Langmuir 2009, 25, 55-58
55
Electrochemical Deposition of Silver Nanoparticle Arrays with Tunable Density Bingjie Yang,† Nan Lu,*,† Chunyu Huang,† Dianpeng Qi,† Gang Shi,† Hongbo Xu,† Xiaodong Chen,‡ Bin Dong,† Wei Song,† Bing Zhao,† and Lifeng Chi*,†,‡ State Key Laboratory of Supramolecular Structure and Materials, Jilin UniVersity, Changchun, 130012, P. R. China, and Physikalisches Institut and Center for Nanotechnology (CeNTech), Westfa¨lische Wilhelms-UniVersita¨t Mu¨nster, D-48149 Mu¨nster, Germany ReceiVed October 27, 2008. ReVised Manuscript ReceiVed NoVember 20, 2008 We present a new approach to fabricate arrays of silver nanoparticles (SNPs) on predefined positions by means of electrochemical deposition (ECD) combined with nanoimprint lithography (NIL). The SNPs arrangement, especially the density can be tuned by varying the pattern design and the preparation conditions. By this approach, various arrays of SNPs with feature size down to submicrometer can be fabricated over several square centimeters on one substrate. The SNPs arrays can be readily extended to other conductive substrates, which have potential applications in detecting and sensing fields, such as surface-enhanced raman scatter.
Introduction Metal nanopatrticles and their arrays have attracted increasing scientific and technological interests due to their unique properties, such as strong electromagnetic enhancement, near-field behavior, and surface plasmon resonances.1 Based on these properties, their applications have been extended to the surface plasmon waveguides,2,3 surface plasmon nanosensors,4 surface-enhanced raman scattering (SERS),5,6 metal-enhanced fluorescence,7-10 and other devices. To fulfill the expected applications, various approaches have been proposed for the fabrication of metal particle arrays, including electron beam lithography (EBL),11,12 microcontact printing (µCP),13,14 photolithography,15,16 pulsed laser lithography,17,18 scanning probe lithography (SPL),19,20 and self* To whom correspondence should be addressed. E-mail: luenan@ jlu.edu.cn;
[email protected]. † Jilin University. ‡ Westfa¨lische Wilhelms-Universita¨t Mu¨nster.
(1) Willets, K. A.; Van Duyne, R. P. Annu. ReV. Phy. Chem. 2007, 58, 267. (2) Maier, S. A.; Kik, P. G.; Atwater, H. A.; Meltzer, S.; Harel, E.; Koel, B. E.; Requicha, A. A. G. Nat. Mater. 2003, 2, 229. (3) Zou, S. L.; Schatz, G. C. Phys. ReV. B 2006, 74, 125111. (4) Haes, A. J.; Van Duyne, R. P. J. Am. Chem. Soc. 2002, 124, 10596. (5) Kaminska, A.; Inya-Agha, O.; Forster, R. J.; Keyes, T. E. Phys. Chem. Chem. Phys. 2008, 10, 4172. (6) Broglin, B. L.; Andreu, A.; Dhussa, N.; Heath, J. A.; Gerst, J.; Dudley, B.; Holland, D.; El-Kouedi, M. Langmuir 2007, 23, 4563. (7) Malicka, J.; Gryczynski, I.; Lakowicz, J. R. Anal. Chem. 2003, 75, 4408. (8) Lakowicz, J. R.; Malicka, J.; Gryczynski, I.; Gryczynski, Z.; Geddes, C. D. J. Phys. D 2003, 36, R240. (9) Miao, X. Y.; Wilson, B. K.; Lin, L. Y. Appl. Phys. Lett. 2008, 92, 124108. (10) Sweatlock, L. A.; Maier, S. A.; Atwater, H. A.; Penninkhof, J. J.; Polman, A. Phys. ReV. B 2005, 71, 235408. (11) Corbierre, M. K.; Beerens, J.; Lennox, R. B. Chem. Mater. 2005, 17, 5774. (12) Werts, M. H. V.; Lambert, M.; Bourgoin, J. P.; Brust, M. Nano Lett. 2002, 2, 43. (13) Santhanam, V.; Andres, R. P. Nano Lett. 2004, 4, 41. (14) Porter, L. A.; Choi, H. C.; Schmeltzer, J. M.; Ribbe, A. E.; Elliott, L. C. C.; Buriak, J. M. Nano Lett. 2002, 2, 1369. (15) Yin, D. h.; Horiuchi, S.; Morita, M.; Takahara, A. Langmuir 2005, 21, 9352. (16) Lee, J. Y.; Yin, D. H.; Horiuchi, S. Chem. Mater. 2005, 17, 5498. (17) Hung, W. C.; Cheng, W. H.; Tsai, M. S.; Chung, W. C.; Jiang, I. M.; Yeh, P. Appl. Phys. Lett. 2008, 93, 061109. (18) Kim, H.; Shin, H.; Ha, J.; Lee, M.; Lim, K. S. J. Appl. Phys. 2007, 102, 083505. (19) Maoz, R.; Cohen, S. R.; Sagiv, J. AdV. Mater. 1999, 11, 55. (20) Garno, J. C.; Zangmeister, C. D.; Batteas, J. D. Langmuir 2007, 23, 7874.
assembly.21,22 For these approaches, the nanoparticles need to be synthesized before the fabrication of nanoparticle arrays. As an alternative, template-assisted electrochemical deposition (ECD) has been applied for the generation of metallic nanostructures due to its facile process. On the substrates bearing template structures generated with EBL and LB techniques,23,24 metallic structures have been successfully constructed by ECD. However, the combination of high throughput, low cost, and easy control is still a key issue in creating metal nanoparticle arrays in the scale of submicro to nanometer, which is the basic of applications of the particle arrays. In the present study, we fabricate arrays of silver nanoparticles (SNPs) by means of electrochemical deposition on predefined positions, generated with nanoimprint lithography (NIL).25 The SNPs arrangement, especially the density can be tuned by varying the pattern design and the preparation conditions. By this approach, various arrays of SNPs with feature size down to submicrometer can be fabricated over several square centimeters on one substrate.
Experimental Section Materials. ITO (20 Ω/sq) was obtained from Jingbo Glass and Display Company, Shenzhen, China. Chloroform (HPLC grade), ethanol (HPLC grade), and acetone (HPLC grade) from Guangfu Fine Chemical Research Institute (Tianjin, China) were used as solvent for cleaning the substances and the lift-off process without further purification. Silver nitrate (AR grade) was from Beijing Chemical Works, China. 4-Mercaptopyridine (4-MPy) used as SERS probe molecule was purchased from Sigma-Aldrich, USA. Mr-I 7030E (Micro Resist Technology GmbH, Germany) was used as resist polymer for thermal nanoimprint lithography. The silicon stamp was fabricated by photolithography, followed by an anisotropic (21) Huang, J. X.; Kim, F.; Tao, A. R.; Connor, S.; Yang, P. D. Nat. Mater. 2005, 4, 896. (22) Shimomura, M.; Sawadaishi, T. Curr. Opin. Colloid Interface Sci. 2001, 6, 11. (23) Menke, E. J.; Thompson; Xiang, M. A. C.; Yang, L. C.; Penner, R. M. Nat. Mater. 2006, 5, 914. (24) Zhang, M. Z.; Lenhert, S.; Wang, M.; Chi, L. F.; Lu, N.; Fuchs, H.; Ming, N. B. AdV. Mater. 2004, 16, 409. (25) Chou, S. Y.; Krauss, P. R.; Renstrom, P. J. Appl. Phys. Lett. 1995, 67, 3114.
10.1021/la803559c CCC: $40.75 2009 American Chemical Society Published on Web 12/08/2008
56 Langmuir, Vol. 25, No. 1, 2009
Letters
Figure 1. Schematic illustration of the procedures for fabricating the arrays of SNPs. (a) Spin-coating resist polymer onto ITO substrates; (b) imprinting; (c) RIE etching process; (d) ECD of SNPs; (e) “lift-off” of the resist layer.
etching.26 The stamp is of 300 nm in depth, consisting of various patterns, such as dots and squares arrays with different feature size, and space of micro to submicro meters. The stamp was treated with an antiadhensive layer (tridecafluoro-1,1,2,2,-tetrahydrooctyltrichlorosilane) by the vapor phase deposition, resulting in a very low surface energy before use. This treatment can prevent the resist polymer from sticking to the stamp during the imprinting process, facilitating the separation of the stamp and the imprinted patterns.27 Substrate Patterning. ITO substrates (2 × 2 cm2 slices) were sonicated consecutively in the bath of acetone, chloroform, ethanol, and water (Millipore, resistance of 18.2 MΩ · cm) for 5 min each. These slices were used as counter electrodes for the ECD process. A 300 nm Mr-I 7030E layer was spin-coated onto the clean ITO substrate, followed by annealing at 90 °C for 5 min to remove residual solvent. The imprinting was carried out on a 2.5-in. Nanoimprinter (Obducat AB, Malmo¨, Sweden). Imprint process was performed at 130 °C, under 40 bar for 500 s. After peeling off the stamp from the substrate at 70 °C, the residual polymer was removed by Reactive Ion Etching (O2) (PVA TePla O-Plasma System 100) at 100 W, 460 mTorr for 2 min. The patterns of the stamp were transferred onto the ITO substrate. ECD of SNPs. The ECD process was performed in a self-designed three-electrode system, see Supporting Information (SI) Figure S1. The cell was filled with silver nitrate electrolyte (0.05 mol/L) without any other supporting materials, such as stabilizer or additives. The ECD process was carried out on the BAS100W Electrochemical Analyzer (Bioanalytical Systems, West Lafayette, IN) on 300 mV for 400 s after the optimization. After the ECD process, the resist polymer was removed by rinsing the substrate in acetone, followed by drying under a flow of pure nitrogen. Different arrays of SNPs were produced on the same substrate by a single ECD process. Measurement. Scanning electron microscopy (SEM) was carried out on the field emission scanning electron microscope (FE-SEM, JSM-6700F, JEOL Co., Japan). The monolayer of 4-MPy was prepared on the arrays of SNPs by immersing the substrate into the 4-MPy ethanol solution (1.0 × 10-4 mol/L) for 15 min. The superfluous solution of 4-MPy on the sample was removed by ethanol and the sample was dried under a flow of pure nitrogen. SERS spectra were collected on a Raman spectrometer (Renishaw RM1000 Confocal Micro-Raman Spectrometer, U.K.) using a 514.5 nm Ar+ laser as the excitation light source. Laser power at the sample stage is about 4.2 mW and exposure time is 10 s for the SERS spectra detection. Particle density analysis was performed by means of a self-developed program written in the PW-Wave developing environment and checked by counting the particle number of the SEM images. (26) Bogaerts, W.; Wiaux, V.; Taillaert, D.; Beckx, S.; Luyssaert, B.; Bienstman, P.; Baets, R. J. Sel. Top. Quant. Electron. 2002, 8, 928. (27) Dong, B.; Lu, N.; Zelsmann, M.; Kehagias, N.; Fuchs, H.; Sotomayor Torres, C. M.; Chi, L. F. AdV. Funct. Mater. 2006, 16, 1937.
Figure 2. FE-SEM images of the arrays of SNPs. (a) 1.3 × 1.5 µm dots with a spacing of 3.8 µm; (b) 1.3 × 1.5 µm dots with a spacing of 1.9 µm; (c) 4.8 µm squares with a spacing of 3.8 µm; (d) 4.8 µm square with a spacing of 1.9 µm. Scale bar ) 1 µm.
Results and Discussion The fabrication process is schematically shown in Figure 1. A typical fabricating course includes the following steps. The thermoplastic resist polymer (Mr-I 7030E) was spin-coated onto the substrate first, which was subsequently patterned with NIL, as revealed by the typical FE-SEM image (see SI Figure S2a). After removing the residual resist polymer with the RIE process, ECD was carried out on the polymer patterned ITO substrate to produce the SNPs. The FE-SEM image shows that the SNPs are formed and confined by the previously fabricated patterns (see SI Figure S2b). Thus the arrays of the SNPs can be obtained on the ITO substrates with the template-assisted ECD. It is known that the formation of SNPs includes two steps: nucleation and growth. The potential and duration for ECD are two crucial parameters.33 Besides, the pH value of the electrolyte of ECD could affect the nucleation process. For obtaining the perfect SNPs arrays, the potential, duration and the pH value are optimized. The SNPs’ density of the array changed slightly constant when the potential ranging from 200 mV to 400 mV, which decreases when the potential is out of the range, such as below 200 mV or above 400 mV. With a certain potential, prolonging the duration can lead to a higher density when it is shorter than 500 s; extending the duration to longer than 500 s, the density dramatically decreases due to the formation of particle aggregates. If pH < 4, the density of the SNPs decreases with increasing the pH value. If the pH is higher than 4.0, the SNPs will extend onto the resist polymer, and the sedimentation will happen in the silver nitrate electrolyte when pH is higher than 4.2. The SEM images and the statistic data on densities of the SNPs with the variation of potential, duration and pH value are shown in SI Figure S3. As shown in Figure 2, with the optimized potential (300 mV), deposition duration (400 s) and pH (3.5), various SNPs arrays were created on one ITO substrate via a single ECD process. The densities of SNPs in each image, as we count, are 125.8, 105.7, 32.0, and 23.4 no/µm2, respectively (no: number of SNPs). The SEM images shown here are from one piece of substrate bearing different patterns, including different feature sizes and different spaces, which reveal that the density of the SNPs is dependent on the feature size and the space. With the same space, the SNPs’ density is higher when the feature size of the patterns is smaller. Figure 2 depicts the correlation of SNPs’ density and the pattern design. The space of the SNPs pattern shown in panels a and c
Letters
Langmuir, Vol. 25, No. 1, 2009 57
Figure 3. Correlation of the density of the confined SNPs and the feature size of the pattern. Curve a: with 1.9 µm space, curve b: with 3.8 µm space.
of Figure 2 (panels b and d of Figure 2) are same. The density of the SNPs in Figure 2a is higher than that in Figure 2c. Similarly, the density of SNPs in Figure 2b is higher than that in Figure 2d. With the same feature size, the density of SNPs is higher on the pattern with larger space than the one with smaller space. For example, the space in Figure 2a is larger than that in Figure 2b, resulting in the higher density of SNPs in Figure 2a. Similarly, density of SNPs in Figure 2c is higher than that in Figure 2d. It should be noted that the density of SNPs on bare ITO surface is much lower than that on patterned one under the same conditions (14.4 no/µm2), as shown in SI Figure S4. The correlation of the density of SNPs with the feature size is shown in Figure 3. The particle density on the pattern increases exponentially with decreasing the feature sizes. For curve a presented in Figure 3, the feature sizes can be described in the following equation:
density(x) ) y0 + Ae(-x⁄t) ) 14.50 + 230.46e(-x⁄1.4)
(1)
where x is the variation of the feature size (µm) and y0 and A are constant. It is shown that, with the same space of the arrays, the density of the SNPs increases dramatically with the decrease
of the feature size. The variation in density induced by changing the predefined pattern is larger than that resulted by adjusting the ECD parameters, the former is more than 100 no/µm2, the later is less than 20 no/µm2 (see SI Figure S3). The above results indicate that the effect of pattern on the density of SNPs is stronger than that of other fabrication parameters. Moreover, comparing the two curves in Figure 3, the array with larger space can result in higher density than the one with smaller space. The enlarged images and size distributions of SNPs arrays were shown in SI Figure S5, which presents double peaks on the smaller feature. The particle sizes of Figure 2a-d are 108 ( 32, 103 ( 28, 101 ( 24, and 96 ( 19 nm, which shows that the particle sizes are larger on the higher density SNPs arrays. The 500 nm SNPs dots were successfully fabricated using this method, whose density is 137.2 no/µm2, see SI Figure S6. The measured average distance of the interparticles shown in Figure 2a-d are 8 ( 10, 18 ( 17, 52 ( 27, and 86 ( 44 nm, and it means that the higher particle density can lead to shorter interparticle distances. To gain insight into the effect of the pattern feature on the density of SNPs, we employed finite element modeling of the electric field distribution within the cell. Quickfield (version 5.5, Tera Analysis Ltd.) was used to model the electrical field distribution near the ITO surface. Under an applied potential of 300 mV, for the bare ITO substrate, the strength of the electric field is uniform on the ITO surface and the electric field strength is homogeneous (Figure 4a). However, for the patterned one, the electric field strength near the ITO surface is higher than that near resist polymer surface (panels b and c in Figure 4). Several theories have been proposed to explain the effect of that sawtooth curve electric field.28 Also, the electric field strength increases with reducing the feature size (Figure 4d) or increasing the space. The curve in Figure 4d is calculated with the average peak value on the ITO patterns from strength profile of electric field (Figure 4c); the error bar is the deviation to the average one. The field strength increases exponentially with decreasing the feature sizes, as described in eq 2
Figure 4. Cross-sectional 2-D finite element model of (a) unpatterned and (b) patterned ITO substrate at an applied bias of 300 mV. For the patterned ITO, the electrode and dielectric size is 2 µm. (c) The strength profile of electric field on the surface near to ITO substrate. (d) Dependence of field strength on the feature size of the pattern.
58 Langmuir, Vol. 25, No. 1, 2009
Letters
Figure 5. (a) SERS spectra of 4-MPy adsorbed on different SNPs arrays, the spectra of a-d correspond to the samples a-d shown in Figure 2, e is collected on bare ITO substrates; (b) The SERS intensities of SNPs at 1099 cm-1 on different dot arrays and bare ITO substrates.
strength(x) ) y0 + Ae(-x⁄t) ) 1.46 × 105 + 1.41 × 105e(-x⁄1.4) (2) where, x is the variation of the feature size (µm), y0 and A are constant. The correlation of density of SNPs and electric field strength on ITO patterns can be described with eq (3) by combining eq (1) and (2):
density(x) ) 1.65 × 10-3strength(x) - 232.50
(3)
The density of SNPs and the electric field strength exhibits linear correlation. The electric field strength increases uniformly on the unpatterned surface by increasing the applied potential. However, the electric field strength on the patterned surfaces is like sawtooth, as shown in Figure 4c, which is a gradient induced by the features of the pattern. The gradient strength can lead to an easier ion transfer and seeds formation than increasing the applied potential. The field strength on the edges of the smaller features is higher than that on the centers, which results in a higher density and larger sizes of SNPs on the edges than that on the centers (SI Figure S5). The SERS activity of the arrays of SNPs was subsequently characterized by detecting the adsorbed 4-MPy. The feature-size dependence of the SERS was observed, as seen in Figure 5a. The SERS intensities increase slightly with enlarging the space between the dots with same features, as shown in spectrum a and spectrum b (spectra c and d) in Figure 5a. With the same space between the dots (squares), the Raman signal increases dramatically with decreasing the feature size of the SNPs’ array. As shown in Figure 5b, the peak value of the Raman signal recorded at 1099 cm-1 increase dramatically. As a reference, Raman spectrum of SNPs on a bare ITO surface is collected (spectrum e in Figure 5a), which shows that the intensity of the Raman signal is much lower than that on the patterned Ag arrays. The array dependent enhancement might be induced by the increase
of the density of the SNPs, which changed the spacing of the SNPs and inceased the “hot spots” among the SNPs.29-31
Conclusion In conclusion, we reported an approach for fabricating the arrays of SNPs using NIL and ECD with low cost and high throughput. The various arrays of SNPs with different densities can be created using a single ECD process on one piece of substrate. The average gap of the SNPs can be controlled by the pattern design, such as feature size and density. Although the feature size of the arrays in this work is limited by the stamp that we used, it should be possible to scale down the feature size to several 10 nanometers by using proper stamps of NIL technique.25 Moreover, no additive or stabilizer involved in the fabrication process can extend the applications of the arrays of SNPs to the high sensitive catalysis, sensing and detecting. The signal intensity of SERS exhibits pattern-dependence on the arrays of SNPs, which increases with decreasing the feature size or increasing the spacing of the pattern. This may lead to a possible selective increasing of the SERS of different molecules on different arrays, which may be applied for sensor arrays. Acknowledgment. We thank Mr. Hirtz Michael for the particle density analysis program. Financial support was given by the National Natural Science Foundation of China (20773052, 20373019, and 50520130316), the Program for New Century Excellent Talents in University, the National Basic Research Program (2007CB808003, 2009CB939701), and the Program for Changjiang Scholars and Innovative Research Team in University (PCSIRT Grant No. IRT0422). Supporting Information Available: Three-electrode system, optimization of the ECD condition, the SEM image of SNPs on bare ITO surface and the SNPs arrays with 500 m feature size. This material is available free of charge via the Internet at http://pubs.acs.org. LA803559C
(28) Mehdizadeh, S.; Dukovic, J. O.; Andricacos, P. C.; Romankiw, L. T.; Cheh, H. Y. J. Electrochem. Soc. 1992, 139, 78. (29) Lu, Y.; Liu, G. L.; Lee, L. P. Nano Lett. 2005, 5, 5.
(30) Quinten, M. Appl. Phys. B: Laser Opt. 2001, 73, 245. (31) Felidj, N.; Aubard, J.; Levi, G. J. Chem. Phys. 1999, 111, 1195.