Mode-coupling theory (MCT) predicts arrest of colloids in terms of their volume fraction, and the range and depth of the interparticle attraction. We discuss how effective values of these parameters evolve under cluster aggregation. We argue that weak gelation in colloids can be idealized as a two-stage ergodicity breaking: first at short scales (approximated by the bare MCT) and then at larger scales (governed by MCT applied to clusters). The competition between arrest and phase separation is considered in relation to recent experiments. We predict a long-lived 'semi-ergodic' phase of mobile clusters, showing logarithmic relaxation close to the gel line.
PACS numbers:Hard-sphere colloids with short-range attractions can undergo several types of arrest. At high densities they show two distinct glass transitions (repulsion-driven and attraction-driven), with a re-entrant dependence on attraction strength [1]. This scenario was first predicted by mode coupling theory (MCT) [2,3,4], and depends on both the attraction range δ (in units of particle diameter) and well-depth ε (in units of k B T ). MCT is remarkably successful, at least for large volume fractions φ 0.4.At lower volume fractions, however, there is no comparable theoretical framework. Yet 'weak gelation', in which bonding is strong but not so strong as to be irreversible, can lead to nonergodic soft solids, of nonzero static elastic modulus, at volume fractions of just a few percent [5]. It might be argued that a finite modulus requires a percolating network of bonds whose lifetime exceeds that of the experiment. However this is simplistic: as shown by the case of repulsive glasses, a finite modulus can arise with no bonding at all. We argue here that the rigidity of weak gels arises not from bond percolation but from kinetic ergodicity breaking [6], just as it does in glass formation. This suggests that an MCT-like approach to weak gelation could be fruitful.MCT takes its structural input from equilibrium liquid state theory; it cannot address states of arrest where this structure is strongly perturbed [7]. This matters relatively little at φ 0.4, where each particle interacts with many others, and not much room is left for structural development upon a quench. But more severe consequences must be expected at low volume fractions where strongly nonuniform, ramified gels arise. Here the pathway to complete nonergodicity (starting from a homogenized fluid sample, say) must involve a nontrivial episode of structure formation, akin to irreversible cluster aggregation (ICA). Such kinetics certainly dominates for irreversible ("strong") gelation (ε −1 = 0), where particles aggregate on contact into clusters, with various kinetic universality classes [8]. Relative simplicity is restored at low φ thanks to the invariance of this aggregation process under coarse graining in the (ordered) limit δ, ε −1 → 0 and φ → 0. This scaling limit is controlled by an ICA 'fixed point', where details of the short-ranged attraction are irrelevant. The resulting fractal cluster...