
the oxygen can get rid of oxygen vacancies, which will reduce free
carriers. The results match with the XPS analysis mentioned before. As
a result, it can efficiently improve the performance of transistors. As we
can see in the trap densities, the value of oxygen flow rate 0% to 4%

58
decreases from 3.62×1012 (cm-2) to 1.76×1012 (cm-2) when the oxygen
flow rate is 6%. The ID-VD characteristics of 0% and 6% oxygen flow
rate are shown in Fig. 5-12 and Fig. 5-13. As a result, changing the
oxygen flow rate can really influence the performance of thin-film
transistors. The characteristics of different oxygen flow rate are listed
in Table 5-2.

Fig. 5-10. ID-VG characteristics of 40W/20W 0% AGTO TFT

59
Fig. 5-11. ID-VG characteristics of 40W/20W 6% AGTO TFT

Fig. 5-12. ID-VD characteristics of 40W/20W 0% AGTO TFT

60
 Fig. 5-13. ID-VD characteristics of 40W/20W 6% AGTO TFT

Table 5-2. Characteristics of different oxygen flow rate AGTO TFT

Oxygen SS(V/dec) μ(cm2V-1s-1) On/Off Vt(V) Dit(cm-2)

flow rate ratio
0% 0.757 0.35 2.14×101 8.04 3.62×1012

4% 0.539 7.23 2.83×103 4.53 2.25×1012

6% 0.380 1.9 2.86×101 -9.18 1.76×1012

5.3.3 Double-channel thin-film transistor

In these years, many papers are published to introduce the new
methods of improving thin-film transistors. For example, there are some
studies working on depositing a passivation layer in order to elevate the
stability [15]. In this paper, to improve the performance, we decide to
achieve the goal by replacing the single-layer active layer structure with
double-channel active layer one. Many thin-film transistor’s
61
performance and control ability of gate can be strengthened with the
new structure [16].
For the experimental process, most of the steps are same with the
original one. The only different part is that we deposit an extra active
layer by RF sputtering. In previous chapter, we can see the improvement
results of increasing the oxygen flow rate, and the best oxygen flow rate
is 4%. At the same time, the one with the lowest oxygen vacancies
happens in 0% oxygen flow rate parameter, which has the lowest
carriers. Therefore, in order to upgrade the number of free carriers, we
deposit an active layer which is 40W/20W and no oxygen is introduced
during process under the original layer. Both of the layer is 15(nm) so
that the total thick of layer stays same. The structure is illustrated in Fig.
5-14.
In this part, the basic theory of double-channel layer thin-film
transistors is introduced. The layer with 0% oxygen flow rate owns
much carriers because there are more oxygen vacancies in it. As a result,
it can give a hand to open the channel for providing carriers, which can
reduce the threshold voltage, upgrade the on/off ratio. On the contrast,
the one with 4% oxygen flow rate has smaller number of vacancies,
which plays a role in closing the channel of the device, and expect to
get lower leakage current and better gate control ability due to lower
vacancies. After implementing the structure, the On/Off ratio increases
from 2.83×103 to 1.13×104 and the threshold voltage decreases from
4.52(V) to 3.87(V). The ID-VG characteristics is illustrated in Fig. 5-15.
(a) and the ID-VD characteristics is in Fig. 5-15. (b). The comparison
between changing structures is listed in Table 5-3.

62
 Fig. 5.14 Schematic of AGTO double-layer thin-film transistor

Fig. 5-15. (a) ID-VG characteristics of AGTO Double-Channel thin-film

transistor
63
 Fig. 5-15. (b) ID-VD characteristics of AGTO Double-Channel thin-film
transistor

Table 5-3. Comparison between Single-layer TFTs and Double-layer

TFTs
Structure SS(V/dec) μ(cm2V-1s-1) On/Off Vt(V) Dit(cm-2)
ratio
Single- 0.539 7.23 2.83×103 4.53 2.25×1012
channel
Double- 0.300 10.45 1.13×104 3.87 1.37×1012
channel

5.4 Elemental Analysis of AGTO Thin-film Transistors

5.4.1 Transmission Electron Microscopy(TEM) Analysis
64
 Transmission Electron Microscopy(TEM) is used to observe
the device’s structure in detailed. The structure of TFT and the point
chosen to be analyzed is shown in Fig. The cross-image of the specific
point is shown in Fig. Also, we can make sure the real thickness of each
layer is close to the value we set. The TEM cross section image of
40W/20W AGTO TFT is in Fig. 5-16.
5.4.2 Energy Dispersive Spectra(EDS) Analysis
This analysis can clearly show the composition of AGTO TFTs. As
we can see in Fig.5-17 and Fig. 5-18, all the components deposited in
each layers are demonstrated. To conduct the analysis, we choose three
points at the AGTO thin-film layer in Fig., and we can get the results at
point 1, 2, 3 in Fig. 5-19,respectively. To conclude, the only elements
we can find in the thin films are gallium, aluminum, tin and oxygen.

Fig. 5-16. The TEM cross section image of 40W/20W AGTO TFT

65
Fig. 5-17. Cross section image of 40W/20W AGTO TFT

Fig. 5-18. Elements in 40W/20W AGTO TFT

66
 Fig. 5-19 Cross section image of 40W/20W AGTO TFT

Fig. 5-20. EDS spectrum of AGTO 40W/20W AGTO TFT

5.5 Characteristics of AGTO Thin-film Phototransistors

As mentioned above, AGTO is a wide-bandgap material and
appropriate for UV phototransistors [17]. In this part, we choose

67
 40W/20W AGTO TFT as the sample, and is illuminated across the
range from visible light at 500nm to UV light at 220nm. During the
measurement, the applied gate voltage (VG) ranged from -10 (V) to
15(V), and the drain voltage(VD) is fixed at 8V. The ID-VG
characteristics under different light wavelengths are illustrated in Fig.
The photo responsivity is illustrated in Fig. When VG is -2.1(V), the
lowest dark current can be gained. The current increases from
6.29×10-8 (A) to 8.30×10-7(A) under illumination of 220(nm). The
photo to dark ratio is 13.19. This result matches with the characteristics
of photodetectors mentioned in the previous chapter.
The basic principle of thin-film phototransistors is introduced below.
When we put the transistors under UV illumination, the light will excite
the electron-hole pairs in channel layer. This will make the drain current
due to appearance of extra carriers. The formula below can describe
relation between drain current, drain voltage, and gate voltage under
illumination:
I`D = ID +Iph (5 − 1)
In this formula, ID is the total current of phototransistors, ID is current
from the applied drain voltage and Iph is the increased part of current,
which comes from the excited electron-hole pairs. Just like the thin-film
transistors, when the gate voltage(VG) is smaller than the threshold
voltage(Vth),the total current is dominated by the one comes from excited
electron-hole pairs. This situation is defined as Off state, which means
total drain current is occupied with Iph mostly. However, when we enlarge
the gate voltage gradually, the Fermi level in channel layer descends and
brings about the so-called On state. The procedure of illumination is
displayed below in Fig.
Then, we calculate the photo responsivity of AGTO phototransistors
by formula below. Obviously, we can find that the higher responsivity
68
appears at the range of 260(nm) to 220(nm). The rejection ratio is
1.06×101.

Fig.5-21. ID-VG Characteristics of 40W/20W AGTO Phototransistor

Fig.5-22. Photo responsivity of 40W/20W AGTO phototransistor at

different applied gate voltage

69
(a) (b)

(c)

Fig.5-23. Energy-band diagram of AGTO phototransistor under

illumination as well as different applied gate voltage: (a) thermal
equilibrium (VG=0V), (b) VG<Vth, and (c) VG ˃ Vth
5.6 Time-Resolved Response of AGTO Thin Film
Phototransistors
The result of the time-resolved response of the thin film
phototransistors is illustrated in Fig. During the test, each switching
on/off possesses the interval of 150 seconds, and the transistor is tested
for 5 on/off cycles. Besides, the gate voltage is fixed at -10(V) and the
drain voltage keeps at 10(V) during all the experiment. To conclude,
although the falling time is set to be longer than 150 seconds, the current
still cannot reach to the saturation and return to the initial state. The rising

70
and falling time of 40W/20W AGTO phototransistor are listed in Table 5-
4.

Fig. 5-25. Time-resolved response of 40W/20W AGTO phototransistor

Table 5-4. Rising and falling time of 40W/20W AGTO phototransistor

Rising Time Falling Time
40W/20W ˃150 ˃150

71
5.7 Summary
To conclude, this chapter mention about the characteristics of AGTO
thin-film transistors. For the devices, we use RF-sputtering with
different parameters including changing SnO2 power and oxygen flow
rate in order to figure out the relation between the performance of
devices and different variables. Among all the experiment, the AGO
power is set as 20W. Increasing tin can effectively upgrade the on and
off current, and results in the better performance. When we elevate the
SnO2 power from 20W to 50W, the best performance takes place at 40W,
which has the biggest On/Off ratio 2.83×103 as well as the smallest
subthreshold swing 0.539(V/dec).
Subsequently, we change the oxygen flow rate from 0% to 6% in
order to understand the effect of it. During the experiments, the SnO2
power is fixed at 40W and the AGO power is set as 20W. Rising the
oxygen flow rate results in the declining On/Off ratio because the
oxygen can get rid of oxygen vacancies. As a result, it can efficiently
improve the performance of transistors. According to the data, the
On/Off ratio increases from 2.14×101 to 2.83×103 when we elevate the
oxygen flow rate from 0% to 4%. From the trend of trap density, the
effect of filling vacancy by oxygen can be seen. To conclude, the best
performance takes place at 4%. The value of device’s On/Off ratio is
2.83×103 and the subthreshold swing 0.539(V/dec). The results match
with our expectation.
Next, we change the single-layer thin-film transistor to double-layer
in order to improve the performance. Using the new structure, more
carriers will be provided by the 0% layer so that the gate control ability
will be enhanced. As a result, the On/Off ratio increases from 2.83×103
to 1.13×104. Also, the threshold voltage decreases from 4.52(V) to
72
 3.87(V). At the end, we mention about the performance of visible-blind
phototransistors, which is fabricated with 40W/20W and the rejection
ratio is 1.39×102. At the end, time-resolved response of AGTO MSM
photodetectors is discussed and its result is shown. Its rising and falling
time is over 150 seconds. In other words, all of the devices cannot react
to the light quickly.

Reference

[1] Ting-Hao Chang, et al. “Bandgap-Engineered in Indium–Gallium–

Oxide Ultraviolet Phototransistors.” IEEE Photonics Technology

Letters, vol. 27, no. 8, 15 Apr. 2015, pp. 915–918,

https://doi.org/10.1109/lpt.2015.2400446. Accessed 22 Aug. 2022.

[2] Xu, Xin, et al. “Amorphous Indium Tin Oxide Thin-Film Transistors

Fabricated by Cosputtering Technique.” IEEE Transactions on

Electron Devices, vol. 63, no. 3, 1 Mar. 2016, pp. 1072–1077,

https://doi.org/10.1109/ted.2015.2513421. Accessed 2 Aug. 2024.

[3] Kim, Dong‐Ho, et al. “Sputter‐Deposited Ga–Sn–Zn–O Thin Films

for Transparent Thin Film Transistors.” Physica Status Solidi (A), vol.

208, no. 12, 31 Aug. 2011, pp. 2934–2938,

https://doi.org/10.1002/pssa.201127213. Accessed 27 Jan. 2025.

[4] Paine, David C., et al. “Amorphous IZO-Based Transparent Thin

Film Transistors.” Thin Solid Films, vol. 516, no. 17, July 2008, pp.
73
 5894–5898, https://doi.org/10.1016/j.tsf.2007.10.081. Accessed 29

Apr. 2022.

[5] Chen, Shangxin, et al. “Analysis of Backlight White Spots on the

TFT-LCD Screen and Improvement Countermeasures.” Optical

Engineering, vol. 60, no. 08, 9 Aug. 2021,

www.spiedigitallibrary.org/journals/optical-engineering/volume-

60/issue-08/085105/Analysis-of-backlight-white-spots-on-the-TFT-

LCD-screen/10.1117/1.OE.60.8.085105.full,

https://doi.org/10.1117/1.oe.60.8.085105. Accessed 11 Apr. 2025.

[6] Lu, Hsueh-Ping, et al. “Combination of Convolutional and

Generative Adversarial Networks for Defect Image Demoiréing of

Thin-Film Transistor Liquid-Crystal Display Image.” IEEE

Transactions on Semiconductor Manufacturing, vol. 33, no. 3, 1 Aug.

2020, pp. 413–423, https://doi.org/10.1109/tsm.2020.3005164.

Accessed 26 Aug. 2024.

[7] Cao, Yuanzhi, et al. “High‐Resolution Monolithic Integrated

Tribotronic InGaZnO Thin‐Film Transistor Array for Tactile

Detection.” Advanced Functional Materials, vol. 30, no. 35, 9 July

74
 2020, p. 2002613, https://doi.org/10.1002/adfm.202002613.

Accessed 7 Feb. 2021.

[8] Xue, Xianyang, et al. “Flexible Dual‐Parameter Sensor Array without

Coupling Based on Amorphous Indium Gallium Zinc Oxide Thin

Film Transistors.” Advanced Materials Technologies, vol. 7, no. 3, 5

Oct. 2021, https://doi.org/10.1002/admt.202100849. Accessed 22

Apr. 2024.

[9] Pintor-Monroy, Maria Isabel, et al. “Tuning Electrical Properties of

Amorphous Ga2O3 Thin Films for Deep UV Phototransistors.” IEEE

Sensors Journal, vol. 21, no. 13, 22 Apr. 2021, pp. 14807–14814,

ieeexplore.ieee.org/document/9410280,

https://doi.org/10.1109/jsen.2021.3074623. Accessed 11 Apr. 2025.

[10] Woo, Kelly, et al. “From Wide to Ultrawide-Bandgap

Semiconductors for High Power and High Frequency Electronic

Devices.” Journal of Physics: Materials, 23 Jan. 2024,

https://doi.org/10.1088/2515-7639/ad218b.

[11] Goyal, Priyanshi, and Harsupreet Kaur. “Investigating Viability of

Split-Stepped Gate Field Plate Design on Ga2O3 MOSFET for High

Power Applications.” Journal of Electronic Materials, vol. 53, no. 8,

75
 8 June 2024, pp. 4544–4552, https://doi.org/10.1007/s11664-024-

11225-3. Accessed 11 Apr. 2025.

[12] Chung, Sheng-Ti, et al. “Study on Metal-Oxide Field Effect

Transistors of β-Gallium Oxide with AlGaO Spacer Layer Grown on

Sapphire for High-Power Device Applications.” ACS Applied

Electronic Materials, vol. 7, no. 7, 19 Mar. 2025, pp. 2767–2775,

https://doi.org/10.1021/acsaelm.4c02230. Accessed 11 Apr. 2025.

[13] Lin, Yung-Hao, and Ching-Ting Lee. “Stability of Indium Gallium

Zinc Aluminum Oxide Thin-Film Transistors with Treatment

Processes.” Journal of Electronic Materials, vol. 46, no. 2, 21 Oct.

2016, pp. 936–940, https://doi.org/10.1007/s11664-016-4851-4.

Accessed 16 Apr. 2025.

[14] .“Improvement of Self-Heating of Indium Gallium Zinc Aluminum

Oxide Thin-Film Transistors Using Al2O3 Barrier Layer.” Journal

of Electronic Materials, vol. 47, no. 2, 22 Nov. 2017, pp. 1467–1471,

https://doi.org/10.1007/s11664-017-5946-2. Accessed 16 Apr. 2025.

[15] Kim, Donggyu, et al. “Controllable Doping and Passivation of ZnO

Thin Films by Surface Chemistry Modification to Design Low-Cost
and High-Performance Thin Film Transistors.” Applied Surface
Science, vol. 509, Apr. 2020, p. 145289,
76
 https://doi.org/10.1016/j.apsusc.2020.145289. Accessed 1 May 2022.

[16] Ahn, Cheol Hyoun, et al. “Double-Layer Channel Structure Based

ZnO Thin-Film Transistor Grown by Atomic Layer Deposition.”

Physica Status Solidi (RRL) - Rapid Research Letters, vol. 8, no. 4, 7

Mar. 2014, pp. 328–331, https://doi.org/10.1002/pssr.201409044.

Accessed 22 May 2019.

[17] Cao, Fa, et al. “Wide Bandgap Semiconductors for Ultraviolet

Photodetectors: Approaches, Applications, and Prospects.” Research,

23 Apr. 2024, https://doi.org/10.34133/research.0385.

77
 Chapter 6
Conclusion and Future Work
6.1 Conclusion
This study is mainly about AGTO thin films including the way to
fabricate and applications for PDs, TFTs and phototransistors.
Characteristics of this materials are discussed thoroughly. We have
mentioned several ways to change the performance in devices as well
as the characteristics of thin films, such as the double-layer TFTs and
introducing oxygen flow rate.
The AGTO thin films under different parameters are investigated. At
first, we demonstrate the optical, structural and elemental analysis
results of samples. The changed parameters contain SnO2 power and the
oxygen flow rate. As we can see, the AGTO possess high transparency
which is over 90% in the visible light range. We can also get the
bandgap value from Tauc plots. To conclude, there isn’t an obvious
variation under different oxygen flow rate. For changing SnO2 power, a
slightly decrease of bandgap from 5.22 (eV) to 5.02 (eV) when
increasing SnO2 power from 20(W) to 50(W) can be observed. Totally,
the bandgap is around 5 (eV). The XRD spectra shows all of samples
are amorphous, and the AFM images tells that no great change between
different parameters no matter oxygen flow rate or SnO2 power. The
root mean square surface roughness is below 2 (nm). Then, considering
the XPS results, we can find that the oxygen vacancies decreases from
48.11% to 37.11%. Oxygen vacancies are eliminated partly when
introducing oxygen flow rate.
The AGTO MSM PDs are discussed subsequently. In the devices.
The thin films are deposited by RF sputtering. We worked on figuring
out the relation between different SnO2 power, oxygen flow rate and the

78
performance of PDs. When upgrading SnO2 power, the currents
increase. On the other hand, both of dark and photo current decreases
when elevating the oxygen flow rate. When the oxygen flow rate is 4%,
the PD shows the best performance above all the parameters. The photo
and dark current are 4.37×10-6 and 2.46×10-6 respectively. The rejection
ratio is 3.35 and the EQE is 73.6%, which is the largest among all the
samples. At the end, the time-resolved response is shown, and both of
rising and falling time is over 150(s), which means the PD cannot
switch correctly in the short time.
Last but not the least, the AGTO thin-film transistors are produced by
RF sputtering and discussed in this part. First, we mention about two
different parameters that may influence the transistors-changing oxygen
flow rate and SnO2 power. First, changing SnO2 power from 20W to
50W is mentioned, and each of them remains 4% oxygen flow rate. To
conclude, the best performance occurs at 40W/20W, with on/off ratio
2.83×103. Then, we changed the oxygen flow rate from 0% to 6% to
find out the relation between it and performance. According to the
results, it is highly related between both of them. The threshold voltage
decreases from 8.04(V) to -9.18(V) when increasing oxygen flow rate
to 6%. Finally, we can find that the totally best performance takes place
at 40W/20W with oxygen flow rate 4%. Furthermore, in order to
improve the performance, we use double-layer thin-film transistor
schematic to achieve the goal. For the layer which is close the gate, we
deposit 40W/20W with 0% oxygen flow rate AGTO thin film. On the
other hand, we deposit 40W/20W with 4% oxygen flow rate thin film
just under the source and drain electrode. Finally, this method really
works, and it enhances the On/Off ratio from 2.83×103 to 1.13×104,
Simultaneously, the threshold voltage is also decreases from 4.52(V) to
3.87(V). At the end, we use 40W/20W with 4% oxygen flow rate as
79
phototransistor. The threshold voltage shifts to the left side when we
adjusting different wavelength from 500(nm) to 220(nm), which
matches with the theory.

6.2 Future Work

In this study, we have implemented several parameters to control the
performance of AGTO thin-film transistors, including changing SnO2
power and oxygen flow rate. Also, we use the double-layer thin-film
transistor schematic to improve the device. However, there are still
many schematic of thin-film transistors that can try to improve the
performance, such as FinFET and multi-gate, which may show the
better gate control ability.
What’s more, there are many different high-k materials including
HfO2, ZrO2 and so on. As we know, the leakage current problem is
inescapable for the devices, and may be solved by changing the
dielectric layer material. On the other hand, there are still different
structures that can help with upgrading the gate control ability, such as
double-gate thin-film transistor [1] and insert the passivation layer to
elevate the stability [2].
Furthermore, the AGTO may be used on different devices like
resistive random-access memory(RRAM) and gas sensor. Also, the
flexible electronic devices can be an application.

Reference
[1] Li, Yuanyuan V, et al. “Low-Voltage Double-Gate ZnO Thin-Film
Transistor Circuits.” IEEE Electron Device Letters, vol. 34, no. 7,
12June2013,pp.891–893,

80
 https://doi.org/10.1109/led.2013.2263193. Accessed 1 Mar. 2025.
[2] Hwang, Seong-Hyun, et al. “Effects of Al2O3 Surface Passivation
on the Radiation Hardness of IGTO Thin Films for Thin-Film
Transistor Applications.” Applied Surface Science, vol. 578, 1 Dec.
2021,p.152096,
www.sciencedirect.com/science/article/pii/S0169433221031305?
via%3Dihub, https://doi.org/10.1016/j.apsusc.2021.152096.

81