Water directs the self-assembly of both natural1,2 and synthetic3,4,5,6,7,8,9 molecules to form precise yet dynamic structures. Nevertheless, our molecular understanding of the role of water in such systems is incomplete, which represents a fundamental constraint in the development of supramolecular materials for use in biomaterials, nanoelectronics and catalysis10. In particular, despite the widespread use of alkanes as solvents in supramolecular chemistry11,12, the role of water in the formation of aggregates in oils is not clear, probably because water is only sparingly miscible in these solvents—typical alkanes contain less than 0.01 per cent water by weight at room temperature13. A notable and unused feature of this water is that it is essentially monomeric14. It has been determined previously15 that the free energy cost of forming a cavity in alkanes that is large enough for a water molecule is only just compensated by its interaction with the interior of the cavity; this cost is therefore too high to accommodate clusters of water. As such, water molecules in alkanes possess potential enthalpic energy in the form of unrealized hydrogen bonds. Here we report that this energy is a thermodynamic driving force for water molecules to interact with co-dissolved hydrogen-bond-based aggregates in oils. By using a combination of spectroscopic, calorimetric, light-scattering and theoretical techniques, we demonstrate that this interaction can be exploited to modulate the structure of one-dimensional supramolecular polymers.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This project received funding from the European Union’s Horizon 2020 research and innovation program (no. 705701) and the Dutch Ministry of Education, Culture and Science’s Gravitation program (no. 024.001.035). The X-ray diffractometer was financed by the Netherlands Organization for Scientific Research. N.J.V.Z. acknowledges M. L. Ślęczkowski for synthesizing the chiral amine precursor for 1, B. F. M. de Waal for assistance with Karl Fischer titrations and R. A. A. Bovee for performing MALDI–TOF MS measurements. We also thank S. C. J. Meskers, R. J. M. Nolte and A. J. Markvoort for discussions.Reviewer information
Nature thanks C. Hunter, D. Miyajima and D. Pantos for their contribution to the peer review of this work.
Extended data figures and tables
a, DSC trace of 1 (cooling in blue, second heating in red). b, Polarized optical microscopy image of 1 with crossed polarizers at 135 °C after cooling from the isotropic melt. c, Wide-angle X-ray scattering trace of bulk 1 at room temperature (top) with a schematic of the hexagonal columnar morphology (bottom left) and tabulated parameters (bottom right). d, CD signal (top) and absorbance (bottom) of a thin film of 1 at 20 °C. e, FTIR spectrum of bulk 1 at 20 °C after cooling from the isotropic melt. f, Comparison of the FTIR spectra of bulk 1 after cooling from the isotropic melt and A ( = 2.0 mM in dry MCH, labels in cm−1).
a, Supramolecular fibres of 1 stained with Cage-552 photoactivatable dye imaged by super-resolution fluorescence microscopy (left) and corresponding thickness analysis (right). The indicated zone in the microscopy image is depicted in Fig. 1b. A discussion of the thickness analysis and a comparison to fibres of 6 is provided in the Supplementary Information. b, Supramolecular fibres of 1 imaged by AFM in non-contact tapping mode. The indicated zone is depicted in Fig. 1c.
a, Schematic of experimental design. The CD spectrometer was purged with nitrogen at a rate of 20 l min−1. b, CD signal (top) and absorbance (bottom) at 258 nm as a 30 µM solution of 1 is dried over 100 min in the sample holder of the CD spectrometer. All water content measurements are reported as mean ± s.d. (n = 2).
a, van ’t Hoff plot of ln(Ke) versus 1/Te (left) with tabulated Te data (right). b, van ’t Hoff plot of ln(Khyd,A) versus 1/TA→B. The points represent ln(Khyd,A) calculated using the mean of the water content determined for each respective measurement. The error bars correspond to the spread of ln(Khyd,A) as a result of the experimental uncertainty of each respective water content measurement. c, van ’t Hoff plot of ln(Khyd,B) versus 1/TB→C. The points represent ln(Khyd,B) calculated using the mean of the water content determined for each respective measurement. The error bars correspond to the spread of ln(Khyd,B) as a result of the experimental uncertainty of each respective water content measurement. d, Determination of TA→B and TB→C from the second derivative of the corresponding VT-CD curves presented in Fig. 1f (labels in °C).
The CD intensity was monitored at 258 nm while cooling from 95 °C to −5 °C and then immediately heating back to 95 °C with scanning rates of 15 (left), 30 (middle) and 60 °C h−1 (right). Samples were prepared with as-received MCH ( = 30 µM, [H2O] = 35 ± 2 p.p.m.).
A 0.51 µM solution of 1 in as-received MCH was characterized by VT-CD spectroscopy (top), micro-DSC (middle) and light scattering (bottom, mean ± s.d. (n = 5) are shown). In the micro-DSC plot, only the endothermic transitions corresponding to B→A with scan rates of 15 and 30 °C h−1 had baselines suitable for integration (labels in kJ (mol 1)−1).
a, Chemical structure of 2. b, Displacement ellipsoid plot (50% probability level) of 2 in the crystal. C–H hydrogen atoms and chloroform solvent molecules are omitted for clarity. Only one of two independent molecules is shown. The other independent molecule is located on an inversion centre. c, Packing of 2 in the crystal. The two independent molecules are shown in black and red, respectively. Hydrogen atoms and chloroform solvent molecules are omitted for clarity. The structure shows pseudo-translational symmetry in the b-direction.
Extended Data Fig. 8 Molecular model of water binding to an aggregate of biphenyl tetracarboxamide molecules.
a, Chemical structure of 3. b, Molecular models based on density functional theory calculations for the incorporation of four water molecules into a hexameric aggregate of 3. Hydrogen atoms, apart from those engaged in hydrogen bonding, are omitted for clarity. The structures are colour coded as follows: hydrogen bond, dashed lines; carbon, black; oxygen, red; nitrogen, blue; water molecules, green.
a, CD signal (top), ultraviolet absorbance (middle) and light-scattering counts (bottom, mean ± s.d. (n = 5) are shown) acquired while cooling solutions of 6 in wet, as-received or dry MCH. b, Typical AFM picture (left) and height profiles (right) of a sample of 6 (30 µM in wet MCH) that was drop-cast on mica in a water-saturated environment. The indicated zone is depicted in Fig. 4d. c, Typical AFM picture (left) and height profiles (right) of a sample of 6 (30 µM in dry MCH) that was drop-cast on mica under dry conditions in a glovebox. The indicated zone is depicted in Fig. 4e.
This file contains Supplementary Figures 1-13 and Supplementary Table 1.