利用共鍍催化金屬與不同間距高度比之奈米碳管柱列改善氣體游離式感測器之特性研究

i

國 立 交 通 大 學

電 子 工 程 學 系 電 子 研 究 所

碩 士 論 文

利用共鍍催化金屬與不同間距高度比之奈米碳管柱列

改善氣體游離式感測器之特性研究

Study on the improvement of carbon nanotube gas ionization

sensors via co-deposited catalyst and pillar array with different

spacer / height ratios

研 究 生：黃 均 宇

指導教授：鄭 晃 忠 博士

ii

利用共鍍催化金屬與不同間距高度比之奈米碳

利用共鍍催化金屬與不同間距高度比之奈米碳

利用共鍍催化金屬與不同間距高度比之奈米碳

利用共鍍催化金屬與不同間距高度比之奈米碳

管柱列改善氣體游離式感測器之特性研究

管柱列改善氣體游離式感測器之特性研究

管柱列改善氣體游離式感測器之特性研究

管柱列改善氣體游離式感測器之特性研究

Study on the improvement of carbon nanotube gas

ionization sensors via co-deposited catalyst and pillar

array with different spacer / height ratios

研 究 生：黃均宇 Student: Chun-Yu Huang

指導教授：鄭晃忠 博士 Advisor: Dr. Huang-Chuang Cheng

國立交通大學

國立交通大學

國立交通大學

國立交通大學

電子工程學系

電子工程學系

電子工程學系

電子工程學系

電子研究所碩士班

電子研究所碩士班

電子研究所碩士班

電子研究所碩士班

碩士論文

碩士論文

碩士論文

碩士論文

A Thesis

Submitted to Department of Electronics Engineering & Institute of Electronics College of Electrical and Computer Engineering

National Chiao Tung University

In Partial Fulfillment of the Requirements for the Degree of Master in

Electronics Engineering 2011

Hsinchu, Taiwan, Republic of China

iii

利用共鍍催化金屬與不同間距高度比之奈米碳管柱列改善

利用共鍍催化金屬與不同間距高度比之奈米碳管柱列改善

利用共鍍催化金屬與不同間距高度比之奈米碳管柱列改善

利用共鍍催化金屬與不同間距高度比之奈米碳管柱列改善

氣體游離式感測器之特性研究

氣體游離式感測器之特性研究

氣體游離式感測器之特性研究

氣體游離式感測器之特性研究

學生:黃均宇 指導教授: 鄭晃忠 博士

國立交通大學

電子工程學系 電子研究所碩士班

摘

摘

摘

摘 要

要

要

要

氣體游離式感測器是一種以氣體分子各自獨特的物理特性來分辨不同氣體 的元件，傳統上，氣體游離式感測器受限於過大的結構（如火焰游離式感測器及 光游離式感測器）、危險的高電壓操作並其伴隨而來的高功率消耗等因素。因此 在本篇論文的實驗中，吾人嘗試利用奈米碳管較低的功函數、尖銳的特點以及在 適當電場下能獲得極佳之游離待測氣體能力與穩定性等來改善氣體游離式感測 器。 在本篇論文實驗的起頭，首先會討論不同表面型態的奈米碳管薄膜所造成的 氣體崩潰特性之差異。由無定向之碳管薄膜的量測結果發現，其崩潰電壓十分不 穩定且在高電壓區域的誤差有將近 100 伏特的變動。這些結果被認為與其表面碳 管的長度不一有很大的關係。因為相對來說，均勻垂直之碳管薄膜就有較穩定的 氣體崩潰特性。但是，對於這兩種表面型態不同的碳管薄膜來說，在經過穩定性 測試的高電壓處理之後，它們的崩潰電壓漂移的情形都十分嚴重。無定向之碳管 薄膜在經過 1000 次重複的穩定性測試之後，其崩潰電壓由起初的 365V 上升到 605V，相當於上升了 68%。而均勻垂直之碳管薄膜在經過相同 1000 次重複的穩

iv 定性測試之後，其崩潰電壓則由 395V 上升到 575V，也就是上升了 45%。並且 我們從掃描式電子顯微鏡圖中可觀察發現，崩潰電壓上升的主要原因跟碳管在高 電壓下會有被拔除與燒結的現象有關。 因此，為了增進奈米碳管氣體游離式感測器之穩定性，吾人嘗試以鈷-鈦催 化劑金屬共鍍的方式來改善碳管與基板之間的附著力及接觸阻抗。並且由實驗的 結果可發現，以此方式合成之碳管薄膜確實有更穩定的崩潰特性，在經過同樣 1000 次重複的穩定性測試之後，其崩潰電壓僅由 375V 上升到 435V，只上升了 16%，與先前兩種碳管薄膜比較起來可說大有改進。 另外，為了改善氣體游離式感測器的功率消耗，減低其操作電壓是首先需要 研究的。在這部份，則使用不同間距高度比的奈米碳管柱列來探討在多少的間距 高度比下有最理想的表面電場分佈，以期達到有最好的拉電子能力並可最早達到 氣體崩潰；也就是有最低的崩潰電壓。在實驗中，吾人嘗可在量測結果的統整中 發現，在間距高度比約 2.91 附近有最低的崩潰電壓。因此此理想的陣列間距高 度比可應用於降低氣體游離式感測器的操作電壓以及功率消耗。 接下來，這些理想化過後的碳管柱陣列被使用來探討在不同氣體環境下的氣 體游離特性。這些不同的氣體因為具有不同的平均自由路徑、游離能及再結合率， 因此會有各自獨特的 Paschen’s curve。利用這些 Paschen’s curve 並加上適當地選 擇氣體壓力與間距的乘積值，則可製作出既操作在低電壓，又能有足夠寬的間隔 來分辨不同氣體的崩潰電壓。最後，吾人探討不同比例的氬氣、二氧化碳與一般 空氣混和之後的崩潰電壓變化：以間距高度比為 2.91 碳管柱陣列為例，當二氧 化碳在空氣中的比例到達 15%時，則崩潰電壓上升會 60V，當氬氣在空氣中的比 例到達 11%時，則崩潰電壓會下降 100V。

v

Study on the improvement of carbon nanotubes gas

ionization sensors via co-deposited catalyst and pillar array

with different spacer / height ratios

Student: Chun-Yu Huang Advisor: Dr. Huang-Chung Cheng

Department of Electronics Engineering & Institute of Electronics

National Chiao Tung University

Abstract

Gas ionization sensors are physical devices that work by fingerprinting the ionization characteristics of distinct gases. Conventional ionization sensors were limited by the huge and bulky architecture (ex: FID, PID), risky high-voltage operation and high power consumption. In this thesis, carbon nanotubes (CNTs) with relatively low work function, extremely sharp nanotips, and structural and chemical stability under high electrical field were therefore used to improve these issues of gas ionization sensors.

In the beginning of this thesis, the effects on gas breakdown characteristics of different surface morphology of CNTs film are presented. For the Random oriented CNTs film, the variations of the breakdown voltages are especially large at high voltage region and their error bars in the high voltage region are as wide as 100 volts. These variations are associated with the nonuniformity of the CNTs’ length. On the other hand, the gas breakdown characteristics of the Uniform CNTs film were relatively stable from the measurement results. However, for both of the two samples,

vi

the shift-up of their breakdown voltages (Vbr) were fairly severe after the high-voltage process in stability tests. One could find that the Vbr of the Random oriented CNTs film lifts up from 365V to 605V after 1000 cycles, i.e., 68% increase. And for the Uniform CNTs film, it lifts up from 395V to 575V after 1000 cycles, i.e., 46% increase. Observed from the SEM images, the pull-off and evaporation of CNTs resulted from the high local electric field difference were considered as the main reason for the shift-up of breakdown voltages.

In order to acquire a better stability in the CNTs gas ionization sensor, the improvement of the adhesion and the contact resistance between CNTs and substrate under high electric field was obtained using Co-Ti co-deposited catalyst structure. The Vbr of the CNTs film synthesized from Co-Ti co-deposited catalyst lifts up from 375V to 435V after 1000 cycles, i.e., only 16% increase, which is much more reduced than that of the first two conventional CNTs film.

In addition, to improve the issue of high power consumption, pillar arrays of vertical aligned CNTs bundles with different spacer height ratios (R/H) were utilized to investigate the optimal local electrical field on the nanotubes that has the most efficient field emission, namely, the earliest gas breakdown and lowest breakdown voltage. In this thesis, the lowest breakdown voltages were approached by changing H while maintaining R and the optimal R/H ratio was around 2.91. This optimal R/H ratio would lessen the high operating-voltage and thus improve high power consumption issues of the ionization sensors.

Next, the optimized samples were exploited to explore their gas ionization characteristics under different gases environment. From the experiment, dissimilar trends of Paschen’s curve for distinct gases was obtained due to that different gas molecules have different mean free path, ionization energy and recombination rate. With a proper selection of the p×d product value, CNT gas ionization sensor can not

vii

only operate under low voltages but also provide enough space to distinguish between different gases.

Finally, the breakdown voltages of Ar and CO2 gases in mixture with air as a

function of concentration were investigated. Take the R/H = 2.91 optimized patterned sample for example. It was found that the Vbr increases 50V as the concentration of CO2 in the mixture with air reaches 15 %, and decreases 100V as the concentration of

viii

Acknowledgments

一眨眼碩士班兩年的時間就這樣飛過去了，在即將完成這篇論文的同時心中 浮現了許多想要感謝的人事物，首先最要感謝的是不辭辛勞費心指導我的恩師鄭 晃忠老師，老師不僅在學術上給予許多寶貴的意見，也在為人處世上處處提點學 生，尤其在我碩一為是否找尋共同指導教授時，老師的真知灼見也一針見血的讓 學生明白了問題的癥結點，能夠順利完成碩士學業，實在是要感謝老師的諄諄教 誨。 再來要感謝 309B 的學長姐、同學、學弟妹們，不管是逸哲、柏宇、昱智、 宏顯、還是昭龍等各位學長，雖然我們不是同一個組，但在大咪時你們的意見都 使這篇論文更加的完全，還要謝謝俊諭學長許多學術外的看法，每次和學長聊天 都讓學弟我獲益良多。謝謝佳信常常幫忙我這愚拙的實驗室管理員，我實在應該 要把薪水分你一半才對。也謝謝柏鈞在即將畢業的這段時間裡幫忙我許多。再來 要感謝奈米組的各位，首先謝謝萬霖學長給予本論文多有專業的見解，許多的問 題也在和學長的討論中才豁然開朗。接著感謝奈米組兩朵花，筠珊學姐及育荏， 謝謝你們總是帶給大家歡樂，看見你們我總是能不自覺得嶄露笑容。也要感謝即 將一起畢業的俊賢，那些在實驗中遇到的困難擊倒不了我們的，一起努力吧。還 有冠宇跟湛宇，在二樓孤單的日子有你們陪伴真好，你們在研究上認真反倒是我 的榜樣，還有瑋萱、桓民，加油吧，你們一定也可以順利畢業的！ 還要謝謝指導我實驗並提供意見的加聰學長，我是個個性頑固的人，也常常 不尊重學長，感謝學長許多的包容，不僅在研究上一步一步的帶領著我，還要忍 受我一些不合理的作法跟主意，感謝學長不計較這些，還繼續不斷帶領著我。 並且要感謝我的家人，我親愛的父親黃明國先生跟三個弟弟，俊絪、俊福、 偉峰，和姑姑、姑丈及其他所有的家人，你們的全力支持是我完成碩士學業的最 大後盾。還有屬靈上的家人，新竹市召會的弟兄姊妹們，謝謝你們在主裡的扶持，

ix 特別是 409 區的弟兄姊妹們，謝謝你們把我當親生兒子一樣的照顧，我永遠會想 念在這裡的教會生活。希伯來書 12:1「所以，我們既有這許多的見證人，如同 雲彩圍著我們，就當脫去各樣的重擔，和容易纏累我們的罪，憑著忍耐奔那擺在 我們前頭的賽程。」 最後，最要感謝的，是那以永遠的愛愛我的神-親愛的主耶穌，謝謝主在我 還年輕的時候就來到我的生命中，使我得以享受祂一切的豐富。這份信仰伴隨著 我走過高山低谷，不管是身體上的軟弱或心靈上的虛空，親愛的主都陪伴著我渡 過。我知道，未來的人生祂也會陪伴我走下去。我所有的，所得的，都是恩典。 願一切的榮耀和讚美都歸給祂，主必擴增，我必衰減，感謝主，阿們。 哥林多後書 4：16~18「所以我們不喪膽，反而我們外面的人雖然在毀壞， 我們裡面的人卻日日在更新。因為我們這短暫輕微的苦楚，要極盡超越的為我們 成就永遠重大的榮耀。我們原不是顧念所見的，乃是顧念所不見的，因為所見的 是暫時的，所不見得才是永遠的。」

x

Contents

Abstract (in Chinese)...i

Abstract (in English)...iii

Acknowledgments (in Chinese)...vi

Contents...viii

Table Lists...xii

Figure Captions...xii

Chapter 1：

：

：

：Introduction

1-1 Overview of Gas sensors………1

1-2 Overview of Carbon nanotube Gas sensors……….4

1-2-1 Structure and properties of Carbon nanotubes………..4

1-2-2 The synthesis methods of Carbon nanotubes………6

1-2-3 Carbon nanotubes as the Chemical gas sensor………..7

1-2-4 Carbon nanotubes as the Gas ionization sensor……….8

1-3 Theory background………9

1-3-1 The mechanism of electron emission………....9

1-3-2 Electron field emission………...11

1-3-3 Operation principles of electron impact ionization………13

1-3-4 The fundamental mechanism of gas ionization breakdown………...14

xi

1-4 Motivation……….……19

1-4-1 Stability issue………..………19

1-4-2 The reduction of breakdown voltage………..…21

1-5 Thesis organization………..…22

Chapter 2：

：

：Experiment

：

2-1 Experimental procedures………25

2-1-1 Sample fabrication and CNTs synthesis……….…25

2-1-2 Material analysis and Breakdown characteristics measurement…………28

2-1-3 Stability test………29

2-1-4 Gas sensing measurement ………..……29

2-2 Experimental design………29

2-2-1 The comparison between the Uniform CNTs film and the Random Oriented CNTs film………..……29

2-2-2 Effects of the utilization of co-deposition catalyst structure………..30

2-2-3 Finding optimum spacer / height ratio of the pillar array of vertical aligned CNTs bundles ………..…30

xii

Chapter 3：

：

：Results and Discussions

：

3-1 The comparison of breakdown characteristics between the Uniform CNTs film and the Random Oriented CNTs film……….32

3-1-1 The Gas breakdown characteristics of the Random oriented CNTs film and the uniform CNTs film………...32 3-1-2 The stability test of the Random oriented CNTs film and the uniform CNTs film……….35 3-2 The stability improvement of the gas breakdown characteristics using CNTs

film with co-deposition catalyst structure……….….37 3-2-1 Enhancing the adhesion and contact resistance between CNTs and

substrate using co-deposited catalyst structure……….37 3-2-2 The improvement of stability and gas breakdown characteristics using

CNTs-based film synthesized from the co-deposited catalyst structure...38 3-3 Reduction of the breakdown voltages using pillar array of vertical aligned

CNTs bundles with different spacer height ratios……….39 3-3-1 Finding the optimal R/H ratio of pillar arrays of vertical aligned CNTs

bundles that has lowest gas breakdown voltages. ………..….39 3-3-2 Discussions of the gas breakdown characteristics of patterned sample...42 3-4 The gas breakdown characteristics of the optimized CNTs film gas ionization sensors under different gas environments……….……….44

3-4-1 The Paschen’s curve of different gases and its application to gas

ionization sensors ………44 3-4-2 Breakdown characteristics of carbon dioxide and argon in a mixture with

xiii

Chapter 4：

：

：

：Summary and Future prospects

4-1 Summary and conclusions………..…….48

4-2 Future prospects………..………….50

Figures

………..52

References

………..119

xiv

Table Lists

Chapter 1

Table 1-1 Comparison between analytical instruments and gas sensors………..…52

Chapter 2

Table 2-1 The parameters of thermal CVD to grow the Uniform CNTs film and the Random oriented CNTs film………52

Chapter 3

Table 3-1 Gas breakdown voltages of the Random oriented CNTs film……....…..53 Table 3-2 Gas breakdown voltages of the Uniform CNTs film………....53 Table 3-3 The parameters of thermal CVD to grow pillar arrays of vertical aligned

CNTs bundles with different spacer height ratio (R/H)……….…….54

Figure Captions

Chapter 1:

Figure 1-1 Trends in global markets for gas sensors [Frost and Sullivan ＆ BCC Research]……….55 Figure 1-2 A gas sensor is a device which outputs the appropriate signals for

detection and measurement when specific gas was released…………..55 Figure 1-3 Schematic cross section of the FET devices. A single nanotube of either

multi-wall or single-wall type bridges the gap between two gold

electrodes. The silicon substrate is used as back gate [2]………56 Figure 1- 4 Histogram detailing the number of CNT publications per year between

xv

1991 and 2007 (data obtained from ISI Web of Knowledge) [7]………56 Figure 1- 5 High-resolution transmission electron microscopy images of (a) SWNTs,

and (b) MWNTs. Every layer in the image (fringe) corresponds to the edges of each cylinder in the nanotube assembly [9]………..57 Figure 1- 6 Molecular models of SWNTs with (a) chiral vector (b) the categories of

the configuration [12][13]………...57 Figure 1-7 Energy band diagrams of vacuum-metal boundary (a) electron tunneling

via thermionic emission and (b) electron tunneling via field emission...58 Figure 1-8 Energy diagrams of vacuum-metal boundary: (a) without external

electric field; and (b) with an external electric field………59 Figure 1-9 (a) Localized discharge spot can be seen at the gap with applying low

voltage, (b) plane-to-plane silicon electrodes with narrow gap. (c)

micro-discharge in the gap of a comb actuator [39]………60 Figure 1-10 The breakdown discharge process of gas ionization sensor with CNTs

as positive electrode………...………...61 Figure 1-11 The breakdown discharge process of gas ionization sensor with CNTs

as negative electrode………..61 Figure 1-12 Schematic initially very small amount of free electrons, accelerated by a

sufficiently strong electric field, give rise to electrical conduction through a gas by avalanche multiplication………....62 Figure 1-13 The Amplified mechanism of the electron flux in Townsend’s

discharge………...62 Figure 1-14 The Paschen’s curve for air, two flat parallel copper electrodes,

separated by 1 inch, for pressure between 3×10-2 torr and 760 torr…..63 Figure 1-15 The carbon nanotubes gas ionization sensor proposed by Modi et al...64

xvi

Figure 1-16 The carbon nanotubes gas ionization sensor proposed by S J Kim et al……….………..64

Chapter 2:

Figure 2-1 The schematic flowchart for the fabrication of the Uniform CNTs film and Random oriented CNTs film……….65 Figure 2-2 (a) Schematic picture and (b) photograph of thermal CVD. The process

gases used here is hydrogen, nitrogen and ethylene………66 Figure 2-3 The process parameters to synthesize CNTs………...…67 Figure 2-4 The schematic flowchart for the fabrication of CNTs-based film

Synthesized from the Co-deposited Catalyst………..68 Figure 2-5 The schematic flowchart for the fabrication of pillar array of vertical

aligned CNTs bundles using co-deposited catalyst structure…………..69 Figure 2-6 Mask design shows the array of 80µm in inter-pillar distance and 50µm

in circle diameter defined in 1cm × 1 cm area………70 Figure 2-7 The micrographs of samples were taken by Scanning electron

microscope (SEM, Hitachi S-4700I)………...70 Figure 2-8 The high resolution transmission electron microscope (HRTEM; JEOL

JEM-2000EX) was used to examine the structure of CNTs………71 Figure 2-9 High Resolution Confocal Raman Microscope (HOROBA, Lab RAM

HR) was also applied to analyze the crystallinity of the CNTs………..72 Figure 2-10 The gas ionization sensor measurement setups……….73 Figure 2-11 Definition of the R/H ratio, where R is the inter-pillar distance (spacer)

xvii

Chapter 3:

Figure 3-1 The pictures shown here was the tilted images (about 45°) taken by scanning electron microscope (SEM) for the Uniform CNTs film grown by thermal CVD………..75 Figure 3-2 The pictures shown here was the cross-sectional image taken by SEM for

the Uniform CNTs film grown by thermal CVD. The height of the vertical aligned CNTs is about 11.9µm………...75 Figure 3-3 The pictures shown here was the tilted images (about 45°) taken by SEM for the Random oriented CNTs film grown by thermal CVD………….76 Figure 3-4 The pictures shown here was the magnified (10000 times) tilted images

(about 45°) taken by SEM for the Random oriented CNTs film grown by thermal CVD………...76 Figure 3-5 The pictures shown here was the cross-sectional image taken by SEM for the Random oriented CNTs film grown by thermal CVD………..77 Figure 3-6 The Raman spectra analysis of the Uniform CNTs film and the ID/IG

ratio which indicates the graphite crystallinity of the samples is 1.753 and 1.725……….77 Figure 3-7 The Raman spectra analysis of the Random oriented CNTs film and the

ID/IG ratio which indicates the graphite crystallinity of the samples is

1.789 and 1.755………...78 Figure 3-8 (a)The micrographs of CNTs taken by Transmission electron microscopy (TEM) obtained from the Uniform CNTs film and (b) their multiwalled structure can be found with higher resolution……….78 Figure 3-9 (a)The micrographs of CNTs taken by TEM obtained from the Random

oriented CNTs film and (b) their multiwalled structure can also be found with higher resolution………..79

xviii

Figure 3-10 The gas breakdown characteristics of the Radom oriented CNTs film under different gas pressures in the nitrogen environment. The

pressures are (a) 0.0023 torr, (b) 0.0052 torr, (c) 0.079 torr, (d) 0.1 torr, (e) 0.2 torr, (f) 0.51 torr, (g) 0.8 torr, (h) 1.0 torr, (i) 2.1 torr and (j) 5.1 torr. They are integrated into (k)………81 Figure 3-11 Breakdown voltages vs. p×d characteristics of the Random oriented

CNTs film under nitrogen environment (Paschen’s curve)…………...81 Figure 3-12 The gas breakdown characteristics of the Uniform CNTs film under

different gas pressures in the nitrogen environment. The pressures are (a) 0.0024torr, (b) 0.0049torr, (c) 0.078torr, (d) 0.1 torr, (e) 0.21torr, (f) 0.51 torr, (g) 0.82torr, (h) 1.0 torr, (i) 2.0torr and (j) 4.9torr. They are integrated into (k)………..83 Figure 3-13 Breakdown voltages vs. p×d characteristics of the Uniform CNTs film

under nitrogen environment (Paschen’s curve)……….84 Figure 3-14 The stability test of gas breakdown characteristics under nitrogen

environment with the Random oriented CNTs film and the Uniform CNTs film………..85 Figure 3-15 The SEM images before and after stability tests: (a), (c) and (d) are the

images of the Random oriented CNTs film before stability test, after 500 cycles stability tests and after 1000 cycles stability tests. And (b), (d) and (f) are the images of the Uniform CNTs film before stability test, after 500 cycles stability tests and after 1000 cycles stability tests…...86 Figure 3-16 The various surface energy of different metals, where one can find that

the surface energy of cobalt is familiar with that of titanium…………87 Figure 3-17 The diagram of different surface energy metals reacting with Cobalt as

xix

Figure 3-18 The images of SEM displayed the roots of the CNTs for both (a) the conventional samples without co-deposited catalyst and (b) the proposed samples with co-deposited catalyst. The proposed samples were cleaved across the patterned area and a CNT immersed partially into the co-deposited metal layer on the cleaved edge was marked by a circle in (c).

Figure 3-19 The catalyst after pretreatment in reducing gas environment: (a) without Al supporting layer and (b) with Al supporting layer, where

nanoparticles with small sizes could be achieved with Al supporting layer………...89 Figure 3-20 The Transmission Electron Microscopy (TEM) images of (a) using

Co/Ti/Al catalyst structure and (b) using Co-Ti/Al co-deposited catalyst structure. It’s obvious that the diameter of CNTs becomes smaller by using co-deposited catalyst structure……….90 Figure 3-21 The gas breakdown characteristics of the CNTs-based film synthesized

from the co-deposited catalyst structure under different gas pressures in the nitrogen environment. The pressures are (a) 0.0021 torr, (b) 0.0049 torr, (c) 0.082 torr, (d) 0.1 torr, (e) 0.18 torr, (f) 0.41 torr, (g) 0.8 torr, (h) 1.0 torr, (i) 2.1 torr and (j) 5.3 torr. The above is integrated into (k)….92 Figure 3-22 Breakdown voltages versus p×d characteristics of the CNTs-based film

synthesized from the co-deposited catalyst structure under nitrogen environment (Paschen’s curve)………..93 Figure 3-23 The stability test of gas breakdown characteristics under nitrogen

environment with the Random oriented CNTs film, the Uniform CNTs film and the CNTs film with co-sputter catalyst………...94 Figure 3-24 The SEM images before and after stability tests: (a), (b) and (c) are the

xx

images of the CNTs film synthesized from Co-Ti co-deposited catalyst structure before stability test, after 500 cycles stability tests and after 1000 cycles stability tests, respectively……….96 Figure 3-25 Simulation of the equipotential lines of the electrical field for tubes of

different distances between each other………..97 Figure 3-26 (a) Simulation of the equipotential lines of the electrical field for tubes

of 1 µm height and 2 nm radius, for distances of 4, 1, and 0.5 µm between tubes; along with (b) the corresponding changes of the field enhancement factor β and emitter density, and (c) current density as a function of the distance……….98 Figure 3-27 Definition of the R/H ratio, where R is the inter-pillar distance (spacer)

and H is the height of a CNT pillar………...99 Figure 3-28 The cross-sectional view of CNTs film synthesized from Co-Ti

co-deposited catalyst structure for 50.8 µm in height………..99 Figure 3-29 The pictures shown here were the tilted image (a) (about 45°)and the

cross-sectional image (b) taken by the SEM for the 48.8 µm high pillar-like CNTs synthesized from the Co-Ti co-deposited catalyst structure grown by thermal CVD. Here the R/H ratio is 0.61……….100 Figure 3-30 Breakdown voltages vs. p×d characteristics of (a) the film sample (50.8

µm) and (b) the pattern sample (48.8 µm, R/H =0.61) under nitrogen

environment (Paschen’s curve)………102 Figure 3-31 The cross-sectional view of CNTs film synthesized from Co-Ti

co-deposited catalyst structure for 11.3 µm in height………..103 Figure 3-32 The pictures shown here were the tilted image (a) (about 45°)and the

cross-sectional image (b) taken by the SEM for the 12.5 µm high pillar-like CNTs synthesized from the Co-Ti co-deposited catalyst

xxi

structure grown by thermal CVD. Here the R/H ratio is 2.40……….104 Figure 3-33 Breakdown voltages vs. p×d characteristics of (a) the film sample (11.3

µm) and (b) the pattern sample (12.5 µm, R/H =2.40) under nitrogen

environment (Paschen’s curve)………105 Figure 3-34 The cross-sectional view of CNTs film synthesized from Co-Ti

co-deposited catalyst structure for 13.1 µm in height……….106 Figure 3-35 The pictures shown here were the tilted image (a) (about 45°)and the

cross-sectional image (b) taken by the SEM for the 10.3µmhigh pillar-like CNTs synthesized from the Co-Ti co-deposited catalyst structure grown by thermal CVD. Here the R/H ratio is 2.91……….107 Figure 3-36 Breakdown voltages vs. p×d characteristics of (a) the film sample (13.1

µm) and (b) the pattern sample (10.3µm, R/H =2.91) under nitrogen

environment (Paschen’s curve)………108 Figure 3-37 The cross-sectional view of CNTs film synthesized from Co-Ti

co-deposited catalyst structure for 5.95 µm in height……….109 Figure 3-38 The pictures shown here were the tilted image (a) (about 45°)and the

cross-sectional image (b) taken by the SEMfor the 5.95µmhigh pillar-like CNTs synthesized from the Co-Ti co-deposited catalyst structure grown by thermal CVD. Here the R/H ratio is 5.04……….110 Figure 3-39 Breakdown voltages vs. p×d characteristics of (a) the film sample (5.95

µm) and (b) the pattern sample 5.95 µm, R/H =5.04) under nitrogen

environment (Paschen’s curve)………111 Figure 3-40 The comparison of the breakdown voltages characteristics of the film

samples with different height………...112 Figure 3-41 The comparison of the breakdown voltages characteristics of the pattern samples with different R/H ratio………..113

xxii

Figure 3-42 (a)The local electrical field distribution and (b) the emission corner of the film sample and the patterned sample for a constant

anode-to-cathode distance and the same gas pressure……….114 Figure 3-43 Corresponding changes of the field enhancement β as a function of the

R/H ratio when considering (a) the screening effect, (b) the aspect ratio effect and (c) both of the two effects………..115 Figure 3-44 The breakdown characteristics of different gases for (a)the film sample with 11.3 µm in CNTs’ height and (b)the patterned sample with R/H = 2.91………..117 Figure 3-45 Discharge current versus breakdown voltage curves for Ar, N2, Air, O2

and CO2 of the film sample with 11.3 µm in CNTs’ height at p×d

product value around (a)8×10-4 torr cm and (b) 8×10-3torr cm, showing distinct breakdown voltages; carbon dioxide displays the highest and argon the lowest………..118 Figure 3-46 Discharge current versus breakdown voltage curves for Ar, N2, Air, O2

and CO2 of the patterned sample with R/H = 2.91 at p×d product value

around (a)8×10-4 torr cm and (b) 8×10-3torr cm, showing distinct breakdown voltages; carbon dioxide displays the highest and argon the lowest………...119 Figure 3-47 Breakdown voltages of Ar and CO2 gases in mixture with air as a

function of concentration for (a) film sample with 11.3 µm in CNTs’ height and (b) patterned sample with R/H = 2.91………...120

Chapter 4:

Figure 4-1 Extension of the linear region in the right side of the Paschen’s

1

Chapter 1

Introduction

---

1-1 Overview of Gas sensors

In recent age, interests in industry safety and environmental pollution have been growing in our life. The needs of prevention and control of air pollution and detection for toxic gases have gradually increased. Thus, researches and developments on the gas sensors are carried out rapidly. According to Frost and Sullivan, global gas sensor markets were worth $48.5 million in 2005; their forecast for 2012 is $80.6 million. And based on a new technical market research report (BCC Research) (Fig. 1-1), the global market for gas sensors and gas metering is worth an estimated $3.9 billion in 2010, but is expected to increase to nearly $5.2 billion in 2015, for a 5-year compound annual growth rate (CAGR) of 5.9%.These data show how large, and still expanding, this market is. A gas sensor is a device which outputs the appropriate signals (the change of current, voltage and conductance) for detection and measurement when specific gas was released (Fig. 1-2). In our daily life, most of the gases are colorless and smell-less. Moreover, the sense of smelling of human beings was not able to identify the amount and content of certain gases. Hence, people always become conscious that some toxic and harmful gases run out due to the inhalation of overdose that causes uncomfortableness. Therefore, an accurate, fast-reacting and stable gas sensor plays an important role to improve the safety in our life.

In the analysis of gases, conventional analytical instruments have many usage restrictions while gas sensor was more convenient. As shown in Table. 1-1, the data

2

precision of analytical instrument is absolute value, yet its data acquisition time is too long and the size is much larger, which means the low-cost gas sensor attracts much attention from researchers.

Generally, gas sensor was classified into the following kinds: 1) Catalytic combustion gas sensor,

2) Semiconductor-absorbing Gas sensor, 3) Electrochemical Gas sensor,

4) Field-effect transistor Gas sensor, 5) Infrared Gas sensor and

6) Gas ionization sensor.

The first three kinds of gas sensors must be heated to high temperature before detecting gas leakage. Besides, they have different reactive sensitivities and response time of distinct gases, which cause mistakes or inability in sensing specific gases.

The mechanism of Field-effect transistor Gas sensor is as follows. When target gas contacts with catalyst metals, chemical reaction occurs and the product (take H2

for example) will tunnel through the catalyst metal layer to affect charge density in the channel. Sander J.Tans et al. [1] and R.Marel et al. [2] have used carbon nanotubes (CNTs) as the channel of field-effect transistor (Fig. 1-3). By applying a voltage to a gate electrode, the nanotube can be switched from a conducting state to an insulating state. Generally, Source and Drain were fabricated by noble metals like Pt and Au, and doped Si was used as the back gate of transistor. Using the gate electrode, the conductance of a SWNT-FET could be modulated by more than 5 orders of magnitude. The sensitivity of this kind of gas sensor is three times higher than that of common gas sensors and the former has a good response time of two to ten seconds. However, there are still many limitations on Field-effect transistor Gas sensor. For example, it

3

takes high temperature to recover to the initial resistance after each gas sensing measurement. Also, the fabrication process is more complicated, and the improved response time is still too long to detect specific toxic gases.

The Infrared gas sensor, however, measures target gases by determining the absorption of an emitted infrared light source through a certain air sample [3]. Infrared gas analyzers usually have two chambers that one is a reference chamber while the other is a measurement chamber. Infrared light is emitted from some type of source on one end of the chamber, passing through a series of chambers that contain given quantities of the various gases in question. Target gases found in the atmosphere get excited under specific wavelengths found in the infrared range. The concept behind the technology can be understood when considering the greenhouse effect. As sunlight hits the earth surface, the incoming short wave radiation gets turned into long wave infrared radiation that is reflected back into space. If the planet has a thick atmosphere, much of this radiation is absorbed by the "greenhouse gases" in the atmosphere which acts as an isolative blanket. The infrared gas sensor operates on the similar principle. However, the response time of infrared gas sensor is still quite long. A gas sensor with very short response time was required to detect some poison gases.

Among all of these sensors, Gas ionization sensor meets the requirement of response time with a stable discharge current in microseconds to ppm-level gases by its breakdown effect. It shows good sensitivity and selectivity, and is unaffected by extraneous factors such as temperature, humidity, and gas flow [4]. The fundamental operation principle of Gas ionization sensor is as follows. When applying a high voltage between two parallel electrodes, the electrical-field-induced band bending of the Vacuum level forces electrons to tunnel through the electrode material onto the Vacuum level which is the so-called Field emission effect.Since there are neutral molecules in the path of electrons which are moving to the positive electrode by the

4

traction of electric field, impact ionization reaction would take place when the electrons receive enough kinetic energy to ionize the molecule from the electrical field. A positive gas ion and a negative electron would be generated after an effective impact ionization process, and the number of electrons gets doubled after each time of effective collision. The whole reaction could be described via (eq. 1-1) :

e− +X₂ →X₂+ +2e− (1-1) The number of charged particles between two electrodes would increase rapidly when the reaction repeats perpetually. Once it reaches a certain amount, the original poor electrical conducting gases would turn out to cause the electrical breakdown and an unusual high current could be measured afterward. Due to different molecular physical properties of different gases, distinct breakdown voltage could be obtained when breakdown effect occurs.Although gas ionization sensor work by fast response time and fingerprinting the ionization characteristics of distinct gases, they are, however, limited by their huge architecture, risky high-voltage operation and high power consumption. Hence, the miniaturization of device size, the reduction of breakdown voltage, and the improvement of power consumption are the most important issues for this kind of gas sensor.

1-2 Overview of Carbon nanotubes Gas sensors

1-2-1 Structure and properties of Carbon nanotubes

Since the discovery of carbon nanotubes (CNTs) by Iijima in 1991, [5] CNTs have attracted considerable interests because of their unique physical properties and many potential applications [6]. From 1991 until the end of 2007, roughly 30 000 scientific reports have been published on this topic [7]. This explosion of CNT reports is

5

illustrated in (Fig. 1-4) with a histogram detailing the number of CNT publications per year. This large volume of CNT literature includes over 1200 reports that deal specifically with the application of CNTs in a sensing capacity. CNTs have numerous potential applications in nanoelectronics, nanometer-scale structural materials, hydrogen storage, field-emission devices, gas sensors, and so on. Among these applications, CNTs seem to be very promising as electron emitters.

CNTs can be divided into two categories. The first is called multi-walled carbon nanotubes (MWNTs). MWNTs are close to hollow graphite fibers [8], except that they have a much higher degree of structural perfection. They are made of sheets of carbon atoms with a cylindrical shape and generally consist of co-axially arranged 2 to 20 cylinders (Fig. 1-5 (b)). The interlayer spacing in MWNT (d(002) = 0.34 nm) is slightly

larger than that in single crystal graphite (d(002) = 0.335 nm) [9]. This is attributed to a

combination of tubule curvature and van der Waals force interactions between successive garphene layers. The second type of the nanotube is made up of only a single layer of carbon atoms. These nanotubes are called the single-walled nanotubes (SWNTs) and possess good uniformity in diameter about 1.2 nm (Fig. 1-5 (a)). They are close to fullerenes in size and have a single-layer cylinder extending from end to end [10,11].

Most experimentally observed CNTs are multi-walled structures with outer most shell diameters exceeding 10 nm. Since current conduction in a MWNT is known to be mostly confined to the outermost single-walled nanotube and since band gap of a SWCNT varies inversely with its diameter, MWNTs are metallic in nature. SWNTs can be either metallic or semiconducting depending on the way the roll-up of the graphene sheet occurs - an aspect termed as Chirality, and if all the roll-up types are realized with equal probability, 1/3 of the SWNTs end up being metallic and 2/3 semiconducting. The structure of a SWNT can be conceptualized by wrapping a

6

one-atom-thick layer of graphite called graphene into a seamless cylinder. The way of the graphene sheet is wrapped is represented by a pair of indices (n,m) called the chiral vector. The integers n and m denote the number of unit vectors along two directions in the honeycomb crystal lattice of graphene. If m=0, the nanotubes are called "zigzag". If n=m, the nanotubes are called "armchair". Otherwise, they are called "chiral". (Fig. 1-6) depicts these structures of a SWNT [12,13].

CNTs have been attracting much attention for their unique physical and chemical properties such as high mechanical strength, chemical stability, high aspect ratio, super-thermal conductivity, and electron emission properties [14,15]. CNTs could be one of the strongest and stiffest materials known, in terms of tensile strength and elastic modulus respectively. This strength results from the covalent sp2 bonds formed between the individual carbon atoms. The highest tensile strength an individual multi-walled carbon nanotube has been tested to be is 63 GPa [16]. Under excessive tensile strain, the tubes will undergo plastic deformation, which means the deformation is permanent. This deformation begins at strains of approximately 5% and can increase the maximum strain the tube undergoes before fracture by releasing strain energy. For the thermal conductivity of CNTs , it is predicted that carbon nanotubes will be able to transmit up to 6000 watts per meter per kelvin at room temperature; compare this to copper, a metal well-known for its good thermal conductivity, which only transmits 385 W/m/K. The temperature stability of carbon nanotubes is estimated to be up to 2800 degrees Celsius in vacuum and about 750 degrees Celsius in air [17].

1-2-2 The synthesis methods of Carbon nanotubes

Carbon nanotubes (CNTs) have been extensively investigated for the synthesis using arc discharge, laser vaporization, pyrolysis, solar energy, and plasma-enhanced

7

chemical vapor deposition (CVD), for its unique physical and chemical properties and for applications to nanoscale devices. However, common methods of CNT synthesis include: (1) arc-discharge [18], (2) laser ablation [19], (3) thermal CVD [20-22], and (4) plasma enhanced CVD [23,24].

The laser ablation can synthesize pure carbon nanotubes in high fabrication temperature, but large scale display panel cannot be fabricated in the high fabrication temperature above the melting point of glass substrate. The arc discharge can synthesize carbon nanotubes in shorter fabrication times, but it has some issues, such as (1) poor purity, (2) hard to control growth orientations of carbon nanotubes, and (3) poor emission uniformity.

Compared to laser ablation and arc discharge, using CVD for carbon nanotube growth has some features, such as (1) high purity carbon nanotubes, (2) selective growth only for catalyst metal, (3) controlling growth direction, and (4) much suitable to semiconductor fabrication procedure.

1-2-3 Carbon nanotubes as the Chemical gas sensor

Carbon nanotubes (CNTs) are expected as a new material which has an outstanding high sensitivity and selectivity at relatively low temperature with a fast response time. Several kinds of gas sensors using CNTs have been proposed. They can be generally classified into two types: the absorption type gas sensor (chemical type) [25,26] and the ionization type gas sensor (physical type) [27].

The mechanism of a chemical type gas sensor using CNTs as the gas absorption material is based on the fact that CNTs show its semiconductor property, which their electrical resistance can be modified by the charge transfer between CNTs and

oxidizing or reducing gas molecules absorbed on the CNT surfaces. Owing to the high surface area, nanometer size and hollow centre of CNTs, it seems to be an ideal

8

material for absorption and detection of gases. Nonetheless, this type of CNTs gas sensor has some limitations [28] such as the inability to detect gases with low absorption energy or low electro-negativity like inert gases. It’s also difficult to distinguish between gases and gas mixtures with electrical resistance. (Gases in different concentrations could produce the same amount of net change in conductance as produced by a single gas.) In addition, nanotube conductance is very sensitive to environmental conditions like temperature, moisture and gas-flow velocity. Besides, chemisorptions could cause irreversible change in nanotube conductivity. All the limitations mentioned above make us pay more efforts in selecting ionization type gas sensor in order to overcome these disadvantages.

1-2-4 Carbon nanotubes as the Gas ionization sensor

The Gas ionization sensors working by fingerprinting the ionization characteristics of distinct gases that can ignore the magnitude of gas absorption energy have been reported recently [29]. The sharp tips of CNTs generate very high electrical fields at relatively low voltage, about several hundred volts, which is several-fold lower than the traditional metal electrodes [30]. The most crucial part of the operating mechanism of this type of gas sensor is the Field emission. And the field emitted electrons from a cold cathode have often played an important role in vacuum electronics as an alternative to the thermionic emitted electrons from a hot cathode with advantages such as higher efficiency, less scattering of emitted electrons, faster turn-on times and more compactness. As we have mentioned above, the Gas

ionization sensor suffers from their risky high-voltage operation, high power consumption and huge architecture. However, with the employment of CNTs as the electrode, the sharp tips of CNTs can generate very high electric fields at relatively low voltage that hastens the breakdown process. CNTs also have high chemical

9

stability, high emission current density and good emission stability that make it an ideal material to fabricate the Gas ionization sensor.

Recently, this physical type CNT gas sensors has been proposed, which indicates that CNT gas ionization sensors (with proper calibration) show promising potential for room-temperature gas detection at extremely low percentage in mixtures with air and for fast response toward the application of the breakdown electric field resulting in a stable discharge within 20 µs [29,31].

1-3 Theory background

When using CNTs as electrode material of the Gas ionization sensor, the first mechanism needed to be realized is electron field emission. Electrons would receive enough energy to tunnel through the bended potential barrier caused by the high electric field around sharp tips of CNTs. After their tunneling, electrons might have the impact ionization with neutral molecules and generate lots of charged particles between two electrodes. And electrical gas breakdown would take place when achieving enough amounts of electrons. The total operation principles of gas ionization sensor can be described through the derivation of Townsend’s discharge [32,33] and Paschen’s law [34,35], which illustrate the relationship between Gas ionization sensor breakdown voltages and the gas pressure along with the distance between anode to cathode.

1-3-1 The Mechanism of electron emission

Generally speaking, it’s not straightforward for an electron to escape from the surface of materials. A potential barrier (so-called Fermi level) exists at the surface of materials which prevents the electrons from escape unless certain conditions are satisfied so that electrons upon the metal could emit into the surrounding vacuum or

10

gases environment. That is, the binding electron is not able to flee away from the binding energy band of the atom’s surface unless sufficient external energy is provided. The supplied external energy must be higher than the work function of electrode surface material (defined as the work needed to remove an electron from the surface). As electrons receive specific external energy to cross the potential barrier of the work function and reach the vacuum, this phenomenon is called electron escape or electron emission.

In general, the electron emission mechanism can be classified into two types, thermionic emission and field emission. Their operating mechanisms are to use either temperature or electrical field to provide the external energy of electron emission, respectively. These two mechanisms can be described by the band diagram (Fig. 1-7).

Thermionic emission is the heat-induced flow of charge carriers from a surface or over a potential-energy barrier. This happens because the thermal energy applied to the carrier overcomes the potential barrier, also known as work function of the metal (Fig. 1-7 (a)). In the beginning, all electrons were bound under the Fermi level (EF) when the temperature is at 0°K. However, part of electrons might acquire kinetic energy from the thermal heating as the temperature increases gradually. When the temperature is high enough, electrons might escape from Fermi level to Vacuum level with high probabilities. This mechanism accomplishes the electron emission without applying any bias voltages at the cost of the thermal energy supplied. In most cases, the thermionic electron is emitted under considerably high temperature, which is based on the different work functions of different materials. The average temperature is about 1500 to 2000 ℃.

On the other hand, the field emission (FE) (also known as Fowler-Nordheim tunneling) is an emission mechanism of electrons induced by external electric fields, which implies that heating was not necessary for the cathode materials. Ab initio,

11

most of electrons would remain under the Fermi level at low temperatures. The external electric field would lead to band bending of the Vacuum level, which induces gradual narrowing of the effective potential barrier width (Fig. 1-7 (b)). As soon as it was thin to some extent, the electron tunneling effect occurred and the electron might tunnel to the Vacuum level, namely the well-known Fowler-Nordheim tunneling. Because this mechanism provides the main escape energy of emitting electrons to the vacuum level by electrical field, it is thus called the field emission.

The throughout thesis will focus on this mechanism of electron emission as our main research issues.

1-3-2 Electron field emission

In quantum mechanical, electron field emission is a tunneling phenomenon of electrons extracted from the conductive solid surface, such as a metal or a

semiconductor, where the surface electrical field is extremely high.

If a sufficient electrical field is applied on the emitter surface, electrons will be emitting through the surface potential barrier into vacuum, even under a very low temperature. On the other hand, thermionic emission is the hot electron emission under high temperature and low electric field. (Fig. 1-8 (a)) demonstrates the band diagram of a metal-vacuum system.

Here W0 is the energy difference between an electron at rest outside the metal

and an electron at rest inside, whereas Wf is the energy difference between the Fermi

level and the bottom of the conduction band. The work function

φ

is defined as

φ

= W0 - Wf. If an external bias is applied, vacuum energy level is reduced and the

potential barrier at the surface becomes thinner as shown in (Fig. 1-8 (b)).

12

the surface barrier. Fowler and Nordheim derive the famous F-N equation (eq. 1.2) as follow [36]:

( )

exp[ 2 ( )/ ] 3 2 2 E y v b y t aE J

φ

φ

− = , (1-2)

where J is the current density (A/cm2). E is the applied electric field (V/cm),

φ

is the work function (in eV), a = 1.56×10-6, b = -6.831×10-7, y = 3.79x10-4×10-4E1/2/φ, t2(y)~1.1 and v(y) can be approximated as [37]

v(y)=cos(0.5πy), (1-3) or

v(y)=0.95−y2. (1-4)

Typically, the field emission current I is measured as a function of the applied voltage V. Substituting relationships of J = I/α and E = βV into (eq. 1.2), where α is the emitting area and β is the local field enhancement factor of the emitting surface, the following equation can be obtained

exp[ ( ) ] ) ( 2 3 2 2 2 V y bv y t V A I β φ φαβ − = . (1-5)

Then taking the log. form of Eq. (1-5) and v(y) ~ 1

] 2.97 10 ( ( )) ) ( 10 54 . 1 log[ ) log( 2 3 7 2 2 6 2 V y v y t V I

β

φ

φ

αβ

_{− ×} × = − , (1-6)

from Eq. (1-6), the slope of a Fowler-Nordheim (F-N) plot is given by

2.97 10 ( ) 2 3 7 β φ × = ≡slopeFN S , (1-7)

13

The parameter β can be evaluated from the slope S of the measured F-N plot if the work function

φ

was known

2.97 10 ( ) 2 3 7 S

φ

β

=− × (cm-1), (1-8)

The emission area α can be subsequently extracted from a rearrangement of (eq. 1-6)

exp( 9.89)exp(6.53 10 ) 10 4 . 1 ) ( 2 3 7 2 6 2 V V I

β

φ

φ

β

φ

α

− × × = ₋ (cm2). (1-9)

For example, the electric field at the surface of a spherical emitter of radius r concentric with a spherical anode (or gate) of radius r+d can be represented analytically by ( ) d d r r V E= + , (1-10) Though a realistic electric field in the emitter tip is more complicated than above equation, we can multiply (eq. 1-10) by a geometric factor β` to approximate the real condition. Etip ≡function of (r,d) =β` ( ) d d r r V + , (1-11) where r is the tip radius of emitter tip, d is the emitter-anode(gate) distance and β` is a geometric correction factor [38].

1-3-3 Operation principles of electron impact ionization

For the application of Gas ionization sensor, an electron would move along the electric field to the anode after it was emitted from the cathode. However, if the distance from anode to cathode is much larger than the average mean free path of an

14

electron, the electrical field may induce accelerated electrons to collide with neutral molecules or atoms in its path to anode. Normally, there are three kinds of reactions for an electron to impact on a neutral molecule, which are the dissociation reaction (eq. 1-12), the activation reaction (eq. 1-13) and the ionization reaction (eq. 1-14). e X_  2X  e (1-12) e X  X e (1-13) e X_  X_＋ 2e (1-14)

The most notable reaction among the above is the ionization reaction (eq. 1-14), which depicts the increment of the number of electrons after the reaction. An ionized electron would be produced whenever an effective collision occurs. Moreover, an ionized positive ion would receive kinetic energy from the acceleration of electric field and collide with the cathode, which would produce more ionized electrons. And the electrical gas breakdown would take place when the amount of ionized electrons was up to a certain quantity.

What's more, the excited atom of (eq. 1-13) would emit photons and then return to ground state which could be expressed as (eq. 1-15),

X  X  hν (1-15) where h is the Plank Constant and ν is the frequency of radiation light. That’s the reason why spark and glare can be seen when the electric gas breakdown occurs on the microstructure of electrodes with high voltages applied (Fig. 1-8)[39].

1-3-4 The fundamental mechanism of gas ionization breakdown

The magnitude of breakdown voltage of gas ionization sensors is based on Paschen’s law, which points out that the breakdown characteristics of a gap between two electrodes are a function (generally not linear) of the product of the gas pressure and the gap length, which is usually denoted as Vbr= f( pd ). Nevertheless, before this

15

law was deduced, the mechanism of applying high voltage among anode and cathode has to be described below.

From (Fig. 1-10)[40], it shows that the processes of gas ionization via positive nanotip could be separated into three parts, which are molecule ionization via nanotip, impact ionization effect, and electron recombination along with γ-process[41]. As one can see in (Fig. 1-10), in process 1, when CNTs were used as positive electrode, the high voltage among anode and cathode would generate a strong local electric field around nanotips, where neutral molecule was ionized and an electron was released simultaneously. The ion would move to the negative electrode under the electrostatic force and the electron would be absorbed concurrently on the positive nanotip. Here suppose that the ionization rate induced by the ion’s moving to the negative electrode is small so that it could be ignored. On the surface of the negative electrode, the ion would recombine with an electron and revert to molecule. However, the release energy of recombination and the kinetic energy of the ion would promote the electron emission from the negative electrode in the process III. This process is the so-called

γ-process. The electron released from the negative electrode would get enough energy

to make other gas molecule ionized during its moving to the positive electrode. This is the electron impact ionization effect (process II). Thus, the pre-breakdown current was generated mainly by the molecule ionization via positive nanotips when utilizing CNTs as the positive electrode. The current could be continued and magnified in processes I and II to result in high current to cause electric breakdown. Therefore, the application of CNTs as positive electrode could enhance the gas ionization process, lower the working voltage, and improve the sensitivity of the sensor.

When CNTs were used as negative electrode, as shown in (Fig. 1-11), in process I, electrons would emit via negative nanotips. Then they move to the positive

16

enough energy to make a molecule ionized, the impact ionization occurs and produces more electrons and ions as desired. This is the process which we mentioned above (process II). Similarly, the ions also induce recombination of electron and γ-process (process III). These three processes would duplicate unceasingly and the breakdown would occur when the number of generated electrons was sufficient. Therefore, the application of CNTs as negative electrode would enhance the emission current, lower the working voltage, and improve the sensitivity of the sensor.

1-3-5 Townsend discharge and Paschen’s law

As mentioned above, an electron moves beyond the average mean free path along the electrical field might acquire sufficient kinetic energy to ionize a neutral molecule into a positive ion and a free electron. This means that the number of free electrons get doubled if we take the original colliding electron into account. As this mechanism proceeds again, these two electrons might collide with two neutral molecules and become two positive ions and four electrons in totality. Therefore, the number of charged particles increases exponentially like a snow avalanche as it repeats continually, which is the well-known Avalanche breakdown effect.

As depicted in (Fig. 1-12), assume the number of electrons after impact ionization is Ne, then

Ne  2 、2、2、2、2、2、2、2… (1-16)

Note that the number of collisions is related to average mean free path λ. Suppose that the distance of electron to the anode is X, then Ne can be described as

Nex  2



λ_

 e



λ_

 e

α

_(1-17), where α is the Townsend’s first ionization coefficient, which tells the average number of ionizing collisions are made by an electron as it travels 1cm in the direction of the

17 electric field. And

α



 λ_

 Ap exp

 ! "



(1-18).

Here A and B are the constants which depend on gaseous species. E is electrical field strength and p is gaseous pressure. From (eq. 1-17), one can easily realize the

exponential relationship between Ne and α. That is, Ne will grow exponentially as the value of α increase. This is the well-known Townsend discharge which illustrates that initially a very small amount of free electrons, accelerated by a sufficiently strong electric field, give rise to electrical conduction through a gas by the avalanche multiplication.

In order to describe the mechanism of gas breakdown, we have to assume that

γeα#$ 1 electrons will be generated when eα#$ 1 positive ions impact on the

cathode. They are referred to the Secondary electrons.

Now, if the electrical field flux is Γ_&, the unit volume density is n_& and ν_& is the electron velocity, then

Γ_&  n_&ν_& (1-19), The unit of Γ_& is [m· s], n_& is [m], ν_& is [m · s], and

ν_&  µ_&· E (1-20), where µ_& is the electron mobility, E is the magnitude of electrical field between two electrodes.

Therefore, the current between two electrodes (I) is

I  $|e|n_&µ_& $|e|Γ_& (1-21), One can easily understand the physical meaning through (Fig. 1-13),

The total electron flux on positive electrode Γ_&d is the combination of Γ_&eα# and

18

Γ_&d  Γ_&eα#+Γ_&eα# (1-22), and

Γ_&eα#  γ/0Γ_&eα#$ Γ_&1  Γ_&eα#$ Γ_&2 (1-23), Put (eq. 1-22) into (eq. 1-23), we have

Γ

_&

＝

γΓ34&α5

/γ&α5_2

(1-24).

Substitute (eq. 1-24) back into (eq. 1-22),

Γ

_&

d＝

Γ34&

α5

/γ&α52 (1-25),

Thus, the current at x=d is

i

_#



78&

α5

/γ&α5_2 (1-26),

it implies that i_# will approach infinity and the electrical breakdown occurs when

91 $ γ0eα#$ 11:  0. γ0eα#$ 11  1 in that eα# < 1 (1-27), so eα#  γ 1 (1-28), which means αd  ln  γ 1. For α Ap exp ! "  (1-18), we have ln >_γ 1?  pdAexp !_" . After rearrangement ln @ln >_γ 1?A  lnpdA $ !_". If E=B #, then

19

ln @ln >_γ 1?A  lnpdA $ !#_B . The breakdown voltage is therefore

V

_DE



!#

!#F@>4_γG?A

(1-29),

which can be simplified to

V

_DE



!#

/!#GH2 (1-30),

with C=ln / F

>4_γG?2.

This is the Paschen’s law which was first stated in 1889, named after Friedrich Paschen. It depicts the breakdown characteristics of a gap are a function (generally not linear) of the product of the gas pressure and the gap length, usually written as V= f( pd ). As we plot (eq. 1-30) with x axis being the breakdown voltages versus the products of pressure and distance as y axis, one can obtain graphs like (Fig. 1-14), which is the so-called Paschen’s curve.

1-4 Motivation

1-4-1 Stability issue

Gas sensors operate on a variety of different fundamental mechanisms [42], and they play an important part in monitoring the environmental changes, controlling chemical processes, preventing from terrorism, and in the application of medical and agricultural behaviors. Gas sensors can be classified into two types, a chemical type operated by gas absorption and a physical type operated by ionization [43-44,27].

Since the electrical conductance of CNTs is highly sensitive to certain gas molecules, they have been used to fabricate the chemical sensors with a fast response

20

time than conventional materials like metal-oxide, polymer, porous silicon, etc. [45-48]. The sensing mechanism involves detecting conductance change of CNTs induced by charge transfer from gas molecules adsorbed on their surfaces. However, all gas-adsorptive types of sensors have several limitations. For instance, they are unable to detect gases with low adsorption energy, and also challenging to use electrical conductance measurement to distinguish between gases in a mixture, i.e., gases with different concentrations can produce the same output signal as that for a single pure gas. Also, gas sensors of chemical type are sensitive to environmental conditions like moisture, temperature, and gas flow velocity. Besides, chemisorptions can cause irreversible changes in nanotube conductivity [26]. Thus, CNT-based gas ionization sensors are expected to overcome these disadvantages.

Gas ionization sensors are physical mechanisms that work by fingerprinting the ionization characteristics of distinct gases. However, conventional ionization sensors are limited by the huge and bulky architecture, risky high-voltage operation and high power consumption. Many investigations have studied on the improvement of these issues. Carbon nanotubes with relatively low work function, very sharp nanotips, and structural and chemical stability under high electrical field, were known to be the best field emitters over many conventional field emitting metals like Mo and W. The usage of CNTs for the improvement of the characteristic of gas ionization sensors has been addressed in recent years [49-52]. Modi et al. [4] proposed the fabrication and successful testing of ionization micro-sensors (Fig. 1-15) featuring the electrical breakdown of a range of gases and gas mixtures at carbon nanotube tips. The sharp tips of nanotubes generate very high electric fields at relatively low voltages, lowering breakdown voltages several-fold in comparison with traditional electrodes. Moreover, S J Kim et al. [50] fabricated a physical type gas sensor (Fig. 1-16) based on an electrical discharge theory known as Paschen’s law. The gas sensor works by figuring