Introduction

Graphene is one or a few layer graphite sheet which consists of sp2 carbon. Since it was isolated in 2004, graphene has been paid much attention due to its peculiar properties such as high mobility as much as 200000cm2/V ・ s, anomalous quantum hall effect at RT and edge state originating from zigzag edge. Moreover, graphene is expected new catalytic material for its high surface area. When we consider its application to devices, however, graphene has several problems to be solved. A new method for high-through put preparation is needed, replacing the present exfoliation method which gives the best quality graphene. CVD growth on metal substrates, thermal annealing of SiC and chemically preparation via graphene oxide (GO) are expected as promising candidates. I have focused on chemical exfoliation of graphene in my thesis. Chemically prepared graphene is suitable for industrial application because it does not need vacuum and high temperature. However, its typical domain size was smaller than 1pm and reduction process is necessary to be used as graphene. In my thesis I investigated relationship between conductivity and reduction process, which was not understood in detail. The other problem for graphene is about its electric property. Graphene is originally a zero-gap semiconductor and it causes low on/off ratio of graphene transistor. Thus opening of band-gap is crucial for device application. Theoretically speaking, when nitrogen and boron atom are doped into graphene network, band gap will open. Moreover doping of nitrogen is reported to enhance catalytic properties. However, there have been only a few experimental reports about nitrogen doping to graphene. Therefore in my thesis I examined to apply CVD method to fabricate nitrogen doped graphene (NG) and analyzed NG by scanning tunneling microscopy.

Fabrication and reduction process of graphene oxide

First I examined to synthesize large size GO sheets. GO was prepared by modified Hummers methods. I succeeded in preparing GO on substrate without ultra-sonication which would have made graphene sheet small. I could obtain large size monolayer GO sheets (Fig. 2) easily and repeatedly. I constructed the equipment for in situ conductivity measurement, in which electrical properties could be measured in vacuum during hydrazine (the most powerful reductant for GO) reduction or during thermal annealing. Since the conductivity of graphene is sensitive to adsorbates, in situ conductivity measurement is essential to extract intrinsic properties of graphene oxide. Evolution of conductivity was pursued during two different reduction processes; thermal annealing after or before hydrazine treatment. In the both cases current increased drastically at a specific temperature; 390 K for hydrazine and 465 K for only thermal annealing. In order to clarify the difference of two processes, the transfer characteristic of GO field effect transistor (FET) was investigated and the result is shown in Fig. 3. The specimen reduced by hydrazine after thermal annealing shows lower mobility than the specimen reduced only by hydrazine. In order to elucidate reduction process of GO, XPS spectra were measured during thermal annealing. According to XPS spectra, the peak attributed to C-O bond started to decrease at around 433K and GO was reduced gradually with increasing temperature (Fig. 4). From these results around 465 K the current path was formed by reduction and decrease of C-O bond triggered increase of conductivity. However, the highest mobility of GO FET was 3.0cm2/ V ・ s in the present experiment. It is much lower than the exfoliated graphene mainly because GO might have a lot of defects and disorder structure.

In order to elucidate atomic structure during reduction process, STM image of GO was observed. The reduced GO (rGO) on Si(100) by thermal annealing at 1300 K showed only a disordered structure and honeycomb lattice intrinsic to graphene could not be found at all. Since graphene is known to grow on several metal substrates by using their catalytic properties, reduction of GO were examined on Pt(111). Fig. 5 indicates an STM image of GO on Pt(111) after thermal annealing at 1300K. Some flat area appeared on GO/Pt(111) and honeycomb lattice could be observed there. This image means that a Pt substrate catalytically restores GO defects and disorder. A moire structure means monolayer graphene sheet grown on Pt(111).

C 1s XPS spectra of rGO on various substrates were measured (Fig. 6). The FWHM of C 1 s peak on Pt(111) is as narrow as HOPG, while that on Si(100) is broader reflecting its disorder structure. It can be said that thermal annealing of GO on Pt(111) gives so high quality graphene by utilizing catalytic properties of Pt(111) .

Doping of nitrogen atoms in graphene

Since graphene is known to grow on metal substrates from hydrocarbon (HC) molecules, it can be expected that nitrogen will be doped by using HC molecules which contain nitrogen atoms. First, Pt(111) surface was exposed to 200L(1L =1.3 X 10-4 Pa/sec) of pyridine at various substrate temperatures. At 850 K a lot of a fewnm size domed-shaped structures appeared. Some domains have a honeycomb structure for graphene formed at 1000 K, while domed-shaped domains coexisted. Fig. 7 indicates an STM image of graphene after CVD at 1150 K. A lot of large domains with honeycomb lattice grew on Pt(111) and some moire patterns also appeared. In some atomic resolution STM images of graphene, a few atoms seem brighter than other atoms and the brighter area is over a few honeycomb rings. (Fig. 8) Nitrogen atom is expected to induce change of electronic state at surrounding carbon atoms in a honeycomb lattice. Therefore existence of bright area images suggests nitrogen atom(s) incorporated into graphene network.

Next, adsorption process of oxygen on nitrogen doped graphene was investigated. Fig. 9 shows STM image at the same position before and after exposure to 10L 02 at RT. Comparing with two images, around defects become brighter. Moreover according to other STM images oxygen also adsorbed at domain boundaries preferably. XPS and STM measurement suggests that doped nitrogen atoms tend to exist at pyridinic sites (around at defects and domain boundary). Thus it can be considered that the nitrogen atoms around defects and domain boundary adsorb oxygen preferentially. The exposure to 600L 02 at 680 K results in much oxygen adsorption at domain boundary and enhanced modulation amplitude (Fig. 10).

Conclusion

I have investigated chemically modified graphene such as graphene oxide and nitrogen doped graphene. I elucidated relationship between composition and electric conductivity of GO in detail and found that thermal reduction method generates a lot of defects in graphene. Moreover, I proved that annealing of GO on Pt(111) at 1300 K provides a graphene without defects and disorder due to catalytic property of Pt. It is a new method for obtaining a high quality graphene from GO. I succeeded in synthesizing a nitrogen doped graphene on Pt(111) from pyridine and investigated atomic structure with STM. Brighter areas in honeycomb lattice indicated nitrogen doping into carbon network. In addition I also investigated oxygen adsorption on a nitrogen doped graphene. Oxygen tends to be adsorbed around defects and domain boundary of graphene where nitrogen atoms may exist selectively. The change of moire pattern by oxygen indicates adsorption was affected by moire pattern, corresponding to modulation of electronic state caused by interaction with Pt(111) substrate.

Fig. 1 Graphene Oxide

Fig. 2 SEM image of GO

Fig. 3 FET transfer characteristic of each reduction methods

Fig. 4 XPS spectra of Cls during annealing

Fig. 5 Moire structure of GO on Pt(111) 17nm x 17nm I1:5 nA V,:10 mV

Fig. 6 XPS spectra of C 1 s

Fig. 7 Moire and honeycomb Structure ofNG on Pt(111) 4nm x4nm I, :7 nA Vs:10,13 mV

Fig. 8 Bright area in honeycomb lattice 6.313nm x6.313nmI,:7nAV,:-10mV

Fig. 9 (a) Before 02 exposure at RT NG was prepared by 1150 K exposure 20.8nm x 20.8nm It: 1nA V,: 0.003 V

Fig. 9 (b) After 02 exposure at RT 20.55nm x 20.55nm It: 1nA V,: 0.003 V

Fig. 10 After 02 exposure at 680K NG was prepared by annealing at 1150 K after 200 L pyridine exposure at RT 22.16nm X22.16nmIt:1 nAV,: 10mV

本論文は6章から成る.

第1章は序論であり,本論文の研究対象であるグラフェンの特徴および解決すべき問題点を述べている.作製方法やドーピングによる物性への影響など多くの具体例を挙げて説明し,本論文の主題である化学的に修飾・合成されたグラフェンの有用性,意義が示されている.

第2章では本研究で用いた装置に関して述べている.in situ 伝導度測定の目的で論文提出者が作製した真空装置,および光電子分光 (XPS),走査トンネル顕微鏡 (STM) 測定装置等の構成,特徴を説明している.

第3章では巨大な酸化グラフェン (GO)の大量合成,伝導度とGOの組成の関連性,還元法によるGOの構造・伝導度への影響,に関して述べている.成膜プロセスを工夫することで100 μm 100 μmを超える巨大なGOを大量に合成する方法を確立し,さらにin situ 伝導度測定とXPSによる組成解析から,hydroxyl 基やepoxideの還元により伝導パスが生成し,伝導度の急峻な立ち上がりが起きていることを明らかにした.またSTMでの局所構造解析とin situでの伝導度測定から,ヒドラジン還元と真空中での加熱による還元では還元中の反応過程が異なり,ヒドラジン還元後では部分的にハニカム構造が回復しているが,真空還元の場合は乱れた構造をとっていることを原子レベルでの解析によって解明した.

第4章ではGOの新規還元方法に関して述べている.既存の還元剤を用いた還元方法では欠陥のない高品質なグラフェンを得ることは困難であったが,論文提出者はPt(111)の触媒作用に注目し,GOをPt(111)基板上で加熱することでGOから欠陥の非常に少ないグラフェンの作製に成功した.STMにより50nm以上の長周期に渡るシングルドメインのグラフェンが観察され,XPSによっても酸素の完全消失が確認されている.この方法はグラフェンの新規大量合成法につながるものと期待される.

第5章では窒素ドープグラフェンの作製・局所構造解析・酸素吸着特性に関して述べている.1150 K, 1300 KのPt(111)基板上へのピリジン曝露により,一様なグラフェンの生成に成功した.さらに窒素がグラフェンネットワーク内にドープされている事を強く示唆するSTM像からも,ピリジンからの窒素ドープグラフェンの作製に初めて成功したといえる.XPS測定の結果を併用することで窒素のドープ位置に関しても考察し,さらに,窒素原子の酸素吸着特性への影響に関する実験から,窒素ドープグラフェンの触媒への応用について検討している.このように窒素ドープグラフェンに関し,作製条件による局所構造の差異やドープ位置に関して研究された例はなく,本成果は当該分野の今後の研究指針を与えるものである.

第6章では本論文についての総括がなされている.

なお,本論文は佐藤裕樹氏(第3章),田中弘成氏(第3-4章),斉木幸一朗氏(第3-5章)の共同研究であるが,論文提出者が主体となって実験,解析,考察を行なったものであり,論文提出者の寄与が十分であると判断する.

以上のように本論文では,STMによる局所構造解析や伝導度のin situ 測定などを用いて,化学的に合成・修飾されたグラフェンの合成法・物性を詳細に調べた.これらの成果はグラフェンの大量合成法の確立や窒素ドープによる新規物性の発現などに多大な寄与を与え,グラフェン分野の基礎・応用両面に大きな貢献が期待される.これら研究成果のオリジナリティを審査委員会一同で高く評価した.

したがって博士(理学)の学位を受けるのに十分な資格を有すると認める.