Hauke
Carstensen
,
Anne
Krämer
,
Vassilios
Kapaklis
and
Max
Wolff
*
Department of Physics and Astronomy, Uppsala University, Box 516, SE-75120, Uppsala, Sweden. E-mail: max.wolff@physics.uu.se
First published on 18th July 2022
We study the self-assembly of branching-chain networks and crystals in a binary colloidal system with tunable interactions. The particle positions are extracted from microscopy images and order parameters are extracted by image processing and statistical analysis. With these, we construct phase diagrams with respect to particle density, ratio and interaction. In order to draw a more complete picture, we complement the experiments with computer simulations. We establish a region in the phase diagram, where bead ratios and interactions are symmetric, promoting percolated structures.
Colloidal systems consist of microscopic beads dissolved in a carrier liquid. Typically, the size of these beads is large and thermal energy becomes negligible. The systems’ phase behaviour is defined by the interactions between the beads, making them interesting for the study of phase formation and self-assembly. By coating the bead surface with polymers, soft steric interactions may occur.4 More long-ranged interactions that are possible to realize are Coulomb interactions. The self-assembly of charged colloids was studied for beads of different sizes,5 aspect ratios6 and opposite charge.7 Even more rich in behaviour are magnetic interactions, since these are of dipolar nature and may be altered by external magnetic fields.8 The anisotropy of the interaction leads to the formation of clusters in the form of rings or chains with a head to tail orientation of the individual particles.9 With decreasing temperature or increasing density the size of such clusters increases and networks may form. These are challenging to study for nanometer sized particles as frozen samples are required for studies with transmission electron microscopy10 or computer simulations.11,12 For larger particles, the assembly process is easier to follow13 and rings, chains and small networks were reported. Illustrative in this context is the study of cm-sized particles with different magnetic moments and their pattern formation and segregation.14,15 It was shown that by shaking and quenching magnetic particle systems larger open networks may be formed.16,17 All the above mentioned studies were performed without an externally applied magnetic field. Time dependent magnetic fields, however, allow precise tuning of the interactions18 and control of the self-assembly process19 as well as the study of the kinetics of self-assembly20 and even the motion of self-assembled magnetic surface swimmers.21 By dispersing the magnetic beads in a ferrofluid matrix, an effective moment may be introduced, leading to tunable particle interactions. The local self-assembly of such systems was studied22 and a phase diagram of the local coordination was presented.23 For two dimensional systems with an in plane magnetic field, depending on the interaction between the beads, cubic or hexagonal phases may form.24
In certain applications, magnetic liquids are applied, for example, in loud speakers,25 tunable photonic26–28 or plasmonic structures29 and tunable shock absorbers.30 These applications require knowledge on phase formation, not only on the local length-scales but also global, e.g. the formation of clusters or extended networks. On intermediate length-scales, colloidal crystals or networks may form, for which micrometer-sized particles allow tuning of the viscosity in the so-called magneto-rheological materials.31,32
An important concept in this context is jamming, which allows relating changes in macroscopic material properties, e.g. viscosity, to topological constrains.33 In the case of attractive interactions a phase diagram of jamming has been established34 and it was shown that the jamming transition in systems with attractive and repulsive interactions fall into different universality classes.35,36 Similar to jamming, percolation may affect the macroscopic properties of a system. In the percolated state, all parts of a system are connected to each other. The jamming and percolation transition are characterised by critical densities.37 The critical density for percolation was calculated for different coordinations in two and three dimensions38 and is very similar to jamming happening at a critical random-close-packing density.33 The influence of a long range repulsion and short range attraction on the percolation transition was investigated systematically using Monte-Carlo simulations.39 These show that long-ranged repulsion increases the average cluster size and lowers the volume fraction for the percolation threshold.
In this communication, we report on the self-assembly and percolation of a binary, two dimensional system of magnetic beads with tunable interactions in an out-of-plane magnetic field. We classify the cluster formation with respect to the coordination of the individual beads, the crystal symmetry and percolation. We find that the critical density for percolation depends on the interaction between the beads and the bead ratio and report the lowest critical density for a bead ratio of 1:1 with attraction between the two species of beads and repulsion among beads of the same type.
In the limit, χM ≪ 1 and χFF ≪ 1, with χM and χFF being the susceptibilities of the magnetic beads and the ferro-fluid, respectively, the effective bead susceptibilities χM,eff and χN,eff, for the magnetic and non-magnetic beads, respectively, are:22–24,43
χM,eff ∝ χM − χFF | (1) |
χN,eff ∝ −χFF | (2) |
If the beads are confined along one dimension and a magnetic field is applied along the confinement direction, the interaction between the beads can be classified as follows:
• The interaction between identical beads is repulsive but tunable in strength, by changing the concentration of magnetic nano-particles in the ferrofluid.
• For χM,eff = 0 and χN,eff = 0 the magnetic and non-magnetic beads, respectively, are non-interacting.
The interaction between the two types of beads can be further modified in the following ways:24
• For χN,eff < 0 < χM,eff the coupling is anti-ferromagnetic and an attractive force is present between magnetic and non-magnetic beads.
• For χM,eff < 0 and χN,eff < 0 the interaction between all particles is repulsive.
This tunable interaction allows the systematical study of the structure formation in this binary system, with the ground states in the local ordering having been established previously.23,24,42,44 Here, we focus on the structure formation on larger length-scales and percolation phenomena.
The samples were prepared by mixing 20 μL of ferrofluid, 5 μL (1% w/w) of magnetic bead dispersion and 1 μL (2.5% w/w) of non-magnetic bead dispersion. This results in approximately equal numbers of magnetic and non-magnetic beads present in the sample. However, the beads may be inhomogeneously distributed and the local densities and ratios vary spatially within each sample. This enables the systematical study of the effect of density and bead ratio on self-assembly. To vary the interaction between the microbeads, a series of samples with varying FF concentrations were prepared. All of the samples were confined between two glass slides separated by an oil-covered 25 μm thick Mylar ring and studied by transmission optical microscopy. An out-of-plane magnetic field, with a maximum strength of 14 mT, was applied using a pair of Helmholtz coils. The magnetic flux density was calculated with finite element methods, implemented in FEMM.24,45Fig. 1(a) shows a sketch of the experimental setup and the upper right panel results of the field calculation, indicating that the magnetic field is homogeneous over the area of the sample. The sample was loaded into the sample cell at zero field and all microscopy images were taken at an out-of-plane field of 0.14 mT. After loading the samples and before acquiring the images with the microscope, the field is cycled (switched on and off) repeatedly in order to reach an equilibrium structure. Subsequently, microscope images are recorded and bead positions are automatically extracted, identifying circular structures using image processing routines in MATLAB. The bead types are distinguished by their apparent brightness in the images. In addition, automatic detection was made more robust, utilizing a neural network algorithm trained on three images with labeled data (bead positions and types).
To complement the experiments, we performed computer simulations. Magnetic and non-magnetic beads are placed sequentially at random positions (without overlaps). The size of the simulated rectangular area was chosen to match the size of a typical microscopy image (50 × 37 bead diameters) and periodic boundary conditions were applied. We further assume zero temperature, justified by the large size of the beads and the negligible Brownian motion as well as significant viscosity of the ferrofluid (zero bead momentum), no inertia and an overdamped equation of motion. The beads’ magnetic moment is described as out-of-plane dipoles. To avoid overlapping of particles a hard sphere potential was used and particles are placed next to each other once they touch. In each simulation step the bead displacements are calculated from the force acting upon each bead, resulting from the presence of all other beads, according to:
(3) |
Fig. 2 Microscope images (top row) and simulated data (bottom row) for varying bead densities ρ, bead ratios r, and interactions χFF/χM. |
To analyse the coordination of individual beads in more detail, Fig. 3(d) shows the symmetry around beads. Blue, green and red colors indicate regions in which predominantly 6-fold, 5-fold and 4-fold symmetry is found, respectively. The intensity of the color represents how often the respective symmetry is found and quantified by the bond order parameter46 for an s-fold symmetry, defined as , for each bead j having 3 or more neighbors, where θk are the angles of the connecting lines to direct neighbors, with respect to a reference vector. For beads with less than 3 neighbors, the parameter is zero (ϕj = 0). The average over all beads within one image is Φs = 〈|ϕj|〉 and is presented in the colormap. For the apparent moments of the beads being almost equal but of opposite direction four-fold symmetries are found. For highly asymmetric moments hexagonal packing is dominant. Interestingly the symmetry is almost independent of the bead ratio, as individual beads, which are left over, remain isolated. This relates well to the fact that large bond order parameters are found in regions of low average dipole moments. Another interesting observation is the five-fold symmetry found in the intermediate region. This symmetry hinders the formation of large crystals and should promote the formation of open networks and percolated structures. Fig. 3(b) depicts the percolation plotted as a map with respect to the bead ratio and relative susceptibility. To extract a percolation order parameter the average cluster size is weighted by the number of beads in each cluster, to become statistically more robust and is normalized to the total number of beads. This quantity can be used as an order parameter and varies between zero (isolated beads) and one (all beads connected). Clearly, highly asymmetric bead ratios do not form percolated structures, which are predominantly found for bead ratios of 1:1 and 1:2.
Fig. 3 presents results for a bead density of 38.2% but data have been evaluated for bead densities between 12.7% and 50.9%. Fig. 4 depicts the percolation order parameter plotted on a grey-scale versus bead density and interaction, for ratios of 1:1 (panel a) and 1:2 (panel b). The figure was created from simulations. Regions of predominantly hexagonal and cubic symmetry are identified for bond order parameters Φ6 and Φ4 larger than 0.4, indicated by the blue and red colors, respectively. Fig. 4(c) depicts simulation images extracted for χFF/χM = 0.5 at three densities, indicated by the purple circles in Fig. 4(a). For low densities, clusters and/or chains form but do not connect to each other, as their mass is large and their total moment compensated. At higher densities crystallites form and become connected. Fig. 4(d) depicts the simulation images extracted for a density of 38.2% and three values of χFF/χM indicated by the green bars in Fig. 4(b). It turns out that in the intermediate region, where neither cubic nor hexagonal symmetry dominates, open percolated structures form, promoted by the large defect density (centre panel, Fig. 4(d)). The transition to percolated structures with respect to density is relatively sharp for all interaction parameters (Fig. 4(a and b)), allowing the extraction of the critical density for percolation, which is plotted versus χFF/χM in Fig. 5. Bead ratios of 1:1, 1:2 and 1:3 are marked by purple red and green colors, respectively.
Fig. 5 Critical densities for percolation plotted versus interaction of magnetic and non-magnetic beads for bead ratios of 1:1, 1:2 and 1:3. |
The critical density for percolation has been calculated for triangles, squares, honeycomb and kagome lattices in two dimensions38 and is in all cases close to 45%. For a bead ratio of 1:1 we find a lower value of the critical density for percolation, of about 30%, for χFF/χM = 0.5 (top right panel, Fig. 5). Here the interaction between the two species of beads is attractive and one type acts as glue for the other. The fact that the lowest critical density for percolation is found for values of χFF/χM < 0.5 may be related to crystal tilling effects. At χFF/χm = 0.5 large cubic (ρ ≈ 79%) crystals form leading to larger critical densities for percolation, as it critically depends on the links between crystals. At small values of χFF/χM, the critical density for percolation becomes large (>60%), as the repulsion between magnetic beads is dominating. A qualitatively similar behaviour is found for a bead ratio of 1:2, however, with the lowest value of the critical density for percolation found at lower χFF/χM. For a bead ratio of 1:3 percolation is only found for densities larger than 50%. The strong repulsion between the magnetic beads requires three non-magnetic beads to compensate, favouring hexagonal dense packed (ρ > 90%) arrangements.
This journal is © The Royal Society of Chemistry 2022 |