Langmuir 2008, 24, 821-825
821
Cheap and Robust Ultraflat Gold Surfaces Suitable for High-Resolution Surface Modification Bruno Pattier, Jean-Franc¸ ois Bardeau, Mathieu Edely, Alain Gibaud, and Nicolas Delorme* Laboratoire de Physique de l’Etat Condense´ , UMR CNRS 6087, UniVersite´ du Maine, AVenue OliVier Messiaen, 72085 Le Mans Cedex 9, France ReceiVed September 9, 2007. In Final Form: October 16, 2007 A simple procedure to elaborate robust ultraflat gold surface without clean room facilities is presented. Selfassembled 3-mercaptopropytriethoxysilane (MPTMS) on silicon was used as a buffer layer on which gold was sputtered using a common sputter-coating apparatus. The optimization of the sample position into the chamber of the sputtering machine yielded the formation of a thin (∼8 nm) gold layer. The characterization of the resulting gold surface (i.e., AFM, X-ray reflectivity, and diffraction) has demonstrated its high smoothness ( 10 nm). Likewise, if the sample is placed outside the plasma flame, gold does not bind on the MPTMS surface (i.e., the gold layer can be easily removed using an ultrasound bathing). The resulting two-step procedure initially consists of sputtering gold on the sample placed inside the plasma flame for a short period of time (e.g., t ) 30 s), allowing the binding of a thin gold layer onto the MPTMS surface. In the second step, the sample is moved outside the flame and oriented perpendicularly to the flame to continue the gold deposition for several minutes. Outside the plasma flame gold atoms are less energetic, and the orientation of the sample is known to improve the smoothness of a thin metallic film.2,19 Figure 4 shows the AFM topography image of the surface of the substrate after the deposition of gold (30 s inside the plasma flame + 4 min outside the flame). As can be seen, the surface is homogeneous and formed by very small domains with a diameter between 15 and 20 nm. (21) Chan, C.-M. Polymer Surface Modification and Characterization; Hanser Publishers: New York, 1994.
In order to quantify the roughness of the substrate, images were taken with the same tip at increasing scanning areas. Usually, the roughness parameters start to remain independent of the scanning length area only when the scanning size is large enough to include the largest surface features. In order to make consistent quantitative comparisons between images at different scales, we kept the image point density constant.22 Table 1 shows that roughness (mean and rms roughness) remains below 0.7 nm, even for an 10 × 10 µm2 image, demonstrating the high smoothness of the gold surface. Moreover, it can be seen that between 0.5 × 0.5 and 10 × 10 µm2 the roughnesses are similar, indicating the good homogeneity of the gold surface. Figure 5 shows the X-ray reflectivity (XR) curve of the gold surface and the best reflectivity fit as the solid line that describes the four well-defined Kiessig fringes. The presence of these fringes is an indication of the smoothness and the homogeneity of the gold film on a large lateral scale (∼cm). The full analysis of the XR curve can give a complete description of both orientation and density packing of a multilayer assembly.23 All the parameters found in this analysis and summarized in Table 2 were obtained from a four-layer model: a silicon substrate, a native silicon oxide layer, the MPTMS layer, and the gold film. The silicon substrate was supposed to have an infinite thickness and an electron density of 0.710 e Å-1.24 The electron density and the thickness of silicon oxide were fixed to 0.670 e Å-1 using the parameters obtained after the fit of the XR curve of a bare silicon wafer. Finally, the electron density of the gold layer was fixed to 4.39 e Å-1.25 The best fit (Figure 5) was obtained by refining the electron density of the MPTMS layer, the thickness of the gold film, and the roughness of both layers. Note that the calculated curve is convolved with a Gaussian function that takes into account the instrumental resolution. Therefore, the fitting analysis of the (22) Mendez-Vilas, A.; Bruque, J. M.; Gonzalez-Martin, M. L. Ultramicroscopy 2007, 10, 617-625. (23) Baptiste, A.; Gibaud, A.; Bardeau, J. F.; Wen, K.; Maoz, R.; Sagiv, J.; Ocko, B. M. Langmuir 2002, 18, 3916-3922. (24) Tidswell, I. M.; Ocko, B. M.; Pershan, P. S.; Wasserman, S. R.; Whitesides, G. M. Phys. ReV. B 1990, 41, 1111. (25) Gibaud, A.; Daillant, J. X-ray and Neutron ReflectiVity: Principle and Applications; Springer: Paris, 1999.
824 Langmuir, Vol. 24, No. 3, 2008
Pattier et al.
Figure 6. X-ray diffracted intensity of the gold film and theoretical refinement.
X-ray reflectivity curve provides a reliable measurement of the thickness of the gold film, which is close to 8 nm. At this range of thickness the gold film remains transparent and may enable transmission spectroscopy experiments if deposited on a transparent substrate. X-ray diffraction data were collected at grazing incidence geometry, where background minimization is achieved by fixing the grazing angle of incidence below the grazing angle for total external reflection θC ) 0.55° for gold.26 XRD patterns (Figure 6) show intense Au(111) reflections for the substrate compared to polycrystalline gold,9 indicating mainly (111) textured surfaces. Other typical reflections of the Fm3hm cubic structure are identified as the (200), (220), (311), and (222) scattering planes. XRD were refined in the cubic Fm3hm crystalline structure using the MAUD Rietveld program written by Luterrotti.27 The lattice parameter was found to be a ) 4.0796 Å, in good agreement with the value reported by Couderc (a ) 4.08 Å).28 The preferred orientation on the (111) scattering planes was adjusted during the refinement, yielding a March-Dollase value of 0.810, which well-reflects the observation made in the XRD pattern.29 This result is interesting for further surface modifications, since thiols are known to link stronger on Au(111) than on other gold crystalline structures.30 In addition, the size of the crystallites was also adjusted using an isotropic line broadening. This yielded after the convolution with the instrumental resolution a value of 9.6 (1) nm, which is about half of the value obtained by AFM. This discrepancy is likely related to the effect of the tip convolution. Such a small grain size is suitable, if we want to use these surfaces as substrates for high-resolution surface modification. Indeed, the roughness of the edges of the template are minimized when the grain size is small.31 The robustness of the gold substrates was tested by dipping surfaces into boiling acetone or by heating the sample in an oven until 100 °C. In both cases, no roughness variation was observed by AFM. The adhesion strength between the gold film and the silanized silica surface was also confirmed by a peel test using Scotch tape. In order to verify that the ultraflat gold film yields a suitable substrate for high-resolution surface modification, 1-octadecanelthiol (ODT) has been deposited by dip pen lithography (DPN). The AFM tip was “inked” with ODT, which is known (26) Daillant, J.; Alba, M. Rep. Prog. Phys. 2000, 63, 1425-1777. (27) Ferrari, M.; Lutterotti, L. J. Appl. Phys. 1994, 76, 7246-7255. (28) Couderc, J. J.; Garigue, G.; Lafourcade, L.; Nguyen, Q. T. Metallkd 1959, 50, 708. (29) Dollase, W. A. J. Appl. Crystallogr. 1986, 19, 267-272. (30) Camillone, N.; Chidsey, C. E. D.; Liu, G.; Scoles, G. J. Chem. Phys. 1993, 98, 4234-4245. (31) Love, J. C.; Estroff, L. A.; Kriebel, J. K.; Nuzzo, R. G.; Whitesides, G. M. Chem. ReV. 2005, 105, 1103-1169.
Figure 7. AFM friction image and scale drawing of an ultraflat gold surface modified by ODT using the DPN method with two different linear velocities (i.e., 3 and 6 µm s-1).
to self-assemble on gold substrate.32 The tip was then contacted to the surface involving the formation of a water meniscus. ODT molecules, which are adsorbed on the tip, were transferred to the gold surface when the probe is moved along the surface with a controlled linear velocity.32 Figure 7 shows the AFM friction image of the gold surface functionalized by ODT with two different linear velocities. The inner darker square (1 × 1 µm) was scanned at 3 µm s-1, and the outer lighter square was scanned at 6 µm s-1. Both patterns were realized using the same tipsurface force (F ) 0.9 mN) under a relative humidity of 42%. The AFM image was obtained in friction mode (using the same tip just after the DPN process) with a high linear velocity (e.g., ν ) 4 Hz). As previously shown, high-speed imaging is necessary in order to limit the transfer of molecules onto the surface,33 but it leads to a fair image quality. However, as can be seen in Figure 7, a clear difference of contrast is observed between the two deposited squares. During the DPN experiment, tip linear velocity is known to govern the quantity of transferred molecules.34 As a consequence, at high linear velocity (e.g., low thiol concentration), thiols preferentially form a liquid-like phase. However, when the linear velocity is decreased (e.g., increase of thiol concentration), the thiols can reorient to a more densely packed, standing alignment, as illustrated in Figure 7.35 As expected, the high-concentration standing phase, appearing as a dark square in the center of the image, has a relatively low friction coefficient with the AFM tip (covered by ODT molecules). Because of its lower packing density, the surrounding liquidlike region has a higher friction coefficient and appears as a lighter halo.36,37 Finally, the (32) Piner, R. D.; Zhu, J.; Xu, F.; Hong, S.; Mirkin, C. A. Science 1999, 283, 661-663. (33) Ivanisevic, A.; Mirkin, C. A. J. Am. Chem. Soc. 2001, 123 (32), 78877889. (34) Ginger, D. S.; Zhang, H.; Mirkin, C. A. Angew. Chem., Int. Ed. 2004, 43, 30-45. (35) Sheehan, P. E.; Whitman, L. J. Phys. ReV. Lett. 2002, 88, 156104.
Ultraflat Gold Surfaces for Surface Modification
unmodified gold substrate appears as a zone with high friction coefficient because of its strong interaction with the ODT-covered tip.37 The high resolution observed for the edges of the deposited ODT squares is only achievable because of the low roughness of the gold substrate. Indeed, it was previously demonstrated that roughness decreases the resolution of DPN experiments.38 This result can be explained by the increase of the water meniscus size (allowing the transfer of the molecules between the AFM tip and the surface) when the surface is rough due to multiple asperities contact.39 It was also proved that the best resolution in DPN is achieved with a gold surface with very low roughness (i.e., 0.7 nm).38
Conclusion In this study, self-assembled MPTMS on silicon was used as a platform to demonstrate the possibility to elaborate a robust ultraflat gold surface without clean room facilities. In a first step, the MPTMS silanization process was optimized. In a second step, the gold deposition by a common sputtering apparatus was optimized. The characterization of the resulting gold surface (i.e., AFM, X-ray reflectivity) has demonstrated its high smoothness (
