2006 AIChE Annual Meeting
(682a) Aluminum Nanowire Polarizing Grids Via Block Copolymer Lithography: Fabrication and Analysis
Authors
EXPERIMENTAL: To fabricate the nanowire grid, we employed as a template a thin film of a polystyrene-b-poly(n-hexyl methacrylate) diblock copolymer, PS-PHMA, synthesized by living anionic polymerization. This diblock has block molecular weights of 21 and 64 kg/mol, and self-assembles into cylindrical domains of PS in a matrix of PHMA, with a cylinder diameter of 17 nm and a periodicity of 33 nm. PS-PHMA was spin-coated to form a 30 nm thick film, which contained only a single layer of cylindrical domains. The substrate onto which this PS-PHMA film was spun consisted of a fused quartz wafer onto which were previously deposited a 45 nm thick layer of polyimide, followed by 22 nm of silicon nitride. The supported PS-PHMA film was then shear-aligned [6] at 150C. A CHF3/O2 plasma was then used to selectively etch through the PHMA matrix. PS etches 2.5 times slower under these conditions, which effectively transfers the pattern into the SiNx, yielding SiNx stripes on the polyimide layer. Then an O2 plasma increased the aspect ratio of this pattern by etching the polyimide between the stripes down to the quartz substrate, while leaving the polyimide underneath the SiNx stripes untouched. Aluminum (20-25 nm) was then deposited onto the specimen by electron-beam evaporation, creating Al wires on both the quartz substrate and on top of the SiNx stripes. The polyimide layer was then lifted off in an NMP bath, taking the SiNx and overlying Al stripes with it, and leaving an Al nanowire grid supported on the quartz substrate.
RESULTS AND DISCUSSION: Scanning electron micrographs of the aluminum nanowire grid reveal a highly ordered and oriented structure, over the entire 2 cm2 area of the specimen. The wires have a very regular 33 nm periodicity and long-range alignment, with the wires pointing along the original shear direction. Some meandering of the wires and an occasional "dislocation" (a wire which ends) can be observed, but these defects do not have a substantial impact on the polarization performance of the grid, since they do not fundamentally alter the strong anisotropy of the grid necessary for it to function as a polarizer. The spectral performance of the nanowire grid was measured using a UV-visible spectrometer, with a calcite prism as analyzer. The transmission was measured from 220 to 800 nm, with the specimen oriented with its visible-range polarization either parallel or perpendicular to the analyzer's. From the parallel and perpendicular transmission values we obtain the polarization efficiency vs. wavelength curve. The polarization efficiency of the grid reaches 50% in the visible, and significant polarization is observed down to the shortest wavelengths measured, 220 nm. Interestingly, the polarization direction of the transmitted radiation shows an unexpected reversal near 300 nm. To our knowledge, this is the first time that such a polarization reversal phenomenon has been reported. To understand this phenomenon, we begin by considering wavelengths much longer that the grid periodicity d, wherein a highly conducting parallel wire array can be viewed in the DC limit. A DC electric field parallel to the wires, and incident from above upon the grid plane at z = 0, is "shorted out" and decays exponentially as exp(-2àz/d) below the grid plane. For z >> d, the effective transmission is zero and the reflection coefficient is unity. By contrast, a perpendicular electric field produces opposite charges across the wire width, creating a charged capacitor array with macroscopic voltage/length unchanged from that in free space. Thus, the perpendicular field goes straight through the grid, making it entirely transparent. The ratio of parallel to perpendicular transmission is zero, so the polarization efficiency is 100%. More involved calculations [8] consider how much of the radiation is transmitted and how much is reflected when the frequency, metal conductivity, and grid thickness are finite; in addition, to account for the fact that the wavelength of light is only one order of magnitude greater than the grid pitch, a full numerical solution can be calculated via Sheng's exact matrix transfer formalism [9]. For an Al grid of 33 nm period, this exact numerical calculation predicts a crossover near 170 nm, rather than the 300 nm observed. This difference is a consequence of oxide in the evaporated Al, which forms by reaction with trace oxygen in the evaporator. A 6:1 by volume ratio of Al to Al2O3 was estimated by fitting transmission vs. wavelength curves of uniform aluminum films deposited under the same conditions; when the average dielectric function [10] of this 6:1 mixture is employed, the exact numerical solution is in excellent agreement with the measured spectral polarization curves. Thus, we have a solid quantitative understanding of, and model for, the factors which influence the performance of such grids. The best polarization efficiency is obtained with a clean metal having a high plasma frequency and a grid pitch an order of magnitude (or more) below the wavelengths of interest. The thickness of the grid need only be on the order of the grid pitch to approach the limiting polarization performance, which is a significant asset in fabrication, since nanowires of circular or roughly square cross-section are perfectly satisfactory.
ACKNOWLEDGEMENTS: This work was supported by the National Science Foundation (MRSEC Program) through the Princeton Center for Complex Materials (DMR-0213706), and by Toshiba.
REFERENCES: [1] Park, M.; Harrison, C.; Chaikin, P.M.; Register, R.A.; Adamson, D.H. Science 1997, 276, 1401. [2] Park, M.; Chaikin, P.M.; Register, R.A.; Adamson, D.H. Appl. Phys. Lett. 2001, 79, 257. [3] Li, R.R.; Dapkus, P.D.; Thompson, M.E.; Jeong, W.G.; Harrison, C.; Chaikin, P.M.; Register, R.A.; Adamson, D.H. Appl. Phys. Lett. 2000, 76, 1689. [4] Wang, R.; Grobman, W.; Reich, A.; Thompson, M. in 21st Annual BACUS Symposium on Photomask Technology; Dao, G.T. and Grenon, J. eds. (SPIE Proceedings 4562, 2002), pp. 406-417. [5] Harrison, C.; Adamson, D.H.; Cheng, Z.; Sebastian, J.M.; Sethuraman, S.; Huse, D.A.; Register, R.A.; Chaikin, P.M. Science 2000, 290, 1558. [6] Angelescu, D.E.; Waller, J.H.; Adamson, D.H.; Deshpande, P.; Chou, S.Y.; Register, R.A.; Chaikin, P.M. Adv. Mater. 2004, 16, 1736. [7] Angelescu, D.E.; Waller, J.H.; Register, R.A.; Chaikin, P.M. Adv. Mater. 2005, 17, 1878. [8] Pelletier, V.; Asakawa, K.; Wu, M.; Adamson, D.H.; Chaikin, P.M. Appl. Phys. Lett. 2006, accepted. [9] Sheng, P.; Stepleman, R.S.; Sanda, P.N. Phys. Rev. B 1982, 26, 2907. [10] Hashin, Z.; Shtrikman, J. J. Appl. Phys. 1962, 33, 3125.