Nanospring Pressure Sensors Grown by Glancing Angle Deposition

NANO LETTERS

Nanospring Pressure Sensors Grown by Glancing Angle Deposition

2006 Vol. 6, No. 4 854-857

S. V. Kesapragada, P. Victor, O. Nalamasu, and D. Gall* Department of Materials Science & Engineering, Rensselaer Polytechnic Institute, Troy, New York 12180 Received January 19, 2006; Revised Manuscript Received February 28, 2006

ABSTRACT Arrays of Cr zigzag nanosprings and slanted nanorods, 15−55 nm and 40−80-nm-wide, respectively, were grown on SiO2/Si substrates by glancing angle deposition. The arrays exhibit a reversible change in resistivity upon loading and unloading, by 50% for nanosprings and 5% for nanorods, indicating their potential as pressure sensors. The resistivity drop is due to a compression of nanosprings (by a measured 19% for an applied external force of 10-10 N per spring), which causes them to physically touch their neighbors, providing a path for electric current to flow between nanosprings. Repeated loading and unloading at large loads (g1 MPa) results in irreversible plastic deformation and a degradation of the pressure sensitivity.

Thin films deposited from oblique angles (>70°) by physical vapor deposition (PVD) exhibit highly underdense columnar morphologies,1-3 which are due to atomic shadowing effects under limited adatom mobility conditions. Glancing angle deposition (GLAD), a technique developed by Robbie and Brett,1,2 exploits this effect to nanoengineer thin-film morphologies and create arrays of uniquely shaped nanostructures including zigzags,2 pillars,4 chevrons,5,6 spirals7,8 and Y shapes.9 Potential applications include optical polarizers,10 high birefringence biaxial films,11 thin-film wave plates,12 optical humidity sensors,13 magnetic storage media,14,15 nanoemitters,16 and actuators.17 The mechanical response of GLAD nanospring structures has been investigated using finite element analyses,18 nanoindentation measurements on an array of SiO springs,19 and with a novel tip-cantilever assembly attached to a conventional atomic force microscope to measure the deformation response of individual Si nanosprings.20 In this letter, we demonstrate that GLAD nanostructures may find applications as sensitive pressure sensor arrays. The envisioned sensors consist of metallic zigzag nanosprings or slanted nanorods that are grown on insulating substrates. On the application of load, the nanostructures compress to yield an electrical contact between individual nanorods. Hence, changes in a localized load can be sensed by the resistance measured across an array of contacts. We report on the growth and electrical characterization of Cr nanosprings and nanorods, 15-55 nm and 40-80-nm-wide, respectively, grown on SiO2/Si substrates. Scanning electron microscopy (SEM) indicates that the nanostructures compress upon the application of an external load. They exhibit a * Corresponding author. Phone: 1-518-276-8471. Fax: 1-518-276-8554. E-mail: [email protected]. 10.1021/nl060122a CCC: $33.50 Published on Web 03/16/2006

© 2006 American Chemical Society

reversible resistivity decrease when compressed but also show degradation and reduced sensitivity after multiple loading-unloading cycles. All Cr nanostructures were grown in a load-locked ultrahigh vacuum magnetron sputter deposition system, described in detail in ref 21, onto Si(001) substrates that were thermally oxidized to form a 1.5-µm-thick insulating SiO2 surface layer. A 7.5-cm-diameter Cr target (99.95% pure) was positioned 10 cm from the substrate with the target center within the plane of the substrate surface. A collimating plate covering the substrate surface was used to prevent any nondirectional flux from striking the substrate and to control the azimuthal deposition angle to be 84°. Slanted Cr nanorods were grown with stationary substrates while the zigzag pillars were obtained by rotating the substrate by 180° about its polar axis between successive stationary deposition steps. The sample heating due to the deposition plasma was monitored by a thermocouple within the sample stage and was below 125 °C for all depositions. Deposition was done at 0.26 Pa (2.0 mTorr) in 99.999% pure Ar that was further purified using a Micro Torr purifier. Sputtering was carried out at a fixed power of 500 W, yielding a deposition rate of ∼2 nm/min. Cu contacts, 8 mm apart, were deposited through a mask in a planar four-point probe configuration on top of the nanostructures. Contact resistances were found to be negligible so that current-voltage measurements were carried out in a two-point probe measurement, using a computerinterfaced DC voltage source and pA meter (HP4140). Pressure was applied to the nanostructures between the electrical contacts using rectangular-shaped, fine-polished glass blocks with calibrated masses of 10.5 and 14.9 g. A thin supporting glass plate was positioned on the sensor sample prior to loading investigations to ensure uniform

Figure 1. Cross-sectional scanning electron micrographs of Cr nanosprings (a) as-deposited and (b) after loading experiments. (c) Average column-tilt angles (β) of the four arms of the nanosprings for the as-deposited and post-load samples. The schematic in the inset shows the nanosprings constructed with measured β values.

distribution of the load. The plate was kept on the sample during multiple load-unload sequences. After each loading and unloading step, the voltage was scanned from 0 to 20 V for nanorods and 0 to 90 V for nanosprings, respectively, and the corresponding current was recorded. The resulting resistance values were obtained by averaging over the voltage sweep and correcting for specimen geometry by subtracting the resistance originating from inactive regions of the sensor, and were then plotted as a function of loading cycle. Film morphologies were investigated using a scanning electron microscope (JEOL SUPRA 55) operated at 5 kV and an extractor current of 183 µA. Cross-sectional specimens were prepared by cleavage of the substrate both prior to and after mechanical loading experiments. Figure 1a is a cross-sectional scanning electron micrograph from as-deposited Cr nanosprings grown on SiO2/Si. The zigzag nanosprings were grown in a four-step process, with a polar substrate rotation of 180° after each step, resulting in springs with four alternately tilted arms. The average nanospring height is 207 ( 15 nm. Their width increases from 15 to 55 nm with increasing height, as visualized by the dotted line around a spring in Figure 1a. This increase in width is attributed to a competitive columnar growth mode, which is common for PVD layers and is expected particularly for GLAD nanostructures grown on flat substrates.22,23 The nanospring area density, as determined from plan-view micrographs (not shown), is 62 µm-2. Figure 1b is an SEM micrograph from the same sample as in part a; however, it Nano Lett., Vol. 6, No. 4, 2006

is obtained after the nanosprings have been exposed to multiple loading and unloading cycles. The springs in b exhibit a comparable area density to those in a, indicating that their adhesion to the substrate was sufficient to withstand the mechanical testing. However, the mechanical testing caused plastic deformation, as indicated by the average height, 167 ( 11 nm, which is 19% smaller than the value of the as-deposited springs. The plastic deformation due to loading is also illustrated in Figure 1c, which is a plot of the measured tilt angle, β, of the nanospring arms before and after mechanical testing. The plotted angles represent average values from 15 springs, with the error bars corresponding to the standard deviation of each dataset. The β value is largest for the first arm, β ) 56°, and drops to 38°, 38°, and 42° for the following arms. We attribute this decrease to column expansion with increasing layer thickness. The abrupt substrate rotation by 180° during the deposition causes multiple nucleations on the surface of the previous arm, leading to an increase in the width of the growing arm and a decrease in β. Similar arm-broadening has been observed during the growth of MgF2 nanosprings.24 All β values are considerably smaller than the deposition angle R ) 84°, as expected based on various experimental and theoretical studies that have reported a relationship between the deposition angle and the column tilt in PVD layers.25-28 Our β values are also smaller than what would be expected from the tangent rule (tan β ) 1/2 tan R),28 consistent with a reported breakdown of this rule at large R,29 and supporting reports that β may also depend on the surface diffusion length,30 material properties,31 and crystal structure.32 The tilt angles of the zigzag arms after mechanical testing are consistently larger in comparison with the as-deposited springs, as plotted in Figure 1c and also illustrated in the inset, which is a schematic of the spring shapes drawn using the measured average tilt angles. The maximum change of 15 ( 2° is observed in the uppermost arm (no. 4) of the nanosprings, which is the most accessible arm. These results clearly indicate a plastic deformation of the nanosprings during multiple loading-unloading cycles, which results in a degradation of the reversible pressure sensing functionality, as discussed below. The deformation of these Cr nanosprings is similar to that reported for SiO nanosprings,19 which show an elastic-to-plastic transition in their deformation behavior with increasing load. Figure 2 is a plot of the measured resistance R of a 5 × 3.5 mm2 Cr nanospring array during a sequence of 10 loading-unloading cycles. The as-deposited sample has a resistance of 2.5 × 1010 Ω, corresponding to a resistivity of this spring array of 3.5 × 105 Ω‚cm. This is 10 orders of magnitude higher than the reported room-temperature bulkresistivity of Cr of 13 µΩ‚cm.33 The high resistivity is attributed to the highly porous microstructure, exhibiting open gaps between the individual nanosprings as observed in Figure 1. The first solid circle in Figure 2 corresponds to the resistance R ) 1.4 × 1010 Ω after initial loading with 8.0 × 105 Pa. This load corresponds to an average compressive force of 0.1 nN per nanospring and results in a decrease of R by 56%. That is, the nanospring pressure sensor becomes 855

Figure 2. Resistance of a Cr nanospring array when exposed to multiple loading-unloading cycles with a pressure of 8.0 × 105 Pa.

more conductive under compressive load. This is attributed to nanosprings physically touching their neighbors when compressed, providing a path for electric current to flow between nanosprings and, consequently, reducing the resistance parallel to the substrate surface. Successive unloading results in a more than twofold increase in R, to 3.3 × 1010 Ω. This resistance after the first unloading cycle is higher than that of the as-deposited sample, which is likely due to sample degradation (like scraping off of some springs) caused by the initial loading experiment. The plot of the 10 loadingunloading cycles in Figure 2 exhibits overall an oscillating function with the sample having lower R when loaded than unloaded, indicating a (at least partially) reversible pressure sensing functionality. The resistance is nearly constant R ) 15 ( 2 × 1010 Ω for all measurements with the sample loaded, indicating that the nanosprings reach a certain level of compression for a constant load of 8.0 × 105 Pa. In contrast, R values vary widely for the unloaded sample. In particular, R is high for cycles 1, 3, and 5 but remains nearly at the level of the loaded resistance for cycles 2, 4, and 6. We attribute the high resistance to cycles where the springs completely relax while in other cases the springs remain compressed. This indicates that the nanosprings deform as an ensemble rather than individually, which can be explained by the interlocking of neighboring spring arms, consistent with the interpretation that neighboring springs touch when compressed. For higher cycle numbers, the nanosprings lose their ability to fully relax. In particular, the resistance change in cycles 8-10 is less than 20%. This corresponds to a degradation in the pressure sensor’s sensitivity, which can be attributed to plastically deformed nanosprings. That is, after 10 loading cycles, the nanosprings remain compressed, nearly independent of the applied load. This is consistent with the SEM micrograph in Figure 1b, showing an array of compressed springs after multiple loading-unloading cycles, even if no load is applied when taking the micrograph. Figure 3a is a cross-sectional SEM micrograph from slanted Cr nanorods grown on SiO2/Si. The rods are 40-80 nm wide, 1.12 ( 0.05 µm tall, and are tilted toward the direction of the Cr deposition flux with an angle β ) 61°, which is slightly larger than 56°, the first arm tilt of the 856

Figure 3. Cross-sectional scanning electron micrographs of (a) as-deposited and (b) post-load Cr nanorods grown on SiO2.

nanosprings. Near the substrate surface is a high density of 37 ( 12-nm-wide nanorods, which initially nucleate but die out and are overgrown during a competitive columnar growth.23 We estimate that the layer density (in g‚cm-3) of this nanorod array is approximately two times larger than that for the nanosprings shown in Figure 1, based on the 4× larger average nanorod cross-sectional area and the 2× smaller nanorod area density, 28.5 µm-2, as determined from plan-view micrographs (not shown). The higher layer density of the nanorods versus the nanosprings yields an electrical resistivity that is approximately 6 orders of magnitudes smaller, as shown below. Figure 3b is the corresponding SEM micrograph for the sample after testing as a pressure sensor. The nanorods are delaminated from the SiO2 surface, which is attributed to the repeated loading and unloading, and causes the degradation of the pressure sensor as described below. The mushy area in the micrograph below the nanorods is due to the narrow rods that die out during the growth, as described above. Figure 4 is a plot of the resistance of a Cr nanorod array during a loading-unloading experiment. R for the asdeposited layer is 58.9 kΩ, corresponding to a resistivity of 20.6 kΩ‚cm, which is 4.2 × 105 times smaller than that for the Cr zigzag array, indicating that the slanted nanorods exhibit a much higher degree of connectivity, that is, they touch their neighbors already in the as-deposited state. The application of a 5.6 × 105 Pa load leads to a ∼5% decrease in R. This relative decrease is much smaller than the 56% for the nanosprings, which can be attributed to the initial rod connectivity and higher density of the nanorod layer. The inset in Figure 4 is a blow-up of the first loading cycles, showing a reversible decrease and increase in R upon loading and unloading, respectively. However, in addition the Nano Lett., Vol. 6, No. 4, 2006

References (1) (2) (3) (4) (5) (6) (7) (8) (9)

Figure 4. Resistance of Cr nanorods during loading-unloading cycles with increasing loads of 0.56, 0.80, and 1.13 MPa. The inset is a blow-up of the resistance values at 0.56 MPa.

observed R value also increases continuously, which is attributed to sample degradation as shown in Figure 3b. The steep increase in R after cycle 10 in Figure 4 indicates that the sample degradation is exacerbated when the applied load is increased to 8.0 × 105 Pa and 1.3 × 106 Pa. Despite the dramatic increase in R and the related detachment of nanorods from the substrate, the R versus cycle curve indicates a continued reversible change ∆R upon loading and unloading. ∆R ) -60 ( 30 Ω remains constant through the experiment. That is, it is independent of the applied load to within the accuracy and repeatability of the measurement. In conclusion, both nanospring and slanted nanorod arrays exhibit a reversible change in resistivity upon loading and unloading, indicating their potential as pressure sensors. They both exhibit distinct degradation behavior: nanosprings plastically compress leading to a decrease in R for the unloaded samples and a decreasing pressure sensitivity. Slanted nanorods detach from the substrate after multiple loading-unloading cycles, leading to an increase in resistivity that is larger in magnitude than the reversible ∆R upon loading and unloading. Acknowledgment. This research was supported by the National Science Foundation, Division of Manufacturing and Industrial Innovation, under grant no. DMII-0423358.

Nano Lett., Vol. 6, No. 4, 2006

(10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) (25) (26) (27) (28) (29) (30) (31) (32) (33)

Robbie, K.; Brett, M. J.; Lakhtakia, A. Nature, 1996, 384, 616. Robbie, K.; Brett, M. J. J. Vac. Sci. Technol., A 1997, 15, 1460. Azzam, R. Appl. Phys. Lett. 1992, 61, 3118. Dick, B.; Brett, M. J.; Smy, T. J. Vac. Sci. Technol., B 2003, 21, 2569. Harris, K. D.; Westra, K. L.; Brett, M. J. Electrochem. Solid-State Lett. 2001, 4, C39. Dick, B.; Brett, M. J.; Smy, T. J.; Freeman, M. R.; Malac, M.; Egerton, R. F. J. Vac. Sci. Technol., A 2000, 18, 1838. Kennedy, S. R.; Brett, M. J. J. Vac. Sci. Technol., B 2004, 22, 1184. Kennedy, S. R.; Brett, M. J.; Toader, O.; John, S. Proc. SPIE Int. Soc. Opt. Eng. 2002, 5023, 101. Wang, J.; Huang, H.; Kesapragada, S. V.; Gall, D. Nano Lett. 2005, 5, 2505. Suzuki, M.; Taga, Y. J. Appl. Phys. 1992, 71, 2848. Robbie, K.; Brett, M. J.; Lakhatia, A. J. Vac. Sci. Technol., A 1995, 13, 2991. Hodgkinson, I. J.; Wu, Q. H. Opt. Eng. 1998, 37, 2630. Hrudey, P. C. P.; Van Popta, A. C.; Sit, J. C.; Brett, M. J. Proc. SPIE Int. Soc. Opt. Eng. 2005, 5931, 593113. Lisfi, A.; Lodder, J. C. Phys. ReV. B 2001, 63, 174441. Dick, B.; Brett, M. J.; Smy, T. J.; Freeman, M. R.; Malac, M.; Egerton, R. F. J. Vac. Sci. Technol., A 2000, 18, 1838. Singh, J. P.; Tang, F.; Karabacak, T.; Lu, T.-M.; Wang, G.-C. J. Vac. Sci. Technol., B 2004, 22, 1048. Singh, J. P.; Liu, D.-L.; Ye, D.-X.; Picu, R. C.; Lu, T.-M.; Wang, G.-C. Appl. Phys. Lett. 2004, 84, 3657. Zhang, G.; Zhao, Y. P. J. Appl. Phys. 2004, 95, 267. Seto, M. W.; Robbie, K.; Vick, D.; Brett, M. J.; Kuhn, L. J. Vac. Sci. Technol., B 1999, 17, 2172. Liu, D.-L.; Ye, D.-X.; Khan, F.; Tang, F.; Kim, B.-K.; Picu, R. C.; Wang, G.-C.; Lu, T.-M. J. Nanosci. Nanotechnol. 2003, 3, 492. Kesapragada, S. V.; Gall, D. Thin Solid Films 2006, 494, 234. Karabacak, T.; Singh, J. P.; Zhao, Y.-P.; Wang, G.-C.; Lu, T.-M. Phys. ReV. B 2003, 68, 125408. Dick, B.; Brett, M. J.; Smy, T. J. Vac. Sci. Technol., B 2003, 21, 23. Messier, R.; Venugopal, V. C.; Sunal, P. D. J. Vac. Sci. Technol., A 2000, 18, 1538. Dirks, A. G.; Leamy, H. J. Thin Solid Films 1977, 47, 219. Meakin, P.; Krug, J. Phys. ReV. A 1992, 46, 3390. Leamy, H. J.; Dirks, A. G. J. Appl. Phys. 1978, 49, 3430. Nieuwenhuizen, J. M.; Haanstra, H. B. Phillips Tech. ReV. 1966, 27, 87. Tait, R. N.; Smy, T.; Brett, M. J. Thin Solid Films 1993, 226, 196. Lichter, S.; Chen, J. Phys. ReV. Lett. 1986, 56, 1396. Hodgkinson, I.; Wu, Q. H.; McPhun, A. J. Vac. Sci. Technol., B 1998, 16, 2811. Paritosh; Srolovitz, D. J. J. Appl. Phys. 2002, 91, 1963. Fine; Greiner; Ellis J. Met. 1951, 191, 56.

NL060122A

857