The temperature dependent structure of liquid 1-propanol as studied by neutron diffraction and EPSR simulations.

The structure of liquid 1-propanol is investigated as a function of temperature using neutron diffraction together with Empirical Potential Structure Refinement modelling. The combined diffraction and computer modelling analysis demonstrates that propanol molecules form hydrogen bonded clusters with a relatively wide size distribution, which broadens at lower temperatures. We find that the cluster size distribution is well described by a recently proposed statistical model for branched H-bonded networks [P. Sillrén, J. Bielecki, J. Mattsson, L. Börjesson, and A. Matic, J. Chem. Phys. 136, 094514 (2012)]. The average cluster size increases from ~3 to 7 molecules, whilst the standard deviation of the size distribution increases from 3.3 to 8.5 as the temperature is decreased from 293 to 155 K. The clusters are slightly branched, with a higher degree of branching towards lower temperatures. An analysis of the cluster gyration tensor (R(mn)) reveals an average elongated ellipsoidal shape with axes having proportions 1:1.4:1.9. We find that the average radius of gyration has a cluster size dependence consistent with that of fractal clusters, R(g) ∝ n(1/D), with a fractal dimension D ≈ 2.20, which is close to D = 2.00 expected for an ideal random walk or D = 2.11 expected for reaction limited aggregation. The characteristic angles between the H-bonded OH-groups that constitute the clusters show only a weak temperature dependence with O-H···O angles becoming more narrowly distributed around 180° at lower temperatures.

A statistical model of hydrogen bond networks in liquid alcohols.

We here present a statistical model of hydrogen bond induced network structures in liquid alcohols. The model generalises the Andersson-Schulz-Flory chain model to allow also for branched structures. Two bonding probabilities are assigned to each hydroxyl group oxygen, where the first is the probability of a lone pair accepting an H-bond and the second is the probability that given this bond also the second lone pair is bonded. The average hydroxyl group cluster size, cluster size distribution, and the number of branches and leaves in the tree-like network clusters are directly determined from these probabilities. The applicability of the model is tested by comparison to cluster size distributions and bonding probabilities obtained from Monte Carlo simulations of the monoalcohols methanol, propanol, butanol, and propylene glycol monomethyl ether, the di-alcohol propylene glycol, and the tri-alcohol glycerol. We find that the tree model can reproduce the cluster size distributions and the bonding probabilities for both mono- and poly-alcohols, showing the branched nature of the OH-clusters in these liquids. Thus, this statistical model is a useful tool to better understand the structure of network forming hydrogen bonded liquids. The model can be applied to experimental data, allowing the topology of the clusters to be determined from such studies.