Stepwise Self-Assembly of Ordered Supramolecular Assemblies

Stepwise Formation of Ruthenium(II) Complexes by Direct Reaction on Organized Assemblies of Thiol-Terpyridine Species on Gold. Nunzio Tuccitto , Vanna...
0 downloads 0 Views 196KB Size
Download PDF
2082

Langmuir 2006, 22, 2082-2089

Stepwise Self-Assembly of Ordered Supramolecular Assemblies Based on Coordination Chemistry Daniel R. Blasini,†,‡ Samuel Flores-Torres,† Detlef-M. Smilgies,‡ and He´ctor D. Abrun˜a,*,† Department of Chemistry and Chemical Biology, Baker Laboratory, Cornell UniVersity, Ithaca, New York 14853-1301, and Cornell High Energy Synchrotron Source (CHESS), Wilson Laboratory, Cornell UniVersity, Ithaca, New York 14853 ReceiVed September 20, 2005. In Final Form: NoVember 29, 2005 Ultrathin multilayers based on transition metal complexes have been prepared by successive deposition and selfassembly. Dendrimer layers were deposited onto SiO2 wafers by alternately immersing the substrate into a solution of terpyridyl (tpy)-pendant poly(amido amine) (PAMAM) dendrimers (dend-n-tpy; n ) 8, 16) dissolved in CH2Cl2, followed by the interfacial coordination reaction with cobalt (Co2+) from aqueous solution. The films derived from this simple assembly method have been characterized by electrochemical methods, synchrotron-based X-ray reflectivity (XRR), and X-ray fluorescence (XRF) recorded under grazing incidence. XRR analysis revealed a linear thickness dependence with an increase of 9.9 ( 0.5 Å and 10.8 ( 0.4 Å per growth cycle for both dend-8-tpy and dend-16-tpy, respectively, indicative of a layer-by-layer (LbL) growth of single dendrimer/Co2+ layers. XRF and electrochemical results showed that the amount of Co2+ increases linearly as more layers are deposited, and that the Co2+ concentration (mol/L) in dend-8-tpy/Co2+ films decays slowly as the number of growth cycles (l) increases. Moreover, a preliminary kinetics analysis indicated that the growth of a dendrimer layer in a deposition cycle is a self-limiting process.

Introduction Supramolecular assemblies, constructed from electro- or photoactive molecules as building blocks, represent an attractive way to create materials with deliberate architectures suitable for the development of novel catalytic systems1,2 and electrochromic devices,3,4 as well as for energy conversion schemes.5,6 These systems also have potential applications as electron-transfer mediators,7 in controlled delivery systems,8,9 in biosensors,10 and in chromatographic separations.11 Moreover, suparamolecular assemblies capable of incorporating paramagnetic transition metal ions, such as Co2+ and Ni2+, present interesting magnetic properties and could find applications in numerous fields.12,13 The use of dendrimers as molecular building blocks for such assemblies is of particular interest because their size, geometry, and chemical functionality can be exquisitely controlled.14-16 Because of this ability, dendrimers offer a unique opportunity * Corresponding author. Phone: (607) 255-7568. Fax: (607) 255-9864. E-mail: [email protected]. † Baker Laboratory. ‡ Wilson Laboratory. (1) Cheng, L.; Pacey, G. E.; Cox, J. A. Electrochim. Acta 2001, 46, 4223. (2) Heechang, Y.; Crooks, R. M. J. Am. Chem. Soc. 2005, 127, 4930. (3) Newkome, G. R.; Moorefield, C. N.; Vo¨gtle, F. Dendritic Molecules: Concepts, Synthesis, PerspectiVes; VCH: Weinheim, Germany, 1996. (4) Newkome, G. R. AdVances in Dendritic Macromolecules; JAI Press: Greenwich, CT, 1994. (5) Balzani, V.; Campagna, S.; Denti, G.; Juris, A.; Serroni, S.; Venturi, M. Acc. Chem. Res. 1998, 31, 26. (6) Balzani, V.; Ceroni, P.; Juris, A.; Venturi, M.; Campagna, S.; Puntoriero, F.; Serroni, S. Coord. Chem. ReV. 2001, 219-221, 545. (7) Gorman, C. B.; Smith, J. C.; Hager, M. W.; Parkhurst, B. L.; SierzputowskaGracz, H.; Haney, C. A. J. Am. Chem. Soc. 1999, 121, 9958. (8) Khopade, A. J.; Caruso, F. Biomacromelecules 2002, 3, 1154. (9) Khopade, A. J.; Caruso, F. Nano Lett. 2002, 2, 415. (10) Yoon, H. C.; Kim, H. Anal. Chem. 2000, 72, 922. (11) Castognola, M.; Cassiano, L.; Lupi, A.; Messana, I.; Patamia, M.; Rabino, R.; Rossetti, D. V.; Giardina, B. J. Chromatogr., A 1995, 694, 463. (12) Waldmann, O.; Hassmann, J.; Mu¨ller, P.; Hanan, G. S.; Volkmer, D.; Schubert, U. S.; Lehn, J.-M. Phys. ReV. Lett. 1997, 78, 3390. (13) Waldmann, O.; Hassmann, J.; Mu¨ller, P.; Volkmer, D.; Schubert, U. S.; Lehn, J.-M. Phys. ReV. B 1998, 58, 3277. (14) Tomalia, D. A.; Taylor, A. M.; Goddard, W. A. Angew. Chem., Int. Ed. Engl. 1990, 29, 138. (15) Frechet, J. M. J. Science 1994, 263, 1710. (16) Zeng, F.; Zimmerman, S. C. Chem. ReV. 1997, 97, 1681.

for the fabrication of structurally controlled and well-defined films. The preparation of dendrimer-based ultrathin multilayer assemblies has relied mainly on self-assembly, using electrostatic1,8,9,17-19 and nonelectrostatic principles.10,20-24 In the electrostatic mode of self-assembly (ESA), first demonstrated by Decher,25,26 ordered multilayers could be prepared in a layer-by-layer (LbL) fashion by alternating the adsorption of oppositely charged polyelectrolytes.27-29 In such LbL growth mode, a complete deposition cycle consists of a bilayer, obtained from the adsorption of two oppositely charged polyelectrolytes. This growth method exhibits a linear increase in the total film thickness as a function of the number of deposition cycles. The thickness of the bilayer depends on the polyelectrolytes used, the concentrations, and the ionic strength of the solution.29 This technique has been successfully extended to the preparation of a wide variety of systems of dendrimerdendrimer,17 dendrimer-polymer,8,9,18 dendrimer-conductive polymer,19 and dendrimer-inorganic assemblies.1 Although the assemblies prepared by this method exhibited layer-by-layer (LbL) growth behavior, such films were generally found to have poor resistance to polar solvents and limited stability. Among the nonelectrostatic approaches employed for the formation of dendrimer based multilayer assemblies, the use of covalent bonding chemistry,10,20-22 biospecific recognition,23 and (17) Tsukruk, V. V.; Rinderspacher, F.; Bliznyuk, V. N. Langmuir 1997, 13, 2171 and references therein. (18) Khopade, A. J.; Caruso, F. Langmuir 2002, 18, 7669. (19) Li, C.; Mitamura, K.; Imae, T. Macromolecules 2003, 36, 9957. (20) Liu, Y.; Brueninig, M. L.; Bergbreiter, D. E.; Crooks, R. M. Angew. Chem., Int. Ed. Engl. 1997, 36, 2114. (21) Zhao, M.; Liu, Y.; Crooks, R. M.; Bergbreiter, D. E. J. Am. Chem. Soc. 1999, 121, 923. (22) Zhong, H.; Wang, J.; Jia, X.; Li, Y.; Qin, Y.; Chen, J.; Sheng Zhao, X.; Cao, W.; Li, M.; Wei, Y. Macromol. Rapid Commun. 2001, 22, 583. (23) Anzai, J.; Kobayashi, Y.; Nakamura, N.; Nishimura, M.; Hoshi, T. Langmuir 1999, 15, 221. (24) Watanabe, S.; Regen, S. L. J. Am. Chem. Soc. 1994, 116, 8855. (25) Decher, G.; Hong, J. Makromol. Chem. Macromol. Symp. 1991, 46, 321. (26) Decher, G.; Hong, J. Ber. Bunsen-Ges. Phys. Chem. 1991, 95, 1430. (27) Decher, G. Science 1997, 277, 1232. (28) Arys, X.; Fischer, P.; Jonas, A. M.; Koetse, M. M.; Laschewsky, A.; Legras, R.; Wischerhoff, E. J. Am. Chem. Soc. 2003, 125, 1859. (29) Lvov, Y. M.; Decher, G. Crystallogr. Rep. 1994, 39, 625.

10.1021/la052558w CCC: $33.50 © 2006 American Chemical Society Published on Web 01/31/2006

Self-Assembly of Ordered Supramolecular Assemblies

Langmuir, Vol. 22, No. 5, 2006 2083

crystal surfaces.30-32 Moreover, it was demonstrated that such highly ordered domains arose from the two-dimensional packing of one-dimensional strands (akin to a pearl necklace) where the repeat unit in the case of dendrimers reacting with M2+ (M ) Fe or Co) was (tpy-dend-(tpy′)n-2-tpy-M2+)x where tpy ) terpyridine which bridges two dendrimer molecules (via coordination to a M2+ center) and tpy′ ) terpyridine not coordinated to a M2+ center. It was also established that not all the tpybearing arms in the dendrimers were coordinated to a metal center and that the amount of metal coordinated in the monolayers varied with dendrimer generation. This behavior was due, at least in part, to steric hindrance and because the coordination reaction was constrained to the interface between two immiscible liquids (CH2Cl2/water) as mentioned earlier.30,31 Here, we report on the formation of multilayers of terpyridyl (tpy)-pendant PAMAM dendrimers (dend-n-tpy, n ) 8, 16) using a similar approach (i.e. interfacial coordination chemistry) with cobalt ions (Co2+). The dendrimer multilayer films were prepared by the repetitive and sequential deposition of the terpyridylpendant PAMAM dendrimers (dend-n-tpy) (from CH2Cl2 solution) onto SiO2, followed by the interfacial coordination reaction with an aqueous solution of cobalt ions (Co2+). Thin films derived by this simple method using two different dendrimer generations (dend-n-tpy, n ) 8, 16) were characterized by synchrotron-based X-ray reflectivity (XRR) and X-ray fluorescence (XRF) techniques. Experimental Section

Figure 1. Structures of (A) dend-16-tpy and (B) dend-8-tpy.

coordination chemistry24 has been demonstrated. The use of coordination chemistry for the preparation of such assemblies introduces the versatility of a wide selection of ligands and transition metal ions that can be used as building blocks, thus providing the potential of tuning the film properties by the judicious combination of its components. Furthermore, this approach offers the possibility of controlling surface structure at the molecular level via synthesis of deliberately designed ligands. We have previously shown that the interfacial coordination reaction (at the CH2Cl2/water interface) of terpyridyl (tpy) pendant poly(amido amine) (PAMAM) dendrimers (dend-n-tpy, where n represents the number of tpy bearing arms) (see Figure 1)30,31 and other terpyridine-containing ligands32 (dissolved in CH2Cl2) with transition metals such as cobalt (Co2+) and iron (Fe2+) (in aqueous solutions) led to the formation of highly ordered twodimensional structures on surfaces such as HOPG and Pt single(30) Diaz, D. J.; Storrier, G. D.; Bernhard, S.; Takada, K.; Abrun˜a, H. D. Langmuir 1999, 15, 7351. (31) Diaz, D. J.; Bernhard, S.; Storrier, G. D.; Abrun˜a, H. D. J. Phys. Chem. B. 2001, 105, 8746.

Synthesis. Dend-n-tpy (where n ) 8, 16) were prepared as described previously33,34 via peptide coupling of the carboxylic acidpendant terpyridine ligand with the appropriate poly(amido amine) (PAMAM) starburst dendrimer obtained from Aldrich Chem. Co. PAMAM starburst dendrimers of generations 1 (G1) and generation 2 (G2) (i.e. with 8 and 16 pendant groups, respectively) were employed for the synthesis of dend-8-tpy and dend-16-tpy, respectively. The structure of dend-16-tpy and that of dend-8-tpy are presented in Figure 1A,B. Materials. High-purity solvents CH2Cl2 (HPLC grade, Burdick & Jackson) and CH3OH (HPLC grade, Sigma-Aldrich) and highpurity reagents CoSO4‚6H2O (99.998%, Aldrich), 30% H2O2 (CMOS grade, J.T. Baker), NH4OH (reagent grade, Mallinckrodt), and H2SO4 (99.999%, Sigma-Aldrich) were used as received. NaClO4 (anhydrous, 98.0-102.0% (assay), Alfa Aesar) was recrystallized three times from methanol and dried under vacuum for 72 h before use. Aqueous solutions were prepared using Millipore water of at least 18 MΩ‚cm resistance. Silicon wafers (Si(111) with a 1000 Å thermal oxide layer, (Virginia Semiconductor, Inc.) were used as substrates. The wafers were cut into 2 × 2 cm pieces, degreased in a methanol (CH3OH) hot bath, cleaned with the standard RCAcleaning process (SC-1),35 and stored under Millipore water prior to use. All the glassware and Teflon tweezers used during the experiments were cleaned with the SC-1 solution. The platinum (Pt) electrodes employed in the electrochemical characterization were prepared by sealing a 1 mm Pt wire (99.997%, Alfa Aesar) in glass. The electrodes were mechanically polished using 800 and 1200 polishing grit (Buehler), as well as 1 µm diamond paste (MetadiBuehler), between experiments. Afterward the electrodes were sonicated for 15 min in Millipore water and cleaned electrochemically in 0.1 M H2SO4 by cycling the potential between the oxygen and hydrogen evolution potentials until the characteristic cyclic voltammogram (CV) of a clean polycrystalline Pt electrode was obtained. The active electrochemical area was obtained by integrating the (32) Bernhard, S.; Takada, K.; Diaz, D. J.; Abrun˜a, H. D.; Murner, H. J. Am. Chem. Soc. 2001, 123, 10265. (33) Storrier, G.; Takada, K.; Abrun˜a, H. D. Langmuir 1999, 15, 872. (34) Takada, K.; Storrier, G.; Moran, M.; Abrun˜a, H. D. Langmuir 1999, 15, 7333 and references therein. (35) Kern, W. J. Electrochem. Soc. 1990, 137, 1887.

2084 Langmuir, Vol. 22, No. 5, 2006 charge under the hydrogen adsorption waves (typical roughness factors of 30-50% were observed). The clean Pt electrodes were rinsed and stored under Millipore water prior to use. Film Preparation. Films were prepared using a modification to the procedure presented in ref 31. In detail, either the silicon substrates or the Pt electrodes were dried under a stream of nitrogen and placed in a solution of either dend-n-tpy (n ) 8 or 16) in dichloromethane (CH2Cl2) for 10-15 min. Typical solution concentrations were between 0.05 and 0.10 mM. The samples were slowly removed from the solution, so that the solvent could drip from the surface, and the surface was subsequently rinsed with CH2Cl2, to remove excess reagent and blown dry with nitrogen. The sample was then immersed in an aqueous solution of 0.1 M CoSO4 and 0.1 M NaClO4 for 2-5 min. The sample was then removed from this solution and rinsed thoroughly with water and methanol and dried under a stream of nitrogen. This represented a complete growth cycle (l) in which the tpy-bearing arms of two different dendrimers molecules are linked through the formation of a [Co(tpy)2]2+ complex, and the repeat unit is (tpy-dend-(tpy′)n-2-tpy-M2+)x (where tpy denotes a terpyridine group which bridges two dendrimer molecules via coordination to a Co2+ center and tpy′ denotes a terpyridine group not coordinated to a Co2+ center).30,31 Multilayers were obtained by the repetition of these steps, with a new layer being generated at the end of each growth cycle (l). In the kinetics studies, the solution concentrations of both dend8-tpy and cobalt were kept constant at 0.10 mM and 0.1 M, respectively, and the reaction time with the cobalt ions was kept constant at 2 min. The dendrimer deposition time was varied for the formation of four layers; all four growth cycles (l) were carried out with the same deposition time within each data set. All the samples were prepared at room temperature, under atmospheric conditions, and stored in a desiccator in the dark. The samples were characterized within a few days of being prepared. Samples reanalyzed after a few months of storage yielded similar results, indicating that the films are stable to long-term storage under ambient conditions. Characterization. Synchrotron radiation based X-ray reflectivity (XRR) using 10.4 keV incident radiation was used to determine the normal density profile of the films grown onto SiO2. The data were analyzed with the formalism developed by Parratt,36 using the software package “Parratt32” (Christian Braun, Hahn-MeitnerInstitut, Berlin, 1998). From such analysis the thickness, the rms (root-mean-square) roughness and laterally averaged electron density profiles for the films were determined as a function of the number of deposition cycles. Specular XRR profiles were measured at the G-2 station of the Cornell High-Energy Synchrotron Source (CHESS). G-line is supplied with photons from a 50-pole wiggler in the 5.2 GeV Cornell storage ring. In the G-2 station, a high-flux X-ray beam from a multilayer monochromator is split by means of a beryllium single crystal into a broad-band-pass transmitted beam and a monochromatic side-bounce beam for G2. The incident beam was collimated with two sets of slits of 1 mm height and 5 mm width to a maximum divergence of about a few hundredths of a degree vertically. Kiessig fringes37 from films up to 200 Å could be easily resolved, while maximizing the signal-to-noise ratio for the weakly scattering organic films. Samples were mounted on a motorized height translation stage of a vertical four-circle Huber diffractometer. The reflected intensity was recorded as a function of the glancing incident angle, R, for the incident 10.4 keV X-ray radiation. A motorized attenuator wheel with aluminum attenuators of different thicknesses was employed in order to be able to cover the full range of the reflectivity signal of 8 orders of magnitude with a scintillation detector (Crysmatec). The characteristic Co KR and Kβ X-ray fluorescence signals under grazing incidence were recorded simultaneously with the reflectivity (i.e. as a function of the glancing incidence angle) for selected samples grown onto SiO2. An energy dispersive germanium detector (Princeton-Gamma Tech) with a resolution of 417 eV and a standard (36) Parratt, L. G. Phys. ReV. 1954, 95, 359. (37) Kiessig, H. Ann. Phys. 1931, 5, 769.

Blasini et al. multichannel data acquisition system (Ortec, Trump-PCI) were used. The detector was placed at 90° with respect to the incident beam, as well as at a grazing exit angle from the sample’s surface at a distance of a few centimeters from the sample so that the solid-angle of collection was maximized. In this geometry, elastic scattering from bulk defects and inelastic Compton scattering are strongly suppressed due to the grazing exit geometry and the polarization of the X-ray beam, respectively. Furthermore, by the choice of the incident angles around the critical angle (Rc) of the sample, the X-ray penetration depth could be controlled to less than 200 Å, which made it possible to achieve monolayer sensitivity. Moreover, at the critical angle (Rc) the fluorescence signal can be enhanced by up to a factor of 4.38 The characteristic angular response of the fluorescence yield makes it possible to distinguish between surface and bulk contributions. The Co fluorescence yield was normalized by dividing the integrated intensity from the Co KR line by the measured XRR intensity at the critical angle (Rc) for different samples. The beam energy used during these experiments was 9.8 keV. Cyclic voltammetry (CV) was employed to characterize the electrochemical properties of the films. These experiments were carried out in a 0.1 M NaClO4 aqueous N2 degassed solution, using a standard three compartment electrochemical cell, with a Ag/AgCl electrode as the reference electrode and a large area Pt wire coil as the counter electrode. Cyclic voltammetry was carried out on a BAS CV-27 potentiostat/galvanostat (Bioanalytical Systems) and recorded on a pentium class computer through a DAQ card (National Instruments). The current density was obtained by dividing the measured current by the Pt electroactive area. Surface coverages were calculated by integrating the charge under the cobalt-based oxidation wave at a sweep rate of 100 mV/s. The Co concentration (mol/L) was determined using the surface coverage data obtained from the electrochemical measurements and the thickness of the films obtained from the XRR measurements. All potentials are referenced to a Ag/AgCl electrode without regard for the liquid junction potential.

Results and Discussion As mentioned earlier, the interfacial coordination reaction of dend-n-tpy (n ) 8, 16, Figure 1) and Co2+ generates a monolayer in which two different dendrimer molecules are linked through the formation of a [Co(tpy)2]2+ complex. The repetitive unit in these films is (tpy-dend-(tpy′)n-2-tpy-M2+)x, where tpy denotes a terpyridine group that bridges via coordination to a Co2+ center two dendrimer molecules and tpy′ denotes a terpyridine group not coordinated to a Co2+ center.30,31 XRF spectra under grazing incidence were recorded simultaneously with XRR to establish the presence of cobalt in the films. Typical reflectivity profiles and fluorescence yield curves for the case of a film prepared through different growth cycles (l ) 1, 5, and 7) prepared with dend-8-tpy are presented in Figure 2A (top and bottom panels). The Co fluorescence yield data show a dependence of the intensity as the wavevector transfer (qz) is varied (by changing the angle of incidence (R)) across silicon’s critical angle (Rc), with a maximum in the fluorescence yield at the critical angle (arrow, Figure 2A, qz/qc ) 1). (Note that the angle of incidence (R) is related to qz by the expressions qz ) 4π/λ sin(R) and qc ) 4π/λ sin(Rc).) Such a behavior is characteristic of a fluorescence source (cobalt ions in this case) located at the surface38 (within the penetration depth of the incident beam). Figure 2A (bottom panel and inset), also shows an increase of the cobalt fluorescent yield as a function of the number of growth cycles (l), indicating that the amount of Co2+ in the films is increasing as more layers are deposited. The signal intensity for the diffuse elastic scattering also showed a variation with qz (not shown). In this case the source of the elastic signal is due to defect scattering from the oxide layer and/or the bulk substrate. Figure 2B shows a (38) de Boer, D. K. G. Phys. ReV. B 1991, 44, 498.

Self-Assembly of Ordered Supramolecular Assemblies

Langmuir, Vol. 22, No. 5, 2006 2085

Figure 3. (A) Cyclic voltammograms (CV) for multilayers derived from dend-8-tpy/Co2+, (B) Co2+ surface coverage, and (C) Co2+ concentration (mol/L) as a function of the number of growth cycles (l). The scan rate was 100 mV/s. (The [Co2+] was calculated using the data in Figures 3B and 5B.)

Figure 2. (A) X-ray reflectivity (XRR) profile for different layers (l ) 1, 5, and 7) of dend-8-typ/Co2+ (top panel) and angular dependence for the cobalt X-ray fluorescence yield (bottom panel) recorded near the critical angle (Rc). The inset shows the Co fluorescent yield as a function of the number of growth cycles (l). The solid line (A, top panel) is the fit obtained with Parratt’s formalism. (B) X-ray fluorescence spectra at the critical angle (Rc) for five layers (l ) 5).

fluorescence spectrum close to the critical angle (Rc) in which the characteristic KR and Kβ emission lines of cobalt, as well as the diffuse elastic peak are clearly evident. To quantify the amount of cobalt present in the films, we measured the cyclic voltammetric (CV) response as a function of the number of growth cycles (l). The profiles for films prepared with dend-8-tpy are presented in Figure 3A. The CV profiles exhibit a redox couple with a formal potential value of +0.12 V, which we ascribe to the Co2+/3+ process of the [Co(tpy)2] moieties.31 It can be noticed that as the number of growth cycles (l) increases, the CVs show a diffusional component, indicative of thick layer formation. As more layers are deposited, the electron-transfer kinetics is dominated by the counterion diffusion into the film (perchlorate anions in this case), thus causing the diffusional behavior in the CVs. Figure 3B shows that there is a linear increase in the surface coverage of Co2+ as more growth cycles (l) were carried out, confirming the formation of multiple layers. We were able to estimate the Co2+ concentration ([Co2+], mol/L) in these films by combining the surface coverage data obtained from the electrochemical measurements and the thickness of the films obtained from the XRR measurements (Figure 5B, discussed in the following paragraphs). The variation of [Co2+] with the number of growth cycles (l) is presented in Figure 3C. The average [Co2+] in the films prepared by up to 11 growth cycles (l e 11) is 0.33 mol/L, and it decays slowly

Figure 4. X-ray reflectivity profiles (XRR) for multilayers derived from (A) dend-16-tpy/Co2+ and (B) dend-8-tpy/Co2+ as a function of number of growth cycles (l). The solid lines are the fits obtained using Parratt’s formalism; traces are offset for better viewing.

as more layers are deposited. This behavior will be discussed in more detail below (vide infra). X-ray reflectivity (XRR) profiles were measured for multilayers prepared by successive growth cycles (l) from two dendrimer generations, dend-8-tpy and dend-16-tpy, and representative data are shown in Figure 4A,B, respectively (traces were offset for better viewing). The bare substrate (Figure 4A,B with l ) 0)

2086 Langmuir, Vol. 22, No. 5, 2006

Figure 5. Film thickness and rms roughness (inset) as a function of the number of growth cycles (l) for multilayers prepared with (A) dend-16-tpy/Co2+ and (B) dend-8-tpy/Co2+.

gave rise to a smooth intensity distribution which can be readily simulated by the Fresnel reflectivity of a Si crystal interface with an oxide thickness of about 1000 Å and a surface rms roughness σ ) 3 Å. In this case, Kiessig fringes37 were not resolved, partly because of the very low contrast between silicon and silicon oxide and partly because of the relatively low angular resolution used, to be able to measure the extremely weak signal from the organic layers. After the deposition of multiple dendrimer layers, the XRR profiles revealed intensity oscillations (i.e. Kiessig fringes), indicating the formation of well-defined layers. Moreover, it can be observed that the frequency of the oscillations increased with the number of growth cycles for both dendrimer generations (Figure 4), indicative of an increase in film thickness. As the number of growth cycles increased for dend-8-tpy, the Kiessig fringes were damped faster than for dend-16-tpy (compare Figure 4A,B). We noted an absence of multilayer Bragg reflections for the thicker films which is likely related to a very homogeneous and smooth electron density distribution within the layers and a low electron density contrast between successive layers. Our XRR data can be well simulated using Parratt’s36 formalism. From such an analysis, we were able to determine the thickness, the rms roughness, and the normal electron density profiles of the deposited layers, as shown in Figures 5 and 6. The film thickness data (Figure 5) are well described by a linear dependence with an increase of 10-11 Å/growth cycle (l) indicative of a layer-by layer (LbL) growth. The first layers obtained with dend-16-tpy (Figure 5A) and dend-8-tpy (Figure 5B) both had film thicknesses of 15 and 20 Å, respectively, which are somewhat thicker than for the successive layers. This may be attributed to the different chemical environment of the first dendrimer layer which is dominated by adsorbate-substrate interactions between the dendrimer molecules and the oxide layer on the silicon surface, whereas in successive layers dendrimer-

Blasini et al.

Figure 6. Lateral electron density profiles as a function of number of growth cycles (l) for multilayers derived from (A) dend-16-tpy/ Co2+ and (B) dend-8-tpy/Co2+. Note the first layer has a significantly higher electron density than subsequent layers.

dendrimer interactions are expected to be dominant. This finding is also consistent with the nonzero intercept of the linear plots (Figure 5). Multilayers prepared with both dendrimer generations exhibited a similar film thickness increase/growth cycle (l), with increases of 10.8 ( 0.4 and 9.9 ( 0.5 Å for dend-16-tpy and dend-8-tpy, respectively (Figure 5A,B). These results suggest the formation of densely packed layers along the surface normal, in which the dendrimers appear to be in a highly compressed state, which is consistent with the compression of charged PAMAM dendrimers along the surface normal in mono- and multilayers as reported previously.17 This behavior has been attributed to the high interaction strength between surface groups along with shortrange van der Waals forces as well as long-range capillary forces.17 Reported axial ratios for these highly compressed aniso-diametric dendrimers are about 1:3 for intermediate dendrimer generations (generation 4 (G4) with 64 arms), a value that increases to 1:6 for higher generations (generation 6 (G6) with 256 arms).17 Consequently, the thickness increase/layer deposited varied with dendrimer generation.17 The dendrimers presented here have compression ratios of 1:3 for dend-8-tpy and 1:4 for dend-16-tpy (calculated using the molecular modeling data for dend-n-tpy presented in ref 31). Our results indicate that the dendrimers herein may be more compressed than those used in previous investigations,17 since we employed dendrimers with generations 1 (G1) and generation 2 (G2),33,34 i.e., with 8 and 16 pendant groups, respectively, as precursors for the terpyridine derivatives. These observations can be rationalized, at least in part, by considering the coordination reaction between two adjacent dendrimers and cobalt and recalling that the formation of [Co-

Self-Assembly of Ordered Supramolecular Assemblies

(tpy)2]2+ is kinetically facile and with a large formation constant.39 Thus, the dendrimer tpy arms may rearrange, to maximize the number of tpy bearing arms coordinated to Co2+ metal centers, forcing the dendrimer to adopt a flatter conformation. As mentioned earlier, not all the tpy arms will be coordinated to cobalt, due to steric constraints and/or electrostatic repulsions. These constraints will vary with dendrimer generation, as higher generations are more able to rearrange and compress.17 Therefore, higher dendrimer generations should react with cobalt readily, and one should expect an increase in the amount of cobalt coordinated as the dendrimer generation is increased in the series dend-n-tpy (n ) 4, 8, 16, 32). In fact, this has been found to be the case.31 Moreover, we did not observe a significant variation of the thickness increase/deposited layer with dendrimer generation. This suggests that the increase in thickness/layer deposited is largely determined by the close-packing distance between the [Co(tpy)2]2+ moieties, which is ca. 8.9 Å in crystals of [Co(tpy)2(ClO4)2].40 The variation in rms roughness as a function of growth cycles, for the multilayer films prepared with dend-16-tpy and dend8-tpy, is presented in the insets of Figure 5. The bare substrate surface rms roughness is also presented, i.e., l ) 0. Note that there is no significant variation in the rms roughness for the multilayers prepared with dend-16-tpy (Figure 5A), with the value remaining at about 3 Å, as each layer was built. On the contrary, the rms roughness for multilayers prepared with dend8-tpy (Figure 5B) showed a large increase in rms roughness after the fourth layer was deposited (l ) 4). The value increased from 3 to 8 Å as l increased from the first (l ) 1) to the eighth layer (l ) 8). Profiles of the laterally averaged electron density obtained from the reflectivity data are presented in Figure 6. The profiles were normalized to the electron density of SiO2 (FSiO2 ) 0.66 e/Å3)41 for convenience. The laterally averaged electron density for the multilayers prepared with dend-16-tpy (Figure 6A) varies around 0.62 as the number of growth cycles (l) increases. The films prepared with dend-8-tpy showed a significant variation in the lateral electron density (Figure 6B). The electron density for the first layer (Figure 6B, l ) 1) is 0.39, it decreased to 0.37 for the third layer (l ) 3), and then it decreased monotonically as more layers were deposited. It reached a value of 0.31 for the 12th layer. These variations are consistent with the rms roughness data presented in Figure 5, and the decrease in [Co2+] presented in Figure 3C as the number of growth cycles was increased. We conclude that, as more layers of dend-8-tpy/Co2+ are deposited, the overall film quality degrades. It is also interesting to note that the first layers derived from both dendrimer generations (dendn-tpy, n ) 8, 16) are similar in terms of thickness and rms roughness. However, after the deposition of the first layers, the behavior for the two dendrimer generations is quite different. While the roughness for dend-16-tpy remained essentially constant for the first 10 layers, the roughness in the dend-8-tpy film doubled after the fifth deposition cycle. We calculated the electron densities for the films using the diameter for dend-n-tpy (n ) 8, 16)/Fe2+ determined by molecular modeling,31 assuming a dendrimer hexagonal close-packing and a layer thickness based on the close-packing distance between the [Co(tpy)2]2+ moieties in crystals of [Co(tpy)2(ClO4)2].40 In this calculation we also assumed that half of the dendrimer arms are coordinated with cobalt ions. All these assumptions are (39) Holyer, R. H.; Hubbard, C. D.; Kettle, S. F. A.; Wilkins, G. Inorg. Chem. 1966, 5, 622. (40) Figgis, B. N.; Kucharski, E. S.; White, A. H. Aust. J. Chem. 1983, 36, 1537. (41) Obtained from Center for X-ray Optics (CXRO) at Berkeley.

Langmuir, Vol. 22, No. 5, 2006 2087 Table 1. Electron Density Calculations for the Dendrimer Films dendrimer generation

diameter of the dendrimer (Å)a

growth cycles (l)

F/FSiO2 (expt)

F/FSiO2 (calcd)c

dend-16-tpy/Co2+ dend-8-tpy/Co2+ dend-8-tpy/Co2+

70 43 43

g1 1 >1

0.63b 0.39 0.34b

0.25 0.32 0.32

a Sizes obtained from ref 31. b Average values for all the layers. Calculated assuming hexagonal close-packing, a layer thickness based on the [Co(tpy)2](ClO4)2 crystal structure,40 and considering half of the dendrimer arms coordinated to Co2+.

c

reasonable and are based on previous experimental results.30,31 The experimental averaged electron densities/layer were obtained for each dendrimer generation using the values obtained from Figure 6, and they are presented in Table 1. We found that there is a large discrepancy between the calculated and the experimental values especially for films derived from dend-16-tpy/Co2+ and for the first layer of dend-8-tpy/Co2+, whereas the values obtained for successive layers of dend-8-tpy/Co2+ are in good agreement; within the experimental error. We believe that the main reason for these deviations arises from departures from a hexagonal close-packing. In our analysis, we assumed that the dendrimers are packed in a hexagonal array retaining their disklike shape, with minimal or no interpenetratation of their arms. Takada et al.34 have previously observed that, upon adsorption, [Ru(tpy)2]pendant PAMAM dendrimers generate a significantly compressed monolayer. Other possible explanations for the deviations in the calculated values may be the incorporation of water40 within the solvation sphere, as well as errors in the determination of the experimental values. On the basis of the calculated and experimental values for the electron densities, we believe that the first layer (l ) 1) produced by both dendrimer generations (dend-n-tpy, n ) 8, 16) is significantly compressed, with a high degree of interpenetration. As the dendrimers are flattened, driven by the coordination with cobalt, some of the unreacted arms could get interdigitated with other unreacted arms from neighboring dendrimers. This interpenetration/compression will cause the electron density to be higher than the value expected for an ideal hexagonal closepacked system (see Table 1). Moreover, since both dendrimer generations are subject to different conformational constraints, we can expect them to have different behaviors, as they form multilayers. We observe that all the layers derived from dend16-tpy seem to have this high degree of interpenetration, as evidenced by the small variation in electron density (Figure 6A). In the case of the multilayers prepared with dend-8-tpy, only the first layer (l ) 1) appears to be significantly compressed. The electron density decreased from 0.39 to 0.31 as more layers were deposited (Figure 6B). This decrease of approximately 20% in the electron density is in contrast with an increase in the rms roughness (Figure 5B), suggesting that the overall film quality is degrading as more layers are deposited. However, even after the electron density has decreased approximately 10% (after the deposition of the fourth layer), we detect a significant increase in roughness only after the fifth layer is deposited. Somewhat surprisingly, a linear increase in the thickness is observed up to the 12th layer (Figure 5B). These observations could be rationalized by considering that the electron density for these layers is consistent (within experimental error) with the values predicted by assuming an ideal hexagonal close-packing (see Table 1). These suggest that the structures found in these layers are mostly based on a close-packing of disk-shaped dendrimers, with a small degree of interpenetration of their arms. However, the systematic decrease in electron density and increase in roughness after the deposition of the eighth layer (l ) 8) would

2088 Langmuir, Vol. 22, No. 5, 2006

Blasini et al.

7A) increased as the deposition time increased. This behavior may be due to conformational changes of the dendrimers within the layers as more dendrimers are deposited and more cobalt and counterions are incorporated into the films. Chen et al.42 observed a similar behavior for the layer-by-layer (LbL) growth of nitrocontaining diazoresin (NDR) and poly(sodium-p-styrene sulfonate) (PSS).

Conclusions

Figure 7. Adsorption kinetics for the preparation of four layers (l ) 4) of dend-8-tpy/Co2+: (A) film thickness variation (main panel) and rms roughness (inset); (B) lateral electron density as a function of deposition time per dendrimer layer. The cobalt reaction time was 120 s for each cycle.

suggest that the packing density decreases as more layers are deposited (Figures 5B and 6B). To explore the formation of multilayers derived from dend8-tpy in more detail, we decided to study the effect of the variation in the dendrimer deposition time on the film parameters. We varied the dendrimer deposition time throughout the preparation of four layers, while keeping the cobalt reaction time constant. The kinetic data obtained are presented in Figure 7. We observe that the film thickness reaches a plateau for dendrimer deposition times above 540 s (Figure 7A). Moreover the lateral electron density values (Figure 7B) increased with deposition time as well. This self-limiting behavior is consistent with a layer-bylayer (LbL) growth, in which about 540 s is necessary for the completion of a dendrimer/Co2+ single layer, after which further deposition ceases. In contrast, the incorporation of Co2+ in the second step is practically instantaneous, presumably due to the high permeability of the ions into the film and largely due to the fast kinetics and favorable thermodynamics of coordination of Co2+ with the terpyridine ligands.39 The data in Figure 7A could be fitted using a kinetic control model under Langmuirian adsorption conditions.34 The deviations from the fit and the data are attributed to departures from Langmuirian behavior, most likely due to repulsive interactions of the cobalt centers within the layers. We determined an average rate constant of 3.7 × 102 M-1 s-1 for the adsorption of the dendrimers in each layer. This value is 1 order of magnitude smaller than those determined for the formation of monolayers from homologous generations of [Ru(tpy)2]-pendant dendrimers.34 Moreover, we observed that the rms roughness of the films (Figure

We have prepared ultrathin multilayers based on the terpyridyl (tpy)-pendant PAMAM dendrimers (dend-n-tpy; n ) 8, 16) by the repetitive deposition and self-assembly from CH2Cl2, followed by an interfacial coordination reaction with cobalt ions (Co2+) from aqueous solution. The resulting film thickness showed a linear dependence with the number of growth cycles (l), with a thickness increase of 9.9 ( 0.5 and 10.8 ( 0.4 Å per layer for both dend-8-tpy and dend-16-tpy, respectively, indicative of a layer-by-layer (LbL) growth behavior of single dendrimer/Co2+ layers. XRF and electrochemical results showed that the amount of Co2+ increased linearly, as more layers were deposited, and that the Co2+ concentration (mol/L) in dend-8-tpy/Co2+ films decayed slowly as the number of growth cycles increased. The average cobalt concentration [Co2+] in the films prepared by up to 11 growth cycles (l e 11) was found to be 0.33 mol/L. Kinetic data showed that the maximum amount of dendrimer that could be deposited in a cycle reached an asymptotic limit, i.e., the growth of a single dendrimer layer is a self-limiting process. The kinetics of adsorption for dend-8-tpy appears to be activation controlled with a rate constant of 3.7 × 102 M-1 s-1. This growth mode allows for the preparation of films with a precise number of deposited layers, as opposed to other growth processes also termed layer-by-layer (LbL), in which two or more layers grow at the same time and which do not show any self-limiting behavior in solution.25-29 It is noteworthy that the growth method presented here is closely analogous to the atomic layer deposition (ALD)43 method used for the preparation of ultrathin gate oxides. ALD is also based on a binary chemical reaction which exhibits self-limiting behavior. Comparison of the experimentally determined layer thicknesses and electron densities with the expected dendrimer size (from molecular modeling) and calculated electron densities (based on dimensions determined by molecular modeling and close to those observed by STM31), respectively, suggests that, upon adsorption and coordination, the dendrimers are significantly compressed. Both dendrimer generations showed similar packing properties for their first layer but behaved differently afterward. Subsequent layers of the films derived from dend-8-tpy displayed a lower packing density, with values closer to that expected from hexagonal packing. Moreover, the layer thickness increase per growth cycle (l) of 10-11 Å for dend-8-tpy and dend-16-tpy was very similar despite the much larger size of dend-16. We attribute this behavior to the difference in the ability of each dendrimer generation to rearrange and compress, to meet the geometric constraints imposed by the coordination reaction with cobalt. Ongoing and future work is focused on varying the composition of the layers and to address the question of lateral ordering in multilayers. Acknowledgment. This work was supported by the Cornell Center for Materials Research (CCMR) and NSF Grants DMR(42) Chen, J.; Huang, L.; Ying, L.; Luo, G.; Zhao, X.; Cao, W. Langmuir 1999, 15, 7208. (43) Lim, B. S.; Rahtu, A.; Gordon, R. G. Nat. Mater. 2003, 2, 746.

Self-Assembly of Ordered Supramolecular Assemblies

9970838 (G-line facility) and DMR 0114094 (G2 monochromator). D.R.B. acknowledges support by a NSF Graduate Fellowship. D.R.B. thanks Dr. Kazutake Takada for useful discussions. We express our thanks to the CHESS staff and, in

Langmuir, Vol. 22, No. 5, 2006 2089

particular, to the G-line staff and students. CHESS is a national user facilitysupportedbyNSF/NIH/NIGMSawardDMR-0225180. LA052558W