Manuscript submitted July 3, 2014; revised manuscript received July 28, 2014. Published August 21, 2014.
Sustaining Moore’s law requires constant transistor scaling, boost-ing the creation of new materials for future nanoelectronics appli-cations. Several emerging materials, such as Si nanowires, carbon nanotubes (CNTs), and III–V semiconductor field effect transistors (FETs), are potential components in this continuous shrinking pro-cess. In particular, CNTs are expected to overcome the physical lim-itation of current Si transistors and Cu interconnections in molecular electronics.1–5 However, their integration into Si-based metal-oxide-semiconductor field effect transistors (MOSFETs) or new nanoelec-tronics remains challenging when developing these transistors and interconnections. They are naturally deposited as bundles in a vertical direction because they tend to adhere to each other vertically. Verti-cally aligned nanotube field-effect transistors (VCNTFETs) have been proposed to yield the Si device characteristics required for 2016, as set by the International Technology Roadmap for Semiconductors.6–12 The feasibility of this vision depends on direct approaches to achieve selective depositions in the trenches or holes of Si wafers. The de-position of CNTs bundles in the trenches and holes as channels and conductors, respectively, can provide sufficient current density. The manipulation of CNTs orientation in either horizontal or vertical di-rection also plays a key role in manufacturing. This study systemati-cally evaluates the synthesis of CNTs by microwave plasma chemical vapor deposition (MPCVD) using an Fe catalyst, a CoSi2 film, and
Ni islands, which frequently serve as gate electrodes and contact materials in Si microelectronics. The selective growth of CNTs in trench/hole/planar forms is also examined in conjunction with their morphology and nanostructures. The field emission characteristics of CNTs deposited in trenches and holes are examined to determine elec-tronic performance. Moreover, the elecelec-tronics properties of nanocrys-tals and tubular structures are compared. The growth mechanism and electronic properties of nanostructured materials are addressed.
Experimental
Figure1schematically shows relative positions between plasma and sample in a MPCVD system. Figure2compares various carbon-based materials synthesized on Si wafers using the same MPCVD system. Process parameters are divided into three groups, i.e., nan-otubes, nanowires, and nanocrystals, according to the synthesized material structures. Main parameters include temperature, reactive gas type (CH4/H2,CH4/N2,CH4/H2/N2), catalyst (Fe, Co Ni),
addi-tional Si source, and patterning design for selective CNTs growth. Deposition temperatures are estimated by placing a thermocouple un-der a substrate holun-der. Nanotube morphologies and microstructures were identified by scanning (SEM) and transmission (TEM) electron
zE-mail:[email protected]
microscopes. Field emission properties were evaluated by I-V mea-surements at 10−6 Torr for electrode separations of 50 and 100μm. TableIlists the detail parameters of each sample and its corresponding morphology.
Results and Discussion
Nanostructured material synthesis by MPCVD.— Figure2shows the linkages among various carbon-based materials synthesized on Si wafers using the same MPCVD system at different parameters, such as temperature, gas types (CH4/H2, N2), deposited buffer layer (Co
and Ti),13additional Si source, and pattern design for selective CNTs growth. Three routes (1, 2 and3 in Fig.2) were compared for the catalyst-assisted synthesis of carbon nanowires and nanorods as well as selective CNTs growth. The results reveal that formation and properties of CNTs can be manipulated by applying catalysts with H2reduction gas (CH4/H2ratio= 10 sccm /100 sccm, temperature at
500◦C) leading to CNTs formation (Fig.3). In contrast, Condition 2 (route2 in Fig.2) is under lower CH4/H2 ratio (CH4/H2ratio= 1
sccm /100 sccm, temperature at 450◦C) leading to nanowire forma-tion. (Fig.4). The CH4/H2ratio influences the formation of tubular and
crystalline structures. A high CH4/H2ratio favors the formation of
C-sp2bonding (graphite structure), whereas, a low CH
4/H2ratio favors
the formation of C-sp3 bonding (diamond structure). Therefore,
car-bon atoms surround the catalysts and later precipitate from them with different CH4/H2ratios to form hollow tubes or solid nano-wires.
Un-der Condition 3 (route3 in Fig.2), CNTs were selectively deposited on patterned wafers, such as (a) parallel Fe-coated line arrays and (b) CoSix-coated hole arrays. This novel method is compatible with
Si microelectronic device manufacturing, as shown in Figs.5and6. In addition, Fig.5bshows 18μm-long CNTs selectively deposited on Fe-coated line arrays at 3μm/min. These CNTs are essentially well aligned, uniform in size, and perpendicular to the substrate. Fig. 6shows that CNTs are also selectively deposited in the holes of Si wafers patterned with hole arrays (aspect ratio 6). SEM micro-graphs reveal the 5 nm-diameter CNTs are wrapped inside the holes rather than forming well-aligned CNTs (Fig.5) under similar condi-tions, suggesting the high selectivity of this process. The wrapping of the CNTs in the holes may result from the local circular flow of gases in each hole.
To summarize, nanostructured carbon nanowires and nanorods were successfully synthesized on patterned and unpatterned Si wafers in the presence of a catalyst by varying the process parameters, such as catalyst materials, source gases, gas ratios, and deposition tempera-tures. This result also offers a different perspective on the mechanism of the catalyst-assisted MWCNTs growth. The CH4/H2 ratio
Figure 1. Relative positions between plasma and the sample in a reactor.
Figure 2. Process roadmaps of forming various carbon based materials.
Figure 3. Orientated CNTs formation under condition 1 by applying catalysts with H2reduction gas (CH4/H2ratio= 10 sccm /100 sccm, temperature at
500◦C).
Figure 4. Nanowire formation under condition 2 by applying catalysts with H2 reduction gas (CH4/H2 ratio = 1sccm /100 sccm, temperature
at 450◦C). (a) Catalyst assisted nanowire formation (b) Catalyst assisted nanorodformation.
Figure 5. Process flows of forming CNTs on Fe trench arrays, oxide film deposition→ Fe film deposition → dry etching → CNTs growth Fe assisted CNTs deposition on trench array in low and high magnification. Fe assisted CNTs deposition (b) trench arrays (c) high magnification.
TABLE I. Sample designations and conditions.
Route Catalyst Source gases (sccm/sccm/sccm) Additional Si source Plasma power, (W) Morphology
1
Fe, Co, Ni CH4/H2= 10/100 No 500∼900 nanotubes
2
Fe, Co, Ni CH4/H2= 1/100 No 500∼900 Nanowire/nanorodes
3
Fe, Co CH4/H2= 10/100 No 500∼900 Selective nanotubes growth
4
Fe, Co, Ni CH4/N2= 10/100 No 500∼900 Bamboo nanotubes
5
Fe, Co, Ni CH4/N2/H2= 10/100/100 No 500∼900 Bamboo nanotubes
6
– CH4/N2= 10/100 Yes 500∼900 Crystals
7
Co CH4/N2= 10/100 Yes 500∼900 Crystals
8
Figure 6. (a) Process flows of forming CNTs on CoSi2 holes, oxide film
deposition→ dry etching → Co film deposition → RTP (CoSi2formation)→
unreacted metal removal→ CNTs growth. CoSi2assisted CNTs deposited in
holes in low and high magnification. CoSi2 assisted CNTs (b) deposited in
hole arrays(c) high magnification.
this CH4/H2ratio promotes C-sp2bonding (graphite structure) when
high but C-sp3 bonding when low (diamond structure). Therefore,
carbon atoms surround the catalysts and later precipitate from them to form hollow tubes or solid nanorods at different CH4/H2ratios.
In addition, Conditions 4 and 5 (route4 in Fig.2) demonstrate the importance of nitrogen in bamboo NT formation. In the previ-ous studies, bamboo-like nanotubes using different catalysts, such as Fe, Ni, or Ni: Cu: Al alloy can be produced in many ways, includ-ing arc discharge,14microwave plasma CVD15and thermal pyrolysis methods.16Therefore, formation of bamboo-like structures seems to be independent of the deposition method and catalyst type. The pro-posed formation mechanisms of bamboo structures, including open-ended growth17 and stress-induced catalyst jumping,18seem unable to explain the results presented here. We propose the nitrogen plasma exhibits greater bombardment energy than hydrogen plasma because of larger atomic size of nitrogen compared to hydrogen. Therefore, the presence of nitrogen during CNTs growth keeps the upper catalyst surface clean and active to prolong surface passivation and enhance carbon bulk diffusion. In our previous study, we adopted a series of treatments including H2, NH3 and N2 treatment and concluded
ni-trogen based treatment can lead to nanoparticle agglomeration.19A higher bombardment energy of nitrogen plasma facilitates agglom-eration effects during catalyst pretreatment and initial CNTs growth stages, producing larger size nanoparticles. On the other hand, in-troducing N atoms into the carbon nanotube structure may induce distortion by changing its bonding to pentagonal, heptagonal, or other crystal lattices and increasing bending stress. Fig.7shows the bamboo NT TEM image and its corresponding SEM morphology.
Conditions 6 and 7 (route6 in Fig.2) show that appropriate
tim-ing of an additional solid source, such as Co-coated Si columns, can be used to vary SiCN film compositions, morphologies, structures, and properties20 By comparing the conditions of forming catalyst-assisted SiCN nanotubes without forming SiCN films (Condition 8, route8), Figure8shows the growth models of SiCN crystals
forma-tion and SiCN nanotubes formaforma-tion. Figures9aand9bshow SEM images of crystalline SiCN and tubular SiCN, respectively. The tubu-lar SiCN electron energy loss spectroscopy (EELS) spectrum and its corresponding X-ray photoelectron spectroscopy (XPS) spectrum are shown in Fig.10. Growth models for SiCN crystals and nanotubes suggest that additional solid Si sources are the main silicon
contribu-Figure 7. The formation of bamboo-like CNTs includes the presence of ni-trogen and keeping an active and clean top surface of the catalyst particles. (a) TEM image of bamboo-like CNTs (b) SEM morphology.
Si
Si
CH4+N2 plasma(b)
(a)
H2 CH4+H2+N2 plasma H2 H2Figure 8. Growth models of (a) SiCN crystals (b) SiCN nanotubes, the shaded particle indicates the catalysts.
Figure 9. SEM morphologies of (a) SiCN crystals and (b) SiCN nanotubes formed using Co catalyst film.
tors to SiCN crystals and nanotubes. The tubular structure may stem from the introduction of H2gas during the deposition, which may
de-lay the so-called catalyst poisoning and keep the tube end open during growth. Although some Si could be derived from the Si substrate, the plasma ionized solid Si columns actively participate in the reaction. The nano-sized catalysts promote the formation of tubular, wire, or rod morphologies. Catalytic functions in the H2-free process differ
from those in the presence of H2 gas. Catalysts have been proposed
to provide nucleation sites for SiCN crystal nucleation and effec-tively reduce the energy of formation during the initial stages. This catalytic function is lost when the growing film covers the catalytic particle. In contrast, the role of the catalyst during SiCN tube forma-tion is similar to that described in the vapor–liquid–solid mode.21The tube grows by precipitating of graphite sheets from a supersaturated catalytic droplet. The formation of a curved graphite basal plane is energetically favorable, giving rise to a tubular structure.
Horizontal and vertical CNTs growth.— The preferred growth ori-entation of CNTs was examined in detail. Field alignment using ap-plied direct current bias (−200 V) can direct CNTs growth perpen-dicularly to the substrate (Fig.11). Moreover, the catalyst density is also important in determining CNTs growth direction. Many vertically grown, dense MWCNTs are found to protrude from a single catalyst particle, which may be associated with the lower temperatures during H2-mediated reduction and CNTs deposition stages. Gaseous sources
also play crucial role in controlling CNTs growth orientations. When the flow gas is guided horizontally with respect to the substrate, CNTs grow in the same direction. Two catalysts plates produce horizontal CNTs growth while separate catalyst islands leads to horizontal CNTs formation. The CNTs are deposited horizontally between splitting cat-alyst islands. (Fig.12).
Electronic behaviors of nanostructured materials.— The current density–electric field (J–E) curves of SiCN crystals and tubes are compared in Fig.13. Nanotubesανδ crystals show a turn-on electric field (Eturn-on) of 2 and 2.5 V/μm, respectively, at J = 1μA/cm,
in-dicating that nanotubes display better field emission properties than crystals. According to the Fowler–Nordheim equation, the
relation-Figure 10. (a) EELS spectrum of SiCN nanotube recorded from the tube walls. (b) The corresponding XPS spectrum of SiCN crystals Si(2p) core level, C (1s) core level, and N (1s) core level.
ship between current (I) and applied voltage (V) can be expressed as follows:22 I = aV2exp −b V a= αAβ 2 1.1φ exp β(1.44 × 10−7) φ1/2 b= 0.95Bφ 3/2 β
Figure 11. (a) Vertically grown CNTs (DC bias -200V) and (b) their corre-sponding TEM.
Figure 12. (a) Schematic diagram of horizontal CNTs growth on parallel Ni islands. (b), (c) Corresponding SEM images of horizontal CNTs growth in between two islands.
Figure 13. J-E curves of SiCN tubes and SiCN crystals. The corresponding current density of 1mA/cm2is obtained at 2.56 and 7.98V/μm.
Where A and B are constants (A= 1.54 × 10−6, B= 6.87 × 107),
is the work function, α is the effective emitting area, and β is the field enhancement factor. Threshold voltage and emission cur-rent strongly depend on the work function of the material, effective emitting area, and field enhancement factor, which is related to the microstructure sharpness. Therefore, SiCN nanotubes present a lower threshold voltage than the nanocrystals because of their higher field enhancement factor. Figure14compares J–E curves obtained using Fe assisted CNTs at trenches and CoSi2assisted CNTs at holes. At a
current density of 1μm/cm2, turn-on electrical fields are obtained at
2.71 and 4.03 V/μm for CNTs in trenches and holes, respectively. Al-ternatively, a corresponding current density of 1 mA/cm2is obtained
at 3.97 and 6.30 V/μm for CNTs in trenches and holes, respectively. The low turn-on electrical field of the CNTs at high current density indicates a robust electrical performance. Figure15compares the J-E curves obtained from horizontal CNTs and vertical CNTs. The electri-cal fields at a current density of 1mA/cm2are 1.57 and 4.32 V/μm for
CNTs in horizontal and vertical to substrate direction, respectively. The horizontal CNTs show better field emission property than ver-tical CNTs. This may be due to the fact that the field emission of the vertical CNTs is more restricted by the catalysts at the tips and their effective emission area from defects of the CNTs body is ef-fectively diminished by the neighboring CNTs. The CNTs body size, which can contribute to field emission, is greater in horizontal CNTs in comparison to vertical CNTs. Given that the latter are restricted by neighboring CNTs, their effective area for field emission is rather limited.
Figure 14. J-E curves of CNTs at trenches and holes. The corresponding current density of 1mA/cm2is obtained at 3.97 and 6.30 V/μm.
Figure 15. J-E curves of CNTs with preference orientation horizontal and vertical to the substrate. The corresponding current density of 1mA/cm2is obtained at 1.57 and 4.32 V/μm, respectively.
Conclusions
On account of Moore’s law which relies on constant transistor scal-ing, emerging materials such as Si nanowires, carbon nanotubes, and III–V semiconductor FETs, are expected to transform future nanoelec-tronics applications validating this theory. This study demonstrates selective CNTs deposition methods that lead to vertically and hori-zontally oriented growth. In addition, several SiCN and carbon-based nanostructures are successfully synthesized using the same MPCVD system. The development of nanostructured materials with unique electrical properties may expand nanoelectronic device applications.
8. V. Derycke, R. Martel, J. Appenzeller, and P. Avouris,Nano Lett.1, 453 (2001). 9. J. Chaste, L. Lechner, P. Morfin, G. Fve, T. Kontos, J. M. Berrorir, D. C. Glattli,
H. Happy, P. Hakonen, and B. Plaais,Nano Lett.8, 525 (2009).
10. R. V. Seidel, A. P. Graham, J. Kretz, B. Rajasekharan, G. S. Duesberg, M. Liebau, E. Unger, F. Kreuppl, and W. Hoenlein,Nano Lett.5, 147 (2005).
11. Sander J. Tans, Alwin R. M. Verchueren, and Cees Dekker,Nature393, 49 (1998). 12. W. Hoenlein, F. Kreupl, G. S. Duesberg, A. P. Graham, M. Liebau, R. Seidel, and
E. Unger,Mater. Sci. & Eng. C23, 663 (2003).
13. H. L. Chang and C. T. Kuo,Diamond Relat. Mater.10, 1910 (2001). 14. Y. Saito and T. Yoshikawa,J. Cryst. Growth134, 154 (1993).
15. H. Murakami, M. Hirakawa, C. Tanaka, and H. Yamakawa,Appl. Phys. Lett.76, 1776 (2000).
16. D. C. Li, L. Dai, S. Huang, A. W. H. Mau, and Z. L. Wang,Chem. Phys. Lett.316, 349 (2000).
17. D. Zhou and S. Seraphin,Chem. Phys. Lett.238, 286 (1995).
18. M. Endo, S. Iijima, and M. S. Dresselhays(Eds.) Carbon Nanotubes, BPC Press, UK, pp. 37, 159 (1999).
19. C. H. Lin, H. L. Chang, C. M. Jsu, A. Y. Lo, and C. T. Kuo,Diamond Relat. Mater.
12, 1851 (2003).
20. H. L. Chang, J. S. Fang, and C. T. Kuo, Rev. Adv. Mater. Sci. 5, 432 (2003). 21. M. K. Sunkara, S. Shama, and R. Miranda,Appl. Phys. Lett.79, 1546 (2001). 22. C. A. Spindt, I. Brodie, I. Humphery, and E. R. Westerberg,J. Appl. Phys.47, 5248
