Formation and Stabilization of Silver Nanoparticles with Cucurbit [n

Feb 15, 2011 - ... and not threaded along a metal-linked axle).(20, 21) Also, to the best of our knowledge, there has been no previous report of intim...
0 downloads 0 Views 2MB Size
Download PDF
ARTICLE pubs.acs.org/Langmuir

Formation and Stabilization of Silver Nanoparticles with Cucurbit[n]urils (n = 5-8) and Cucurbituril-Based Pseudorotaxanes in Aqueous Medium Xiaoyong Lu and Eric Masson* Department of Chemistry and Biochemistry, Ohio University, Athens, Ohio 45701, United States

bS Supporting Information ABSTRACT: A series of silver/cucurbituril nanoparticles and aggregates have been prepared upon reduction of silver nitrate with sodium borohydride in the presence of different cucurbit[n]uril (CB[n]) macrocycles; CB[7] and CB[8] allow the formation of stable solutions of monocrystalline, narrowly dispersed nanoparticles (5.3 and 3.7 nm, respectively), while CB[5] and CB[6] induce rapid aggregation and sedimentation. The rigidity of CB[5] and CB[6], and their possible lack of suitable arrangement at the silver surface, may explain the poor stabilization of these silver assemblies, while the more flexible CB[7] and CB[8] may undergo some minor distortions and better adapt to the requirements of the metallic surface; computer modeling supports the existence of interactions between the silver nanoparticles and the oxygen atoms of the CB[n] carbonylated rim. The optimal silver nitrate/CB[7] ratio for the formation of stable nanoparticles is 1:1-2:1, while large excesses of silver or CB[7] trigger aggregation. Masking the portals of CB[7] by adding a bulky, positively charged guest into its cavity has a surprisingly minor effect on the stability of the silver/CB[7] assemblies; in such a case, the CB[7] rim is still expected to interact with the NPs, albeit via a fraction of its carbonyl oxygen atoms.

’ INTRODUCTION Cucurbit[n]urils (CB[n]s) are increasingly popular macrocycles, which display outstanding recognition properties in aqueous medium toward a wide range of guests, in terms of both affinity and selectivity. As their Latin-derived name indicates, CB[n]s bear a pumpkin shape, with n glycoluril motifs linked by methylene bridges forming two hydrophilic carbonylated portals and a hydrophobic cavity (see Figure 1).1 Although mixtures of CB[n]s had been prepared more than 100 years ago unbeknownst to their investigator,2 one had to wait until the beginning of the present millenium for the X-ray characterization of four of its congeners (CB[5], CB[7], CB[8], and CB[10]).3,4 Since then, CB[n] chemistry has been blossoming at an impressive rate, with an average of approximately 85 publications and patents per year since 2005. CB[n]s are ideal hosts for positively charged amphiphilic guests, such as alkane-1,ω-diammonium cations and adamantylpyridinium,1 whose positive charges interact with the carbonylated rim of the macrocycle through ion-dipole stabilization and possibly hydrogen bonding, while the hydrophobic moiety sits inside the CB[n] cavity.1 Common binding affinities5 are 105 to 1012 M-1 and can even reach or surpass the benchmark biotin-streptavidin interaction (3.0  1015 M-1 vs ∼1015 M-1!).6 On several occasions, the carbonyl rim of CB[n]s has been shown to interact with metallic cations7,8 such as alkali and alkali earth metals, various transition metals, and some lanthanides. We r 2011 American Chemical Society

recently showed that it could also interact with organometallic species (such as transient π-alkynylsilver intermediates), and stabilize them.9 In that case, this effect led to the catalysis of the silver-promoted desilylation of trimethylsilylalkynyl derivatives. We thus considered the possibility that CB[n] might even be able to interact with silver(0) atoms or silver nanoparticles (Ag NPs). This supposition has been confirmed in a concurrent study by Geckeler et al., who, very recently, stabilized Ag NPs with CB[7] and studied their cytotoxicity toward two cancer cell lines.10 The preparation of new nanoscaled Ag assemblies is particularly attractive, since among other applications, Ag NPs are known to exhibit antibacterial, anti-inflammatory, antiviral, and wound healing properties.11,12 Ag NPs and nanowires are commonly obtained by reduction of a Ag(I) salt (sodium borohydride being the most frequent reducing agent),13,14 and can be stabilized with ligands such as polyacrylates (1c),15 polyacrylonitrile (1d),16 polyacrylamide (1e),17 and poly(vinyl pyrrolidone) (PVP; 1f)18 in order to inhibit aggregation. In this study, we report the unique and contrasted effects of CB[5], CB[6], CB[7], and CB[8], in the absence and presence of encapsulated guests, on the formation, structure, and stabilization of Ag NPs and nanoaggregates in aqueous medium. The Received: November 27, 2010 Revised: January 11, 2011 Published: February 15, 2011 3051

dx.doi.org/10.1021/la104729j | Langmuir 2011, 27, 3051–3058

Langmuir

Figure 1. Structure and preparation of CB[7], from glycoluril (1b) and formaldehyde. 1,3-Dimethylimidazolidin-2-one (1a) is considered a mimic of the key repeating unit in CB[7] (see yellow oval). Structures of various stabilizers for silver nanoparticles: polyacrylate (1c), polyacrylonitrile (1d), polyacrylamide (1e), and poly(vinyl pyrrolidinone) (PVP; 1f).

effect of the four most readily available CB[n] macrocycles on the structure and stability of metallic NPs has been systematically studied on only one occasion in the case of gold NPs,19 and examples of “intimate” ternary metal/guests ⊂ CB[n] nanostructures are scarce (i.e., with CB[n]s in direct contact with the metal surface, and not threaded along a metal-linked axle).20,21 Also, to the best of our knowledge, there has been no previous report of intimate ternary metal/guests ⊂ CB[n] NPs. This lack of systematic study and of a possible link between the macrocycle size and the properties of the metallic nanoaggregates prompted our investigation. Preparation of Ag NPs and Ag Aggregates. Ag/CB[n] NPs. CB[n] (n = 5, 6, 7, 8; 2.0 μmol) was added to an aqueous solution of silver nitrate (40 μL, 50 mM), which was subsequently diluted with water (1.9 mL). The mixture was sonicated for 2 min, until CB[5] or CB[7] dissolved, and CB[6] or CB[8] formed a slightly turbid mixture. A freshly prepared solution of sodium borohydride in water (80 μL, 50 mM) was then added rapidly under vigorous stirring, and the resulting mixture was stirred for an additional 5 min at 25 °C (total silver nitrate and CB[n] concentrations, 1.0 mM; final sodium concentration, 2.0 mM). Ag/1,3-dimethyl-2-imidazolidinone and Ag/glycoluril systems (Ag/1a and Ag/1b, respectively) were prepared similarly, in the presence of 1,3-dimethyl-2-imidazolidinone (1a; 28 μmol) and glycoluril (1b; 14 μmol) instead of CB[n]s. Ag/Guest ⊂ CB[n] NPs. CB[7] (0.20 mL, 10 mM in water), guest 2a (0.20 mL, 10 mM), silver nitrate (40 μL, 50 mM), and water (1.5 mL) were combined and sonicated until a clear solution was obtained (total concentrations of silver, CB[7], and guest 2a: 1.0 mM). A freshly prepared solution of sodium borohydride (80 μL, 50 mM) was then added rapidly under vigorous stirring, and the resulting mixture was stirred for an additional 5 min at 25 °C. Other Ag assemblies were prepared similarly in the presence of guests 2b, 2c, 3a, 3b, 4a, and 4b.

’ RESULTS AND DISCUSSION Effect of the CB[n] Dimensions on the Formation and Stability of Ag/CB[n] NPs and Aggregates. In a first series of

ARTICLE

experiments, Ag/CB[n] assemblies were prepared by reducing a 1:1 mixture of silver nitrate and CB[n] (1.0 mM) with sodium borohydride (2.0 equiv) at 25 °C. In the case of CB[5] and CB[6], gray and black fine suspensions were obtained, respectively, upon rapid addition of the reducing agent, and full sedimentation took place within 2 h. When the reactions were repeated in the presence of CB[7] and CB[8], the initial grayish slurries turned into clear, dark orange solutions after only 30 s upon stirring. The solutions were stable for at least 6 months at room temperature and were insensitive to ambient light and moderate heating (sedimentation did take place at 70 °C after 2 h). The reduction was also performed in the presence of 1,3dimethylimidazolidin-2-one (1a), which we consider a mimetic of the repeating urea unit of CB[n]s, as well as with glycoluril (1b; a building block of the CB[n] macrocycles). Fourteen equivalents of imidazolidinone 1a and 7.0 equiv of glycoluril (1b) were used, in order to mirror CB[7] conditions. While the stabilities of Ag/1b and Ag/CB[6] aggregates were comparable, imidazolidinone 1a was found to better stabilize the NPs (the obtained yellow turbid mixture did not undergo rapid sedimentation). Transmission electron micrographs were recorded, and the size distributions of the assemblies were assessed. Aggregation was found to affect Ag assemblies according to the sequence Ag/CB[7] ≈ Ag/CB[8] < Ag/1a < Ag/CB[5] < Ag/ 1b ≈ Ag/CB[6] (see Figure 2, micrographs a-g). Water-stable Ag/CB[7] (Figure 2e and g) and Ag/CB[8] NPs (Figure 2f) were the smallest on average, and were narrowly dispersed (5.3 ( 1.9 nm and 3.7 ( 1.7 nm, respectively), while all other systems were significantly larger (>10 nm) with a wide size distribution. Amazingly, the treatment of silver nitrate with 2.0-12 equiv sodium hydroxide in the presence of CB[7] by Geckeler et al. afforded almost identical NP sizes, despite a very different reducing method.10 The high-resolution TEM micrograph of Ag/CB[7] NPs indicated that most assemblies were single crystals with 2.33, 2.02, and 1.42 Å d-spacings, corresponding to the (111), (200), and (220) planes of face-centered cubic silver, respectively (Figure 2h and i). Small Ag NPs are expected to absorb visible light, with an absorption maximum at approximately 415 nm, which corresponds to the surface plasmon resonance (SPR) of the metallic NPs.13-18 Aggregation has been shown to cause red shifts, lower absorbances, and significant signal broadening.18 Ag/1a systems displayed an absorption maximum at 394 nm (the most energetic absorption among all samples), thereby indicating the presence of small NPs (in the 3-5 nm range); however, a shoulder between 420 and 550 nm and a high value for the full width at half-maximum (fwhm = 124 nm vs 70 nm in the case of Ag/ CB[8] assemblies; see Figure 3a) indicated the formation of polydisperse nanoaggregates, in accordance with TEM experiments (see micrograph 2a). The extremely broad and weaker absorption of Ag/1b, Ag/CB[5], and Ag/CB[6] samples indicated significant aggregation (see Figure 3, spectra b-d); in the case of Ag/1b, a second absorbance maximum was detected at 566 nm (spectrum b), and is very probably caused by the longitudinal plasmon resonance of 1D Ag aggregates, as depicted in micrograph 2b.22 Samples prepared in the presence of CB[7] and CB[8] displayed absorption maxima at 423 and 411 nm with the narrowest fwhm values (97 and 70 nm, respectively; see Figure 3, spectra e and f), corresponding to 5.3 and 3.7 nm average diameters and monodisperse size distributions, as depicted in TEM micrographs (see Figure 2e and f). We also note that (1) at least 2.0 equiv sodium borohydride were required 3052

dx.doi.org/10.1021/la104729j |Langmuir 2011, 27, 3051–3058

Langmuir

Figure 2. Transmission electron micrographs of (a) Ag/1,3-dimethylimidazolidin-2-one (Ag/1a), (b) Ag/glycoluril (Ag/1b), (c) Ag/CB[5], (d) Ag/CB[6], (e) Ag/CB[7], and (f) Ag/CB[8] assemblies, prepared upon reduction of silver nitrate (1.0 mM) with sodium borohydride (2.0 equiv) in the presence of the above-mentioned stabilizers (1.0 mM in the case of CB[5]-CB[8], 7.0 mM glycoluril 1b or 14 mM imidazolidinone 1a) at 25 °C. (g) Scanning and (h) high-resolution transmission electron micrographs of Ag/CB[7] NPs. (i) Diffraction pattern of Ag/CB[7] NPs, corresponding to face-centered cubic Ag. We note that some much larger prisms (5-10 μm), which most probably correspond to traces of undissolved CB[6] and CB[8], were detected in Ag/CB[6] and Ag/CB[8] preparations.

Figure 3. UV-vis spectra and photographs of (a) Ag/1a, (b) Ag/1b, (c) Ag/CB[5], (d) Ag/CB[6], (e) Ag/CB[7], and (f) Ag/CB[8] samples prepared as described above. Solutions and suspensions were diluted 10 times in water immediately before UV-vis analysis; suspensions were stirred during irradiation and data acquisition.

ARTICLE

in order to maximize absorbances of Ag NPs solutions, (2) the addition of sodium chloride to our Ag NP solutions did not result in the formation of insoluble silver chloride, thereby indicating the absence of Ag(I) cations, (3) the initial concentration of sodium borohydride was found to have no significant effect on the size distribution and stability of the Ag NPs (this observation is in stark contrast with experiments performed in the absence of stabilizer),23,24 and (4) much lower concentrations of silver nitrate (0.25 mM) are necessary in order to obtain stable Ag NP solutions without the addition of stabilizers.23,24 The effect of CB[n]s on the stability of Ag NPs and aggregates can be rationalized if one addresses the mechanisms of their formation and growth. Three mechanisms are usually considered:23 (1) a nucleation and growth process, where metal atoms add to small nanocrystalline seeds (the classical LaMer mechanism); (2) an aggregative mechanism, where the small nanocrystals coalesce to form larger assemblies; and (3) an Ostwald ripening process, where small particles disintegrate, and the debris are captured by the larger growing particles. In a set of carefully described studies,23,24 Zukoski et al. showed that the most probable mechanism for Ag NP formation from silver nitrate, sodium borohydride, and no additional stabilizer (initial silver nitrate concentration 0.25 mM) is an aggregation process. Reduction of Ag(I) occurs during the first few seconds of the reaction, and is possibly even diffusioncontrolled. It affords 1-3 nm seeds, which subsequently aggregate to form larger, 8-20 nm NPs during the course of 1 h. Aggregation kinetics can be evaluated using the classical DerjaguinLandau-Verwey-Overbeek model,25a which states that the interaction between two particles is the sum of electrostatic repulsions controlled by the NP surface potentials, and of all interatomic van der Waals attractions controlled by the Hamaker coefficient for the bulk metal.25b Aggregation rates of Ag nanoseeds were found to be extremely sensitive to the NP surface potential, which is controlled by the presence of borohydride anions in the Stern layer of the NPs: high concentrations of borohydride anion at the beginning of the reaction lead to a positive surface potential, which favors the formation of small NPs and inhibits aggregation, while the degradation of sodium borohydride in aqueous solution, affording hydrogen and sodium borate,26 lowers the NP surface potential and triggers NP growth. Stabilizing agents, such as polyacrylamide (1e) and PVP (1f), interact with the NPs or the smaller nanocrystalline seeds, thereby increasing the NP surface potential and limiting collisions and aggregation (i.e., the dipoles of the stabilizing agent induce image dipoles at the surface of the NPs).18,23,24 The interactions between Ag NPs and amides, such as PVP (1f), have been evaluated on several occasions, using infrared (IR) and X-ray photoelectron spectroscopy (XPS); the interaction between Ag NPs and CB[7] has also been recently assessed.10 However, conclusions drawn from these studies remain controversial. In the case of Ag/1f adducts, C-N stretching vibrations, and in one case CdO stretches, were found to undergo blue shifts in the presence of Ag, thereby suggesting that both nitrogen and oxygen atoms of PVP (1f) may interact with the metal surface.27,28 XPS experiments are even less conclusive, since in one study, binding energy increases were detected in N 1s electrons,28 and on another occasion in O 1s electrons.18 While Geckeler et al. indicated that a 7 cm-1 shift at higher wavenumbers was observed for the CB[7] CdO stretching vibration in the presence of Ag NPs, we could not detect any significant difference between the spectra of CB[7] and Ag/CB[7] NPs, most probably because both Ag/CB[7] 3053

dx.doi.org/10.1021/la104729j |Langmuir 2011, 27, 3051–3058

Langmuir

Figure 4. Interaction between a Ag14 cluster and 1,3-dimethylimidazolidin-2-one (1a), as determined using the B3LYP density functional method and a combined 6-31G(d)∪LANL2DZ basis set.

preparation methods and chemical environments are different. Similarly, no conclusion could be drawn from XPS measurements we performed on (1) CB[7], (2) 1:2 mixtures of CB[7] and sodium (from sodium borohydride), and (3) Ag/CB[7] NPs, since deviations of the O 1s and N 1s signals were insignificant ((0.1 eV). Unable to reach a clear conclusion from IR and XPS measurements, we optimized in silico the interaction between imidazolidinone 1a and a Ag14 cluster using the Gaussian 09 software and the B3LYP density functional method with the 6-31G(d) basis set for all atoms except Ag, for which the Los Alamos LANL2DZ effective core potential was used. This type of combined basis has been used on several occasions with organometallic species29 and small metallic clusters interacting with organic compounds.30 We first prearranged a face-centered cubic cell of 14 Ag atoms, and optimized the assembly without constraints; although the quasiface-centered cubic structure we obtained for Ag14 may31 or may not32 be the most stable packing depending on the applied computation method, it is at least a local energy minimum according to our model, with a lattice constant in reasonable agreement with the experimental value (4.01 vs 4.07 Å, respectively).33 The Ag14/1a assembly was then optimized while keeping the Ag14 cluster frozen. Regardless of the initial location of ligand 1a, the same optimized structure was obtained repeatedly and clearly showed a favorable interaction between the Ag surface and the carbonyl oxygen of 1,3-dimethylimidazolidin-2one (1a) with a Ag-O distance of 2.59 Å (see Figure 4); the carbonyl-Ag angle C-O-Ag was 134° and the two N-methyl substituents were approximately parallel to the edges of the cell; no Ag-nitrogen interaction could be detected. Also, in accordance with previous experimental results,18,23,24 Bader charge analysis34 indicated that the carbonyl dipole induced a strengthening of the partial positive charge at the vicinal silver atom (þ0.11 |e|). Although in silico calculations should be considered with utmost care, we can reasonably propose that CB[n]s interact with the Ag surface via their carbonylated portal, and not via the nitrogen atoms at their periphery, especially since the entropically favorable macrocyclic effect (i.e., the greater affinity of a guest toward a cyclic host bearing n identical binding sites, compared to its affinity toward the n separate fragments)35 even enhances the Ag-carbonyl interactions. A similar hypothesis was proposed by Geckeler et al.10 and Scherman et al.36 in the case of Ag/CB[7] and gold/CB[5] NPs, respectively. However, this conclusion should be nuanced: in the case of the rigid CB[5] and CB[6] macrocycles, the 5 (or 6) oxygen atoms may not be ideally positioned for Ag binding (an enthalpic impediment) and do not contribute to the stabilization of Ag NPs to the same extent as 5 (or 6) equiv imidazolidinone 1a. Also, the proximity of the 5 (or 6)

ARTICLE

partially negative oxygen atoms at the CB[5] or CB[6] rim may enhance partially positive mirror charges on several adjacent silver atoms at the metal surface, as suggested by the Bader charge analysis (see above); such a possible ligand-induced repulsive Ag-Ag interaction may in fact prevent the ligand from properly interacting with the NP. When the larger and more flexible CB[7] and CB[8] are used, their carbonyl oxygens can better adapt to the structural and electronic requirements of the Ag surface, and the entropic gain attributed to the macrocyclic effect overcompensates the enthalpic penalty for a slight deformation of the CB[n] unit (Ag NPs prepared in the presence of CB[7] and CB[8] are much better stabilized than those in the presence of 7, 8, or even 14 equiv of imidazolidinone 1a). We do not consider that the affinity of CB[5], CB[7], and CB[8] for the sodium cation, as well as a possible competition between the cation and Ag NPs for interactions with the CB[n] portals, has a significant impact on the lack of stability of Ag/ CB[5] aggregates: the binding affinity of sodium toward CB[5]37 (which does not stabilize Ag NPs) is lower than its affinity for CB[7] and CB[8], which we determined by 1H NMR titration (71 M-1 vs 770 and 420 M-1, respectively). Since the affinity of sodium toward CB[6] is significantly higher (3100 M-1),38 a similar conclusion for CB[6] is hypothetical. It should be noted that (1) Ag/CB[5] and Ag/CB[6] aggregates, as well as Ag/ CB[7] and Ag/CB[8] NPs, could be obtained in the presence of sodium nitrate (50 mM), similarly to reductions performed in the absence of the salt, and (2) the stability of Ag/CB[n] (n = 5-8) aggregates follows the same trend when reduction is performed with lithium borohydride instead of its sodium analogue, although samples are generally less stable (sedimentation occurs after 2 h in the case of Ag/CB[7]); it is therefore clear that the alkali metals play a role in the overall stability of Ag aggregates in aqueous medium, yet this role does not seem to be connected to their interactions with CB[n]s, and is beyond the scope of this study. Analogously to the silver nitrate/sodium borohydride systems described by Zukoski et al.,23,24 no Ag(I) cation could be detected after borohydride reduction in the presence of CB[n]s, and therefore, CB[n] favorable interactions with Ag(I) did not inhibit reduction; we propose a similar aggregation mechanism for the formation of our Ag/CB[n] assemblies, with the macrocycles competing (advantageously in the case of CB[7] and CB[8]) against the borohydride anion for Ag interactions, and causing contrasted differences in the surface potentials of the Ag NPs. We also note that precipitated Ag/CB[5] and Ag/ CB[6] aggregates contain only 15 mol % CB[n] relative to Ag; the surface coverage of Ag/CB[7] and Ag/CB[8] assemblies in solution cannot be assessed, since it depends on the unknown binding affinity of the NP toward the macrocycle. Effect of the Ag/CB[7] Ratio. We subsequently assessed the effect of the CB[7]/Ag ratio on the structure and stability of Ag/ CB[7] NPs. While reductions carried out in the presence of 0.10 equivalent CB[7] led to the concomitant formation of 14 ( 4.3 nm particles and 1D aggregates (see Figure 5a), which readily sedimentated in the course of 10 h, reactions with 0.50 and 1.0 equiv CB[7] afforded stable orange solutions, with well-defined and narrowly distributed spherical nanoparticles (6.1 ( 1.9 and 5.3 ( 1.9 nm, respectively; see micrographs 5b, 2e and 2f). Larger particles were obtained in the presence of 2.0 and 5.0 equiv CB[7] (13 ( 6.1 and 21 ( 7.4 nm, respectively), and a uniform, 8.0-nm-thick coating of some particles, or surrounding a group of particles, was observed in the latter case (Figure 5d). We suspect that the coating consists of several layers of CB[7]. 3054

dx.doi.org/10.1021/la104729j |Langmuir 2011, 27, 3051–3058

Langmuir

ARTICLE

Figure 7. CB[7] guests: substituted p-xylylenes 2, 1,ω-alkane diammoniums 3, and adamantyl derivatives 4.

Figure 5. Transmission electron micrographs of Ag/CB[7] NPs and aggregates prepared with CB[7]/Ag ratios = (a) 0.10, (b) 0.50, (c) 2.0, and (d) 5.0.

Figure 6. UV-vis spectra and photographs of Ag/CB[7] adducts prepared with CB[7]/Ag ratios = (a) 0.10, (b) 0.50, (c) 1.0, (d) 2.0, and (e) 5.0. Initial concentrations of the reducing agent were 2.0 mM in all cases. Solutions and suspensions were diluted 10 times in water immediately before UV-vis analysis; suspension (a) was stirred during irradiation and data acquisition.

The interpretation of the UV-vis spectra generally mirrors TEM observations (see Figure 6): a second absorption band at 530 nm was observed when the reduction was carried out in the presence of 0.10 equiv CB[7], thereby indicating the presence of 1D aggregates (Figure 6, spectrum a, and micrograph 5a); with similar diameters (6.1 ( 1.9 vs 5.3 ( 1.9 nm), Ag NPs formed in the presence of 0.50 and 1.0 equiv CB[7] displayed surface plasmon resonance maxima at similar wavelengths (418 vs 423 nm); however, the usual trend (i.e., smaller particles absorbing at shorter wavelengths) was not observed in this case. Since these NPs were similarly distributed, signal widths at half-maximum were comparable (93 vs 97 nm). Ag/CB[7] NPs prepared with 2.0 equiv CB[7] showed an absorbance maximum at a wavelength as low as 408 nm; this is plausible if one considers the broad size distribution of the NPs (13 ( 6.1 nm), and as a consequence, the possible presence of very small particles. When 5.0 equiv CB[7] were used, only a very weak SPR band could be observed (spectrum e). As described above, the optimum Ag/CB[7] ratios for stable Ag NPs were found to be 1:1 to 2:1; in those cases, frequent interactions between NPs and CB[7] could efficiently prevent NP growth. Less CB[7] (0.10 equiv, for example) could not hamper aggregation efficiently, and larger particles as well as 1D

structures were obtained; it is unclear at the moment why these nanorod-like assemblies were observed, although CB[7] may interact preferentially with facets along the rods, compared to facets at their ends, thereby allowing longitudinal growth.38 Higher concentrations of CB[7] (>1.0 mM) should stabilize Ag NPs even better and favor very small NPs, yet larger aggregates, often linked to neighbors (see micrographs 5c and 5d) were obtained. The reduced absorbance at 415 nm indicates either aggregation, or the presence of very small NPs (1-3 nm), especially if a UV absorption band is observed at approximately 220 nm.23,24 In our case, we noticed some significant absorbance between 290 and 340 nm, compared to samples with lower CB[7]/Ag ratios; this may indicate the presence of small clusters. Therefore, we suspect that interactions between CB[7] units, possibly mediated by water, force small Ag/CB[7] clusters to agglutinate into larger, polycrystalline assemblies, which can even be surrounded by additional layers of ligand. Effect of Positively Charged Guests Encapsulated in the CB[7] Cavity. We then determined the effect of CB[7] guests 2-4 on the structure and stability of Ag/CB[7] NPs (see Figure 7). Cations 2 and 3 were chosen in order to mask both portals of CB[7] (guests 2b and 2c bearing the bulkiest susbtituents), and adamantyl derivatives 4 were selected in order to shield only one carbonylated rim. All guests were used with trifluoromethanesulfonate as their counteranion, since it does not form insoluble salts with Ag(I) cations (contrary to the chloride anion, for example). Guests 2-4 bind strongly to CB[7] (binding constants Ka between 108 and 1012 M-1).5 In a first set of experiments, a CB[7] guest (1.0 equiv) was combined with a 1:1 mixture of silver nitrate and CB[7] (1.0 equiv each), and the reducing agent was added subsequently (i.e., Ag nanoaggregates were formed in the presence of pseudorotaxanes 2 ⊂ CB[7], 3 ⊂ CB[7], and 4 ⊂ CB[7]); in a second set of experiments, Ag NPs were first formed upon reduction of silver nitrate in the presence of CB[7] (1.0 equiv), and guests 2-4 were added to the resulting solution. Since adding guests to Ag/CB[7] assemblies shields one or both portals of the macrocycle, we expected that they could disrupt the interaction between the NPs and CB[7], and induce aggregation followed by precipitation. We also considered the unlikely possibility of very slow exchanges between Ag NPs and CB[7], which may prevent the organic guest from binding to the macrocycle. The latter hypothesis happened to be invalid indeed, since in all cases, regardless of reagent addition sequences (sodium borohydride added to a solution of silver nitrate, CB[7], and guests 2-4; or guests added to a mixture of silver nitrate, CB[7], and sodium borohydride), 1H NMR experiments 3055

dx.doi.org/10.1021/la104729j |Langmuir 2011, 27, 3051–3058

Langmuir

Figure 8. 1H NMR spectra of (a) guest 2c, (b) pseudorotaxane 2c ⊂ CB[7] in the presence of an equimolar amount of silver nitrate, (c) a solution of Ag/2c ⊂ CB[7] assemblies; all measurements performed in deuterium oxide.

indicated that CB[7] encapsulated all organic guests quantitatively (see Figure 8 for an example): hydrogen atoms located inside the CB[7] cavity underwent a strong upfield shift (1.02 and 0.35 ppm in the case of hydrogens a and b of guest 2c, spectra a and b), while hydrogens at the periphery of the portal were shifted downfield moderately (0.06, 0.08, and 0.36 ppm in the case of hydrogens c-e). Reduction of Ag(I) to the corresponding Ag NPs did not result in any significant alteration of the 1H NMR spectra (see spectrum c, Figure 8). However, regardless of the addition sequence, guests 2-4 did affect the stability, size, and shape of Ag aggregates, compared to Ag/CB[7] NPs, although not to the point of causing rapid aggregation and precipitation: Ag/guest ⊂ CB[7] systems did display the characteristic absorption pattern at approximately 415 nm, and could remain stable in solution during an average of 20 h before some sedimentation was observed (solutions of Ag/CB[7] NPs are stable even after 6 months). UV-vis absorptions were weaker (54-99% relative to Ag/CB[7] NPs; see Figure 9), and bands were broader (full widths at halfmaximum ranged from 102 to 148 nm, vs 97 nm in the absence of guest, spectrum h). TEM measurements showed three typical structural modifications compared to Ag/CB[7] NPs: (1) small NPs (