Auguste
Tetenoire
a,
Anna
Omelchuk
a,
Volodymyr
Malytskyi
b,
Ivan
Jabin
b,
Victor
Lepeintre
bc,
Gilles
Bruylants
c,
Yun
Luo
d,
Arnaud
Fihey
a,
Mikaël
Kepenekian
*a and
Corinne
Lagrost
*a
aUniv Rennes, ENSCR, CNRS, ISCR (Institut des Sciences Chimiques de Rennes) – UMR 6226, F-35000 Rennes, France. E-mail: mikael.kepenekian@univ-rennes.fr; corinne.lagrost@univ-rennes.fr
bLaboratoire de Chimie Organique, Université libre de Bruxelles (ULB), Avenue F. D. Roosevelt 50, CP160/06, B-1050 Brussels, Belgium
cEngineering of Molecular NanoSystems, Ecole Polytechnique de Bruxelles, Université libre de Bruxelles (ULB), Avenue F. D. Roosevelt 50, CP165/64, B-1050 Brussels, Belgium
dUniversité Paris Cité, CNRS, Laboratoire de Chimie et de Biochimie Pharmacologiques et Toxicologiques, F-75006 Paris, France
First published on 16th August 2024
The interface robustness and spatial arrangement of functional molecules on metallic nanomaterials play a key part in the potential applications of functional nano-objects. The design of mechanically stable and electronically coupled attachments with the underlying metal is essential to bring specific desirable properties to the resulting hybrid materials. In this context, rigid multipodal platforms constitute a unique opportunity for the controllable grafting of functionality. Herein, we provide for the first time an in-depth description of the interface between gold nanorods and a chemically-grafted multipodal platform based on diazonium salts. Thanks to Raman and X-ray photoelectron spectroscopies and theoretical modeling, we deliver insights on the structural and electronic properties of the hybrid material. More importantly, it allows for the accurate assignment of Raman bands. The combination of experimental and theoretical results establishes the formation of four carbon–gold anchors for the calix[4]arene macrocycle leading to the exceptional stability of the functionalized nano-objects. Our results lay the foundations for the future design of robust and versatile platforms.
The reduction of an aryldiazonium function into the corresponding aryl radical enables a robust grafting of the aryl moiety onto a wide variety of materials (insulating, semi-conductor or metal), either as flat massive materials or nanomaterials (Scheme 1).11,12 This strategy however often leads to the formation a multilayered coating. Interestingly, the controlled formation of monolayers has been demonstrated through a unique strategy developed by some of us, based on the reductive grafting of calix[4]arene tetra-diazonium salts.13,14 Calix[4]arenes are macrocyclic molecules made up of four methylene-bridged aromatic subunits. In our approach, the calix[4]arene diazonium salts are constrained in a cone conformation, orienting the four diazonium moieties on the large rim of the calixarene. This rigid structure is exploited to design interesting molecular platforms for diazonium surface grafting, yielding well-packed and dense monolayers coating the surfaces. Thus, the reductive grafting of calix[4]arene tetra-diazonium salts onto gold, silver and platinum nanoparticles15–17 has led to the preparation of electrocatalysts with excellent performances in oxygen reduction reaction (ORR) and methanol oxidation reaction (MOR) processes, notably because of their very good stability.17,18 Similarly, this strategy enabled the development of various ultrastable calix[4]arene-coated silver and gold nanomaterials that were used for sensing applications in the biomedical field.19,20
Yet, little is known about the nature of the interface between the grafted calix[4]arene and the nano-objects. In particular, the multiple nature of the anchoring mode of the calix[4]arene to the metallic surfaces remains unclear. This aspect is of importance because multipodal ligands are particularly interesting building blocks since they allow for a better control of molecular interactions.21–24 Notably, the perpendicular orientation of terminal functional groups with respect to the surface is more likely to be guided by the binding mode of a rigid scaffold. In addition, a multisite anchor should lead to significantly stronger adsorption, hence larger stability, and maximize electronic coupling between the ligands and the metal.23,25,26 Rigid multipods are beneficial for controlling their lateral arrangement on the surface along with the spatial arrangement of the terminal exposed functional groups because of the sterically hindered footprint of such ligands. Especially for the calix[4]arene molecules herein, the spatial arrangement of terminal groups is dictated by the small rim geometry. Nevertheless, such architectures are generally built with thiols as anchoring moieties whereas multipodal rigid platforms are scarcely based on the diazonium anchoring function despite its strong interest. A key point is whether a four anchoring mode of the calix[4]arene is possible, especially when considering that the diazonium chemistry is a “one-shot” grafting procedure, rendering reorganization of the interface impossible, unlike the case with thiol-gold self-assembled monolayers.27
To unravel the nature of the interface between the gold surface and the diazonium-derived aryl layer, Raman spectroscopy has been used for gold nano-objects, greatly benefiting from the coupling with quantum chemistry calculations for the assignment of bands.28,29 Unfortunately, these theoretical treatments often require models that represent the metal surface as small size molecular clusters to reduce the computational effort, which can be detrimental to the modeling of spectroscopic signatures.30–33 However, the use of the computationally efficient density functional tight-binding (DFTB) approach can circumvent this limitation.34–36 In particular, DFTB has been shown to accurately describe both ground and excited states of gold–organics hybrid materials37–40 with an adapted set of parameters.41 Interestingly, this framework extends beyond the typical size limitations of standard DFT models for Raman spectra simulation,42 facilitating the comparison with experimental data.
In this work, we present the original synthesis of gold nanorods functionalized with the monopodal calix[4]arene mono-diazonium X4N2+ and the multipodal calix[4]arene tetra-diazonium X4(N2+)4 bearing four carboxyl groups at their small rim (Fig. 1a). Despite, their strong capping to the nanorods, we successfully exchange the CTAB surfactants by the calix[4]arene molecules. Using a combination of UV-visible absorption spectroscopy, high-resolution transmission electron microscopy (HRTEM), X-ray photoelectron spectroscopy (XPS), Raman spectroscopy, and a theoretical description within the DFTB framework, we shed light on the gold–carbon hybrid interface. After characterizing the nature of the bonds, we thoroughly detail the Raman spectra of the mono- and tetrapodal calix[4]arenes grafted on the gold nanorods surface. The combination of experimental and computed spectra allows us to rationalize the shifts of the Au–C bond peaks as a function of the anchoring configuration. Our findings demonstrate the utility of Raman spectroscopy in identifying the number of Au–C bonds per molecule and establish that multipodal grafting is achieved through diazonium chemistry owing to the rigid macrocyclic structure of calix[4]arene. The four-anchoring mode is found to be the most favorable binding mode for the tetrapodal calix[4]arene, demonstrating that such a calixarene coating yields a very stable and well-structured hybrid scaffold, useful as a building block for electrocatalysis or sensing.
The C 1s signal of CTAB coated nanorods can be decomposed into two components that agree with the CTAB structure along with contributions (C–O, CO) due to carbon contamination, which is usually observed. As expected, the C 1s signals for the calix[4]arene capped nanorods are similar and can be decomposed into three main components, at 285.4 ± 0.1 eV for carbon atoms involved in C–C, C–H bonds, at 286.1 ± 0.1 eV for carbon atoms in C–O–C bonds and at 289.9 ± 0.2 eV for carbons atoms involved in the carboxylate termini. A small contribution (2–6%) at 288.9 eV can be assigned to CO functions, probably due to carbonaceous contamination. The atomic ratio for C–O–C and O–CO components are equal to 1.75 and 1.8 for the monopodal and multipodal calix[4]arenes, respectively. These values are close to the expected theoretical value equal to 2 for a calix[4]arene structure. All the XPS observations suggest that the CTAB have been successfully removed from the surface, and efficiently replaced by the calix[4]arenes. However, no further information concerning the exact nature of the gold/ligand interface with the calix[4]arene moieties could be drawn from our XPS measurements, notably because the electronegativity of C and Au are very close (2.55 and 2.54, respectively according to the Pauling scale). Thus we engage in a computational investigation coupled with Raman spectroscopic analysis to reach a fine description of the two types of calix[4]arenes grafting.
Various cases must be considered for the interaction between calix[4]arene and Au surfaces (Fig. S10–S13 ESI†). For X4N2+, after reduction of the diazonium group, the molecule presents one phenyl radical that can form an Au–C bond. However, for X4(N2+)4, four phenyl radical anchors are in principle available. In that case, we consider the creation of 1 to 4 bonds with the gold surface, with hydrogen atoms saturating the unbound radicals in the 1, 2 and 3 bonds configurations. Finally, to cover the entire range of molecule/surface interactions, we also consider the case of a fully hydrogen-saturated calix[4]arene that cannot be involved in any Au–C bond.
Fig. 3 presents the variation of the adsorption energy with respect to the number of anchors for all possible conformers and regioisomers on the four considered facets. In the fully hydrogen-saturated calix[4]arene configuration, the molecule is only simply physisorbed on the gold surface with an adsorption energy varying between −0.74 and −1.24 eV, and a shortest Au–C distance comprised between 3.16 and 3.57 Å, depending on the facet (Tables S2 and S3, ESI†). When one phenyl radical is formed, hence yielding 1 possible anchor, the Au–C bonding distance is shortened to ca. 2.2 Å, and the adsorption energy vary among −2.58 to −3.65 eV. As the number of anchors is increased, the adsorption energy becomes more and more exothermic, reaching values for the 4-anchored molecule between −8.29 eV (Au(110)) and −6.81 eV (Au(111)). It is interesting to note that the adsorption energy per anchor is not constant but increases from an average of −3.13 eV for 1 anchor, to −1.89 eV for 4 anchors (Fig. S15 and Table S3, ESI†). This variation is due to the geometry distortions undergone by the gold surfaces and the molecule in order to create 4 Au–C bonds. However, the formation of 4 bonds from the calix[4]arene tetra-diazonium cations remains energetically very favourable.
We explore further the nature of the Au–C bond by considering the density of states (DOS) projected on the orbitals of the C and Au atoms forming a bond in the case of the monopodal calix[4]arene (Fig. S16, ESI†). We observe overlapping peaks of C(2s,p), Au(6s,p) and Au(5d) orbitals near below and above the Fermi level. The partial charge densities computed at those peaks reveal the σ and σ* orbitals resulting from their hybridization establishing the existence of covalent Au–C bonds.
In the case of gold nano-objects, the appearance of a shallow peak in the 400–550 cm−1 region has also been used as a marker of successful covalent grafting.27,30–32,49–52 The Raman spectra of nanorods functionalized with the mono- and tetrapodal calix[4]arenes both show a peak at 396 cm−1 (Fig. 4). This band is usually assigned to the stretching mode of the Au–C bond (νAu–C). However, this assignment comes from measurements performed on gold complexes with non-aromatic ligands (CO, CN, Me),30,31,48,49,53,54 and not on coated surfaces or nano-objects, which could greatly affect the position of the mode. In the following, we will once more depict the gold nanorod as a series of surfaces. Our approach will be illustrated on the Au(111) surface, a facet primarily found on the tips of the nanorods, where the plasmon-induced electric-field enhancement is the greatest considering the morphology of the nanorod and the excitation wavelength used.55
Within the SCC-DFTB model, we simulate the Raman spectra of three typical complexes used to identify νAu–C: Au(CN)2−, MeAuPMe3, and Me3AuPMe3 (Fig. 5a, S19 and S20, ESI†).53,54 For each molecule, we describe the Au–C stretching mode with excellent accuracy (2% error) between 400 and 550 cm−1 (Fig. 5a, Tables 1 and S4, ESI†), confirming the experimental assignment. However, when we no longer consider a gold complex, but a molecule anchored on a gold surface (e.g. methyl), the Au–C stretching mode is shifted towards lower frequencies, νAu–C = 378 cm−1 (Fig. 5a and Table 1), in good agreement with the experimental assignment for an alkyl chain grafted on gold (νAu–C = 387 cm−1).32 If we consider a system closer to the calix[4]arene molecule, as a phenyl anchored on Au(111), the shift becomes even greater with νAu–C = 218 cm−1 (Fig. 5a and Table 1). Finally, considering the mono- and tetra-anchored calix[4]arene molecules on gold, we find groups of stretching modes at 230–285 cm−1 (Fig. 5a and 6). These contributions fit well with the peak identified at 282 cm−1 in both experimental spectra, that we can assign to the Au–C stretching mode νAu–C. The peak appearing at 396 cm−1, initially assigned to the Au–C stretching mode, corresponds, in our simulations, to the frustrated rotations of the phenyl rings found at 360–410 cm−1. Hence, the bands in the 200–600 cm−1 area of the Raman spectra confirm the covalent grafting of both calix[4]arenes after reduction of their corresponding mono- and tetra-diazonium cations. However, there is no significant difference between the two systems, preventing us from determining the number of anchors that we can obtain from the tetra-diazonium molecule.
Vibration | MeAuPMe3 | Me/Au(111) | Ph/Au(111) |
---|---|---|---|
Au–C stretch (νAu–C) | 546 (535 (ref. 53)) | 378 | 218 |
Ph stretch (νPh) | 1458, 1524 |
A closer look at the simulated Raman spectrum of the phenyl chemisorbed on Au(111) reveals another interesting region around 1500 cm−1 in which intense Raman peaks can be observed, corresponding to stretching modes of the phenyl ring, νPh, indirectly affected by the Au–C bond (Fig. 5b and Table 1). The same modes are obtained in the simulated spectra of the grafted calix[4]arene (Fig. 5b and 6). Indeed, the mono-anchored molecule presents a peak at 1604 cm−1, while the same mode appears at 1576 cm−1 for the tetra-anchored calix[4]arene because of the stiffening of the structure imposed by the multiple anchors. Corresponding bands are observed on the experimental Raman spectra of the grafted mono- and tetrapodal calix[4]arenes (Fig. 4) after reduction of the X4N2+ and X4(N2+)4 diazonium cations at 1590 and 1563 cm−1, respectively. The 27 cm−1 downward shift when going from mono- to tetrapodal macrocycles matches the computed trend and is a signature of the multiple anchoring of the tetrapodal calix[4]arene on Au(111).
Interestingly, when performing the Raman spectroscopy using a lower excitation wavelength, i.e. 638 nm, there is no clear shift of the large peak found at 1580 cm−1 (Fig. S18, ESI†). This can be interpreted based on the simulation of the spectra coming from the different facets. Indeed, under the given conditions, it is expected that the electric-field enhancement is no longer localized at the tips of the nanorod, but is distributed over its body which is composed of larger fractions of (100), (110), and (520) facets.46,47,55 When considering the simulated spectra of those facets (Fig. S24, ESI†), one can see that the phenyl stretching mode appears for both mono- and tetrapodal functionalized surfaces at the same frequency. Hence, a 638 nm excitation leads to overlapping signals, while a 785 nm laser excitation leads to specific features for the gold nanorods coated from the grafting of X4N2+ and X4(N2+)4, respectively.
Let us note that more peaks can be found in the 550–1400 cm−1 region. Unfortunately, these modes cannot be clearly assigned and consist in an undistinguishable blend of CH2 wagging, O–C bond deformation, and phenyl H out-of-plane distortions. As a consequence, we cannot rationalize the evolution of the signals in that frequency window going from the mono- to the tetrapodal system.
The combination of the adsorption energies and Raman spectra establish the formation of 4 Au–C bonds per calix[4]arene when the gold nanorod is coated from the grafting of X4(N2+)4. Indeed, adsorption energies show the significant thermodynamical incentive for a calix[4]arene to form as many Au–C bonds as possible, while the experimentally observed Raman spectrum on the (111) facet correspond to the multipodal grafting of X4(N2+)4. Although, we do not obtain a specific Raman signal for multiple anchoring modes on the other facets, the even stronger adsorption energies found on the (100), (110), and (520) facets, strongly suggests that a multiple anchoring is also found on these planes.
Footnote |
† Electronic supplementary information (ESI) available: Synthesis details; Experimental setups; Computational details; Additional UV-visible spectroscopy, XPS, Raman spectroscopy, and Computational results. See DOI: https://doi.org/10.1039/d4sc02355b |
This journal is © The Royal Society of Chemistry 2024 |