Flying colloidal carpets

Abstract

DNA plays a special role in polymer science not just because of the highly selective recognition of complementary single DNA strands but also because natural DNA chains can be made very long, yet perfectly monodisperse. Solutions of such long DNA chains are widely used as model systems in polymer science. Here, we report the unusual self-assembly that takes place in systems of colloids coated with very long double-stranded DNA. We find that colloids coated with such long DNA can assemble into unique “floating” crystalline monolayers that are suspended at a distance of several colloidal diameters above a weakly adsorbing substrate. The formation of these monolayers does not depend on DNA hybridization. Floating colloidal structures have potentially interesting applications as such ordered structures can be assembled in one location and then deposited somewhere else. This would open the way to the assembly of multi-component, layered colloidal crystals.

---

Introduction

Much of the interest in DNA-coated colloids derives from the fact that such building blocks can be used to make complex, self-assembling materials, because of the high selectivity of hybridization of complementary single-stranded (ss)DNA sequences.^1–11 Most efforts are directed toward binary system of two types of complementary DNA-coated colloids that can bind via DNA hybridization in solution. Here, we take a different approach. Instead of the assembly of complementary colloids in solution, we now consider colloids that cannot bind to each other but that can bind to a supporting surface through either complementary interactions between the sticky end of the DNA polymers to those adsorbed to the surface or via non-specific interactions. In this case, DNA-mediated colloid–colloid binding is not possible as all colloids display the same DNA sequence.

The immobilization of colloids coated with short DNA fragments on a surface coated with complementary DNA was extensively studied by Niemeyer and co-workers.^12,13 In these experiments, the immobilization of the colloids on the DNA micro-array indicated that the surface adsorption proceeds with complete site selectivity, as few immobilized colloids were observed when non-complementary DNA was used. By printing small areas with ssDNA, a pattern consisting of areas with or without colloids could thus be obtained. The colloids within these areas exhibited no two-dimensional ordering, rather they formed an amorphous layer because the colloids tended to be immobilized on the spot where they landed. Niemeyer et al. also explored if two-dimensional ordering of the colloids could be improved by grafting two different sequences on the colloids and introducing additional “linker” DNA.¹⁴ The first sequence on the colloids was again used to bind (hybridize) to the surface. The second type of colloid-bound oligomers was used to establish cross-links to neighboring colloids by means of the “linker” DNA. While the length of the linker DNA controlled the inter-particle spacings, no additional lateral order was observed in this system. Other authors also observed that DNA-coated colloids bound to a substrate tended to exhibit no lateral ordering.^15,16 In fact, ordered colloidal layers were only observed in cases where the surface had been prepared with an ordered pattern of surface grafted ssDNA strands.^15,16 In those cases, the order was induced by the “template” rather than by the colloid–colloid interactions.

There exists an extensive literature on colloidal particles that form two-dimensional crystals upon adsorption to a flat and unstructured surfarce.^17,18 In these systems, the attraction between colloids and wall leads to a high concentration of mobile, adsorbed colloids. At sufficiently high densities, these colloids undergo a (two-dimensional) freezing transition. However, when the attraction between colloids and surface is reduced, these crystals melt again. Density-driven freezing of DNA-coated colloids was observed by Cheng et al.¹⁹ These authors were able to make hexagonally stacked crystal sheets by drying a drop of DNA-coated colloid solution confined in the pores of a super-lattice sheet. By evaporation of the excess solution a freestanding colloidal film appeared. As, in this case, the ordering was density-driven, the colloids used in this study did not need complementary ssDNA ends.

In this article we show that it is possible to obtain two-dimensional colloidal structures that are floating well above a properly prepared but unstructured surface. While previous studies mainly used short oligonucleotides, we choose to work with longer DNA. As we shall show below, DNA-mediated colloid–surface interactions are not responsible for the formation of ordered two-dimensional structures. In fact, in all cases where DNA-mediated colloid–surface interactions are significant, we find that an amorphous layer of adsorbed colloids forms. The formation of floating crystalline monolayers is related to the fact that the DNA grafted onto the colloids behaves as a long, yet monodisperse polymer.²⁰

Materials and methods

Coating glass surfaces with DNA

We use 96 well plates (Sensoplate, Greiner bio-one) with a glass bottom as sample chambers. The glass surfaces are coated with a polymer in a three-step procedure. Firstly, the wells are rinsed with a strong soap solution (Hellmanex, 10% solution) for at least 5 h. After removing all the soap (rinsing with double distilled (dd)H₂O) a poly-L-lysine–poly(ethylene glycol)–biotin polymer solution (PLL–PEG–biotin (0.5 mg ml⁻¹, 50 µl), Surface-solutions) is added. Subsequently we remove the excess of polymer solution. Next, a layer of streptavidin (0.5 µg µl⁻¹, 50 µl, Invitrogen) is added. In a final step short single strands of 12 bases (5′-GGGCGGCGACCT-3′) with a biotin attached to the 3′-end are added (1 µM, 50 µl, Eurogentec). Surface coverage is tested in a Tris–HCl buffer (100 mM, pH 8) with neutravidin-coated polystyrene colloids (PS colloids, Invitrogen, 1 µm diameter) coated with the complementary 12 bases.

Preparation of biotin–DNA solutions

Two different double-stranded DNA polymer lengths were used. The longest polymer was λ-phage DNA, which is predominately circular at room temperature. It was linearized by heating 25 µg ml⁻¹ solutions to 65 °C, rendering the double-stranded (ds)DNA polymers with two complementary 12 base single-stranded (ss) overhangs. The linearized DNA was mixed with a solution of short single strands of 12 bases (5′-AGGTCGCCGCCC-3′) with a biotin attached to the 3′-end (5 µl, 20 µM, Eurogentec). To hybridize the oligonucleotides to the DNA, the solution was heated to 65 °C for ∼30 min and then cooled overnight to room temperature. Subsequently, T4 DNA ligase (New England Biolabs) was added to ligate the DNA backbone. To remove the excess of oligonucleotides and enzyme the samples were centrifuged and washed three times on a Microcon YM100 membrane (Millipore) with Tris–HCl buffer (250 mM, pH = 8). The biotin–DNA solution was then recovered in a clean tube.

The pBeloBac11 plasmid was purchased as a strain (New England Biolabs, ER2420S). To obtain sufficient DNA for coating 50 µl colloids (1% solids), the strain was grown overnight at 37 °C in 60 ml LB medium (for 100 ml: Bacto-Tryptone 1 g, Bacto-Yeast extract 0.5 g, NaCl 1 g and ddH₂O to 100 ml, autoclave sterilized) in the presence of chloramphenicol (20 µg ml⁻¹). With a spin miniprep-Kit (Qiagen) the plasmid DNA was then isolated. The purity of the DNA was checked by gel electrophoresis on a 1% agarose gel. The plasmid contains the same 12 bp site that can be opened by restriction with λ-terminase (BIOzymTC), leaving the same 12-base single strands as on the λ-phage DNA. The pBeloBac11 DNA was then modified with biotin according to the same protocol as described above.

Preparing DNA-coated colloids

To obtain DNA-coated colloids the biotin–DNA was mixed with neutravidin-coated red-fluorescent polystyrene colloids (diameter 1 µm, Molecular Probes) dispersed in a mixture of Tris–HCl buffer (250 mM, pH = 8). In each case the DNA and colloids were reacted overnight during which they were continuously tumbled. The next day the samples were pelleted and washed five times to remove excess of non-conjugated DNA. The supernatant was retained to quantify (by UV-spectrometry at 260 nm) the amount of DNA that had effectively bound to the beads. For λ-DNA this leads to an estimate of some 10 DNA chains per bead. pBelo-coated particles are grafted with around 25 DNA chains (all in mushroom regime). In between these washing steps, samples were heated once for 10 min at 50 °C and washed once in a NaOH solution (0.15 µM) to remove poorly bound DNA. The DNA-coated colloids were then diluted in a fresh mixture of Tris–HCl buffer (100 mM final concentration, pH = 8) to obtain a 0.5% solution. For a typical experiment 15 µl of the colloidal solution were added in 200 µl sucrose–Tris buffer (150 mg ml⁻¹ sucrose, 100 mM Tris).

Confocal imaging

All suspensions were imaged by means of an inverted microscope (DMIRB, Leica) with a confocal spinning disc scan head (CSU22, Yokogawa Electric Corp.) and a 60× water immersion objective. Fluorescence of the colloids was excited at 512 nm. Emission was observed above 600 nm.

Results and discussion

As a sample chamber we used a 96 well plate, allowing us to run many experiments under identical conditions. Their bottom glass surfaces were functionalized with ssDNA “sticky ends” by grafting a monolayer of poly-lysine–poly (ethyleneglycol)–biotin to them. Subsequently we added a layer of streptavidin which has four binding sites for biotin (Fig. 1A). After thorough washing (removing excess material) 12-base single strands with a biotin end could be bound to the streptavidin layer on the surface. To test the surface coverage of this ssDNA monolayer we introduced a solution of polystyrene (PS) colloids that were grafted with the complementary 12-base oligonucleotides. The high number of colloids bound indicates proper surface coating (Fig. 1B). Besides providing a suitable method of linking DNA to a glass surface, the polymer layer also prevents non-specific binding of PS colloids without DNA to the glass surface. Control experiments in which we disperse colloids without a DNA coating show that far fewer colloids are bound (by physisorption) to the surface (Fig. 1C).

Fig. 1 (A) Schematic representation of the experiments shown in B and C. The glass is coated with a poly-lysine–PEG–biotin layer to which streptavidin is connected. Biotinylated-ssDNA is bound to the streptavidin. Colloids coated with complementary DNA can hybridize to this surface. “Bare” PS colloids remain in solution (density matching prevents the colloids from sedimenting in time). Images are not to scale. (B) Coverage of a flat surface coated with single-stranded DNA by colloids carrying complementary short (12 bp) ssDNA. (C) If the same colloids are used, but now without attached ssDNA, no adsorption is observed. This indicates that in this system, non-specific binding is unimportant. Both scale bars are 10 µm.

The main experiments were performed with λ-coated colloids. λ-Phage DNA is monodisperse and has a contour length of 16 µm (485 000 bp; radius of gyration ∼800 nm).¹¹

The resulting colloids carry long dsDNA spacers with ssDNA “sticky ends” that can bind (hybridize) to the complementary ssDNA on the surface. Hybridization between the colloids is not possible as they all display the same ssDNA sequence (5′-AGGTCGCCGCCC-3′). As the radius of gyration (R_g) of λ-DNA is similar to the colloid-size used, the colloids are coated with no more than 8 to 10 strands.¹¹ Sedimentation is minimized by density matching the DNA-coated polystyrene colloids with a sucrose–Tris buffer. In all experiments we work at a colloidal volume fraction of ∼4 × 10⁻⁴.

2 h after injecting a solution of colloids coated with λ-DNA into the sample cell, we observe the appearance of single layers of close-packed colloids. If left at room temperature for longer times, these structures can grow into large colloidal monolayers that span the entire field of view of the microscope (Fig. 2A). Within one sample well up to 40 independent 2D crystals can be found. They appear in a broad size range, ranging from 25 to about 2500 colloids per monolayer. The structures are clearly crystalline, albeit that some crystals contain point defects or even grain boundaries. Indeed, the pair-correlation function of this two-dimensional crystal displays distinct peaks corresponding to those of a 2D hexagonal crystal (black arrowheads, Fig. 2B). Interestingly, these colloidal sheets float above the bottom of the cell, at heights ranging from the R_g of λ-DNA up to 5 µm (Fig. 3A) with an average around 2.5 µm. As the colloidal monolayers are, on average, several particle diameters above the substrate, we refer to these two-dimensional crystals as “colloidal flying carpets”. We only observe the formation of colloidal carpets above the bottom surface of the cell, but not in the bulk of the solution.

Fig. 2 Aggregation of λ-DNA-coated colloids above a “sticky” surface leads to 2D flying carpets. (A) Image of a flying carpet imaged with a confocal microscope. As the structure is slightly bent in the xy-plane a z-projection is shown. Note that not all colloids participate in the assembly. Scale bar is 10 µm. (B) Pair-correlation function of (A). The first neighbor is at contact, the function also shows peaks for the second, third and fourth neighbors. Black triangles: guide to the eye; positions of expected peaks for a perfect hexagonal crystal.

Fig. 3 Flying colloidal carpets can move. (A) The 2D colloidal crystals sit not directly on the surface. Instead they are suspended above the surface at heights ranging from λ-DNA R_g ≈ 0.8 up to 5 µm. The bars indicate the height of the carpets above the surface from three different experiments (wells) performed in the same conditions and concentrations. In black a Gaussian fit is drawn to obtain an estimate of the average height: 2.5 µm. (B) Flying colloidal carpet image taken with confocal microscopy at t = 0 h. (C) Image of the same carpet taken at t = 2 h. Both scale bars indicate 5 µm.

The fact that the height of the carpets is somewhat larger than the effective coil diameter of free λ-DNA (1.6 µm) indicates the grafted DNAs are weakly stretched. This is not surprising as the surface density of DNA grafted on a carpet is inside the overlap regime: if we assume that there are 4 DNA's per colloid on both sides of the carpet, then the reduced areal density of DNA, ρ/ρ* ≈ 9. Hence, we expect moderate stretching by a factor of 2–3, as observed.

λ-DNA-coated colloids contain 8 to 10 strands per colloid. If all strands of DNA would anchor to the surface the carpets would be unable to move. By imaging a single carpet at different moments in time, possible slow movements can be detected. Fig. 3B shows a carpet with an unusual shape at a given time. After leaving the sample for 2 h we imaged the same carpet again. This time there was no sign of growth, but the carpet did rotate roughly 30° counter-clockwise (Fig. 3C). Depending on the carpet size, these movements extended over no more than tens of colloidal diameters. The mobility of the carpet indicates that either not all strands of DNA were hybridized or at the very least that the hybridization at room temperature is sufficiently weak to make rearrangements possible. As room temperature is well below the melting temperature of our 12 base pair bonds, the first probability is the more likely. Another conclusion that can be drawn from Fig. 3 is that the crystallites have some time to anneal during growth, as the 2D crystals can even show clear facets.

When we repeated the above experiments with an intermediate-length pBelo-DNA (7500 bp, R_g ≈ 200 nm, 25 DNAs per colloid)⁸ we again observed the formation of floating 2D crystal structures (see Fig. S1B in ESI†), although in this case the maximum sizes of the colloidal sheets seemed smaller than for the λ-DNA-coated beads. However, when we repeated the experiments with colloids coated with “short” DNA (no double-stranded (ds)DNA spacer; 12 base ssDNA attached directly to the colloids), no crystalline sheets are formed. Rather, we observe an amorphous colloidal layer in direct contact with the polymer layer coating the surface of the sample well (ESI†, Fig. S1C).

As our colloids are designed to have no DNA-mediated attraction to each other, the appearance of close-packed colloids must be due to other (non-specific) interactions. The mechanism of self-assembly depends on the effective interaction between colloids mediated by the environment such as solvent, substrate or external fields. The most common non-specific interactions between colloids are: Coulomb interactions, dispersion forces and depletion forces. In the present case (screened) Coulomb interactions between the like-charged DNA-coated colloids are unlikely to lead to aggregation. Depletion forces are entropic, attractive interactions between colloids that can be caused by non-adsorbing polymers.²¹ In our system, there is no free DNA and hence one might be tempted to dismiss depletion forces as a possible source of attraction. However, if the colloids move in a semi-dilute mesh of long DNA strands, short-ranged depletion interactions cannot be ruled out a priori—even if the DNA is grafted to the colloids.

To test whether depletion forces are responsible for the 2D aggregation behavior in our system, we repeated the experiment with colloids without DNA coating, but with free λ-DNA in solution. Whereas grafted polymers led to ordered 2D aggregates and no aggregation in bulk, the same concentration of DNA free in solution only yields small 3D aggregates of ∼15 colloids at most (see Fig. S2A and B in ESI†). Even if we increase the λ-DNA concentration a hundred times we still see a clear difference in aggregation behavior. Now the aggregates comprise more colloids, but they consist of 3D branched structures (ESI†, Fig. S2C) while no flat 2D crystals are found. Kim and co-workers, who compared depletion forces of colloids of different materials, showed a similar effect for neutravidin-coated colloids.² From the obtained pictures it is clear that depletion forces are not responsible for the formation of the “colloidal flying carpet”. These results indicate that the crystallization mechanism is different from previous observations where 2D crystallization from a very dilute colloid–polymer mixture was observed.²²

On the basis of the above discussion, it seems most likely that short-range dispersion forces play a key role in the formation of the dense colloidal carpets. As the primary minimum of the dispersion interaction between two colloids is extremely deep, we should expect that colloidal carpets once formed are very stable. To test this hypothesis, we examined the behavior of the carpets at elevated temperatures. We first prepared crystalline colloidal carpets at room temperature (Fig. 4A and B). Once these had formed, the sample was heated to 70 °C for several hours (Fig. 4C). The two-dimensional crystal structures were found to be remarkably stable against heating: even after more than 20 h at 70 °C only minor changes can be seen. Moreover, the carpets remained at similar heights compared to room temperature.

Fig. 4 Short and long DNA-coated colloids respond differently to high temperatures. (A) Schematic cartoon of colloids ‘bound’ to the surface via λ-DNA that has been attached to the colloids. (B) The resulting 2D carpet formed at room temperature. (C) Heating the carpet for 7 h at 70 °C does not melt the crystal. (D) Schematic illustration of colloids bound to the PLL–PEG biotin layer via two complementary 12 base ssDNA sequences, one being attached to the colloids. (E) Typical bound, amorphous layer of such a system formed at room temperature. (F) After heating (70 °C) only few particles remain. All scale bars correspond to 10 µm.

These results strongly suggest that the colloidal carpets are held together by van der Waals forces. But to strengthen our conclusion, we tested whether oligonucleotide functionalized colloids (with only the 12 bp ssDNA) could dissociate from a surface with complementary ssDNA coating upon heating to 70 °C (Fig. 4D). As this system only showed few colloids bound to the surface at a low volume fraction (∼4 × 10⁻⁴) a significantly higher volume fraction was used (∼4 × 10⁻³). First the surface coverage is imaged at room temperature (Fig. 4E). Then the sample was heated to 70 °C. As can be seen from the figure, after removal of unbound colloids only few colloids remain bound to the surface (Fig. 4F). This indicates that at 70 °C the 12-base DNA duplex is dissociated.

As increasing the temperature can dissociate the 12-base duplex as shown in Fig. 4F, DNA binding cannot be the only factor responsible for keeping the carpets near the surface. Hence, non-specific DNA binding to the surface may play a role. Whether or not this is the case can be tested by repeating the carpet formation experiments under conditions where DNA–surface hybridization is impossible. Indeed, ref. 23 reports a study on non-specific DNA interactions with PLL–PEG films. In this reference it is shown that DNA can adsorb to the positively charged polymer layer. The strength of this adsorption depends on the ionic strength in solution. To test whether non-specific DNA–surface attraction can facilitate the formation of colloidal carpets, we prepared samples with no DNA coating on the surface while the colloids were grafted with blunt pBelo-DNA. The results (see Fig S3 in ESI†) show that DNA hybridization is indeed not necessary to obtain floating 2D crystal structures. Any weak attraction, here provided by the Coulomb attraction between the positively charged PLL layer and the negatively charged DNA double strands, will lead to the present 2D aggregation. However, attraction is lessened when salt is added to the Tris buffer. Increasing the salt concentrations above 100 mM (Na⁺) prevents the formation of carpets (no ssDNA present at the surface), due to screening of the surface charges.

The colloids coated with oligonucleotides (no dsDNA spacer, 12 bases ssDNA attached directly to the colloids) were unable to form crystalline sheets. Instead, an amorphous colloidal layer in direct contact with the polymer layer coating the surface of the sample well is formed. This randomness as well as the fact that these colloids can be “melted off” above the hybridization temperature of the 12 bp bonds (T_m ≈ 42 °C) suggests that two factors are essential for the formation of crystalline colloidal membranes: (1) weak binding to the surface that allows colloids to diffuse laterally and (2) weak steric stabilization of the colloids against the formation of direct contacts that are stabilized by dispersion forces. If the colloids are too strongly bound to the surface of the cell, as is the case for the colloids coated with many short ssDNA strands, surface diffusion of the colloids is inhibited and hence only amorphous layers can form, since annealing is not possible. In addition, the short DNAs provide steric stabilization against the formation of direct colloid–colloid contacts. Hence, neither condition for the formation of crystalline colloidal carpets is satisfied in the case of colloids coated with short ssDNA.

Colloids grafted with either λ-DNA or pBelo-DNA with the same ssDNA overhangs do not aggregate in the bulk under the conditions used in the present experiments. This implies that in the bulk of the solution, where the colloidal concentration is low, the DNA cloud that surrounds the colloids is sufficient to prevent aggregation due to non-specific van der Waals interactions. However, near the bottom cell surface, this situation changes: the charged DNA strands on the colloids are attracted weakly and non-specifically to the (oppositely charged) poly-lysine layer on the surface. As a consequence of the DNA–surface attraction, the colloidal concentration is significantly enhanced on the surface and colloids come into contact sufficiently frequently to overcome the weak entropic stabilization provided by the long dsDNA. Indeed, in the 2D crystals the colloids are effectively touching each other and, as we argued above, the thermal stability in the colloidal membranes is therefore most likely due to the action of short-range dispersion forces. At the same time, the fact that highly ordered 2D crystals form indicates that crystal growth is slow: aggregation is definitely not diffusion limited.

Conclusions

We have found that colloids coated with long DNA strands can spontaneously form a crystalline 2D colloidal carpet that hovers several microns above the support surface. Our control experiments suggest that the formation of these colloidal membranes is facilitated by the weak, non-specific adsorption of the DNA-coated colloids (or, more precisely, of the DNA coating of these colloids) to a weakly positively charged lower surface of the sample cell.²³ Under these conditions, the steric stabilization of the colloids by the grafted dsDNA is insufficient to prevent the slow formation of dense 2D colloidal crystals that are subsequently stabilized by short-ranged dispersion forces. Colloids that are coated with very short ssDNA strands bind strongly to the surface via specific interactions between the complementary strands and form an amorphous adsorbate rather than a crystalline floating carpet.

The ability to make floating, yet surface-bound structures, could provide an interesting route to make novel colloidal structures: in particular, by making use of suitable DNA linkers, it should be possible to induce the self-assembly of multiple layers of 2D crystals that can be attached on top of the carpets. Another option would be to use the free “sticky ends” to close the carpets, forming a tube of colloids with DNA on both sides, as possible docking sites to bind other molecules or materials. Indeed, preliminary results already suggest that the colloidal carpets can be lifted of the surface using buoyancy forces. They stay stable over many days and can be moved with optical tweezers.

Acknowledgements

We thank M. Dogterom for providing laboratory facilities. We are also grateful for useful discussions with D. Frenkel. This work is part of the research program of the Stichting voor Fundamenteel Onderzoek der Materie (FOM), which is supported by the Nederlandse Organisatie voor Weten-schappelijk Onderzoek (NWO).

Notes and references

1. C. A. Mirkin, R. L. Letsinger, R. C. Mucic and J. J. Storhoff, Nature, 1996, 382, 607–609.
2. A. J. Kim, P. L. Biancaniello and J. C. Crocker, Langmuir, 2006, 22, 1991–2001.
3. D. Nykypanchuk, N. N. Maye, D. van der Lelie and O. Gang, Nature, 2008, 451, 549–552.
4. S. Y. Park, A. K. R. Lytton-Jean, B. Lee, S. Weigand, G. C. Schatz and C. A. Mirkin, Nature, 2008, 451, 553–556.
5. V. T. Milam, A. L. Hiddessen, J. C. Crocker, D. J. Graves and D. A. Hammer, Langmuir, 2003, 19, 10317–10321.
6. M. P. Valignat, O. Theodoly, J. C. Crocker, W. B. Russel and P. M. Chaikin, Proc. Natl. Acad. Sci. U. S. A., 2005, 102, 4225–4229.
7. P. H. Rogers, E. Michel, C. A. Bauer, S. Vanderet, D. Hansen, B. K. Roberts, A. Calvez, J. B. Crews, K. O. Lau, A. Wood, D. J. Pine and P. V. Schwartz, Langmuir, 2005, 21, 5562–5569.
8. N. Geerts, T. Schmatko and E. Eiser, Langmuir, 2008, 24, 5118–5123.
9. A. V. Tkachenko, Phys. Rev. Lett., 2002, 89, 148303.
10. P. L. Biancaniello, A. J. Kim and J. C. Crocker, Phys. Rev. Lett., 2005, 94, 058302.
11. T. Schmatko, B. Bozorgui, N. Geerst, D. Frenkel, E. Eiser and W. C. K. Poon, Soft Matter, 2007, 3, 703–706.
12. C. M. Niemeyer, B. Ceyhan, S. Gao, L. Chi, S. Peschel and U. Simon, Colloid Polym. Sci., 2001, 279, 68–72.
13. S. Peschel, B. Ceyhan, C. M. Niemeyer, S. Gao, L. Chi and U. Simon, Mater. Sci. Eng., C, 2002, 19, 47–50.
14. B. Zou, B. Ceyhan, U. Simon and C. M. Niemeyer, Adv. Mater., 2005, 17, 1643–1647.
15. D. M. Hartmann, M. Heller, C. S. Esener, D. Schwartz and G. Tu, J. Mater. Res., 2002, 17, 473–478.
16. E. M. Puchner, S. K. Kufer, M. Strackharn, S. W. Stahl and H. E. Gaub, Nano Lett., 2008, 8, 3692–3695.
17. P. N. Pusey, A. D. Pirie and W. C. K. Poon, Physica A (Amsterdam), 1993, 201, 322–331.
18. J. J. Cerda, T. Sintes, C. M. Sorensen and A. Chakrabarti, Phys. Rev. E: Stat. Phys., Plasmas, Fluids, Relat. Interdiscip. Top., 2004, 70, 011405.
19. W. Cheng, M. J. Campolongo, J. J. Cha, S. J. Tan, C. C. Umbach, D. A. Muller and D. Luo, Nat. Mater., 2009, 8, 519–525.
20. T. T. Perkins, D. E. Smith and S. Chu, Science, 1997, 276, 2016–2021.
21. S. Asakura and F. Oosawa, J. Chem. Phys., 1954, 22, 1255–1256.
22. U. Gasser, J. Phys.: Condens. Matter, 2009, 21, 203101.
23. R. Schlapak, D. Armitage, N. Saucedo-Zeni, W. Chrzanowski, M. Hohage, D. Caruana and S. Howorka, Soft Matter, 2009, 5, 613–621.

---

Footnote

† Electronic supplementary information (ESI) available: Images to show the dependence of carpet formation on DNA length, that the formation is not depletion driven, and that DNA hybridization is redundant. See DOI: 10.1039/b917846e

---