Dense packings of freely jointed chains of tangent hard spheres are produced by a novel Monte Carlo method. Within statistical uncertainty, chains reach a maximally random jammed (MRJ) state at the same volume fraction as packings of single hard spheres. A structural analysis shows that as the MRJ state is approached (i) the radial distribution function for chains remains distinct from but approaches that of single hard sphere packings quite closely, (ii) chains undergo progressive collapse, and (iii) a small but increasing fraction of sites possess highly ordered first coordination shells. DOI: 10.1103/PhysRevLett.100.050602 PACS numbers: 05.20.ÿy, 61.20.ÿp Maximally random jammed (MRJ) packings of identical spheres has been the subject of extensive analytical and numerical work, and a considerable body of knowledge has been collected over the last two decades (see [1][2][3] and references therein). It seems, however, that work has been concentrated on packings of single spheres while dense packings of chains have received comparatively little attention [4 -6]. In this Letter we present numerical results about the packing of freely jointed chains of tangent hard spheres at volume fractions very close to the MRJ state [2] in three-dimensional Euclidean space. The systems investigated consisted of ensembles of freely jointed chains of tangent hard spheres of unit diameter and lengths N 12 or N 24 with a total of 1200 interaction sites. Both N 12 and N 24 lie deep in the infinite-chain asymptotic regime regarding packing and local structure, although chain dimensions have not reached asymptotic behavior. Persistence length, computed as hR 2 i=2N 0:5 [7] at ' 0:10 is 1.59 for N 12 and 1.80 for N 24.Several algorithms for packing single spheres exist, some of them able to generate random packings up to the MRJ density [8] ' ss 0:64 (subindices ss refer to single spheres, hsc to hard sphere chains). Although computationally demanding, recent algorithms can produce very dense packings with such efficiency that extremely large packings of up to a million spheres, necessary to investigate some subtle structural features, are within reach [9]. It is somewhat surprising that packings of hard-sphere chains have received little attention, although they represent a valuable departing point for perturbation work [10], in addition to having richer structure than single spheres. The main hurdle is the substantial computational difficulty associated with the generation and relaxation of packings of hard spheres which are close to the MRJ state and simultaneously satisfy the holonomic constraints defining chain connectivity. Molecular dynamics (MD) methods are notoriously inefficient at compacting ensembles of linear chains because of increasingly sluggish dynamics and large relaxation times as chain length increases. Monte Carlo (MC) schemes have been shown [11] [12]). However, by using a carefully chosen combination and modification of several state-of-the-art MC algorithms for efficient structural relaxation of both large ...