Supplementary Informations for

Drawing Circuits with Carbon Nanotubes: Scratch-Induced Graphoepitaxial Growth of Carbon Nanotubes on Amorphous Silicon Oxide Substrates

By Won Jin Choi1, Yoon Jang Chung1, Yun Ho Kim1, Jeongho Han2, Young-Kook Lee2, Ki-jeong Kong1*, Hyunju Chang1, Young Kuk Lee1, Byoung Gak Kim1and Jeong-O Lee1*

[1] W. J. Choi, Y. J. Chung, Dr. Y. H. Kim, Dr. K. J. Kong*, Dr. H. J. Chang, Dr. Y. K. Lee, Dr. B. G. Kim, Dr. J-O. Lee*
Advanced Materials Division, Korea Research Institute of Chemical Technology (KRICT)
Daejeon, 305-343, South Korea
E-mail: ,

[2]J. Han, Prof. Y.-K. Lee
Department of Materials Science and Engineering, Yonsei University

Seoul, 120-749, South Korea

*Corresponding authors: Dr.Jeong-OLee, 141Gajeongro, Yuseong, Korea Research Institute of Chemical Technology, Daejeon, Postal # 305-343. Telephone, (82-42)860-7303. Fax: (82-42)860-7508. Corresponding author e-mail: -jeong Kong, 141Gajeongro, Yuseong, Korea Research Institute of Chemical Technology, Daejeon, Postal # 305-343. Telephone, (82-42)860-7367. Fax: (82-42)860-7508. Corresponding author e-mail:

Figure S1: Statistics on the scratch lithography technique according to diamond diameter size. (a) Distribution of scratched lines with the same scratch depth at various diamond particle sizes. It is clearly shown here that a finer control over scratch depth can be achieved when smaller particles are used to polish the substrate. (b) and (c) summarizes these results in terms of scratch depth and density, respectively. As can be seen in the figures, as the diamond particle size increases, scratch density tends to decrease while scratch depth increases. (d) Schematic diagram of scratch lithography. R equals the diameter of diamond particles,  the scratch depth, and 2a denotes the widths of scratches.

Fig. S2. The effect of gas flow direction on the aligned growth of SWNTs. Though there are a few that align with gas flow direction, most of the tubes were aligned with the scratches on substrate.

Fig. S3. The lengths distributions of SWNTs. (a) The lengths distributions of aligned SWNTs grown on scratched SiO2 substrate. (b) The lengths distributions of randomly grown SWNTs on conventional SiO2 substrate. All growth parameters were the same as in the case of aligned SWNTs.

Figure S4. (a) SEM image of horizontally aligned carbon nanotubes on ST-cut quartz crystal. (b) SEM image of randomly grown carbon nanotubes on fused silica substrate. The scale bar in (a) and (b)are25 μm. (c) compares the Raman spectra of the random and aligned SWNTS. (d) average G+ peak shift values in the aligned and random SWNTs.

Figure S5.(a) The AFM image of a conventional silicon dioxide substrate with photolithographically patterned trenches. The trench width was 2 μm with 8 μm spacings. (b) The XPS profiles of the Si 2p peaks from the flat(blue curve as a reference) and patterned SiO2 substrates (wet etched using buffered oxide etcher(BOE)). Here, the 2,4 and 2,8 linespace patterns denotephotolithographic patterns with 2μm wideactive regions (etched regions) with 4 or 8 μm spacings.

Figure S6. The dielectric performance of scratched and normal SiO2 substrates. Nanoscale air gaps may be present in the scracthed structure, causing deterioration in the dielectric properties of the film. However, as can be seen variance is non-existent at very high and low frequencies, and the difference is not very large even at intermediate ranges.

Figure S7. Electrical properties of the aligned SWNTs according to SWNT density. A schematic diagram of how the density of aligned SWNT can influence device performance is shown in (a). Since the effective ‘channel width’ is determined by the number of SWNTs that are connected between the source and drain electrodes, SWNT density is a very important factor in this case. The resistance extracted from the I-V curves of various samples is summarized in (b) as a function of SWNT density. The profile shows that resistance decreases as density increases, which is expectable from the previous assumption. Likewise, the on/off ratio and on/off current values are given in (c) and (d), respectively. In terms of transistor performance, the threshold for proper amplification characteristics is ~10 SWNTs between the source and drain electrodes. When the density is higher than that, the devices fail to show significant modulation and have a more metallic character.

Numerical simulations.

All the density functional theory (DFT) based calculationsare performed by usingperiodic plane-wave approach as implemented in the VASP code.[1,2] Frozen-core projector augmented wave pseudopotentials[3] and Perdew-Burke-Ernzerhofexchange-correlation functionals[4,5]were used. Kohn–Sham wave functions of valence electrons were expanded using a plane wave basis set within a specified energy cutoff of 500 eV. To ensure that theresults of the calculations are directly comparable, identicalconditions are employed for all systems.Geometry relaxations wereperformed with the criterion that ionic forces are less than0.01 eV/Å.In VASP modeling of surface properties, the dipole correction, proposed by Neugabeauer and Scheffler[6], was used. The correction removes the artificial field arising in empty space from the use of periodicboundary conditions in solution of Poisson equation by fast Fourier transform (FFT) method.

At first, α-quartz crystal system is considered as a representative of SiO2 bulk. In spite of the underlying amorphous nature of SiO2, only the crystalline SiO2 isconsidered in this work in order to reduce the computationaleffort.Theintegrations over the Brillouin zone are performed using a 5 × 5 × 4 Monkhorst–Pack k-mesh, sufficiently largefor achieving convergence in bulk calculations. Fitting the Murnaghan’s equationof state to the total energy variation as a function of cell volume, we have obtained following optimized lattice parameters:a=b=4.994 Å,c=5.495 Å, α=β=90°, γ = 120°.

Next, to simulate the unscratchedSiO2surface model, as shown in Fig. 4(c),we took the silicon terminated SiO2 (0001) surface, a repeating slabstructure consisting of seven O–Si–O layers and additional Si monolayer with a 16.5Åvacuum region. The surface Brillouin zone (SBZ) is taken as 2a ×2b unit of α-quartz. The slab contains 32 silicon atoms and 56 oxygen atoms and eight hydrogen atoms are used to passivate the bottom siliconatoms. The k-space integration is replaced by a summation over a 3×3 uniform k-point mesh in the SBZ.As was explained in main text, suboxides were found in scratched substrate. Thus we have used Si-rich surface as a model of scratched substrate. The Si-rich surface was modeled by adding two silicon atoms to the top of the surface cell (2a×2b) of the unscratched surface. The two Si adatoms formed a dimer structure with a Si–Si bond length of 2.31 Å. This Si dimer acts as a preferential adsorption site for sp2 bonded carbon network, gives orinetation dependent growth of CNT.

REFERENCES

[1] Kresse, G., and Furthmüller, J. Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys. Rev. B 54, 11169 (1996)

[2] Kresse, G. and Furthmüller, J. Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set. Comput. Mater. Sci. 6, 15-50 (1996)

[3] Kresse, G. and Joubert, D. From ultrasoft pseudopotentials to the projector augmented-wave method. Phys. Rev. B 59, 1758-1775 (1999)

[4] Blöchl, P. E. Projector augmented-wave method. Phys. Rev. B 50, 17953-17978 (1994)

[5] Perdew, J. P., Burke, K., and Ernzerhof, M. Generalized gradient approximation made simple. Phys. Rev. Lett. 77, 3865-3868 (1996)

[6] Neugebauer, J., and Scheffler, M. Adsorbate-substrate and adsorbate-adsorbate interactions of Na and K adlayers on Al(111). Phys. Rev. B 46, 16067-16080 (1992).