Complementary metal-oxide-semiconductor-compatible and self-aligned catalyst formation for carbon nanotube synthesis and interconnect fabrication Can Zhang, Feng Yan, Bernhard C. Bayer, Raoul Blume, Marleen H. van der Veen et al. Citation: J. Appl. Phys. 111, 064310 (2012); doi: 10.1063/1.3694678 View online: http://dx.doi.org/10.1063/1.3694678 View Table of Contents: http://jap.aip.org/resource/1/JAPIAU/v111/i6 Published by the American Institute of Physics. Related Articles Electrical transport properties of boron-doped single-walled carbon nanotubes J. Appl. Phys. 113, 054313 (2013) Electric field induced needle-pulsed arc discharge carbon nanotube production apparatus: Circuitry and mechanical design Rev. Sci. Instrum. 83, 123907 (2012) Cylindric quantum wires in a threading magnetic field: A proposal of characterization based on zero bias electron transport J. Appl. Phys. 112, 123715 (2012) A doping-free approach to carbon nanotube electronics and optoelectronics AIP Advances 2, 041403 (2012) Magnetic and electrical properties of PbTiO3/Mn-Zn ferrite multiphase nanotube arrays by electro-deposition J. Appl. Phys. 112, 104310 (2012) Additional information on J. Appl. Phys. Journal Homepage: http://jap.aip.org/ Journal Information: http://jap.aip.org/about/about_the_journal Top downloads: http://jap.aip.org/features/most_downloaded Information for Authors: http://jap.aip.org/authors Downloaded 07 Feb 2013 to 129.169.173.162. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions JOURNAL OF APPLIED PHYSICS 111, 064310 (2012) Complementary metal-oxide-semiconductor-compatible and self-aligned catalyst formation for carbon nanotube synthesis and interconnect fabrication Can Zhang,1,a) Feng Yan,1 Bernhard C. Bayer,1 Raoul Blume,2 Marleen H. van der Veen,3 Rongsi Xie,1 Guofang Zhong,1 Bingan Chen,1 Axel Knop-Gericke,2 Robert Schlögl,2 Bernard D. Capraro,4 Stephan Hofmann,1 and John Robertson1,b) 1 Department of Engineering, University of Cambridge, Cambridge CB2 1PZ, United Kingdom Fritz-Haber-Institut der Max-Planck-Gesellschaft, Department of Inorganic Chemistry, 14195 Berlin, Germany 3 imec, Kapeldreef 75, B-3001, Leuven, Belgium 4 Intel Ireland Limited, Collinstown Industrial Park, Leixlip, Ireland 2 (Received 6 December 2011; accepted 14 February 2012; published online 21 March 2012) We have for the first time developed a self-aligned metal catalyst formation process using fully CMOS (complementary metal-oxide-semiconductor) compatible materials and techniques, for the synthesis of aligned carbon nanotubes (CNTs). By employing an electrically conductive cobalt disilicide (CoSi2) layer as the starting material, a reactive ion etch (RIE) treatment and a hydrogen reduction step are used to transform the CoSi2 surface into cobalt (Co) nanoparticles that are active to catalyze aligned CNT growth. Ohmic contacts between the conductive substrate and the CNTs are obtained. The process developed in this study can be applied to form metal nanoparticles in regions that cannot be patterned using conventional catalyst deposition methods, for example at the bottom of deep holes or on vertical surfaces. This catalyst formation method is crucially important C 2012 for the fabrication of vertical and horizontal interconnect devices based on CNTs. V American Institute of Physics. [http://dx.doi.org/10.1063/1.3694678] I. INTRODUCTION Carbon nanotubes (CNTs) have been widely investigated for their exceptional electrical, thermal and mechanical properties. Over the past 20 years, there has been considerable effort to use them in electronic devices, such as fieldemitters, field-effect transistors and thin film transistors.1–4 They have also been extensively investigated as promising candidates to replace copper as interconnects in integrated circuits beyond the 22 nm node.5–12 CNT can carry a current density up to 109 A/cm2 that exceeds the maximum current density of 6 106 A/cm2 of copper before it fails due to electromigration.13,14 Although some progress in CNT integration for vertical interconnect (via) applications has been made, the latest CNT vias are still limited to 70–500 nm in diameter and aspect ratios of below 2.5,6,9–12 One of the major obstacles is the difficulties in patterning the catalyst metal at the bottom of small and deep via holes using conventional physical vapor deposition (PVD) techniques. Similarly, the use of CNTs as horizontal interconnects has been considered. However, the fabrication of those poses equally great and unsolved challenges to catalyst deposition, as the metal catalyst needs to be patterned exclusively onto vertical surfaces with high reproducibility and low levels of contamination.7,15,16 Self-alignment is a very valuable and important technique used in semiconductor industry; it achieves the formation or deposition of material only in regions that are already a) Electronic mail: cz232@cam.ac.uk. Electronic mail: jr214@cam.ac.uk. b) 0021-8979/2012/111(6)/064310/6/$30.00 defined in previous steps without using extra lithography or masking steps.17 Self-aligned techniques have been used in fabricating transistors based on CNT and graphene, and have significantly improved the manufacturability of such devices.18–21 So far, these methods have achieved selfalignment in the device fabrication steps, e.g., formation of source, drain and gate electrodes for transistors. On the other hand, the CNT/graphene synthesis still needs dedicate lithography steps for metal catalyst patterning or, alternatively, devices will be fabricated at random locations on the wafer. Here, we report a process for formation of selfaligned Co catalyst on cobalt disilicide (CoSi2) for aligned CNT growth. This method dramatically increases the CNT yield compared to previous reports where silicide itself (i.e., without deliberately transforming them into metal phase) is used as catalyst.22–27 CoSi2 is a widely used electrical contact material in complementary metal-oxide-semiconductor (CMOS) processing,28 and is used in our process both as electrically conductive electrode and precursor for Co catalyst formation. We also demonstrate that this process can achieve catalyst patterning on vertical surfaces and at the bottom of via holes below the back-end CMOS integration temperature (<450 C). Hence, our process enables aligned CNT growth with ohmic contacts to the substrate in and on previously inaccessible geometrical features. The CNT growth is carried out at 600 C. II. EXPERIMENTAL We started with a 20 nm CoSi2 thin film formed on top of a 200 nm polycrystalline silicon (poly-Si) layer on a 111, 064310-1 C 2012 American Institute of Physics V Downloaded 07 Feb 2013 to 129.169.173.162. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions 064310-2 Zhang et al. single-crystalline Si wafer. The silicide was formed through a self-aligned silicidation (salicidation) process as shown in Fig. 1(a), similar to that reported elsewhere.29,30 First, 15 nm Co was sputtered on top of the poly-Si with TiN as a capping layer. The CoSi2 formation was achieved after a combination of rapid thermal annealing (RTA) and wet etch steps that remove the unused Co and TiN. Then, we performed the self-aligned catalyst formation process on the CoSi2 layer, as illustrated in Fig. 1(b). Reactive ion etch (RIE)31–33 was performed for 15 s under 15 sccm SF6 at 0.15 mbar and 70 W rf power, at room temperature, H2 reduction was achieved by annealing the sample for 3 min under 500 sccm H2 at 15 mbar at 600 C. CNT growth was carried out under 480 sccm H2 and 20 sccm C2H2 at 15 mbar and 600 C. It is noteworthy that cobalt fluorides can start to decompose and form cobalt metal under H2 environment at 320 C,34 below the back-end CMOS integration temperature (<450 C). FIG. 1. (Color online) Schematics showing (a) the self-aligned silicidation process for CoSi2 patterning on Si, (b) the effects of SF6 RIE treatment and CVD on CoSi2 layer. (c)–(e) Scanning and transmission electron microscopy micrographs of CNT forest grown for 5 min. The scale bars are 5 lm in (c), 200 nm in (d) and 10 nm in (e). J. Appl. Phys. 111, 064310 (2012) Deep reactive ion etch (DRIE)33 was performed for 20 seconds with 400 W rf power, 50 W bias power, 0.1 mbar pressure and 250 sccm SF6 etchant, the sample was maintained at 10 C. In situ XPS was performed at the BESSY II synchrotron in the ISISS end station of the FHI-MPG, it allows processstep and time-resolved monitoring of the chemical surface state (information depth 0.7–1.2 nm) during CVD conditions (pretreatment in H2 and growth in C2H2 at 0.2 mbar, 600 C). The spectra were collected in normal emission geometry and background correction was performed using a Shirley background; spectra were fitted following the Levenberg-Marquardt algorithm to minimize the v2 value. Peak shapes were modeled using asymmetric DoniachSunjic functions convolved with Gaussian profiles. III. RESULTS AND DISCUSSION Reactive ion etch (RIE) step at room temperature was carried out on the CoSi2 layer with SF6 as the gas source. The RIE etches away the Si from the silicide by forming SiF4 which is volatile at room temperature (boiling temperature 86 C) and hence easily pumped away.31–33 Simultaneously, a solid phase of cobalt fluoride was formed on the surface by reaction of the SF6 plasma with the left behind Co from the silicide. This is due to the very low volatility of CoxFy (boiling temperature over 1000 C).32 The thickness of the CoxFy can be controlled by the etching time and 1 nm CoxFy is formed after 10 s RIE.32 Then, for CNT growth, samples were annealed in H2 at 600 C, which reduces the Co-fluoride to metallic Co. The Co metal in turn serves as the highly active CNT catalyst during subsequent carbon feedstock gas (C2H2) exposure. Vertically aligned CNT forests were grown from the processed silicide layers in both cold-wall, reduced pressure and hot-wall atmospheric pressure CVD,35,36 as shown in Figs. 1(c)–1(e). The CNTs are typically 7 nm in diameter with 4 walls, and the forest has a density of 17 mg/cm3. In situ x-ray photoemission spectroscopy (XPS) was employed to confirm the chemical changes caused by the SF6 RIE treatment and during the CVD process.37,38 The spectra in Fig. 2 show the evolution of the chemical state of the sample surface. For the as-prepared CoSi2 surface, the Co2p3/2 shows a single peak at 778.9 eV corresponding to CoSi2 and no F signal as expected from our process flow.39 After exposure to SF6 plasma in the RIE step, the peak positions in the Co2p3/2 scan as well as the F1s spectrum change drastically. The Co2p3/2 spectrum now strongly resembles spectra of Co fluorides.32 The formation of fluorides is also confirmed by the emergence of a peak in the F1s spectrum at 685.2 eV, the binding energy representing metal fluorides.32 During subsequent heating of the samples in pure H2 to 600 C, we find that the Co-fluoride is reduced. This is evidenced by the disappearance of the peak in the F1s spectrum and a complete change of the Co2p3/2 spectrum. This spectrum is now dominated by a strong Co metal peak at 778.3 eV accompanied by a smaller peak at 778.9 eV attributed to remaining CoSi2 underneath.40 By replacing the H2 with pure C2H2, CNTs were successfully grown on the Downloaded 07 Feb 2013 to 129.169.173.162. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions 064310-3 Zhang et al. J. Appl. Phys. 111, 064310 (2012) FIG. 2. In situ XP spectra of (a) Co2p and (b) F1s of CoSi2 in as-prepared state, after SF6 RIE treatment and during H2 annealing (from bottom to top). The peak evolution fully confirms our suggested chemical mechanisms during the process flow in Fig. 1(b). samples monitored with in situ XPS. Therefore, the XPS data in Fig. 2 fully confirms the suggested atomistic behavior in our process flow in Fig. 1(b). Atomic force microscopy (AFM) was used to investigate the changes on the surface roughness of the CoSi2 surface during the processing, as shown in Fig. 3. The RIE step does not affect the surface morphology significantly as the cobalt fluoride formed has a similar thickness of the CoSi2 replaced [Fig. 3(b)].32 However, after the H2 annealing step, there are more recognizable nanoparticles formed on the surface [Fig. 3(c)], with diameters below 10 nm, as estimated from the line scans shown in Fig. 3(d). These nanoparticles are the cobalt nanoparticles from which CNTs can nucleate. To demonstrate that the Co catalyst formation is selfaligned to the exposed CoSi2 regions, we fabricated two device structures which have 20 nm CoSi2 thin layers patterned on either planar or vertical surfaces. The first structure [Fig. 4(a)] was fabricated by dry etch processes to form via holes on top of a 20 nm CoSi2 layer [silicide deposition process as for blanket wafers in Fig. 1(a)]. A deep RIE (DRIE)33 step with SF6 was then used to transform the top of the CoSi2 layer into cobalt fluoride at the bottom of the via followed by CNT growth. This structure serves as a test vehicle for CNTs in vertical interconnects. The second structure [Fig. 4(b)] has the CoSi2 layer patterned on a vertical surface that is achieved using a salicidation technique.29,41 First, a 200 nm thick highly doped poly-Si with 50 nm SiON hard mask on the top was patterned as the electrode material. This was followed by the same Co/TiN sputtering and salicidation steps that were carried out for the blanket CoSi2 wafer preparation. Then RIE and CVD were performed. It is worth noting that RIE is isotropic so it can etch the vertical surfaces of the electrode in Fig. 4(b). This second structure is a test vehicle for horizontal CNT interconnects. Figure 4(c) shows the top view of the CNTs grown out of the via holes. The yield is nearly 100% in all the 200 nm diameter via holes of 380 nm depth. There is no CNT growth observed on the top surfaces or on the sidewalls of the via holes [Fig. 4(d)–4(f)], confirming the high selectivity of our self-aligned Co formation. Figure 4(g) shows the horizontally aligned CNTs that were grown from the CoSi2 layer patterned on the vertical surface after the Co formation. The spacing between two opposite electrodes [Fig. 4(h)] is 400 nm; the growth time is controlled to be 20 s to limit the length of the CNTs. Both the vertical and horizontal test vehicles show that CNT growth on predefined geometrical features can be achieved using our self-aligned catalyst formation. We also test the electrical characteristics of our tubes and their interface to the silicide, using the horizontal interconnect test structure. A path for electron transportation is achieved when tubes grown from opposite electrodes meet in Downloaded 07 Feb 2013 to 129.169.173.162. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions 064310-4 Zhang et al. FIG. 3. (Color online) Atomic force microscope (AFM) images of (a) asprepared CoSi2 layer, (b) after RIE treatment, (c) after annealing in H2. The scale bars are 200 nm. (d) Typical line scans of the corresponding images. J. Appl. Phys. 111, 064310 (2012) the space between. The electrical measurement was carried out in air with a two-point probe. The intrinsic resistance (about 2 kX) of the electrode and probe tip-electrode contact was determined from reference samples where the electrodes were short circuited by CoSi2. The approximately linear relationship between voltage and current demonstrates that ohmic contact was achieved (Fig. 5). The small deviation of resistance with voltage is probably due to tunneling between tubes or the charge hopping in the CNT arrays when they are in close contact.42,43 The resistance per tube (assuming 5–10 tubes are connected between the two electrodes) is in the range of 100–200 kX, comparable to recent studies of CNT forest grown on conductive supports.30,38,44 However, our value here includes the resistance of the CNT-silicide contact, CNTs, and the CNT-CNT contact. The CNT-CNT contact resistance is estimated to be above 100 kX for 10 nm diameter CNTs,45 therefore the CNT-silicide contact resistance should be much smaller than 100–200 kX. For CNT via structures, a further and significant resistance decrease can be achieved by chemical-mechanical-polishing (CMP) processes that open tube caps and obtain electrical connection to the inner walls of the CNTs.6,9–12 It is noteworthy that DRIE, as the mainstream technique used in via and through-silicon-via (TSV) fabrication in semiconductor industry, is capable of etching holes or trenches of extremely small width (sub 100 nm) and high aspect ratio up to 50.33,46–48 However, the most recent CNT vias have an aspect ratio below 2 due to the difficulties in catalyst patterning at the bottom of deep via holes.5,6,9–12 In our process, we demonstrated that the Co catalyst formation can be self-aligned to the CoSi2 at the regions exposed to RIE or DRIE process, either on a vertical surface or at the bottom of small via holes. This represents a crucial step forward to realize advanced CNT vias. The demonstration of patterned and horizontally aligned CNT growth from an FIG. 4. (Color online) Cross section of (a) the via hole structure with CoSi2 at the bottom (b) the horizontal device structure with CoSi2 patterned on the vertical surfaces of the electrode. (c) SEM images of CNTs grown from the 200 nm via holes. (d)–(f) Cross section of the via hole after CNT growth. The samples were tilted by 45 under SEM. CNT collapse in (f) is due to sample cutting in order to reveal the clean sidewall of the via. (g) Horizontally aligned CNTs grown from CoSi2 patterned on a vertical surface of a straight line pattern. (h) A two electrode CNT horizontal interconnect. The scale bars are 5 lm in (c) and 1 lm in (g), (h). Downloaded 07 Feb 2013 to 129.169.173.162. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions 064310-5 Zhang et al. J. Appl. Phys. 111, 064310 (2012) 10 FIG. 5. (Color online) Typical I-V and R-V curves measured on a horizontal CNT interconnect with 400 nm spacing; the measured resistance is about 20 kX for 5–10 MWNTs (estimated from SEM). electrode material can also ease the fabrication of planar or three dimensional nanoelectromechanical system (NEMS) and CNT-based sensors at wafer scale.49–51 IV. CONCLUSIONS In conclusion, we have demonstrated a new route of metal catalyst formation for both vertically and horizontally aligned CNT growth. The surface of a conductive and CMOS compatible material, CoSi2, is transformed into Co nanoparticles after a RIE and H2 reduction process. This process is shown to achieve self-aligned catalyst patterning on both planar and vertical surfaces for CNT interconnect fabrications. The process is able to solve the issues involved with catalyst patterning at the bottom of small and deep via or TSV with the geometry limit only placed by the DRIE technique. ACKNOWLEDGMENTS The authors acknowledge funding from the EU Integrated Projects ViaCarbon and Technotubes. 1 S. Fan, M. G. Chapline, N. R. Franklin, T. W. Tombler, A. M. Cassell, and H. J. Dai, Science 283, 512–514 (1999). S. J. Kang, C. Kocabas, T. Ozel, M. Shim, N. Pimparkar, M. A. Alam, S. V. Rotkin, and J. A. Rogers, Nat. Nanotechonol. 2, 230–236 (2007). 3 P. Avouris, Z. Chen, and V. Perebeinos, Nat. Nanotechnol. 2, 605–615 (2007). 4 D. Sun, M. Y. Timmermans, Y. Tian, A. G. Nasibulin, E. I. Kauppinen, S. Kishimoto, T. Mizutani, and Y. Ohno, Nat. Nanotechnol. 6, 156–161 (2011). 5 F. Kreupl, A. P. Graham, G. S. Duesberg, W. Steinhogl, M. Liebau, E. Unger, and W. Honlein, Micro. Eng. 64, 399–408 (2002). 6 D. Yokoyama, T. Iwasaki, T. Yoshida, H. Kawarada, S. Sato, T. Hyakushima, M. Nihei, and Y. Awano, Appl. Phys. Lett. 91, 263101–263103 (2007). 7 G. F. Close, S. Yasuda, B. Paul, S. Fujita, and H. S. P. Wong, Nano Lett. 8, 706–709 (2008). 8 S. Esconjauregui, M. Fouquet, B. C. Bayer, C. Ducati, R. Smajda, S. Hofmann, and J. Robertson, ACS Nano 4, 7431–7436 (2010). 9 N. Chiodarelli, S. Masahito, Y. Kashiwagi, Y. Li, K. Arstila, O. Richard, D. J. Cott, M. Heyns, S. D. Gendt, G. Groeseneken, and P. M. Vereecken, Nanotechnology 22, 085302–085308 (2011). 2 J. Dijion, M. Fayolle, T. Vo, J. Pontcharra, D. Acquaviva, D. Bouvet, A. Ionescu, S. C. Esconjauregui, B. Capraro, E. Quesnel, and J. Robertson, Tech. Dig. IEEE Int. Electron Devices Meeting (IEDM’10), 33.4.1-33.4.4 (2010). 11 A. Kawabata, S. Sato, T. Nozue, T. Hyakushima, M. Norimatsu, M. Mishima, T. Murakami, D. Kondo, K. Asano, M. Ohfuti, H. Kawarada, T. Sakai, M. Nihei, and Y. Awano, Proc. IEEE Int. Interconnect Technol. Conf. 237–239 (2008). 12 M. Katagiri, Y. Yamazaki, M. Wada, M. Kitamura, N. Sakuma, M. Suzuki, S. Sato, M. Nihei, A. Kajita, T. Sakai, and Y. Awano, Jpn. J. Appl. Phys. 50, 05EF01-1 (2011). 13 B. Q. Wei, R. Vajtai, and P. M. Ajayan, Appl. Phys. Lett. 79, 1172–1174 (2001). 14 A. Javey, J. Guo, M. Paulsson, Q. Wang, D. Mann, M. Lundstrom, and H. Dai, Phys. Rev. Lett. 92, 106804–106807 (2004). 15 C. Zhang, D. Cott, N. Chiodarelli, P. Vereecken, J. Robertson, and C. M. Whelan, Phys. Status Solidi B 245, 2308–2310 (2008). 16 C. Santini, D. J. Cott, A. R. Negreira, S. Sanseverino, S. D. Gendt, and P. M. Vereecken, ECS Trans. 18, 845–850 (2009). 17 B. R. Fair, Proc. IEEE 86, 111–137 (1998). 18 A. Javey, J. Guo, D. B. Farmer, Q. Wang, E. Yenilmez, R. G. Gordon, M. Lundstrom, and H. Dai, Nano Lett. 4, 1319–1322 (2004). 19 M. Muoth, T. Helbling, L. Durrer, S. W. Lee, C. Roman, and C. Hierold, Nat. Nanotechnol. 5, 589–592 (2010). 20 L. Liao, Y. Lin, M. Bao, R. Chen, J. Bai, Y. Liu, Y. Qu, K. L. Wang, Y. Huang, and X. Duan, Nature 467, 305–308 (2010). 21 P. Gao, Q. Zhang, H. Li, and M. B. Chan-Park, Small 7, 2195–2200 (2011). 22 J. M. Mao, L. F. Sun, L. X. Qian, Z. W. Pan, B. H. Chang, W. Y. Zhou, G. Wang, and S. S. Xie, Appl. Phys. Lett. 72, 3297–3299 (1998). 23 H. Wen, K. Yang, K. Ou, W. Wu, R. Luo, and C. Chou, Micro. Eng. 82, 221–227 (2005). 24 H. L. Chang, C. H. Lin, and C. T. Kuo, Thin Solid Films 420–421, 219–224 (2002). 25 W. Yang, T. Yang, and T. Yew, Carbon 45, 1679–1685 (2007). 26 H. Qi, C. Qian, and J. Liu, Nano Lett. 7, 2417–2421 (2007). 27 M. Nihei, A. Kawabata, and Y. Awano, Jpn. J. Phys. L721–L723 (2003). 28 J. P. Gambino and E. G. Colgan, Mater. Chem. Phys. 52, 99–146 (1998). 29 S. Buschbaum, O. Fursenko, D. Bolze, D. Wolansky, V. Melnik, J. Nieb, and W. Lerch, Microelectron. Eng. 76, 311–317 (2004). 30 B. C. Bayer, C. Zhang, R. Blume, F. Yan, M. Fouquet, C. T. Wirth, R. S. Weatherup, L. Lin, C. Baehtz, R. A. Oliver, A. Knop-Gericke, R. Schlögl, S. Hofmann, and J. Robertson, J. Appl. Phys. 109, 114314 (2011). 31 H. Jansen, H. Gardeniers, M. Boer, M. Elwenspoek, and J. Fluitman, J. Micromech. Microeng. 6, 14–28 (1996). 32 M. R. Baklanov, S. Vanhaelemeersch, W. Storm, W. Vandervorst, and K. Maex, J. Vac. Sci. Technol. B 16, 164–172 (1998). 33 T. S. Kulmala, A. Colli, A. Fasoli, A. Lombardo, S. Haque, and A. Ferrari, ACS Nano 5, 6910–6915 (2011). 34 I. G. Sharma, A. Pamela, S. Majumdar, J. Kishor, and A. K. Suri, J. Alloy Compd. 437, 231–237 (2007). 35 C. T. Wirth, C. Zhang, G. Zhong, S. Hofmann, and J. Robertson, ACS Nano 3, 3560–3566 (2009). 36 C. Zhang, S. Pisana, C. T. With, A. Parvez, C. Ducati, S. Hofmann, and J. Robertson, Diamond Relat. Mater. 17, 1447–1451 (2008). 37 C. Mattevi, C. T. Wirth, S. Hofmann, R. Blume, M. Cantoro, C. Ducatti, C. Cepek, A. Knop-Gericke, W. Milne, C. Castellarin-Cudia, S. Dolafi, A. Goldoni, R. Schlogl, and J. Robertson, J. Phys. Chem. C 112, 12207–12213 (2008). 38 B. C. Bayer, S. Hofmann, C. Castellarin-Cudia, R. Blume, C. Baehtz, S. Esconjauregui, C. T. Wirth, R. A. Oliver, C. Ducati, A. Knop-Gericke, R. Schlögl, A. Goldoni, C. Cepek, and J. Robertson, J. Phys. Chem. C 115, 4359–4369 (2011). 39 F. Takahashi, T. Irie, J. Shi, and M. Hashimoto, Appl. Surf. Sci. 169–170, 315–319 (2001). 40 XPS fitting parameters: For Co metal at 778.3 eV: Gaussian width (G) ¼ 0.8eV, Lorentzian width (L) ¼ 0.44 eV, asymmetry factor is 0.15. For cobalt silicide at 778.9 eV: G ¼ 1.2 eV, L ¼0.25 eV. For cobalt fluoride for the peak at 783.2 eV: G ¼ 1.6 eV, L¼ 0.5 eV and for the peak at 781.45 eV: G ¼1.55 eV, L ¼ 0.5 eV. For F at 685.2 eV: G ¼1.6 eV, L ¼ 0.25 eV. 41 H. M. Kamal, N. S. Argenti, C. S. Blair, IEEE Trans. Semicond. Manuf. 15, 350–354 (2002). 42 E. Bekyarova, M. E. Itkis, N. Cabrera, B. Zhao, A. Yu, J. Gao, and R. C. Haddon, J. Am. Chem. Soc. 127, 5990–5995 (2005). Downloaded 07 Feb 2013 to 129.169.173.162. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions 064310-6 Zhang et al. J. Appl. Phys. 111, 064310 (2012) 43 48 44 49 Z. J. Han and K. Ostrikov, Appl. Phys. Lett. 96, 233115–233117 (2010). G. D. Nessim, M. Seita, K. P. O’Brein, A. J. Hart, R. K. Bonaparte, R. R. Mitchell, and C. V. Thompson, Nano Lett. 9, 3398–3405 (2009). 45 C. Santini, A. Volodin, C. Van Haesendonck, S. De Gendt, G. Groeseneken, and P. M. Vereecken, Carbon 49, 4004–4012 (2011). 46 F. Laermer and A. Urban, Microelectron. Eng. 67–68, 349–355 (2003). 47 X. Miao, L. Huli, H. Chen, X. Xu, H. Woo, C. Bencher, J. Shu, C. Ngai, and C. Borst, Proc. SPIE 6924, 69240A–69240A-8 (2008). T. Wang, K. Jeppson, L. Ye, and J. Liu, Small 7, 2313–2317 (2011). Y. Hayamizu, T. Yamada, K. Mizuno, R. C. Davis, D. N. Futaba, M. Yumura, and K. Hata, Nature Nanotechnol. 3, 289–294 (2008). 50 D. Acquaviva, A. Arun, S. Esconjauregui, D. Bouvet, J. Robertson, R. Smajda, A. Magrez, L. Forro, and A. M. Ionescu, Appl. Phys. Lett. 97, 233508–233210 (2010). 51 M. D. Volder, S. H. Tawfick, S. J. Park, D. Copic, Z. Zhao, W. Lu, and A. J. Hart, Adv. Mater. 22, 4384–4389 (2010). Downloaded 07 Feb 2013 to 129.169.173.162. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions