Wafer-Scale Nanopatterning and Translationinto High-Performance PiezoelectricNanowiresThanh D. Nguyen,†,§ John M. Nagarah,‡,§ Yi Qi,† Stephen S. Nonnenmann,†Anatoli V. Morozov,† Simonne Li,† Craig B. Arnold,† and Michael C. McAlpine*,†
†Department of Mechanical and Aerospace Engineering, Princeton University, Princeton, New Jersey 08544,United States, and ‡Broad Fellows Program, Division of Biology, California Institute of Technology, Pasadena,California 91125, United States
ABSTRACT The development of a facile method for fabricating one-dimensional, precisely positioned nanostructures over large areasoffers exciting opportunities in fundamental research and innovative applications. Large-scale nanofabrication methods have beenrestricted in accessibility due to their complexity and cost. Likewise, bottom-up synthesis of nanowires has been limited in methodsto assemble these structures at precisely defined locations. Nanomaterials such as PbZrxTi1-xO3 (PZT) nanowires (NWs)swhich maybe useful for nonvolatile memory storage (FeRAM), nanoactuation, and nanoscale power generationsare difficult to synthesize withoutsuffering from polycrystallinity or poor stoichiometric control. Here, we report a novel fabrication method which requires only low-resolution photolithography and electrochemical etching to generate ultrasmooth NWs over wafer scales. These nanostructures aresubsequently used as patterning templates to generate PZT nanowires with the highest reported piezoelectric performance (deff ∼145 pm/V). The combined large-scale nanopatterning with hierarchical assembly of functional nanomaterials could yield breakthroughsin areas ranging from nanodevice arrays to nanodevice powering.
KEYWORDS Nanomanufacturing, energy conversion, piezoelectric nanowires, laser annealing
Nanowires (NWs) have emerged as powerful buildingblocks for functional applications such as nanosen-sors,1 nanogenerators,2 and nanoelectronics.3 Key
to achieving high device performance is the generation ofNWs whose characteristicssincluding geometry, composi-tion, and defect densitysare well controlled at the pointof their growth or fabrication. Further, many applicationsrequire nanowires which are hierarchically patterned overlarge areas and at precisely defined locations. Thus, togenerate NW arrays for practical applications, a fabricationmethod which is high throughput, not exotic, applicable toa broad range of functional materials, and capable ofgenerating highly ordered NWs on a large-scale area isrequisite. For example, bottom-up synthesis of NWs byvapor-liquid-solid (VLS)4 or electrochemical5 methods hassuccessfully produced NWs with broad compositions6 whichcan be assembled via flow alignment,7 Langmuir-Blodgettassembly,8 dry transfer,9 or dielectrophoresis10 over mac-roscopic scales.
Recently, top-down methods, including nanoimprint li-thography (NIL),11,12 superlattice nanowire pattern transfer(SNAP),1,13 and lithographically patterned nanowire elec-trodeposition (LPNE)14 have regained attractiveness in their
ability to pattern NWs that can be deterministically posi-tioned. For example, LPNE has shown the preparation ofNWs with diameters down to sub-50 nm for materials thatcan be electrodeposited (such as noble metals and leadtelluride),14,15 while SNAP is applicable to single-crystallinematerials by etching metal templates into host substrates.16
Here, we report a new top-down methodsinspired by LPNEand SNAP and free of complex processingsfor generatingfunctional NW arrays with accurate placements over waferscales. Our technique can generate NWs of complex com-positions and functionalities. Indeed, we demonstrate itspower by realizing large area fabrication of high-perfor-mance piezoelectric NWs, which may have applications innanodevice power generation.
Our approach, which we term photolithography andetching for nanoscale lithography (PENCiL), produces ametal nanomask template over wafer scales. Figure 1aoutlines the PENCiL process. First, a 100 nm Ni thin filmis deposited onto any substrate. Next, arrays of 1-2 µmwide resist lines are patterned using standard photoli-thography. Third, and key to PENCiL, the Ni layer iselectrochemically etched in concentrated phosphoric acidby “undercutting” the resist windows in order to definethe NW structures. Variation in the applied voltage, pho-toresist patterns/heights, and etching times can be utilizedto directly control the shapes and sizes of various NWfeatures.17 Most significantly, Figure 1b shows that NWspatterned using the PENCiL technique can reproducibly
* Corresponding author: telephone number, (609) 542-0275; fax number, (609)258-1918; e-mail, [email protected].§ These authors contributed equally to this work.Received for review: 07/26/2010Published on Web: 10/12/2010
achieve sub-40 nm diameters, as well as curved and ring-like-shaped NWs (see Supporting Information). BecausePENCiL relies only on photolithography and etching, large-area scalability of this process was achieved by patterninga 3 in. wafer with Ni NW arrays, as shown by thediffraction pattern in Figure 1c. Indeed, ordered, uniformNW arrays are realized on a large scale (Figure 1d) and
monolithically patterned to larger contact pads to formnanodevices from the two-step PENCiL process (Figure1d).
This facile, large-area fabrication of NWs is enabled bythe etching conditions: polished Ni sidewalls and a controlledetch rate are ensured by the concentrated phosphoric acid,such that the rate-limiting step becomes diffusion of waterto the metal surface to dissolve reaction products. Raisedfeatures on the metal surface are thus etched faster becausethey are exposed to a higher local water concentration.18,19
This electrochemical etch process results in nonoxidized Niwires that retain their conductivity (see Supporting Informa-tion). Despite success in producing polished Ni sidewalls, theprocess is susceptible to surface and line edge roughness dueto the contact lithography step. Some critical nanodeviceapplications require ultrasmooth line roughness (<10 nm).For example, a decrease in piezoelectric electromechanicalconversion efficiency due to incomplete charge compensa-tion has been reported due to defects in piezoelectric nano-structures.20
Self-perfection by liquefaction (SPEL) is a laser annealingtechnique that has been previously shown to correct defectsin patterned metal features.21 In the liquid phase, thenanostructure flows under the influence of surface tensionand interaction with the substrate, effectively smoothing outthe structure’s rough edges. Here, the SPEL technique wasused to smoothen nanomask features defined by PENCiL.Figure 2a shows the process: a single ArF excimer laser pulse(λ ) 194 nm) with a duration of 20 ns was employed tolocally melt and resolidify the masking nanowires. To avoidsubstrate heating, a transparent 200 nm SiO2 layer (bandgap ∼9 eV) was deposited by electron beam evaporationbetween the PZT substrate and Ni film. A calibration processwas performed before the laser exposure. By changing thefluence of the laser light gradually, an optimal value of 105mJ/cm2 (NW width ∼85 nm) was identified for smoothingthe metal NWs defined by PENCiL.
Panels b and c of Figure 2 show scanning electronmicroscopy (SEM) images of NWs patterned by PENCiLbefore and after SPEL, respectively. These images weresubsequently used to quantitatively obtain a line-edge rough-ness profile of the NWs before (Figure 2d) and after (Figure2e) the SPEL treatment. A custom-built Matlab (The Math-Works, Inc., Natick, MA) program was used to quantitativelyanalyze the roughness of NWs based on root-mean-squaredeviations of NW line edge profiles. We found that thePENCiL process alone produces NWs that are quite smooth,with a line-edge roughness of 15.4 nm. Significantly, thisroughness can be decreased to 4.2 nm after the SPELtreatment, approaching the 3 nm smoothness limit reportedby the International Technology Roadmap for Semiconduc-tors.22 The pulse must be extremely rapid for the smoothen-ing process, and when combined with the laser spot size of2 × 3 mm2, the SPEL process can ultimately be applied overa 3 in. wafer in a time scale of only minutes.
FIGURE 1. Wafer scale nanomask patterned by PENCiL. (a)Schematic showing nanopatterning by PENCiL: step 1, a Ni thinfilm is deposited on a substrate and photolithography is carriedout on top; step 2, the Ni film is electrochemically etched to yieldNi NWs undercut from the resist pattern; step 3, the resist isremoved, revealing a Ni nanomask. (b) Scanning electron micro-graph (SEM) image of Ni NW arrays with diameters of 35 nm. Theinset shows a single NW at a higher magnification. Scale bar is200 nm. (c) Nanowires patterned via PENCiL across an entirewafer form a grating which diffracts natural light. (d) SEM of NiNW arrays (vertical) patterned over a large scale. The horizontallines are monolithically patterned contact pads.
Because the PENCiL process relies only on Ni metaldeposition, photolithographic patterning, and electrochemi-cal etching, it can be performed on a wide variety ofsubstrates, including materials with inherent functionalities.One particularly interesting class is piezoelectric crystals,which represent promising materials for electromechanicalenergy conversion technologies.2,23 Electromechanicallyderived energy harvesting via low-dimensional piezoelectricnanostructures has garnered substantial research attention,due to the synthesis of various one-dimensional piezoelectricnanostructures including ZnO,24 BaTiO3,25 GaN,26 CdS,27
and PVDF NWs,28 and the possibility of harvesting energyfrom the environment for self-powered nanosystemsand biomedical devices.2,23 Lead zirconate titanate(PbZrxTi1-xO3, PZT) is a piezoelectric material used for high-performance applications due to its exceptionally largepiezoelectric constant (∼250 pm/V).29 However, this poten-tial is greatly restricted by the lack of an efficient methodfor its synthesis at the nanoscale. This is a result of thecomplex stoichiometric composition of PZT, as most tech-niques (e.g., VLS) for producing high-quality NWs cannot beeasily applied due to issues such as phase separation andthe lack of suitable catalysts.30 Synthesis of PZT NWs to datehas thus been limited to sol-gel templates and hydrother-
mal methods,31-34 resulting in polycrystalline wires and/orwires with limited stoichiometric control.
By contrast, the controlled synthesis of stoichiometric,epitaxial thin films of PZT with precise atomic compositionalcontrol has been well studied.35,36 Accordingly, we per-formed the PENCiL process on an underlying layer ofcrystalline PZT. A PZT thin film (300-500 nm thickness, 52/48 Zr/Ti) was first deposited epitaxially on Pt/MgO or on aSi substrate via rf-sputtering with a 52/48 PZT target toensure a composition close to the morphotropic phaseboundary (MPB).37 The film was postannealed at 700 °C for15 min to induce a phase transition to a perovskite struc-ture.23,38 After performing PENCiL, a subsequent dry reac-tive ion etching (RIE) step was performed with a Micro RIE(150 mTorr SF6, 150 W), selectively removing the exposedPZT film and resulting in PZT NW arrays under the PENCiLNW mask (Figure 3a). Finally, the Ni nanomask was re-moved using 63% aqueous nitric acid at 60 °C for 10 minto complete the process and reveal the defined PZT NWs.
Figure 3b shows that even over large areas, the PENCiLnanomask can be directly translated into highly ordered PZTNW arrays (diameter ∼75 nm). Energy dispersive spectros-copy (EDS) was locally performed at multiple points alongthe axial direction of individual PZT NWs (Figure 3c, also seeSupporting Information). Importantly, the EDS spectrumshows that the stoichiometry closely matches that of a bulkstandard PZT sample and confirms that the PENCiL transla-tion process does not leave residual Ni from the nanomask.
Next, functional properties of the defined PZT NWs wereinvestigated using piezo-force microscopy (PFM).39 A 80 nmPt film was deposited prior to PZT film deposition for use asthe bottom conducting electrode in the PFM measurement.An AFM (Asylum Research, Santa Barbara, CA) equippedwith a doped diamond conducting tip (Olympus AC240TMTi/Pt coating, k ) 2.1 N/m, f ) 77 kHz) was operated incontact mode. A lock-in amplifier (Stanford InstrumentsSR830) was used, and an applied ac signal frequency of 283kHz was chosen to avoid unnecessary topographic crosstalknear the cantilever resonance. An intermediate force (2000nN) was applied to ensure that the tip deflection waselectromechanically response dominated.39
Figure 4a shows representative local height (left), piezo-electric amplitude (middle), and phase (right) scans of thePZT NW array collected via the conductive cantilever tip(note that the PZT NWs look wider than they are due to thesize effect of the PFM tip in contact mode). The observedcontrast in the amplitude and phase traces of Figure 4aindicates that the PZT NW array is of uniform response(standard deviation ) 7.8%) and shows distinct differencesfrom the collected character of the substrate base layer. Toillustrate this more clearly, Figure 4b shows a three-dimensional overlay of the induced piezoelectric amplitudeimposed on the topographic features of the PZT NW array
FIGURE 2. Correction of defects via self-perfection by liquefaction(SPEL). (a) Schematic of the SPEL process used to smoothen Ni NWs.(b, c) SEM images show 85 nm Ni NWs before (b) and after (c) laserexposure in SPEL. (d, e) Statistical analysis performed on the SEMimages allowed for quantitative calculation of line edge profiles inNi NWs before (d) and after (e) SPEL.
over a 20 × 20 µm2 area. Again, the induced piezoelectricresponse is found to be quite uniform along the length ofeach NW.
An effective piezoelectric coefficient, deff, was derived bycalculating the average deformation (∼300 pm withoutamplification) per applied ac voltage amplitude (VAC ) 3 V),yielding a value of ca. 100 pm/V for the unpoled PZT NWs.Finally, ramping mode PFM, in which the cantilever tip isheld in direct contact with the PZT NW surface while theapplied ac bias to the tip is continuously increased (VAC )0-2.5 V), was performed to quantitatively determine theeffective piezoelectric coefficient of individual PZT NWs asdefined by PENCiL. Figure 4c shows the collected linearpiezoelectric response of an individual, unpoled PZT NWusing ramping mode PFM, with the resulting slope indicatingan effective piezoelectric coefficient of 114 pm/V. Theobserved value increased to deff ) 145 pm/V following 30min of poling under an applied dc voltage of +10 V, scannedover a selected area of the PZT NW.38 To our knowledge,this value represents the highest piezoelectric coefficientreported to date from any piezoelectric nanostructure.34,40
We attribute this result to the ability of the PENCiL techniqueto define smooth nanowire structures from high-quality thinfilms.
In summary, a combined photolithographic and electro-chemical etching methodswhich we term PENCiLswas ableto pattern highly ordered PZT NW arrays over large areas,
with the highest reported piezoelectric coefficient to date. Itshould be emphasized that this process is universally ap-plicable to a host of metallic nanomask systems and func-tional substrates, including semiconducting, magnetic, andoptoelectronic materials, so long as selective etch reagentsare available. We anticipate that the results demonstratedhere will open exciting avenues in fundamental research onpiezoelectric size scaling in nanomaterials, as well as ap-plications requiring batch fabrication such as memory cells,nanogenerators, and nanoactuators. Yet, despite these ad-vantages, it should be noted the spacing between thenanowires is still limited by photolithographic resolution.Ongoing work will address this by, for example, employingmultiple PENCiL steps to achieve higher densities, as wellas generating addressable crossbar structures.
Acknowledgment. We acknowledge the use of the PRISMImaging and Analysis Center, which is supported by the NSFMRSEC Program via the Princeton Center for ComplexMaterials (No. DMR-0819860). J.M.N. acknowledges supportfrom The Broad Foundations. M.C.M. acknowledges supportof this work by the Young Investigator Award from theIntelligence Community (No. 2008-1218103-000), the De-fense Advanced Research Projects Agency (No. N66001-10-1-2012), and the National Science Foundation (No. NSFCMMI-1036055).
FIGURE 3. Large scale translation into PZT NW arrays. (a) Schematic showing the process to obtain PZT NWs from the Ni NW mask: step 1,PENCiL was performed on top of a PZT thin film, and RIE was used to selectively etch the exposed PZT; step 2, the Ni NW mask is completelyremoved to reveal PZT NWs. (b) SEM image showing a highly ordered PZT NW array translated over a large area from the pattern in Figure 1d.(c) EDS was performed locally at points along a PZT NW (red crosses in inset) to determine the composition of PZT NWs after PENCiL. Si, Pb,Zr, Ti, and Ni peaks are labeled.
Supporting Information Available. Additional figures ofnanostructures with various sizes and shapes fabricated byPENCiL, an etching calibration plot, NW width statistics, I-Vmeasurements, and large-scale PZT NW arrays fabricatedon Pt/MgO substrates, along with EDS characterization.Thismaterial is available free of charge via the Internet at http://pubs.acs.org.
REFERENCES AND NOTES(1) McAlpine, M. C.; Ahmad, H.; Wang, D. W.; Heath, J. R. Nat. Mater.
2007, 6, 379–384.(2) Xu, S.; Qin, Y.; Xu, C.; Wei, Y.; Yang, R.; Wang, Z. L. Nat.
Nanotechnol. 2010, 5, 366–373.(3) Nam, S.; Jiang, X. C.; Xiong, Q. H.; Ham, D.; Lieber, C. M. Proc.
Natl. Acad. Sci. U.S.A. 2009, 106, 21035–21038.(4) Morales, A. M.; Lieber, C. M. Science 1998, 279, 208–211.(5) Singh, M.; Wang, J.; Tian, M. L.; Zhang, Q.; Pereira, A.; Kumar,
N.; Mallouk, T. E.; Chan, M. H. W. Chem. Mater. 2009, 21, 5557–5559.
(6) Tian, B.; Kempa, T. J.; Lieber, C. M. Chem. Soc. Rev. 2009, 38, 16–24.(7) Huang, Y.; Duan, X.; Wei, Q.; Lieber, C. M. Science 2001, 291,
630–633.(8) Whang, D.; Jin, S.; Lieber, C. M. Nano Lett. 2003, 3, 951–954.(9) Fan, Z. Y.; Ho, J. C.; Jacobson, Z. A.; Razavi, H.; Javey, A. Proc.
Natl. Acad. Sci. U.S.A. 2008, 105, 11066–11070.(10) Freer, E. M.; Grachev, O.; Duan, X.; Martin, S.; Stumbo, D. P. Nat.
Nanotechnol. 2010, 5, 525–530.(11) Ahn, S. H.; Guo, L. J. ACS Nano 2009, 3, 2304–2310.(12) Chou, S. Y.; Krauss, P. R.; Renstrom, P. J. Science 1996, 272, 85–87.(13) Melosh, N. A.; Boukai, A.; Diana, F.; Gerardot, B.; Badolato, A.;
Petroff, P. M.; Heath, J. R. Science 2003, 300, 112–115.(14) Menke, E. J.; Thompson, M. A.; Xiang, C.; Yang, L. C.; Penner,
R. M. Nat. Mater. 2006, 5, 914–919.(15) Yang, Y.; Taggart, D. K.; Brown, M. A.; Xiang, C.; Kung, S.-C.;
Yang, F.; Hemminger, J. C.; Penner, R. M. ACS Nano 2009, 3,4144–4154.
(16) Boukai, A. I.; Bunimovich, Y.; Tahir-Kheli, J.; Yu, J.-K.; Goddard,W. A.; Heath, J. R. Nature 2008, 451, 168–171.
(17) Shenoy, R. V.; Datta, M.; Romankiw, L. T. J. Electrochem. Soc.1996, 143, 2305–2309.
(18) Cordeiro, G.; Mattos, O. R.; Barcia, O. E.; Beaunier, L.; Deslouis,C.; Tribollet, B. J. Appl. Electrochem. 1996, 26, 1083–1092.
(19) Du, B.; Suni, I. I. J. Electrochem. Soc. 2004, 151, C375–C378.(20) Dawber, M.; Chandra, P.; Littlewood, P. B.; Scott, J. F. J. Phys.:
Condens. Matter 2003, 15, L393–L398.(21) Chou, S. Y.; Xia, Q. F. Nat. Nanotechnol. 2008, 3, 295–300.(22) InternationalTechnologyRoadmapforSemiconductors;2006;Table77b.(23) Qi, Y.; Jafferis, N. T.; Lyons, K.; Lee, C. M.; Ahmad, H.; McAlpine,
M. C. Nano Lett. 2010, 10, 524–528.(24) Wang, Z. L.; Song, J. H. Science 2006, 312, 242–246.(25) Wang, Z. Y.; Hu, J.; Yu, M. F. Nanotechnology 2007, 18, 235203.(26) Duan, X. F.; Lieber, C. M. J. Am. Chem. Soc. 2000, 122, 188–189.(27) Routkevitch, D.; Bigioni, T.; Moskovits, M.; Xu, J. M. J. Phys. Chem.
1996, 100, 14037–14047.(28) Chang, C. E.; Tran, V. H.; Wang, J. B.; Fuh, Y. K.; Lin, L. W. Nano
Lett. 2010, 10, 726–731.(29) Li, J. F.; Dai, X. H.; Chow, A.; Viehland, D. J. Mater. Res. 1995,
10, 926–938.(30) Han, S.; Li, C.; Liu, Z. Q.; Lei, B.; Zhang, D. H.; Jin, W.; Liu, X. L.;
Tang, T.; Zhou, C. W. Nano Lett. 2004, 4, 1241–1246.(31) Lin, Y. R.; Liu, Y. T.; Sodano, H. A. Appl. Phys. Lett. 2009, 95,
122901.(32) Zhang, X. Y.; Zhao, X.; Lai, C. W.; Wang, J.; Tang, X. G.; Dai, J. Y.
Appl. Phys. Lett. 2004, 85, 4190–4192.(33) Limmer, S. J.; Seraji, S.; Forbess, M. J.; Wu, Y.; Chou, T. P.;
Nguyen, C.; Cao, G. Z. Adv. Mater. 2001, 13, 1269–1272.(34) Luo, Y.; Szafraniak, I.; Zakharov, N. D.; Nagarajan, V.; Steinhart,
M.; Wehrspohn, R. B.; Wendorff, J. H.; Ramesh, R.; Alexe, M.Appl. Phys. Lett. 2003, 83, 440–442.
(35) Blanco, O.; Heiras, J.; Siqueiros, J. M.; Martinez, E.; Castellanos-Guzman, A. G. J. Mater. Sci. Lett. 2003, 22, 449–453.
(36) Takayama, R.; Tomita, Y. J. Appl. Phys. 1989, 65, 1666–1670.(37) Kim, Y. K.; Kim, S. S.; Lee, B.; Shin, H.; Baik, S. J. Mater. Res. 2005,
20, 787–790.(38) Granzow, T.; Kounga, A. B.; Aulbach, E.; Rodel, J. Appl. Phys. Lett.
2006, 88, 252907.(39) Kalinin, S. V.; Bonnell, D. A. Phys. Rev. B 2002, 65, 125408–
125418.(40) Wang, J.; Sandu, C. S.; Setter, N. IEEE Trans. Ultrason. Ferr. 2009,
56, 1813–1819.
FIGURE 4. Characterization of PZT NWs by piezoresponse forcemicroscopy (PFM). (a) (left image) 2D topography of PZT NWs; thewhite lines represent PZT NWs. (middle image) 2D piezoresponse(PR) amplitude. (right image) 2D piezoresponse in phase mode,which reflects the direction of polarization in PZT NW domains.Dimensions of all figures are (20 µm)2. (b) 3D image of the PZT NWpiezoresponse (colored) overlaid on their topographical features. Thevertical scale bar on the right indicates the level of piezoresponseunder 3 V applied ac voltage. (c) Piezoresponse in ramping mode,with an applied ac voltage ramped from 0 to 2.5 V, and theamplitude recorded before and after poling.
