NANO LETTERS
Near-Field Photochemical Imaging of Noble Metal Nanostructures
2005 Vol. 5, No. 4 615-619
Christophe Hubert,* Anna Rumyantseva, Gilles Lerondel, Johan Grand, Sergeı1 Kostcheev, Laurent Billot, Alexandre Vial, Renaud Bachelot, and Pascal Royer Laboratoire de Nanotechnologie et d’Instrumentation Optique, CNRS FRE 2671, UniVersite´ de Technologie de Troyes, 12 rue Marie-Curie, BP 2060, 10010 Troyes Cedex, France
Shih-hui Chang, Stephen K. Gray, and Gary P. Wiederrecht Chemistry DiVision and Center for Nanoscale Materials, Argonne National Laboratory, Argonne, Illinois 60439
George C. Schatz Department of Chemistry and Institute for Nanotechnology, Northwestern UniVersity, EVanston, Illinois 60208 Received December 10, 2004; Revised Manuscript Received February 2, 2005
ABSTRACT The sub-diffraction imaging of the optical near-field in nanostructures, based on a photochemical technique, is reported. A photosensitive azobenzene-dye polymer is spin coated onto lithographic structures and is subsequently irradiated with laser light. Photoinduced mass transport creates topographic modifications at the polymer film surface that are then measured with atomic force microscopy (AFM). The AFM images correlate with rigorous theoretical calculations of the near-field intensities for a range of different nanostructures and illumination polarizations. This approach is a first step toward additional methods for resolving confined optical near fields, which can augment scanning probe methodologies for high spatial resolution of optical near fields.
Near-field optics has attracted a great deal of attention during the past two decades due to the prospect for spatially resolving optical fields at length scales far below the diffraction limit. As a result, many theoretical and experimental efforts have been devoted to the study of the optical near-fields of nanostructures.1-5 Successful experimental demonstrations use scanning probe methodologies, particularly photon scanning tunneling microscopy (PSTM), aperture scanning near-field optical microscopy (SNOM), and apertureless scanning near field optical microscopy (ASNOM). While these are enormously worthwhile research directions, it is a fact that the scanning probe generally modifies the very near field that is of interest. This occurs through a variety of nanoscale interactions, such as dipoledipole coupling or multiple scattering between sample and probe. Overall, the process of image formation with SNOM and the challenge of probe manufacturing still constitute subjects of research. To avoid these issues, other approaches have recently been proposed that attempt to use a photo* Corresponding author. E-mail:
[email protected]; Fax: +33 (0) 3 25 71 84 56; Phone number: +33 (0) 3 25 71 80 23. 10.1021/nl047956i CCC: $30.25 Published on Web 03/09/2005
© 2005 American Chemical Society
sensitive polymer to map the optical near-field of different kinds of particles.6-9 While these efforts have successfully shown topographic features induced by the particles, they have not been able to spatially resolve the optical fields with the precision of scanning probe based methods. Here we report an original approach based on the exposure of lithographically designed metallic nanostructures coated with a photosensitive azo-dye polymer, where each chromophore acts as a probe of the intensity of the electric field. For the first time without scanning optical probes, the dipolar response of excited metal nanoparticles and the lightning rod effect in elongated nanoparticles have been resolved. This work should significantly impact nanophotonics research, providing a new tool for obtaining spatial information of the optical near-field for appropriate structures. Our approach has numerous advantages over other nanooptical characterization methods of noble metal nanostructures. First, topographic modifications of the polymer film surface are directly recorded with an atomic force microscope (AFM) after exposure and without any chemical treatment. Second, removal of the spin coated film with the
polymer solvent enables the same sample to be characterized with this method several times without any degradation of the metallic structures. Finally, in addition to avoiding sample-probe electromagnetic interactions, this method avoids the problem of low signal-to-noise ratios that can plague SNOM measurements. It should be however stressed that our method is not expected to compete with SNOM whose efficiency has been demonstrated over the past 20 years. On the contrary, our method should be viewed as a complementary tool to observe near-field features of nanostructures. To map the optical near-field of metallic nanostructures, we take advantage of the mass transport properties of poly(methyl methacrylate) (PMMA) functionalized with azobenzene-dye chromophores. It is now well established that azodyes undergo trans-cis isomerization following photoexcitation, and that the use of polarized light enables molecular rotation through thermal diffusion within the PMMA matrix to finally produce, because of energy minimization, a full reorientation of the chromophores.10,11 Azo-dye molecules thus play the role of molecular motors, pushing or pulling the polymeric host as reorientation occurs. This produces a photoinduced mass transport from high to low intensity illuminated regions, thus creating a one-step photoinscription of surface relief. This phenomenon has previously been used to reversibly create surface relief gratings in azo-dye polymer films by illuminating the polymer film with an interference pattern of polarized laser beams at a wavelength lying in the absorption band of the azo-dye molecules.12,13 It is known that the mechanism leading to the surface relief gratings is neither swelling due to light absorption nor ablation in the high-intensity regions. Rather, surface deformation is connected with the photoisomerization process of chromophores, with different models indicating that a mass transport phenomenon can occur well below the glass transition temperature.14-18 The technique for photochemical optical near-field imaging begins with the three steps shown in Figure 1. Nanostructures are first fabricated by electron-beam lithography, typically through the lift-off method.19 For this study, a set of differing silver or gold nanostructures is created with a 50 nm thickness (the height of the nanostructures are measured with an atomic force microscope). The second step is the deposition of the optical resist. It consists of the azo-dye molecule Dispersed Red 1 [4-(N-(2-hydroxyethyl)-N-ethyl)amino-4′-nitroazobenzene] (DR1) grafted as a side chain to PMMA in a 30% molar ratio (DR1MA/MMA). DR1MA/ MMA is then dissolved in 1-1-2 trichloroethane or in methylisobutyl-ketone (MIBK) (same experimental results were obtained regardless of the solvent). Polymer films of different thicknesses are usually obtained through spin-coating at varying speeds and different concentrations of copolymer. In our case, the thickness of the polymer film is equal to 80 nm, which is sufficient to fully cover the structures and thin enough to be sensitive to the optical near-field of the particles. No drying was performed after spin coating. The third step consists of illuminating the sample, in this case at normal incidence. To overlap the 400-600 nm absorption 616
Figure 1. Schematic view of the approach. (1) Fabrication of metallic nanostructures on glass, to include arrays of dots and ellipsoids as shown by the two electron microscopy (SEM) images. (2) Spin-coating of an azo-polymer on the metallic structures. The trans and cis isomeric forms of the Dispersed Red 1 azo-dye molecule are also shown. (3) Irradiation of the sample with the 514 nm line of an argon ion laser. Circular or linear polarization of the laser beam can be used.
band of the DR1 molecule, we used the 514 nm or 532 nm lines of an argon-ion laser or a frequency doubled, diodepumped Nd:YAG laser, respectively. As described in detail below, the polarization and irradiation intensity were carefully controlled and correlated with the detected topographic features. Following the illumination process (step 3), the “imaging” of the optically induced topography is performed through atomic force microscopy. Theoretical modeling is used to support the experimental observations. We rigorously calculate the near-field optical intensities using the finite-difference time-domain (FDTD) method20 and show that the negative image of the computed near-field intensities can be correlated with the observed photoinduced topographies. This negative image illustrates the fact that the matter escapes from high-intensity regions to low intensity regions. We also investigated several simple models of the polymer mass transport as it relates to the field intensity. These included the second derivative model of Tripathy and co-workers,18 but found that the models lead to similar localized field distributions. Our proposed model based on the negative intensity actually corresponds to the first-order approximation of the comprehensive photoassisted matter migration model given by ref 14. Nano Lett., Vol. 5, No. 4, 2005
The FDTD calculations were fully three-dimensional with appropriate periodic and absorbing boundary conditions.20 The metals were described by Drude models with parameters fit to the experimental dielectric constant data for the wavelengths of interest, as described in ref 21. The glass substrate and DR1MA/MMA were also included in the calculations, with dielectric constants of 2.25 and 2.89, respectively. Fourier transformations of the time-domain fields on the wavelengths of interest then yield steady-state fields and thus field intensities. As a first example of the near-field imaging capability of this method, Figure 2 shows the results obtained with an array of silver nanoparticles having a diameter of 75 nm, a height of 50 nm, and a periodicity of 500 nm. Since the extinction spectra of the arrays performed before and after spin-coating indicate that the polymer film layer induces a red shift of 80 nm on the silver particles’ plasmon resonance position, a 532 nm irradiation wavelength was used to optimally overlap this resonance. This is also confirmed by an FDTD calculation of the extinction spectrum, which shows a maximum near 540 nm. After irradiation, two holes can be observed (Figures 2a,b) in the polymer that are close to the particles and are oriented with the incident light polarization. The depth of each hole is ≈4 nm, with diameter ≈100 nm, which corresponds approximately to λ/5. The depressions correlate remarkably well with the expected dipolar near-field spatial profile, which to our knowledge has not been resolved with any other method using photosensitive polymers.6-9 Numerical calculations of the electric field intensity distribution around a silver nanoparticle covered with DR1MA/MMA and irradiated with a linearly polarized laser beam show that intensity maxima are located at the same position as the holes observed in the topographic image. Figure 2c displays the theoretical electric field intensity, and Figure 2d displays its negative image. (The experimental AFM results over the metal nanoparticles naturally reflect the particle heights as opposed to just field intensities. Therefore in all our theoretical plots a constant factor is added to the corresponding optical intensity or its negative image over the particle region to reflect this.) The negative image agrees qualitatively with its experimental counterpart, Figure 2b. Figure 2e displays line cuts along the polarization axis and through a particle. The AFM profiles (left) of the surface before (red) and after (blue) irradiation are displayed as well as the corresponding theoretically computed negative of the near-field intensity (right). The positions of the dumps are consistent with molecular transport from the regions of high optical intensity to regions of low optical intensity. Topographic modifications observed after irradiation are thus due to a mass transport phenomenon photoinduced by the optical near-field of the metallic silver particles produced by the plasmon resonance. This is supported by a strong correlation between the negative of the electric field intensity around a silver nanoparticle (Figure 2d) and the AFM image of the polymer surface after irradiation (Figure 2b). The minima located near the particle correspond to dumps on the topographic image. However, there is not a perfect matching of negative field intensity to Nano Lett., Vol. 5, No. 4, 2005
Figure 2. Topographic images and theoretical calculations of optical near fields around silver nanoparticles. AFM image recorded after irradiation of silver particles covered with DR1MA/MMA (a,b). Irradiation wavelength, time, and intensity were, respectively, 532 nm, 20 min and 50mW/cm2. The light polarization direction is indicated within the AFM images. The silver particles have a diameter of 75 nm, a height of 50 nm, and a periodicity of 500 nm. MIBK was used as the solvent. (c) Image of the theoretically computed near-field intensity around a particle indicated in black, and its negative image (d). The color scale in (a)-(d) is such that white is high and black is low. (e) Comparison of line cuts of the AFM image before (red) and after (blue) irradiation and the negative of the calculated field intensity. Line cuts were taken through a particle and in the polarization direction. The calculated field intensity images correspond to the field intensity on the polymer surface.
surface topography: on the theoretical image, the field intensity is more localized near the particle edges, which could be due to tip convolution as well as to molecular diffusion. Next we study the optical near-field around gold ellipsoidal particles under different incident light polarization directions, i.e., parallel or perpendicular to the long axis of the ellipsoids. The results are displayed in Figures 3a-e. The ellipsoids are 50 nm in height, with long and short axis lengths of 1 µm and 60 nm, respectively. Particle to particle distances are 800 nm in the long axis direction. 617
Figure 3. Topographic images and theoretical calculations of optical near fields around gold ellipsoids. AFM images recorded before (b) and after (a, c) irradiation of gold ellipsoids covered with DR1MA/MMA. The light polarization direction is parallel (a) and perpendicular (c) to the long axis of the ellipsoids. Irradiation wavelength, time, and intensity were respectively equal to 514 nm, 30 min, and 100 mW/cm2. 1-1-2-Trichloroethane was used as the solvent. Figures d, e represent the theoretical negative images of the field intensity for parallel and perpendicular polarization, respectively. The color scale in (d-e) is such that white is high and black is low. The black line outline the boundary of the simulated ellipsoid.
Figure 3b corresponds to the preirradiation experimental AFM image. Extinction spectra performed after spin coating of the azo-polymer layer indicate that for polarization parallel to the long axis, the 514 nm irradiation wavelength used is outside the resonance plasmon band, whereas in the case of a polarization direction perpendicular to the long axis, the 514 nm irradiation wavelength is resonant with the plasmon band. The spectral response was also calculated with the FDTD method, and we found that the resonant peak occurs at 510 nm, and 1.28 µm for polarization along the short and long axis, respectively. This agrees well with the above experimental observation. Figure 3a clearly shows that the polymer topography has minima located at the ellipsoid extremities when illuminated with a polarization parallel to the long axis. This is explained by the fact that, for this polarization, an electromagnetic singularity is excited at these extremities. This is consistent with photoinduced mass transport occurring from high to low intensity regions, i.e., optical intensity maxima leading to topographic minima. This is supported by Figure 3d, representing the theoretical negative image of the field intensity and showing dips near the ellipsoid extremities. Since for such a polarization the plasmon resonance associated with the long ellipsoid axis does not overlap with the irradiation wavelength, the observed effect is due to a geometrical singularity called the lightning rod effect rather than a plasmon resonance.22 In the case of polarization along the short ellipsoidal axis, Figure 3c, dips can be observed, located along the long axis of the particles. These dips are located at the same position as the optical near-field intensity maxima around ellipsoids for such a polarization. This is confirmed when looking at the corresponding calculated negative image of the field intensity 618
Figure 4. Topographic images and theoretical calculations of silver nanoparticles illuminated with circularly polarized laser light. AFM images recorded before (a) and after (b) irradiation of silver particles covered with DR1MA/MMA. (c) Blow-up of (b). Irradiation wavelength, time and intensity were equal to 514 nm, 30 min and 100 mW/cm2, respectively. 1-1-2-Trichloroethane was used as a solvent. Silver particles are 50 nm in height, 100 nm in diameter and the spacing is equal to 1 µm. (d) Theoretical negative field image corresponding to (c), and (e) is a surface plot around one of the particles.
around the particles for this illumination condition, Figure 3e, which is in good agreement with experimental observations. As a last example of the ability of this method to spatially resolve complex fields, Figure 4 shows the result obtained with silver nanoparticles covered with DR1MA/MMA but now irradiated with a circularly polarized laser beam. The silver particles are 50 nm in height, 100 nm in diameter and have a periodicity of 1 µm (Figure 4a). In this case, large topographic modifications are again observed at the polymer film surface, as shown in Figures 4b and 4c. The experimental observations can be compared with the computed negative of the electric field intensity, shown as a twoNano Lett., Vol. 5, No. 4, 2005
dimensional image in Figure 4d and as a surface plot in Figure 4e. The AFM image of Figure 4c shows that an inner array of lobes around each particle can be distinguished, as well as an outer array of lobes, whose periodicity is equal to the lattice spacing. These outer lobes probably result from far-field interferences of diffraction orders. Quite encouragingly, the negative computed field intensity of Figures 4d and 4e also shows inner and outer high relief features around the particles, although they are not as structured as the experimental result (Figure 4c). This probably originates from the fact that theoretical calculations do not actually take into account the diffusion of azodye molecules and thus cannot perfectly reproduce their behavior under illumination with very intense localized near-fields. The results presented here confirm that near-field photochemical imaging can be applied to a variety of metallic structures and illumination conditions. Compared to SNOM, this straightforward method appears as an additional tool for near-field imaging. Remarkably, our method has enabled the easy visualization of important near-field phenomena, such as dipolar responses from metallic nanoparticles and light confinement at the end of nanoellipses. Of course, as any imaging technique, our method presents some limits. For example, fluorescence near-field imaging in liquid is for the moment not enabled. So far our method has been used to characterize metal nanostructures exclusively. We believe that it can be extended to other materials and more complex structures. However, to be applied to unknown samples, a better knowledge of the optical response of the photosensitive material is needed. Molecular engineering possibilities will also permit the design of organic materials whose refractive index or absorption band position can be controlled and thus correspond to different structural characteristics (shape, size, material) and irradiation wavelength used. Furthermore, this method may ultimately be applied to the nanooptical manipulation of molecular motors to produce nanoscale mass transport of a material. Of course, there remain significant basic science issues for understanding and theoretically modeling the nanooptical forces and material transport phenomena involved. However, we believe the present paper is an important first step toward a new and efficient micro and nanooptical characterization technique.
Nano Lett., Vol. 5, No. 4, 2005
Acknowledgment. The work at Argonne was supported by the U.S. Department of Energy, Office of Basic Energy Sciences, Division of Chemical Sciences, Geosciences, and Biosciences under DOE contract W-31-109-ENG-38. Work at Northwestern University was supported by DOE grant DEFG02-03ER15487, and by NSF through the Materials Research Science and Engineering Center (MRSEC). References (1) Salerno, M.; Fe´lidj, N.; Krenn, J. R.; Leitner, A.; Aussenegg, F. R. Phys. ReV. B 2001, 63, 165422. (2) Krenn, J. R.; Salerno, M.; Fe´lidj, N.; Lamprecht, B.; Schider, G.; Leitner, A.; Aussenegg, F. R.; Weeber, J. C.; Dereux, A.; Goudonnet, J. P. J. Microsc. 2000, 202, 122. (3) Bouhelier, A.; Huser, T.; Hamaru, H.; Gu¨ntherodt, H. J.; Pohl, D. W.; Baida, F. I.; Van Labeke, D. Phys. ReV. B 2001, 63, 155404. (4) Krenn, J. R.; Wolf, R.; Leitner, A.; Aussenegg, F. R. Opt. Comm. 1997, 137, 46. (5) Hillenbrand, R.; Keilmann, F.; Hanarp, P.; Sutherland, D. S.; Aizpurua, J. Appl. Phys. Lett. 2003, 83, 368. (6) Andre, P.; Charra, F.; Chollet, P. A.; Pileni, M. P. AdV. Mater. 2002, 14, 601. (7) Ikawa, T.; Hasegawa, M.; Tsuchimori, M.; Watanabe, O.; Kawata, Y.; Egami, C.; Sugihara, O.; Okamoto, N. Synth. Met. 2001, 124, 159. (8) Kawata, Y.; Egami, C.; Nakamura, O.; Sugihara, O.; Okamoto, N.; Tsuchimori, M.; Watanabe, O. Opt. Commun. 1999, 161, 6. (9) Kik, P. G.; Maier, S. A.; Atwater, H. A Mater. Res. Soc. Symp. Proc. 2002, 705, 101. (10) Jones, C.; Day, S. Nature 1991, 351, 15. (11) Hall, D. B.; Dhinojwala, A.; Torkelson, M. Phys. ReV. Lett. 1997, 79, 103. (12) Kim, D. Y.; Tripathy, S. K.; Li, L.; Kumar, J. Appl. Phys. Lett. 1995, 66, 1166. (13) Rochon, P.; Batalla, E.; Natansohn, A. Appl. Phys. Lett. 1995, 66, 136. (14) Lefin, P.; Fiorini, C.; Nunzi, J. M. Opt. Mater. 1998, 9, 323. (15) Barrett, C. J.; Natansohn, A.; Rochon, P. L. J. Phys. Chem. 1996, 100, 8836. (16) Barrett, C. J.; Rochon, P. L.; Natansohn, A. J. Chem. Phys. 1998, 109, 1505. (17) Pedersen, T. G.; Johansen, P. M. Phys. ReV. Lett. 1997, 79, 2470. (18) Bian, S.; Li, L.; Kumar, J.; Kim, D. Y.; Williams, J.; Tripathy, S. K. Appl. Phys. Lett. 1998, 73, 1. (19) Grand, J.; Kostcheev, S.; Bijeon, J. L.; Lamy de la Chapelle, M.; Adam, P. M.; Rumyantseva, A.; Lerondel, G.; Royer, P. Synth. Met. 2003, 139, 621. (20) Taflove, A.; Hagness, S. C. In Computational Electrodynamics: The Finite-Difference Time- Domain Method, 2nd edition; Artech House: Boston, 2000. (21) Gray, S. K.; Kupka, T. Phys. ReV. B 2003, 68, 045415. (22) Novotny, L.; Bian, R. X.; Xie, X. S. Phys. ReV. Lett. 1997, 79, 645.
NL047956I
619
