Introduction
Experiments on individual molecules using scanning probe microscopy have opened up avenues for the exploration of chemical, physical and biological phenomena on an individual non-statistical basis. By being able to induce molecular interactions through exertion of mechanical forces and to monitor them in real time with subnanometre resolution, single-molecule force spectroscopy has provided unprecedented insights into the structure and function of many molecular systems7, 8, 9, 10, 11, 12. AFM tips functionalized with end-grafted molecules13 offer the prospect of delivering individual molecules in a single-molecule force spectroscopy experiment if the bond anchoring the molecule to the tip is weaker than the one to be established with the surface (Fig. 1a). In order to achieve this goal, we developed a strategy based on (1) the grafting of macromolecules bearing reactive groups onto the AFM tip and (2) their selective transfer, via a chemical reaction, to a substrate where complementary moieties are present.
Figure 1: Molecule by molecule delivery process.
a, Reactive polymer molecules, attached at one end to an AFM tip, are brought into contact with a substrate to which they can become linked by a chemical reaction. When the tip is pulled away from the surface, the resulting mechanical force causes the weakest bond — the one between the tip and polymer — to break. b, During the tip–sample contact, a chemical reaction occurs between the activated esters of a poly-N-succinimidyl acrylate chain grafted to the tip and the amino groups of the substrate to form an amide bond, which covalently links the chain to the substrate.
Full size image (21 KB)Gold-coated AFM tips were modified by electrografting poly-N-succinimidyl acrylate (PNSA), according to ref. 14. This electro-initiated polymerization is a convenient way to fabricate polymer brushes with a moderate grafting density and results in the direct chemisorption of the polymer onto the tip surface. Surfaces with amino functions were prepared by grafting aminopropyltrimethoxysilane to silicon substrates. The activated esters of the polymer can easily react, at room temperature, with the amino-derivatives. In an N,N-dimethylformamide (a good solvent for PNSA) solution containing 4-dimethylaminopyridine (DMAP, a catalyst), the functionalized AFM tip was slowly brought into contact with the surface. The chemical reaction between the PNSA activated esters and the amino groups of the substrate forms amide bonds and covalently links polymer chains to the substrate (Fig. 1b). Upon retraction of the tip, single chains are stretched until a bond breaks. The Au(tip)–C(polymer) bond is the weakest link in the system and the most likely candidate for breaking. Upon cleavage, the polymer chain remains covalently attached to the substrate. The deposited chains are reactive and can be easily modified, subsequently, by a wide range of nucleophilic compounds.
The experiments strongly rely on the design and accurate modification of the tips. The use of a swollen brush with a low grafting density (less than one chain per 100 nm2; the compression profile shows that we are in the isolated mushrooms regime, see Methods section) for force spectroscopy experiments prevents unspecific physisorption of a bulk three-dimensional structure onto the substrate thanks to steric repulsion15, 16 and the strong adsorption of flat-lying chains (that adsorb through several attachment points) which leads to the appearance of multiple superimposed peaks in the force–distance curves13. The presence of a good solvent prevents interactions within the polymer. Statistically, the bridging interactions should occur with only one isolated single polymer chain. The strategy makes the detection of the rupture of a single tip–polymer bond easy.
The resulting force–distance curves present simple features (Fig. 2a, b; see Supplementary Information for a histogram of the rupture lengths). During the approach, the force continuously and monotonically increases, as is characteristic of the compression of the end-grafted polymer15. In most of the retraction curves, no interaction occurs. In about 15% of the retraction curves, peaks appear as a result of the bridging interaction of the polymer with the NH2-functionalized surface. In most cases, single peaks are observed. The appearance of multiple peaks is much rarer. The single rupture peaks are characteristic of the tail-like bridging of end-tethered polymer chains17. To exclude the possibility that the higher forces are due to the extension of a bundle of chains, the profiles were fitted using entropic elasticity models, which predict the relationship between the extension of a linear polymer and the restoring force generated18, 19. They were then normalized and superimposed (a classical criterion for single molecule detection9; Fig. 2c). We used an extended wormlike chain model (WLC+) to take into account enthalpic elastic contributions19 (see Supplementary Information for details).
Figure 2: Force curves obtained between PNSA-modified gold-coated AFM tips and an NH2-modified silicon substrate in N,N-dimethylformamide.
a, Force–extension curve (grey: approach profile; black: retraction profile) corresponding to the desorption of physisorbed segments. b, Retraction profile corresponding to a covalent bond breaking. The blue curve is a fit according to the WLC+ model. c, Master curve of experimental data obtained by normalizing the force curves with their corresponding contour length. The agreement between the normalized curves (each symbol corresponds to a different experiment) and the WLC+ model (solid black line) is strong evidence that single molecules are stretched9.
Full size image (35 KB)Figure 3a shows a histogram of the rupture forces, constructed from data obtained with 18 tips. The forces fall within two distinct ranges, suggestive of the existence of two types of events. The smaller forces lie between 0.1 and 0.4 nN. The peaks appear at integer multiples of an elementary force quantum of 96 
12 pN, which most probably corresponds to the physisorption of one segmental unit onto the NH2 surface. The rupture forces centred on 187 19, 291 13 and 374 16 pN thus correspond to the desorption of two, three and four units, respectively (see Supplementary Information for a detailed histogram with gaussian fits). The higher forces, observed in 
20% of the cases, lie in a range of 0.8–1.4 nN, with an average value of 1.1 0.15 nN, a force range characteristic of a gold–organic covalent bond rupture20, 21. When no catalyst is added in the medium, the higher forces are observed in only 7% of cases.
Figure 3: Histogram of the rupture forces between PNSA-modified AFM tips and an NH2-modified silicon substrate in N,N-dimethylformamide.
Rupture forces for a, gold-coated tips and b, silicon nitride tips. The forces fall within two distinct ranges, suggestive of the existence of two types of events (lower forces: desorption from the amino substrate; higher forces: rupture of the tip–polymer bond).
Full size image (23 KB)As a control experiment, we measured the interaction of a PNSA brush grafted directly onto a silicon nitride tip with the same surface. The histogram is shown in Fig. 3b. The range of forces corresponding to physisorption is similar to that observed for PNSA-modified gold-coated tips (confirming our hypothesis that these forces correspond to physisorption onto the NH2 surface), but the second peak is shifted to higher values, suggesting that it is not the same bond that breaks. The average value is 2.4 0.3 nN, a force range typical of the rupture of a Si–C covalent bond20, 21. Notice that, in this case, the rupture may occur either at the tip–polymer bond or within the silane layer (no delivery accomplished).
Atomistic calculations show that the binding energy of the Au–C bond linking the chain to the tip is about 12.5 kcal mol-1, in very good agreement with the expected energy for a Au–C interaction22. The calculations were run along the lines of previous successful applications to gold–organic systems23, 24, 25, which showed that the calculated binding energy is reproduced within 1 kcal mol-1 of the experimental values. The rupture should thus selectively occur between Au and C, which is the weakest link in the system (see Supplementary Information for details).
The selective breaking of the bond between the tip and the polymer immobilizes individual polymer chains on the substrate. The diameter of the PNSA backbone is smaller than the roughness of the substrate. The visualization of single organic molecules having a diameter less than 1 nm is not a routine task, even on atomically flat surfaces like mica. On silicon, they are invisible and a decoration step is required to detect them26, 27. To this end, the substrate was dipped into a solution of a branched polyethyleneimine (PEI) in N,N-dimethylformamide, enabling the amino groups of the PEI to react with remaining activated esters of the deposited PNSA chains. The resulting structure is a PNSA backbone with PEI side chains, which greatly increase the diameter of the original chains. AFM topographic images obtained in areas where individual molecules were delivered are shown in Fig. 4.
Figure 4: AFM topography images obtained in air after the delivery.
a, Image obtained in the area where four PNSA chains were deposited one at a time. The original chains were decorated by a branched PEI, rinsed with DMF, and imaged before the residual film of DMF was completely evaporated. The decorated molecules appear in an extended shape. The maximum vertical height is 4 nm. b, c, areas where three and six chains were deposited, decorated by PEI, rinsed with DMF, and dipped in acetonitrile (a bad solvent) to make them appear as collapsed globules. Maximum vertical height is 7 nm.
Full size image (31 KB)The present experiments not only show that we can deliver single molecules onto the substrate and covalently immobilize them, but also that these molecules can be subsequently modified by further chemical reactions. The free succinimide groups on the deposited chains easily react at room temperature with nucleophilic compounds and can, for example, serve as anchoring groups for the one-step immobilization of biological compounds, as reported previously28. By playing with the nature of the (co)polymer grafted to the tip and the nature of the substrate, high-resolution patterns of chemical functionalities on a range of addressable surfaces could be generated. In view of the increasing demand for nano-engineering operations in 'bottom-up' nanotechnology, this method provides a tool that operates at the ultimate limits of fabrication of organic surfaces, the single molecule.
Methods
Force spectroscopy experiments were carried out with a PicoSPM equipped with a fluid cell (Molecular Imaging). We used gold-coated silicon nitride cantilevers with a spring constant between 0.03 and 0.05 N m-1. The spring constant of each cantilever was calibrated following the procedure described in ref. 29. N-succinimidyl acrylate was electrografted following the procedure reported previously14. The average degree of polymerization is 600, which corresponds to a chain length of about 150 nm. The density of the brush is low, and the approach profile15, 16 estimates less than one chain per 100 nm2. The onset of repulsive forces, at about 30 nm, is close to the Flory radius of the polymer coil (RF
lN3/5=33 nm), which means that we are in the isolated mushrooms regime.
Aminopropyltrimethoxysilane monolayers were prepared and characterized following the procedure described in ref. 30. Before starting the experiments, the PNSA modified tips were dipped in dry N,N-dimethylformamide (DMF) for 48 h to let the polymer swell. The tip and substrate were then placed in the liquid cell containing DMF and a catalytic amount of DMAP (DMAP was used to accelerate the amidification and thus increase the probability of forming an amide bond during the tip–sample contact). The approach–retraction curves were recorded at 22 °C, at a pulling rate of 100 nm s-1. The maximum force applied to the sample was fixed at 0.5 nN. Post-decoration of the single molecules grafted to the substrate was carried out by dipping the substrate in a solution of branched PEI (Mn = 25,000 g mol-1) in DMF (0.1 wt%) for 12 h. As PEI is highly soluble in DMF, it does not tend to adsorb onto the surface in the absence of a chemical reaction. The substrate was then copiously rinsed and sonicated in DMF and either imaged before the residual thin film of DMF was completely evaporated, or first dipped in acetonitrile (a bad solvent for PNSA and PEI) to collapse the decorated chains before imaging. The structures were observed only in the area where the delivery was made.
Author contributions
A.S.D. conceived and designed the experiments, S.C. performed the experiments, and S.G. and C.J. prepared the functionalized AFM tips and contributed to the experimental design. S.R. and F.Z. performed the calculations. A.S.D., S.C., S.R. and F.Z. analysed the data and discussed the results. A.S.D. and F.Z. co-wrote the paper.
-phage DNA