Skip to main content
NCBI home page
Scientific Reports logoLink to Scientific Reports
. 2013 Aug 16;3:2451. doi: 10.1038/srep02451

Gelling by Heating

Sándalo Roldán-Vargas 1,a, Frank Smallenburg 1, Walter Kob 2, Francesco Sciortino 1
PMCID: PMC3744795  PMID: 23948858

Abstract

We exploit the concept of competing interactions to design a binary mixture of patchy particles that forms a reversible gel upon heating. Our molecular dynamics computer simulation of such a system shows that with increasing temperature the relaxation dynamics slows down by more than four orders of magnitude and then speeds up again. The system is thus a fluid both at high and at low temperatures and a solid-like disordered open network structure at intermediate temperature. We further discuss the feasibility of realizing a real material with this reversible behavior.

Some of the most versatile and efficient strategies for designing new materials with unconventional behavior are based on the idea of competitive interaction. While in biology the term competitive interaction indicates the result of rivalry between two or more species competing for resources, in physics it stands for the presence of several interaction mechanisms that can stabilize competing local structures, leading to novel and highly interesting features of the system. Recent examples for this mechanism in soft matter systems include, among others, the competition between short-range attraction and long-range repulsion in charged colloids giving rise to cluster phases1,2,3, the competition between chaining and branching in patchy colloids4,5 where a specific design of the inter-patch interactions results in a phase diagram in which the density of the coexisting liquid approaches the density of the gas5, and the design of DNA-coated colloids with two different DNA sequences for the purpose of establishing a competition between intra and inter-particle interactions, favoring crystal formation6. Often the very nature of these competing mechanisms promotes the emergence of a structure controlled by energy (stable at low temperature, T) which competes with a structure stabilized by entropy at intermediate T.

Recent progress in the synthesis of colloids has led to a novel type of particles that have highly directional and selective interactions, thus providing valence to colloids7. In particular the recent developments in large scale synthesis of patchy colloidal particles8 and in DNA nanotechnology have already allowed for the production of versatile constructs in which specificity, valence and interaction strength can be tuned at will9,10. This ability to control the interactions between nano-and meso-sized particles8, and to design the geometric properties of the patches11,12 and/or their functionalization, offers now the possibility to take advantage of the competitive interactions that are usually present in such systems and hence to modulate material properties with external control parameters. In this Report we develop one such possibility: The design of a material that continuously (i.e. without the interference of a first order transition) and reversibly gels upon heating. In contrast to gelation on heating observed in protein solutions13, where kinetic arrest is driven by the irreversible denaturation of the proteins, here we require thermo-reversibility of the process and a simple design of the constituent particles. We show how a binary mixture of limited valence particles can provide a model in which the competition between entropy and potential energy causes the system to show a re-entrant behavior, passing reversibly from a fluid to a gel and again to a fluid if T is varied. The proposed model, amenable to theoretical analysis and experimental realization, shows in simulations a slowing down of the dynamics of several orders of magnitude on heating, providing the first example of controlled reversible gelation.

Patchy colloidal particles with a limited number f of attractive patches progressively cluster when cooled, leading such systems to form a percolating network that at sufficiently low T will contain all particles in the system14,15,16,17,18,19. Due to this aggregation, the dynamics slows down by several orders of magnitude. In the network state the lifetime of the bonds (a T-controlled quantity) fixes the timescale over which the system behaves as a solid. It is known that if f = 4, the particles (here called A species) form a random tetrahedral network which closely resembles the structure of network-forming atomic systems like silica or silicon20. To melt the network at low T, we add a second species (B) with only a single patch which competes for bonding with the patches on the network forming A-species21. The idea is to design a competitive mechanism such that the bonding between A- and B-particles becomes dominant, but only at a temperature which is much lower than the one at which the AA-network is formed. As a result, the stable low-T phase consists of A-particles decorated with f B-particles which are free to diffuse in the sample volume, whereas at intermediate T the system forms a highly viscous AA-network that is progressively fragmented and transformed into a fluid upon heating.

Results

Model

We consider a binary mixture of patchy colloids where each A-particle has f = 4 patches on its surface that are arranged in a tetrahedral geometry and the B-particles have only one patch (see Fig. 1). The patch-patch interaction is modeled via a Kern-Frenkel potential22, a model that has been extensively used over the last decade to compare simulations and experiments on the self-assembly of patchy colloids23,24,25. Each A-patch can interact either with a B-patch with unit energy Inline graphic and bonding volume Inline graphic or with another A-patch with energy Inline graphic and Inline graphic. No BB-bonding is allowed. The bonding volumes are each determined by an interaction range δα and an angular patch width θα (α ∈ {AA,AB}) (see Fig. 1 and Methods). The attractive patch-patch interaction is complemented by an isotropic hard-core repulsion, where the spherical cores have diameters σA and σB = 0.35σA. The size ratio was chosen such that the B-particles can block the A-patches from bonding to other A-patches without significantly increasing the packing fraction of the pure A system. Due to geometric constraints, each patch can be involved in only a single bond.

Figure 1. The Kern-Frenkel model.

Figure 1

Open in a new tab

(a) Schematic of the interaction parameters in the Kern-Frenkel model. An A-patch can bond with either another A-particle (b), or with a B-particle (c). Panel (d) shows a “flower”, i.e. a fully bonded cluster consisting of one A-particle and four B-particles, representing the lowest-energy state of the system. Here the interaction ranges and the angular patch widths are cos θAA = 0.92, δAA = 0.15σA, and cos θAB = 0.99, δAB = 0.2σA. With these choices the bonding volumes are given by Inline graphic and Inline graphic(see Methods). Note that in the binary Kern-Frenkel model, θ and δ of each bond are defined by the species of both bonding partners and are not properties of individual particles.

Using event-driven molecular dynamics simulations26,27,28 (see Methods), we have studied a system of NA = 600 and NB = 2400 particles, corresponding to a total number density Inline graphic, with partial number densities Inline graphic and Inline graphic, for a wide range of T, whose unit is given by Inline graphic, where kB is Boltzmann's constant. The composition of the system is thus fixed at xA = 0.2. The density Inline graphic of A-particles corresponds to the optimal density at which tetrahedral particles form an unstrained fully bonded network29. With this composition, the fully bonded network has an energy of Inline graphic, whereas a configuration in which all the B-particles are bonded to the A-particles has the significantly lower energy of Inline graphic. We also simulate for a low Inline graphic a reference system composed of 600 flowers (see Fig. 1d), i.e. A-particles bonded to four B-particles. At this low T no bond breaking events take place within the simulation time.

Temperature dependence of the bonding probability

Figure 2 demonstrates the effect of the competitive interactions present in our system. It shows the T–dependence of the probability that a patch on an A-particle is bonded to another A-patch (pAA) or to a B-patch (pAB). On cooling, pAA starts to grow, signaling the onset of the network formation, and reaches a maximum around Inline graphic. We recall that within a mean-field description, percolation of particles with valence f = 4 takes place at pAA = 1/330. Since at the maximum we find pAA ≈ 0.9 we can conclude that at this T the A-particles have formed a highly bonded percolating network. Upon further cooling, pAA starts to decrease whereas pAB grows quickly, showing that the AB-bonds are starting to replace the AA-bonds. The entropy associated with the larger bonding volume Inline graphic for the AA interaction is crucial for promoting the formation of a large number of AA-bonds at intermediate T, before the energetically preferred but entropically disfavored AB-bonds set in. Figure 2 also shows the parameter-free theoretical predictions for the bonding probabilities as obtained from the first-order thermodynamic perturbation theory developed by Wertheim31,32,33 (details on the Wertheim calculations are reported in the Methods section). The Wertheim theory nicely captures the mechanism of competing interactions, reproducing the position and height of the maximum of pAA as well as the low T trends of pAA and pAB.

Figure 2. Bonding probabilities.

Figure 2

Open in a new tab

Probability that a patch on an A-particle is bonded to another patch on an A-particle, pAA (black circles), or to a patch on a B-particle, pAB (red squares). The dashed lines are the prediction for these probabilities as calculated from the Wertheim theory.

Structure

Figure 3 shows the unusual T dependence of the structure of the system, related to the non-monotonic behavior of pAA. At high T, the partial structure factor SAA(q) shows the conventional q–dependence found in simple liquids with a main peak around qσA = 7.2. Upon decreasing T the main peak splits into two, one located around qσA = 5.2 and a higher one around 8.4. This double peak feature is typical of liquids that have a local tetrahedral network structure, such as silicon or silica20. The peak at qσA ≈ 8.4 corresponds to the nearest neighbor distance between two bonded A-particles, whereas the one at around 5.2 is associated with the second-nearest neighbors in the tetrahedral network. Note that this double peak structure is most pronounced at Inline graphic, i.e. at the T at which pAA has a maximum (see Fig. 2) and hence the gel is maximally connected. When T is lowered even further the double peak structure disappears and SAA(q) becomes again similar to the structure factor of a fluid composed of flowers (which is also included in Fig. 3 as a reference). Since the size of a flower is larger than that of an A-particle, the peak position at low T is to the left of the one observed at high T.

Figure 3. Structure factor.

Figure 3

Open in a new tab

Partial structure factor SAA(q) for different values of T (solid lines with different symbols). The SAA(q) for a fluid of flowers at Inline graphic (black solid line) is also represented.

Temperature dependence of the diffusion coefficient

We now quantify the effect of the competing interactions on the dynamics of the system and provide evidence that the change of the structure rich in AB-bonds at low T to the highly bonded AA-network generates a slowing down of the dynamics on heating. For this, we calculate the mean squared displacement (MSD) for the particles of both species and then their corresponding diffusion coefficients Dα (α ∈ {A,B}) from the long-time behavior of the MSD via the Einstein relation. We should note that the center of mass (CM) of a single species has a non-zero velocity (which is compensated by the CM motion of the other species). To obtain meaningful results for the MSD of a single species, we subtract the CM drift of the species in question before evaluating the MSD. In addition, to remove the trivial trend originated from the T-dependence of the thermal velocity we divide Dα by a reference diffusion coefficient Inline graphic, where Inline graphic and mA is the mass of an A-particle. Figure 4 shows the T-dependence of Dα in an Arrhenius plot. At high T, Dα is approximately constant for both type of particles, indicating that bonds do not play a significant role. On cooling, DA starts to decrease very rapidly, with a super-Arrhenius T-dependence reminiscent of that observed in molecular networks20, turning into an Arrhenius law with an activation energy approximately equal to Inline graphic(see dashed line in Fig. 4). Similar values of the activation energy are typically found in tetrahedral network-forming systems where most of the particles belong to the percolating cluster, and bond breaking is the bottleneck for relaxation18,34,35. Before the gel starts to decompose at a temperature below Inline graphic, DA has already decreased by four orders of magnitude compared to its value at high T, indicating the formation of a persistent network. For Inline graphic, DA starts to increase. This rising persists down to the lowest T at which we were able to equilibrate the system. We emphasize that this non-monotonic T-dependence is only observed for the A-particles, i.e. the particles which are involved in the formation of the network. In contrast, DB shows only a rather mild T-dependence. Fig. 4 also shows the diffusion coefficient of the fluid of flowers at Inline graphic for which DA = DB, the common value to which both DA and DB converge.

Figure 4. Diffusion coefficient.

Figure 4

Open in a new tab

Arrhenius plot of the normalized diffusion coefficient Dα/D0 (α = A, circles, α = B, squares). Also included is an Arrhenius law with activation energy Inline graphic(dashed line). The diffusion coefficient of a fluid of flowers at low T is represented by a blue diamond.

Relaxation times

A non-monotonic behavior of the characteristic time is also found in the time evolution of the collective- and self-intermediate scattering functions. Their study provides insight into how the relaxation dynamics depends on the considered length scale. Figure 5a shows an Arrhenius plot of the relaxation time τA(q) determined from the time integral of the intermediate scattering function of the A-particles for two different q–vectors: qσA = 5.2 and qσA = 7.2, which correspond, respectively, to the location of the first peak in the network and in the high-T fluid (see Fig. 3). We find that the self and collective relaxation times τA(q), normalized by τ0, show qualitatively the same T-dependence: a plateau at high T, a fast increase within the T-range in which the network is formed, a quick decrease once the network starts to break up again, and a final plateau at low T. This T-dependence is observed for both values of q, indicating that the relaxation mechanism does not depend on the length scale considered.

Figure 5. Relaxation times.

Figure 5

Open in a new tab

(a) Arrhenius plot of the normalized relaxation time τA(q)/τ0 as obtained from the self (dashed lines) and collective (solid lines) scattering functions, where Inline graphic. The different curves correspond to the wave-vectors given by the first two peaks in SAA(q). (b) Arrhenius plot of the normalized bond-persistence time τb/τ0 for the AA (squares) and AB bonds (circles). Also included is an Arrhenius law with activation energy Inline graphic(blue solid line).

To provide further evidence that the system is ergodic on long time scales, i.e. that the structure of the system has completely lost its memory of the initial state, we have investigated the bond persistence function pb(t), i.e. the probability that a bond which is present at time zero is also present at time t (see Supplementary Information). Hence, the relaxation time of pb(t) provides information on the restructuring time of the network connectivity. Figure 5b shows an Arrhenius plot with the T-dependence of the decay time τb, where pb(τb) = e−1. At intermediate and low T, τb is larger than the relaxation times shown in Fig. 5a. We find thus that pb(t) decays to zero only on a time scale that is significantly longer than the relaxation times associated with the scattering functions, confirming that some fraction of spacial decorrelation of the network, as quantified by the collective scattering function, takes place at partially fixed bonding pattern. The extreme case occurs at very low T where the system is a fluid of flowers that relaxes relatively quickly but in which AB-bonds survive for a very long time. Figure 5b also shows that the T-dependence of the lifetime of the AA-bonds differs strongly from the one of the AB-bonds. The latter shows in the whole T–range an Arrhenius dependence with an activation energy very close to Inline graphic(see blue solid line in Fig. 5b). This suggests that the mechanism for the breaking of an AB-bond is not collective in nature but a simple activated process. In contrast, the breaking time for an AA-bond follows an Arrhenius law at high T but becomes super-Arrhenius within the T-region in which the gel forms, showing that this bond-breaking process is of collective nature. At very low T the effective activation energy becomes again an Arrhenius behavior with an activation energy given by Inline graphic.

Discussion

We have demonstrated that a judicious choice of the interaction parameters of a binary mixture of A and B patchy particles shows a non-monotonic and reversible T-dependence of its dynamic properties. We set up a competition between network-forming AA-bonds and network-breaking AB-bonds, which are favored by entropy and energy, respectively. Thus, at intermediate T, entropy stabilizes a viscous network gel, whereas both at low and high T, the network breaks up, leading to a simple fluid state. Our results are robust with respect to a change in the model parameters and we have indeed observed a qualitatively similar phenomenology in our system for a wide range of the interaction energy and density, Inline graphic and Inline graphic. Since our system requires no fine-tuning, we expect that the observed phenomenon can also be realized experimentally. In particular, two experimental systems are very promising candidates to test the ideas presented in this Report: (i) a solution of DNA constructs of valence four36,37 in the presence of competing DNA single strands and (ii) the binary mixture of patchy particles recently synthesized in Ref. 8. Both of these systems offer single-patch and four-patch colloidal particles with controllable interaction strengths that are already amenable for experimentation, and thus have the potential to provide soft materials that reversibly gel upon heating.

Methods

Kern-Frenkel model

The particles investigated in the paper are modeled via the well-known Kern-Frenkel model22. In this model, particles interact by means of a combination of a hard-sphere potential uHS and an attractive directional interaction upatch. The hard-core repulsion between two particles i and j is given by:

graphic file with name srep02451-m1.jpg

here, β = 1/kBT, with kB Boltzmann's constant and T the temperature, rij is the center-to-center distance between the particles, and σij = (σi + σj)/2 denotes the contact distance between the particles, with σi(σj) the diameter of particle i(j). The site-specific attraction between the particles is determined by the circular patches on the surface of each particle, which interact such that two particles form a bond with interaction energy Inline graphic when i) their centers of mass are within a maximum interaction range σij + δij, and ii) the center-to-center vector between the particles passes through a patch on the surface of both particles (see Fig. 1 of main text). The size of the patches is determined by an opening angle θij. The potential energy of two particles i and j is thus given by

graphic file with name srep02451-m2.jpg

where uSW is a square-well attraction, given by:

graphic file with name srep02451-m3.jpg

The function Φ(rij, {pi}) is defined as

graphic file with name srep02451-m4.jpg

where rij = rj − ri, {pi} is a set of normalized vectors pointing from the center of particle i towards the center of each of its patches, and Inline graphic is a unit vector in the direction of r.

In the model used in the Report, all particles are either species A or species B. The A-particles (with hard sphere diameter σA) have four patches each, arranged in a tetrahedral geometry. The B-particles are smaller, with σB = 0.35σA, and have only a single patch. The parameters for the interaction between two A-particles are given by cos θAA = 0.92, and δAA = 0.15σA. For the AB-interactions, cos θAB = 0.99, and δAB = 0.2σA. There are no attractive interactions between the B-particles. The ratio between the two interaction strengths is fixed at Inline graphic.

Event-driven molecular dynamics

To study the relaxation dynamics of our model we use event-driven Molecular Dynamics (EDMD) simulations26,27. The implementation of the EDMD simulation relies on the numerical prediction of bond formation and bond breaking events, and follows the same scheme as described in Ref. 28. In the model under consideration here, the mass m of each particle is taken to be the same, setting the time unit of the simulation Inline graphic. Similarly, the moments of inertia tensors of all particles were also chosen to be the same: Inline graphic.

During the equilibration of the simulations, the temperature is controlled by an Andersen thermostat: periodically, randomly selected particles are given a new velocity and angular velocity, drawn from a Maxwell-Boltzmann distribution. While measuring the diffusion coefficient, no thermostat is used, so that the total energy in the system is constant. To speed up equilibration of the system at low temperature, the initial configuration is taken from a standard canonical Monte Carlo simulation equilibrated at the same temperature T.

Wertheim theory

Wertheim's thermodynamic perturbation theory allows us to obtain an analytical expression for the free energy of pure fluids and fluid mixtures (a detailed description can be found in Refs. 31, 32). In the context of Wertheim's theory, and specialized to our system, the probability pα that an α-site (α ∈ {A, B}) is bonded is obtained through the law of mass action, which in our case takes the form of a set of two coupled equations33:

graphic file with name srep02451-m5.jpg
graphic file with name srep02451-m6.jpg

where xA = 0.2 is the molar fraction of the species A, ρ is the total number density. All interaction parameters needed for describing bonding between AA-and AB enter in ΔAA and ΔAB:

graphic file with name srep02451-m7.jpg
graphic file with name srep02451-m8.jpg

where Inline graphic and Inline graphic are the bonding interaction energies of the AA- and AB-bonds. In the present work we have approximated ΔAA and ΔAB by using the contact values of the partial radial distribution functions33, gAA(σA), gBB(σB), and gAB(σAB) (where σAB = (σA + σB)/2), for a binary mixture of hard spheres as obtained from the Percus-Yevick Equation38:

graphic file with name srep02451-m9.jpg
graphic file with name srep02451-m10.jpg

where Inline graphic, Inline graphic, being ρA and ρB the partial number densities of the different species. The bonding volumes Inline graphic and Inline graphic present in Equations (7) and (8) are given by:

graphic file with name srep02451-m11.jpg
graphic file with name srep02451-m12.jpg

where δγ and θγ (γ ∈ {AA, AB}) are respectively the interaction ranges and the angular patch widths defined in the model section. Once Eqs. (5) and (6) are solved (by using Eqs. (7)–(12)), the probabilities pAA and pAB that an A-site is specifically bonded to another A- or to a B-site are obtained by the relations: pAA = pA − pB and pAB = pB.

Author Contributions

F. Smallenburg and F. Sciortino wrote the code for the EDMD and MC simulations. S. R.-V. and W. K. performed the simulations and the analysis of the data. All the authors contributed to the interpretation of the results and the writing of the paper.

Supplementary Material

Supplementary Information

Gelling by Heating: Supplementary Information

srep02451-s1.pdf (86.2KB, pdf)

Acknowledgments

We acknowledge support from COMPLOIDS and ERC-PATCHYCOLLOIDS-226207. W. Kob acknowledges support from the Institut Universitaire de France. F. Sciortino thanks F. Starr for discussions.

References

  1. Sear R. P. & Gelbart W. M. Microphase separation versus the vapor-liquid transition in systems of spherical particles. J. Chem. Phys. 110, 4582–4588 (1999). [Google Scholar]
  2. Groenewold J. & Kegel W. K. Anomalously large equilibrium clusters of colloids. J. Phys. Chem. B 105, 11702–11709 (2001). [Google Scholar]
  3. Sciortino F., Mossa S., Zaccarelli E. & Tartaglia P. Equilibrium cluster phases and low-density arrested disordered states: The role of short-range attraction and long-range repulsion. Phys. Rev. Lett. 93, 055701 (2004). [DOI] [PubMed] [Google Scholar]
  4. Tlusty T. & Safran S. A. Defect-induced phase separation in dipolar fluids. Science 290, 1328–1331 (2000). [DOI] [PubMed] [Google Scholar]
  5. Russo J., Tavares J. M., Teixeira P. I. C., Telo da Gama M. M. & Sciortino F. Reentrant phase diagram of network fluids. Phys. Rev. Lett. 106, 085703 (2011). [DOI] [PubMed] [Google Scholar]
  6. Angioletti-Uberti S., Mognetti B. M. & Frenkel D. Re-entrant melting as a design principle for DNA-coated colloids. Nature Mater. 11, 518–522 (2012). [DOI] [PubMed] [Google Scholar]
  7. Jones M. R. & Mirkin C. A. Materials science: Self-assembly gets new direction. Nature 491, 42–43 (2012). [DOI] [PubMed] [Google Scholar]
  8. Wang Y. et al. Colloids with valence and specific directional bonding. Nature 491, 51–55 (2012). [DOI] [PubMed] [Google Scholar]
  9. Seeman N. C. DNA in a material world. Nature 421, 427–430 (2003). [DOI] [PubMed] [Google Scholar]
  10. Condon A. Designed DNA molecules: principles and applications of molecular nanotechnology. Nature Rev. Genet. 7, 565–575 (2006). [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Pawar A. B. & Kretzschmar I. Patchy particles by glancing angle deposition. Langmuir 24, 355–358 (2008). [DOI] [PubMed] [Google Scholar]
  12. Romano F. & Sciortino F. Patterning symmetry in the rational design of colloidal crystals. Nat. Commun. 3, 975 (2012). [DOI] [PubMed] [Google Scholar]
  13. Totosaus A., Montejano J. G., Salazar J. A. & Guerrero I. A review of physical and chemical protein-gel induction. Int. J. Food Sci. Tech. 37, 589–601 (2002). [Google Scholar]
  14. Bianchi E., Largo J., Tartaglia P., Zaccarelli E. & Sciortino F. Phase diagram of patchy colloids: Towards empty liquids. Phys. Rev. Lett. 97, 168301 (2006). [DOI] [PubMed] [Google Scholar]
  15. Bianchi E., Tartaglia P., La Nave E. & Sciortino F. Fully solvable equilibrium self-assembly process: Fine-tuning the clusters size and the connectivity in patchy particle systems. J. Phys. Chem. B 111, 11765–11769 (2007). [DOI] [PubMed] [Google Scholar]
  16. Rovigatti L. & Sciortino F. Self and collective correlation functions in a gel of tetrahedral patchy particles. Mol. Phys. 109, 2889–2896 (2011). [Google Scholar]
  17. Russo J., Tartaglia P. & Sciortino F. Reversible gels of patchy particles: Role of the valence. J. Chem. Phys. 131, 014504 (2009). [DOI] [PubMed] [Google Scholar]
  18. Saw S., Ellegaard N. L., Kob W. & Sastry S. Structural relaxation of a gel modeled by three body interactions. Phys. Rev. Lett. 103, 248305 (2009). [DOI] [PubMed] [Google Scholar]
  19. Saw S., Ellegaard N. L., Kob W. & Sastry S. Computer simulation study of the phase behavior and structural relaxation in a gel-former modeled by three-body interactions. J. Chem. Phys. 134, 164506 (2011). [DOI] [PubMed] [Google Scholar]
  20. Binder K. & Kob W. Glassy materials and disordered solids (World Scientific, Singapore, 2011). [Google Scholar]
  21. Barcenas M., Douda J. & Duda Y. Temperature dependence of the colloidal agglomeration inhibition: Computer simulation study. J. Chem. Phys. 127, 114706 (2007). [DOI] [PubMed] [Google Scholar]
  22. Kern N. & Frenkel D. Fluid–fluid coexistence in colloidal systems with short-ranged strongly directional attraction. J. Chem. Phys. 118, 9882–9889 (2003). [Google Scholar]
  23. Mao X., Chen Q. & Granick S. Entropy favours open colloidal lattices. Nature Mater. 12, 217222 (2013). [DOI] [PubMed] [Google Scholar]
  24. Romano F. & Sciortino F. Two dimensional assembly of triblock janus particles into crystal phases in the two bond per patch limit. Soft Matter 7, 5799–5804 (2011). [Google Scholar]
  25. Giacometti A., Romano F. & Sciortino F. in Janus Particles Synthesis, Self-Assembly, and Applications, edited by Granick, S. & Jiang, S. (RCS Publishing, London, 2012). [Google Scholar]
  26. Rapaport D. C. The event-driven approach to n-body simulation. Progress of Theoretical Physics Supplement 178, 5 (2009). [Google Scholar]
  27. de la Peña L. H., van Zon R., Schofield J. & Opps S. B. Discontinuous molecular dynamics for semiflexible and rigid bodies. J. Chem. Phys 126, 074105 (2007). [DOI] [PubMed] [Google Scholar]
  28. Smallenburg F. & Sciortino F. Liquids more stable than crystals. Nature Physics; 10.1038/NPHYS2693 (2013). [Google Scholar]
  29. De Michele C., Gabrielli S., Tartaglia P. & Sciortino F. Dynamics in the presence of attractive patchy interactions. J. Phys. Chem. B 110, 8064–8079 (2006). [DOI] [PubMed] [Google Scholar]
  30. Flory P. J. Principles of Polymer Chemistry (Cornell University Press, 1953). [Google Scholar]
  31. Wertheim M. S. Fluids with highly directional attractive forces. i. statistical thermodynamics. J. Stat. Phys. 35, 19 (1984). [Google Scholar]
  32. Wertheim M. S. Fluids with highly directional attractive forces. ii. thermodynamic perturbation theory and integral equations. J. Stat. Phys. 35, 35–47 (1984). [Google Scholar]
  33. de las Heras D., Tavares J. M. & Telo da Gama M. M. Bicontinuous and mixed gels in binary mixtures of patchy colloidal particles. Soft Matter 8, 1785–1794 (2012). [Google Scholar]
  34. Del Gado E. & Kob W. Structure and relaxation dynamics of a colloidal gel. Europhys. Lett. 72, 1032–1038 (2005). [Google Scholar]
  35. Del Gado E. & Kob W. Length-scale-dependent relaxation in colloidal gels. Phys. Rev. Lett. 98, 028303 (2007). [DOI] [PubMed] [Google Scholar]
  36. Stewart K. M. & McLaughlin L. W. Four-arm oligonucleotide ni(ii)cyclam-centered complexes as precursors for the generation of supramolecular periodic assemblies. J. Am. Chem. Soc. 126, 2050–2057 (2004). [DOI] [PubMed] [Google Scholar]
  37. Biffi S. et al. Phase behaviour and critical activated dynamics of limited-valence DNA nano-stars. Proceeding National Academy of Science, in press. [DOI] [PMC free article] [PubMed]
  38. Lebowitz J. L. & Rowlinson J. S. Thermodynamic properties of mixtures of hard spheres. J. Chem. Phys. 41, 133–138 (1964). [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Information

Gelling by Heating: Supplementary Information

srep02451-s1.pdf (86.2KB, pdf)

Articles from Scientific Reports are provided here courtesy of Nature Publishing Group

RESOURCES