NANO LETTERS
The Ultimate Strength of Glass Silica Nanowires
2009 Vol. 9, No. 2 831-835
Gilberto Brambilla* and David N. Payne Optoelectronics Research Centre, UniVersity of Southampton, Southampton, SO17 1BJ, U.K. Received November 25, 2008; Revised Manuscript Received January 5, 2009
ABSTRACT In the past decade nanowires have attracted an increase interest because of their extraordinary mechanical strength. In fact, material properties in the nanoregime are extremely different from those found in macroscopic samples: few crystalline materials have shown a tensile strength in excess of 10 GPa in the form of nanowires. Still the length of defect-free crystalline nanowires is limited to a few millimeters and the strength of long nanowires is compromised by defects. The strength of glass nanowires is less affected by single defects. In this paper we present the ultimate strength of glass silica nanowires manufactured by a top-down fabrication technique; this is the highest value reported for glass materials. The measured ultimate strength is in excess of 10 GPa and increases for decreasing nanowire diameters. Scanning electron micrographs of the broken fragments showed a fragile rupture.
The mechanical properties of wires are known to increase significantly for decreasing diameters.1 Consequently, the quest for ultrahigh strength materials has investigated several nanomaterials, including polymer,2 gold,3 tungsten disulfide (WS2),4 silicon carbide (β-SiC),5 silicon nitride (R-Si3N4)6 nanowires, carbon nanotubes,7 and graphene single sheets.8 Their ultimate strength has been measured to be 7, 2-8, 11, 53.4, 17-59, 11-63, and 130 Gpa, respectively. The great majority of these nanomaterials are crystalline, and the presence of a single defect eventually decreases their ultimate strength: a single defect involving a missing atom in an otherwise perfect carbon nanotube 30000 km long would have a 20% decrease in the tensile strength with respect to that of a defect-free carbon nanotube.9 This is a major drawback since most of the crystalline nanowires can be fabricated flawlessly only for lengths of the order of millimeters. On the contrary, glasses lack any long-distance atomic periodicity; thus they present a greater flexibility to accommodate defects. This is associated with a lack of melting temperature and a continuous decrease of the viscosity for increasing temperatures. In particular, silica has an extremely flexible local atomic structure: the angles between neighboring atoms (Figure 1) RSi-O-Si and RO-Si-O have mean values of 151° and 109.7° but a standard deviation of σRSi-O-Si ∼ 9-12° and σRO-Si-O ∼ 5°; Si-O bonds have a strength (621.7 kJ/mol)10 comparable to the value observed for CC bonds in sp2 hybridization (680 kJ/mol)11 typical of graphite, graphene, and carbon nanotubes. This results in superior macroscopic properties, including high glass-transition (1410 K) and softening * Corresponding author,
[email protected]. 10.1021/nl803581r CCC: $40.75 Published on Web 01/26/2009
2009 American Chemical Society
Figure 1. Schematic of the local atom bonding in silica glass. RSi-O-Si, RO-Si-O, dSi-O, and dO-O represent the angle between oxygen and two silicon atoms, the angle between silicon and two oxygen atoms, distance between silicon and oxygen and the distance between two oxygen atoms.
temperatures (1940 K), large Young modulus (72 GPa),12 and hardness value (6.5 in the Mohs scale). Moreover, because of its glassy nature, silica nanowires can potentially be manufactured in extremely long lengths: virtually infinitely long tapers can be fabricated from optical fibers.13 Still, in his studies on materials strength, Griffith showed that macroscopic silica samples present a small strength (0.2 GPa) because of the numerous surface imperfections;14 he also related the ultimate strength to the flaw size, showing that bigger flaws are associated with lower strengths. The flaw size can only be a small fraction of the wire diameter, thus
Figure 2. Representation of a silica nanowire manufactured by the top-down “modified flame brushing technique”. The nanowire is connected to two fiber pigtails having 125 µm diameter by two transition regions. The typical length of the nanowires used in these experiments was 6 mm.
the manufacture of silica nanowires allows to minimize the flaw size and maximize its strength. Silica nanowires used in these experiments were fabricated from telecom optical fibers with 125 µm diameter by the so-called “modified flame brushing technique”.15 This is a top-down technique, previously developed to manufacture compound-glass nanowires, which is based on a microheater (NTT-AT, Japan) scanned over an optical fiber being stretched. The repeated scanning movement continuously fire-polishes the nanowire surface, thus decreasing the surface imperfections. Because of mass conservation, in the heated region the diameter decreases. By controlling the stretching speed and the fiber temperature it is possible to accurately
control the taper profile. A schematic of the tapered fiber with a nanowire in the minimum waist region is shown in Figure 2. The nanowire is attached to two fiber pigtails by two transition regions, which allow the easy handling of the nanowire with instrumentation typical of macroscopic experiments. The nanowire diameter has been estimated using the mass conservation law: approximately half of the samples were gold coated (thickness ∼3 nm) and analyzed with a high-resolution field emission gun scanning electron microscope (FEG-SEM) (JEOL JSM 6500F, Japan) or a highmagnification optical microscope to verify the accuracy of the estimation, which resulted to be within the measurement error (∼3%). Figure 3 shows micrographs of the samples in the nanowire region. The silica nanowires contained some impurities; Na, K, Ca, Mg, Al, Fe, and Ti concentrations were estimated to be considerably smaller than 0.0001% mol. Since silica nanowires have been manufactured from optical fibers, the GeO2 used in the fiber core to increase the glass refractive index diffuses all over the nanowire section and contributes as an additional 0.02% mol impurity. The nanowire ultimate strength σf, defined as the maximum stress a material can withstand, has been measured in a simple static experiment by gradually increasing the weight at the lower extremity of the vertically held fiber pigtail until
Figure 3. Micrographs of silica nanowires taken by a FEG-SEM. Nanowire samples were manufactured by the so-called “modified flame brushing technique”, positioned on a metallic stub, and then coated with ∼3 nm of gold to avoid electrostatic charging. Radii estimated from the SEM micrographs were 81 nm (a), 94 nm (b), 87 nm (c), and 96 nm (d), respectively. The related values predicted from mass conservation considerations were 82.5 nm (a), 95 nm (b), 87.5 nm (c), and 99 nm (d). 832
Nano Lett., Vol. 9, No. 2, 2009
Figure 4. Schematic of the setup used to measure the silica nanowire ultimate strength (σf). (a) The nanowire was clamped at one extremity and held vertically; a small container was attached at the bottom extremity and the weight was gradually increased by adding small masses until (b) fracture occurred. (c) The total mass below the fracture line was then measured using a calibrated scale.
Figure 5. Summary of ultimate strength (σf) measurements carried out on nanowires with radius r. The nanowire was held vertically by its pigtails, and the weight applied at its lower extremity was gradually increased until fracture occurred. σf was evaluated from eq 1. Error bars include the uncertainty in the nanowire diameter, in the absolute value of the weight added, and in the value of the mass recorded by the high precision scale.
fracture occurred (Figure 4). The mass m was then measured using a high-precision scale and σf was derived from the simple relation m σf ) 2 πr
(1)
where r is the nanowire radius. Figure 5 shows a summary of the results carried out on nanowires with r in the range 60-300 nm. Generally σf is in excess of 10 GPa and increases for decreasing r. A maximum of 26 GPa has been recorded for r ∼ 62 nm. This value is considerably higher than that recorded for conventional bulk silica and optical fibers. Typical σf values for silica fibers are ∼6 GPa in air at room temperature.16 The theoretical σf for a material can be estimated from Young’s modulus E, the equilibrium atomic Nano Lett., Vol. 9, No. 2, 2009
Figure 6. Comparison of the ultimate strength σf achieved for silica nanowires to values reported in literature for gold (2), tungsten disulfide (WS2) (3), silicon carbide (βSiC) (4), silicon nitride (RSi3N4) (5) nanowires, and carbon nanotubes (6). Bulk silica, high strength steel (A514), and Kevlar are reported on the right for reference.
separation a, and the period of the assumed sinusoidal interatomic force λ as17 σf )
Eλ π 2a
(2)
Assuming a ) λ /2, the theoretical limit for silica has been evaluated to be of the order of E/π.18 Recent measurements showed that E in silica nanowires can exceed 100 GPa,19 bringing the theoretical σf of silica nanowires above 30 GPa. A possible explanation for the increased strength observed in the silica nanowires manufacture by “modified flame brushing technique” is the reduced possibility of having large cracks in nanowires. Using Griffith’s fracture mechanics equation,13 a crack length of ∼2 nm is calculated for σf of the order of 20 GPa. It is reasonable to assume that the small nanowire size does not support cracks which are longer than 833
Figure 7. Micrographs of the fractured surface of silica nanowires taken by a FEG-SEM. The nanowire radii measured with the FEG-SEM were 95 nm (a), 89 nm (b), and 190 nm (c), respectively. Samples were positioned on a metallic stub and then coated with ∼3 nm of gold to avoid electrostatic charging. All samples present a flat fracture surface and there is no evidence of diameter tapering or corrugated fracture surfaces.
a nanometer. Humidity is another cause of degradation of silica samples.20 While nanowire manufactured from optical fibers with a flame contains a considerable amount of hydroxyls (OH) because of the high atomic hydrogen content in the flame fuel, nanowires fabricated by “modified flame brushing technique” are manufactured with a ceramic microheater; thus they contain considerably fewer OH groups than those manufactured with a burner. Figure 6 compares σf of silica nanowires with previous results reported in literature for crystalline Au, WS2, β-SiC, R-Si3N4 nanowires, and carbon nanotubes with r in the range 9-300 nm. Silica nanowires present strength comparable to the best crystalline nanowires. Yet, silica has the fundamental advantage of glasses: it can be processed to provide extremely long lengths of fibers: tens of kilometers of fibers can be continuously manufactured and a similar process for the fabrication of extremely long nanowires can be envisaged. Moreover, Figure 6 shows that the strength of silica in the nanowire form is 2 orders of magnitude larger than that observed in the macroscopic bulk material and considerably larger than the values recorded for conventional high strength materials like steel ASTM A514 (σf ) 0.76 GPa) and Kevlar (σf ) 3.3GPa).2 Figure 7 shows micrographs of the fractured section of silica nanowires taken with an FEG-SEM. All samples present a flat fracture surface, and there is no evidence of diameter tapering or corrugated fracture surfaces. Although it was suggested21 that at ductile fracture can appear as completely brittle (i.e., flat rupture surface) because of the lack of stable plastic regimes, this was observed only for negligible elongations. The neat fracture surfaces jointly with the lack of tapering in the analyzed samples seem to indicate that nanowires experienced a fragile fracture. Still, the ductile fracture explained in terms of glass flow requires for macroscopic samples a σf smaller than 15 GPa in air at room temperature;14,21 the occasionally higher σf observed at small nanowire sizes might indicate that at small sizes a fragile fracture occurs over a short time scale and/or some of the factors facilitating the plastic flow are inhibited. In fact, although data relative to the viscous flow activation energy ∆Gflow at room temperature are not available, measurements at high temperature showed that below 1700 K ∆Gflow can be assumed to be 712 kJ/mol, comparable with the bond strength.22,23 Moreover, because of the relatively large 834
diameter, the melting point depression observed in covalent materials24 and its related decrease in the softening temperature should be negligible. For long-term reliability and practical applications, it is envisaged that silica nanowires should be embedded in a thin protective water-proof polymer layer. In fact, it has been shown that uncoated silica fibers and microwires degrade with time when exposed to humidity25 and coating provides a long-term solution to reliability issues.26 Moreover, because of the fragile rupture, it is reasonable to assume that the fracture toughness of silica nanowires cannot be significantly higher than the value observed in bulk silica samples (∼1 MPa m1/2 27); the use of a protective polymeric coating could have the additional benefit of improving the overall fracture toughness. In conclusion, the high ultimate strength of silica nanowires manufactured by top-down technique together with the possibility to pull silica into extremely long wires can open a wealth of applications in the macroscopic world ranging from security, structural engineering, and aerospace to more futuristic applications like the space elevator or the orbital tower.28,29 Acknowledgment. G.B. gratefully acknowledges the Royal Society (London) for his research fellowship. The authors thank Wei Loh and Dave Richardson for stimulating discussions. References (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)
Dzenis, Y. Science 2004, 304, 1917–1919. Crist, B. Annu. ReV. Mater. Sci. 1995, 25, 295–323. Wu, B.; Heidelberg, A.; Boland, J. J. Nat. Mater. 2005, 4, 525–528. Kaplan-Ashiri, I.; Cohen, S. R.; Gartsman, K.; Ivanovskaya, V.; Heine, T.; Seifert, G.; Wiesel, I.; Wagner, H. D.; Tenne, R. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 523–528. Wong, S. S.; Sheehan, P. E.; Lieber, C. M. Science 1997, 277, 1971. Iwanaga, H.; Kawai, C. J. Am. Ceram. Soc. 1998, 81, 773–776. Yu, M.-F.; Lourie, O.; Dyer, M. J.; Moloni, K.; Kelly, T. F.; Ruoff, R. S. Science 2000, 287, 637–640. Lee, C.; Wei, X.; Kysar, J. W.; Hone, J. Science 2008, 321, 385–388. Pugno, N. M.; Ruoff, R. S. Philos. Mag. 2004, 84, 2829–2845. Silicon chemistry: from the atom to extended systems; Jutzi, P., Schubert, U., Ed.; Wiley-VCH: Weinheim and Cambridge, 2003; ISBN 3527306471. Bhushan, B. Principles and Applications of Tribology; Wiley: New York, 1999. F300 Specifications spreadsheet, Heraues Quarzglas GmbH. Vukovic, N.; Broderick, N. G. R.; Petrovich, M.; Brambilla, G. J. LightwaVe Technol. 2008, 20, 1264–1266. Nano Lett., Vol. 9, No. 2, 2009
(14) Brambilla, G.; Koizumi, F.; Feng, X.; Richardson, D. J. Electron. Lett. 2005, 41, 400–402. (15) Proctor, B. A.; Whitney, I.; Johnson, J. W. Proc. R. Soc. London 1967, 297A, 534–557. (16) Frenkel, J. Z. Phys. 1926, 37, 572–609. (17) Kurkjian, C. R.; Krause, J. T.; Paek, U. C. J. Phys., Colloq. 1982, 43, 585–586. (18) Silva, E. C. C. M.; Tong, L.; Yip, S.; Van Vliet, K. J. Small 2005, 2, 239–243. (19) Griffith, A. A. Philos. Trans. R. Soc. London 1921, 221A, 163–198. (20) Armstrong, J. L.; Matthewson, M. J.; Kurkjian, C. R. J. Am. Ceram. Soc. 2000, 83, 3100–3108. (21) Marsh, D. M. Proc. R. Soc. London 1964, 282, 33–43.
Nano Lett., Vol. 9, No. 2, 2009
(22) Volarovich, M. P.; Leontieva, A. A. J. Soc. Glass Technol. 1936, 20, 139–143. (23) Hetherington, G.; Jack, K. H.; Kennedy, J. C. Phys. Chem. Glasses 1964, 5, 130–136. (24) Farrell, H. H.; Van Siclen, C. D. J. Vac. Sci. Technol., B 2007, 25, 1441–1447. (25) Brambilla, G.; Xu, F.; Feng, X. Electron. Lett. 2006, 42, 517–519. (26) Xu, F.; Brambilla, G. Jpn. J. Appl. Phys. 2008, 47, 6675–6677. (27) Lucas, J. P. Scr. Metall. Mater. 1995, 32, 743–748. (28) Isaacs, J. D.; Vine, A. C.; Bradner, H.; Bachus, G. E. Science 1966, 151, 682–683. (29) Pearson, J. Acta Astronaut. 1975, 2, 785–799.
NL803581R
835
