The creation of three-dimensional, crystalline-ordered nanoparticle (NP) structures linked by DNA has proved experimentally challenging. Here we aim to systematically study parameters that influence the relative thermodynamic and kinetic stability of such crystals. To avoid experimental bottlenecks and directly control molecular-scale parameters, we carry out molecular dynamics simulations of a coarsegrained model in which short DNA strands (6 to 12 bp) are tethered to a NP core. We examine the influence of the number of bases per strand L, number of linking bases l and the number of spacer bases s on the stability of crystal states. We also consider the effect of using a single linking NP type versus a binary linking system. We explicitly compute the free energy, entropy, and melting point T M for BCC and FCC lattices. We show that binary systems are preferable for generating BCC lattices, while a single NP type generates the most stable FCC crystals. We propose a simple model for short DNA strands that can account for T M of all our data. The model also indicates that the heat of fusion between crystal and amorphous phases grows linearly with l, providing a route to maximize the relative crystal stability.

Three-dimensional ordered lattices of nanoparticles (NPs) linked by DNA have potential applications in novel devices and materials, but most experimental attempts to form crystals result in amorphous packing. Here we use a coarse-grained computational model to address three factors that impact the stability of bcc and fcc crystals formed by DNA-linked NPs : (i) the number of attached strands to the NP surface, (ii) the size of the NP core, and (iii) the rigidity of the strand attachment. We find that allowing mobility in the attachment of DNA strands to the core NP can very slightly increase or decrease melting temperature T(M). Larger changes to T(M) result from increasing the number of strands, which increases T(M), or by increasing the core NP diameter, which decreases T(M). Both results are consistent with experimental findings. Moreover, we show that the behavior of T(M) can be quantitatively described by the model introduced previously [F. Vargas Lara and F. W. Starr, Soft Matter, 7, 2085 (2011)].

No abstract