Reversible and Irreversible Binding of Nanoparticles to Polymeric Surfaces

Reversible and irreversible binding of CdSe-nanoparticles and nanorods to polymeric surfaces via a strong, multiple hydrogen bond (= Hamilton-receptor/barbituric acid) is described. Based on ROMP-copolymers, the supramolecular interaction on a thin polymer film is controlled by living polymerization methods, attaching the Hamilton-receptor in various architectures, and concentrations. Strong binding is observed with CdSe-nanoparticles and CdSe-nanorods, whose surfaces are equipped with matching barbituric acid-moieties. Addition of polar solvents, able to break the hydrogen bonds leads to the detachment of the nanoparticles from the polymeric film. Irreversible binding is observed if an azide/alkine-“click”-reaction is conducted after supramolecular recognition of the nanoparticles on the polymeric surface. Thus reversible or irreversible attachment of the nanosized objects can be achieved.


Introduction
Assembly of nanosized objects [1] is an important step towards the generation of functional materials, that is, for applications in microelectronics [2], medicine [3], and sensor-applications [4].As outlined in a series of recent review-papers [5,6] basically four principles exist to direct a selective and specific assembly of nanoparticles onto surfaces or interfaces: (a) interfacial effects in liquid/liquid-systems [7] (b) crystallization of colloids (c) shape-dependent ordering, and (d) template-directed self assembly processes mediated by supramolecular bonds [8].In all described cases an important point concerns the possibility to control the reversibility of the binding/assembly processes, since often physical properties of the final aggregate are strongly dependent on the underlying ordering processes.Both can be achieved by carefully adjusting the size and adsorption energy between the interfacial contact and the nanoparticles surface.Examples of such processes have been described via size-selective control of nanoparticle-binding at liquid/liquid-interfaces as reported by Duan et al. [9], and Böker et al. [10], or the location of nanoparticles at polymeric interfaces [11][12][13][14] due to entropic or interfacial effects.In the first case [9] smaller nanoparticles display lower interaction energies in comparison to large-sized nanoparticles, thus allowing to selectively remove the smaller particles from the interface after addition of the large ones; in the latter case entropic constraints direct smaller, noninteracting nanoparticles to the polymer interface of block copolymers by entropic chain effects [11,13], whereas selective location of larger nanoparticles can be achieved by appropriate surface engineering [12].
The supramolecular recognition concept overcomes the selection between larger-or smaller-sized nanoparticles by introducing directed, specific, and stronger interactions.Thus various hydrogen bonding systems [15] (i.e., Rotellos thymine/triazine-; [16,17] our Hamiltonreceptor/barbituric acid system [18,19,29], or Reinhoudts triazine/barbiturate-system [21]), and Reinhoudts molecular printboards (relying on cyclodextrin-inclusion complexes [22]) have been used to direct the assembly of nanoparticles to surfaces or interfaces.The big advantage of such supramolecular systems over the purely interfacially controlled is their specificity of binding, which does not select between larger or smaller nanoparticles, but just selects the binding event due to the supramolecular interaction between the nanoparticles surface and the underlying scaffold.However, a disadvantage concerns the irreversibility of the binding event, since a relatively strong interaction energy of E > 10 k B T is needed to overcome the translational energy of the smaller nanoparticles [7,19] often leading to irreversible binding processes.Thus, the ideal combination would be an initially strong interaction, which allows reversibility and subsequent irreversible binding.
In the present paper, we describe a different concept of controlling the reversible/irreversible attachment of nanoparticles and nanorods to polymeric surfaces via directed supramolecular interactions and subsequent covalent fixation, as described in Figure 1: thus binding of nanoparticles or nanorods is mediated by distinct multiple hydrogen bonds, attached to a polymeric surface of defined architectures (either a statistical homopolymer or a block copolymer) (case (a) or case (b)).In these cases, the recognition is entirely mediated via the specificity of the multiple hydrogen bond (Hamiltonfreceptor/barbiturate), acting between the surface of barbiturate-modified-CdSenanoparticles or nanorods, and a polymeric scaffold, presenting the Hamilton-receptor in a specific architecture.Thus, the binding is reversible, since the strength of the hydrogen bond can be weakened from its initial association constant of ∼10 5 M −1 ion nonpolar solvents to ∼15 M −1 in polar solvents such as methanol [18].
However, (approach (c)), if a possibility exists to mediate a chemical reaction between the attached nanoparticles and the surface after recognition, then an irreversible attachment would result.The main question in this case concerns the conditions of this chemical attachment, which must take place still preserving the supramolecular interaction between the nanoparticle and the surface.To this purpose, we have chosen the azide/alkin-"click"-reaction, which is a Cu(I)mediated 1,3-dipolar-cycloaddition reaction [23,24], able to mediate reactions under extremely moderate reaction conditions (i.e., without increase in temperature, solvent-and substrate insensitive) [25][26][27].The following paper focuses on these issues, demonstrating the reversible/irreversible attachment of CdSe-nanoparticles and CdSe-nanorods onto specifically engineered polymeric surfaces via supramolecular concepts and subsequent covalent fixation.
GPC spectra were recorded on a Hewlett Packard Series 1100 Chemstation system using Waters Styragel linear columns HR0.5, HR3, and HR4 in THF at 40 • C. Polystyrene  standards (M p = 580-48840 g/mol) were used for conventional external calibration using a Waters RI 2410 refractive index detector.
IR measurements were performed on a Biroad FTS-135 FT-IR spectrometer.AFM Measurement.For the AFM investigations a NanoScope III multimode SPM from Digital Instruments, Veeco Metrology Group, Santa Barbara, CA, has been used.Measurements were performed in tapping mode under ambient air using single-crystal silicon cantilevers (Arrow NC cantilevers, NanoWorld, Switzerland, spring constant 42 N/m, resonance frequency ∼285 kHz).Scanning was accomplished with an E-scanner (maximum scan range 10 μm × 10 μm) operated at a scanning rate of 2 and 4 Hz, respectively, and an image resolution of 512 × 512 pixels.Data evaluation was performed with the NanoScope software version 6.13r1 (Digital Instruments, Veeco).The roughness data used in this paper are root-mean-square (rms) values, derived as standard deviations of all height values in an image.Phase images (i.e., the phase lag between the piezo-oscillation driving the AFM cantilever and its response) have been recorded simultaneously with the topographic images through the second data channel of the instrument.The used Si-wafers were provided by Wacker Siltronic.
DLS measurements were performed in toluene solutions of the NPs after dilution by ∼1/50 with pure solvent on an ALV/CGS-3 compact goniometer using the ALV-5000/E correlator software.
TEM measurements were achieved on a Phillips EM300 transmission mode 100 kV.The nanorod solution (under high dilution of ∼1/50) was applied on standard TEM-Grids (300 mesh copper carbon only (50 ct)) purchased from Electron Microscopy Sciences.

General Procedure for the Synthesis of Azide Functionalized
Copolymers (Polymer 2).Sodium azide was added to a solution of statistical copolymers (polymer 1s5, polymer 1b/2, polymer 1b/1, or polymer 1b/5) in dry DMF (10 mL) and the reaction was stirred for 6-8 hours at room temperature.The solvent was completely removed under reduced pressure, the residue was resuspended in CHCl 3 , filtrated through a syringe filter, concentrated and evaporated to dryness, yielding azide functionalized copolymers as light brown solids.All azide functionalized polymers have to be stored at −20

Synthesis of CdSe Nanoparticles NP5
The resulting pyridine-covered nanoparticles were then precipitated in n-hexane and centrifuged.The precipitation procedure was repeated twice in order to remove all of the free TOPO-ligands.Subsequently, DOPO-N 3 (5) [31] (44 mg, 0.12 mmol) and hex-5-ynoic acid 3-(dioctylphosphinoyl)propyl ester (DOPO-acetylene) ( 4), (53 mg, 0.12 mmol) [31] were added to a solution of the NPs in freshly distilled anhydrous THF and the stirring was continued for 20 hours at 50 • C. The THF was evaporated under reduced pressure and the nanoparticles (NP5) were precipitated twice in 20 mL anhydrous methanol to remove unattached ligands.

Preparation of Smooth Polymerfilms Derived from
Poly(Oxy-Norbornenes).Silicon wafers were cut in squares of 8 × 8 mm using a glass cutter, cleaned from dust and silicon splinters with a stream of nitrogen and incubated into a dust-free ethanol solution (filtrated through a syringe filter) for 5-10 minutes.From this solution they were quickly taken out and immediately dried with a strong stream of nitrogen in order to prevent solvent margins obtained from evaporation.Subsequent UV-irradiation for 15 minutes completely oxidized the surface and removed the last organic impurities on the wafers, which was again followed by cleaning with nitrogen.All wafers were stored in small and clean glass/plastic jars until they were used for coating.
Polymer solutions (0.5 wt%) were prepared directly before dip coating and stored in dust-free jar.Dip coating was performed on a static wafer which is dip coated by a downward moving (3 mm/min) solution jar.In order to prepare polymerfilms via dip coating techniques the wafers were taken from the storage jar and cleaned by a strong stream of nitrogen.Then they were statically affixed with a tweezer using a metal tripod to place them correctly into the dip coating device.Then a jar containing a completely dustfree polymer solution (syringe filter) was attached onto the lifting arm of the dip coater and the dip coating sequence was started.Once the dip coating sequence is ended, the silicon wafers were cleaned by a stream of nitrogen, placed into a clean glass or plastic jar and stored under exclusion of light at 0 • C.
The prepared polymer films can be annealed subsequently.Therefore the wafers were taken from their storage jars, cleaned with a strong stream of nitrogen, and put into a 250 mL round-bottomed flask by carefully sliding them along the glass wall towards the bottom of the vessel.The flask was put in a preheated oil bath (80 • C), evacuated and the wafers were protected from light.After complete annealing (∼48 hours), the wafers were allowed to cool and the vacuum was removed.The annealed polymer films were stored in a clean glass or plastic jar under exclusion of light at 0 • C.
In order to assemble nanoparticles or "clicking" different functionalities onto the polymer surface 0.3% solution of nanoparticle in toluene was prepared.The annealed polymerfilms on Si-wafer were put into nanoparticle solution and stored there for 16 hours.After complete incubation the silicon wafers were thoroughly cleaned with toluene and dried in a strong stream of nitrogen in order to prevent solvent margins obtained from evaporation and stored in a clean and dust-free jar under exclusion of light at 0 • C.

Results and Discussion
3.1.Polymers, Nanoparticles, Surfaces.Basic concept of the presented work is the use of the Hamiltonreceptor/barbiturate-interaction with an association constant of K assn.∼ 10 5 M −1 , (as determined in nonpolar solvents, i.e, chloroform, dichloromethane, hexane), affixed to the sidechains of a poly(norbornene) which serves as scaffold for the presentation of the supramolecular interaction after film formation.As outlined in Figure 1, three basic principles of CdSe-NP/nanorod binding towards these polymer surfaces are investigated: (a) reversible binding towards statistically distributed Hamilton-receptors within the polymer, (b) reversible binding towards blockydistributed Hamilton-receptors within the polymer, and (c) irreversible binding via first recognition and then irreversible fixation of the nanoparticles on the polymer surface.CdSe-nanoparticles (NPs) and-nanorods (NR) with either trioctylphosphineoxide (TOPO)-(NP2, NR1) or barbituric acid moieties (NP4, NR3) [31] have been used for the binding studies.

Preparation of Polymers.
Polymers derived from poly(norbornenes) with defined amounts of Hamiltonreceptor within their chain were prepared via the ROMP/click-methodology described previously [28][29][30].In brief (see Figure 2) the homo-or block copolymerization of the monomers 1a, b furnishes polymer 1 (total degree of polymerization is n + m = 100) bearing pendant bromine-moieties in amounts ranging from 1, 10, 30, or 50 mol% (Scheme 1).Quantitative transformation of the primary-CH 2 -Br moieties furnishes the CH 2 -N 3 -moieties quantitatively, which subsequently are transformed into the affixed Hamilton-receptors via azide/alkine-"click"reactions, furnishing polymer 3 either as statistical or block structures with molar percentages of the Hamilton receptor ranging from 1-50 mol% [28].Data of the polymers are shown in Table 1, revealing low polydispersities and wellmatching molecular weights of polymers 1 and polymer 3. Since lower percentages of the Hamilton-receptor cannot be obtained by direct copolymerization, they were prepared by stoichiometrical mixtures of polymer 3/s5 with a perfluorinated homopolymer (polymer 4).This leads to the generation of molar percentages of Hamiltonreceptor down to 0.01 mol%.Furthermore, a polymer with pendant Hamilton-receptor and pendant azide-moieties was prepared (polymer 5), serving as scaffold for the subsequent nanoparticle-attachment experiment.To this purpose, polymer 5 was prepared starting from a 50/50 mixture of monomer 1a, 1b, which was reacted into the corresponding azides as described before.Subsequent reaction with a substoichiometric amount of Hamilton-receptor 2 furnished polymer 5 quantitatively within this "click"-reaction, yielding polymer 5 with the following composition: (10 mol% cleft, 40 mol% azide-moieties).The presence of the residual azidemoieties is proven by IR-and NMR-spectroscopy.Details are described in the experimental and the supplementary.

Preparation of Nanoparticles and Nanorods.
CdSenanoparticles (sized 5 nm) were prepared analogous to a previously published method (Figure 2) [31], relying on the generation of trioctylphosphineoxide (TOPO)-CdSe-NP, ligand exchange by pyridine, ligand adsorption of the phosphineoxide-ligand 5, and subsequent click-reaction (furnishing barbiturate-modified nanoparticles NP4.CdSenanorods (length: 20 nm; diameter: 7 nm) [32] were prepared in a similar manner, starting from TOPO-modifed CdSe-nanorods and a similar procedure of ligand-exchange and "click"-reactions.A special case involved the preparation of NP5, bearing attached babituric acid ligands and residual alkine groups.To this respect both ligands (namely ligands 4 and 5) were applied as an equimolar mixture, adsorbed to NP2, and reacted with the barbituric-acid-alkene.As proven by IR-spectroscopy (see supplementary), residual azide-moieties are present on the surface of NP6.
3.1.3.Reversible Nanoparticles Binding.Polymers were processed as thin films (thickness approx.20 nm, as determined by ellipsometry) on silicon wafers via dip coating from polymer solutions (approx.0.5% in tetrahydrofurane (THF).The presence of the fluorinated monomer (1a) was found to be crucial in obtaining thin homogeneous films as described previously [29].After drying, the films were incubated with the respective nanoparticle solutions (in toluene or hexane).As shown in Figures 3 and 4, the NP4 (with the barbiturate-ligand) bind strongly to the statistical polymer surfaces with 1 mol% receptor (Figure 3(a), polymer 3/s5), 0.1 mol% receptor (Figure 3(c), polymer 3/s7), and 0.01 mol% receptor (see Figure 4(a), polymer 3/s8), with reduced amounts of nanoparticles with decreasing amount of supramolecular interaction.In order to probe nonspecific binding, also the noninteracting CdSe-NP2 (bearing a TOPO-ligand) were investigated.
As shown in Figures 3(b), 3(d), and 4(b), increased amount of nonspecific binding is observed, being absent in the presence of high supramolecular interactions and increasing when the amount of supramolecular interaction is reduced.This behavior can be explained by increasing nonpolar interactions between the TOPO-NP2 and the hydrocarbon-moieties of the statistical polymers with reduced amounts of polar (= supramolecular) interaction.
With block copolymers as scaffolds, the difference between specific and nonspecific binding in the case of NP4 versus NP2 is less pronounced (see Figure 5).Thus the difference between specific and nonspecific binding is hardly visible (see comparison between Figures 5(a)/5(b) or Figures 5(c)/5(d)).Due to the presence of both blocks in these relatively diluted block copolymers, polar and nonpolar regions are either too small to be discerned by the nanoparticles, or-alternatively-the polar regions of the block copolymer partially point towards the polar silicon wafer surface.Thus, statistical polymers are advantageous over block copolymers in terms of specificity and selectivity of the nanoparticle-binding process.
As the nanoparticles NP4 are bound via a hydrogen bond, its strength is dependent on the nature of the used solvent.Thus we have probed the detachment of the nanoparticles as shown in Figure 6, using methanol as solvent (24hour incubation).Since methanol is a nonsolvent for the polymer, it can be used to reduce the binding constant of the interaction between the Hamilton-receptor/barbituric acid-moieties from approx.∼10 5 M −1 in nonpolar solvents to ∼15 −1 in polar, hydrogen breaking solvents such as methanol.As demonstrated in Figure 6, the initially bound nanoparticles are removed from the surface after washing with methanol.Thus, reversibility of the film is retained, allowing the reversibly attach and remove the nanoparticles from the surface.
3.1.4.Nanorod Binding.Surface-modified nanorods were bound in a similar manner to the polymer films as the nanoparticles.Again, the binding of CdSe-nanorods with barbiturate-moieties (NR3) to those without specific interactions (TOPO-NR1) were compared on statistical polymers (polymer 3/s7, polymer 3s/8) as well as block copolymers (polymer 3/b2 (10 mol% cleft), polymer 3/b5 (1 mol% cleft).As shown in Figure 7, the binding towards the statistically distributed receptors is relatively nonspecific, leading to only small differences in the binding.
Lower and higher amounts of receptors on the polymeric surface (see supplementary for detailed pictures) did not lead to different results.With block copolymers however (see Figure 8), the picture is different, as with 10 mol% receptor (Figures 8(a  binding again is nonspecific, again revealing nonpolar effects between the NR-surface and the polymeric surface devoid of dense receptor layers.In both cases detachment of the nanorods can be achieved by addition of methanol (data not shown).
Significant, however, is the difference to the nanoparticle binding mentioned in the previous section: best and most selective nanorod binding (NR3) is achieved with the block copolymer polymer 3/b2 (10 mol% cleft), whereas in the case of nanoparticles (NP4), the best and selective binding is achieved with the statistical copolymer polymer 3/s5 (1 mol% cleft).This difference may be understood with respect to the number of interactions present on the respective nanosized object; assuming a surface area of the barbiturate of ∼7 nm 2 , about 44 units can be present on NR3, whereas only ∼11 barbiturate-moieties can be present on a ∼5 nm sized CdSenanoparticle NP4.Due to geometrical reasoning, about in the case of NR3 ∼11 barbiturates can interact with the surface, whereas in the case of NP4 ∼3-4 can assume to be directed towards the polymer surface.In case of a 10 mol%covered polymer film, about 1/5 of the volume is covered with the Hamilton-receptor, whereas in a 1 mol%-covered film this amounts to ∼1/50 (if all receptors would point upwards, and not to the silicon surface; V (monomer-unit) ∼ 1 nm 3 ; V (Hamilton-receptor) ∼ 25 nm 3 ).We propose that the nanorods need a binding event over their whole surface (i.e., moiety and residual acetylenic moieties are bound to the surface.After binding, Cu(I)-salts are added to the solution in order to induce the azide/alkine-"click"-reaction between the azide-moieties on the polymer with the alkinemoieties on NP6, thus fixing the nanoparticle to the surface after recognition.The underlying experiment is shown in Figure 10.
The initial smooth polymer film of polymer 5 (Figure 10(a)) is incubated with a solution of NP6, whereupon a dense layer of NP is observed (Figure 10(b)).The concentration of the Hamilton-receptor in polymer 5 is about 1 mol%, consistent with the optimal, most selective, and specific results obtained by the CdSe-NP-binding reported in the previous chapter.After addition of Cu(I)ions to the solution, the nanoparticles cannot be detached by incubation with methanol (24 hours) from the polymer film as seen in Figure 10(c).Thus a stable addition of the NPs due to the "click"-reaction via the triazole-bond can be assumed.

Conclusions
In summary we have reported the reversible and irreversible addition of CdSe-nanoparticles and CdSe-nanorods to polymer films via supramolecular interactions.CdSenanoparticles and the respective nanorods (20 nm; 7 nm) behave differently with respect to the needed density of matching supramolecular interactions on the polymer surface : whereas CdSe-NP's (5 nm) show optimal and specific attachment to films derived from a statistical copolymer with about 1 mol% Hamilton-receptor, in the case of nanorods a block copolymer with about 10 mol% Hamilton-receptor shows optimal results.In both cases, a reversible detachment is possible by use of methanol.Irreversible binding can be achieved by using polymers and nanoparticles with residual azide/alkine-groups on their surface, thus allowing a covalent fixation after the recognition-event by the hydrogen bonding system.The present method is highly versatile, as both-the polymer and the nanoparticle/nanorod-surfaces can be engineered nearly indefinitely.Moreover, in light of the growing interest for CdSe-nanorods for solar cell applications, this method allows to direct binding of highly luminescent CdSe-nanorods to specific locations of polymers and polymeric surfaces.

Figure 1 :
Figure 1: Schematic concept for the reversible and irreversible binding of nanoparticles to polymeric surfaces (a) film formation of block copolymers and reversible nanoparticle-binding.(b) film-formation of statistical copolymers and reversible nanoparticle binding.(c) irreversible attachment of nanoparticles onto polymer films by first supramolecular recognition and second azide/alkine-"click"-reaction.(d) Chemical structure of the used CdSe-nanoparticles and nanorods.

Table 1 :
Physical properties of statistical and block copolymers.