Structural Polymorphism as the Result of Kinetically Controlled Self

Res. , Article ASAP. DOI: 10.1021/acs.accounts.7b00522. Publication Date (Web): January 30, 2018 ... (9) The unifying theme of these strategies is to ...
12 downloads 10 Views 3MB Size
Article Cite This: Acc. Chem. Res. XXXX, XXX, XXX−XXX

pubs.acs.org/accounts

Structural Polymorphism as the Result of Kinetically Controlled SelfAssembly Ryan D. Brown, Steven A. Corcelli, and S. Alex Kandel* Department of Chemistry and Biochemistry, University of Notre Dame, Notre Dame, Indiana 46556, United States CONSPECTUS: Traditionally, the goal of self-assembly and supramolecular chemistry is to engineer an equilibrium structure with a desired geometry and functionality; this is achieved through careful choice of molecular monomers, growth conditions, and substrate. Supramolecular assemblies produced under nonequilibrium conditions, in contrast, can form metastable structures with conformations quite different from those accessible in equilibrium self-assembly. The study of nonequilibrium growth of clusters potentially impacts the study of nucleation in atmospheric aerosols, nucleation in organic crystallization, and mesoscale organization for systems ranging from biological molecules to molecular electronics. In our experiments, we prepare surface monolayers of small organic and organometallic molecules through direct injection of a solution onto a substrate in high vacuum. During this process, the rapid evaporation of small solution droplets in high vacuum can lead to nonequilibrium growth conditions. The resulting structures are then characterized by scanning tunneling microscopy. Among the features observed in these experiments are cyclic, hydrogen-bonded pentamers. For carboxylic acids, the two-molecule ring dimer is the common binding motif. Large, cyclic hydrogen-bonded systems are uncommon, especially so for rings with five members. Despite this, pentagonal clusters appear to be a general phenomenon for systems containing adjacent strong and weak hydrogen-bonding elements on five-member aromatic rings. Regular pentamers have been observed as metastable structures for ferrocenecarboxylic acid, indole-2-carboxylic acid, and isatin (1-H-indole-2,3-dione). Electronic structure calculations confirm the relative stability of these structures with respect to the dimer or catemer conformations which are observed in the solid-state crystal structures. For ferrocenecarboxylic acid, cyclic pentamers undergo further self-assembly, resulting in long-range order in conjunction with local 5-fold rotational symmetry. This system is the first reported self-assembled molecular quasicrystal, and it remains the only example of a hydrogen-bonded quasicrystal. This supramolecular structure forms as a result of the cocrystallization of hydrogenbonded cyclic pentamers with intercalated molecular dimers. The shared bonding to a single dimer is responsible for locking the adjacent pentamers in specific distances and orientations, which produces the quasicrystal. Careful analysis of experimental data provides evidence that, in some cases, metastable clusters are formed in solution and then subsequently adsorb on the surface. This is a unusual mechanism for supramolecular assembly, and it has important implications for understanding questions in crystal growth, namely: what the initial stages of crystal growth are as molecules are first precipitating from solution; what role the solvent plays in determining crystal structure; and whether solvent-mediated clustering is important in the broader phenomenon of solid-state polymorphism.

1. INTRODUCTION The study of self-assembly as a growth process has led to the ability to produce nanostructured materials using a variety of organizing principles.1−4 These strategies have been used to produce metal−organic network materials,5 ordered organic molecular structures,6−8 and ordered nanoparticle arrays.9 The unifying theme of these strategies is to design a system in which the intermolecular interactions produce an equilibrium state with the desired structure. The question of how nonequilibrium growth conditions impacts self-assembly at an interface touches on our fundamental understanding of intermolecular interactions, our ability to control structure on the nanoscale, and polymorphism in crystal structures.10−14 Growth in nonequilibrium conditions can lead to kinetically controlled structure formation, which can allow for the production of metastable features if kinetic barriers prevent access to the most thermodynamically favorable supramolecular conformation. An improved under© XXXX American Chemical Society

standing of growth under nonequilibrium conditions impacts fields ranging from the organization of biomolecules15 to the development of new materials for molecular electronics,16,17 and the study of cluster formation potentially impacts nucleation of atmospheric aerosol particles18 and organic crystals.19 One strategy for studying self-organization is to investigate the formation of metastable clusters of small molecules, specifically using systems in which there is a welldefined hierarchy of intermolecular interactions. Hydrogen bonding comprises a diverse range of potential interactions,20 and hydrogen bonds are common in crystal engineering due to their relatively high bond energies and strong directionality. While one might expect supramolecular structures to adopt configurations maximizing these strong interactions, studies of both bulk crystal structures and twoReceived: October 20, 2017

A

DOI: 10.1021/acs.accounts.7b00522 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research

Figure 1. (a) A 470 Å × 555 Å STM image of a quasicrystalline region of FcCOOH, with the orientation of each pentagonal cluster colorized. The tick marks on the color wheel axis indicate the orientations of each pentamer in the image. (b) High-resolution STM with the molecular model for the pentamers superimposed on the image. The pentagonal cluster consists of FcCOOH molecules with the Cp rings parallel to the substrate, and the pentamers are locked in place by FcCOOH dimers with their Cp rings perpendicular to the surface plane. (c) A 205 Å × 205 Å STM image of FcCH2COOH lacking in any pentagonal clusters. (d) Calculated total cohesion energy of two FcCOOH molecules as a function of the COH bond angle (red) compared with that for formic acid (blue). Adapted with permission from ref 27. Copyright 2014 Nature.

structures in conditions far from equilibrium.25,26 While there are variables not addressed in these studies (including initial concentration, droplet size, solubility, and solvent boiling point) associated with using pulse-deposition as a preparation method, it is sufficient for producing supramolecular structures at the vacuum−solid interface far from equilibrium. Carefully iterating the chemical functional groups between experiments, performing annealing experiments, and comparing results to electronic structure calculations allow for the production of models of intermolecular bonding and the determination of the relative stability of observed molecular clusters.

dimensional assemblies at surfaces instead reveal a rich array of possible structures, with the appearance of metastable species dependent on the growth conditions.21−23 Since metastable species are kinetically trapped local minima in the global potential energy surface for supramolecular conformations, growth under nonequilibrium conditions can be utilized to produce and observe these species. Scanning tunneling microscopy (STM) is a natural technique for studying the formation of heterogeneous surface structures.24 The local nature of this probe can be used to study heterogeneous surfaces with precision not available to ensemble or reciprocal space measurements. The combination of complementary electronic structure theory calculations with STM experiments can produced a detailed understanding of the relationship between intermolecular forces and supramolecular structure. Our main strategy of using direct injection of a solution in to vacuum allows us to produce supramolecular

2. FERROCENECARBOXYLIC ACID QUASICRYSTALS The pulse-deposition of ferrocenecarboxylic acid, FcCOOH, in benzene, on Au(111) substrates is a prime example of how nonequilibrium growth techniques can produce supramolecular structures quite different from those obtained via traditional B

DOI: 10.1021/acs.accounts.7b00522 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research

Figure 2. (a) A 405 Å × 360 Å image of the quasicrystalline FcCOOH assembly which was low-pass filtered to emphasize long-range order. Pentagons are overlaid for each pentagonal cluster, with some highlighted (white) and colored for comparison with features in the Penrose P1 tiling, shown in (b). Interstitial features of the P1 tiling (star, boat, and rhombus) match interstitial features in the image. (c) Two-dimensional spatial correlation function of the quasicrystalline domain from (a). (d) One-dimensional section of the spatial correlation function (blue line in panel c) is on top, with its Fourier transform on bottom. The fundamental reciprocal lattice vector corresponds to a 34.7 Å wavelength, and the higher order vectors are multiples of this wavelength by powers of the golden ratio, φ. Adapted with permission from ref 27. Copyright 2014 Nature.

self-assembly processes.27 Preparation of FcCOOH monolayers produces a variety of structures, including ordered dimer domains, disordered clusters, and a region composed of pentagonal clusters. Further examination of these domains of pentagonal clusters show that they are quasi-crystalline, the first such observation for small organometallic molecules and the only one, to date, for a hydrogen-bonded assembly. However, we will first discuss the intermolecular structure of the pentagonal clusters. An ordered domain of pentamers is shown in Figure 1a; it consists of bright pentagonal pentamers with dim features between these clusters. Figure 1b gives a high-resolution image of these clusters along with a superimposed structural model of five FcCOOH molecules with their cyclopentadienyl rings (Cp) oriented parallel to the surface. The pentamer is a cyclic catemer, which we suggest is stabilized by the bifurcation of the OH···O bond with the weak CH donor on the adjacent Cp ring. Repeating this experiment with ferroceneacetic acid (FcCH2COOH), in which the carboxylic acid group is farther removed from the CH donor on the Cp ring, results in a surface comprised of FcCH2COOH dimers without pentamer formation (Figure 1c). Density functional theory (DFT) calculations of calculated binding energy vs ∠COH, Figure 1d, confirm that the presence of a weak hydrogen bonding

donor creates a shallow well for the geometry necessary for the observed pentamers. These calculations involve a strained dimer, which will underestimate the depth of this local minimum because in a cyclic catemer the dangling bonds present in the strained dimer interact with neighboring molecules. Despite this limitation, it is a useful calculation for determining the qualitative shape of the potential energy surface as a function of ∠COH. Additional DFT calculations were employed to determine the relative stability of the cyclic catemer bonding motif for cyclic clusters of three to six molecules, and to compare it to the cohesive energy of the FcCOOH dimer. The cohesive energy per molecule was the basis for comparison, and depending on the density functional used, in general the pentamer and dimer energies were predicted to be within 2 kJ mol−1 of each other,27 which is less than kBT at room temperature. It is likely these species coexist at room temperature prior to becoming kinetically locked due to cooling to 77 K for imaging. While these calculations are useful for characterizing the relative stability of individual features or clusters and the short-range interactions which define their structure, they do not account for potential stabilization from interactions with neighboring supramolecular species. C

DOI: 10.1021/acs.accounts.7b00522 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research A pentagonal cluster alone does not guarantee for the formation of a quasicrystal, since there are known crystalline packing schemes for pentagonal clusters, albeit without continuously filling the available space.28 Each pentamer is surrounded by five FcCOOH dimers (see Figure 1b) whose Cp rings are oriented perpendicular to the surface plane. The presence of these dimers prevents the adjacent pentamers from organizing into supramolecular structures without 5-fold symmetry. This pentamer and dimer assembly has 10-fold symmetry, lacks translational symmetry, and possesses orientational order, which fits the definition of an icosahedral quasicrystal.29 Quasicrystals have been observed in metallic, colloidal, and soft-matter systems (examples of which are given in ref 27), but the FcCOOH quasicrystal was the first reported system in which hydrogen-bonding interactions play a critical role in the formation of a quasiperiodic structure. This quasicrystal shares certain features of the Penrose P1 tiling, as seen in Figure 2, with deviations in local packing but agreement in intermediaterange orientational and translational order. The FcCOOH quasicrystal’s long-range orientational and translational order is apparent in the two-dimensional translational correlation function (Figure 2c), and the Fourier transform of a high symmetry axis yields reciprocal lattice vectors corresponding to frequencies which are multiples of powers of the golden ratio, φ = (1 + √5)/2, as seen in Figure 2d.

Figure 3. Representative STM image (−1 V, 10 pA, 25 Å × 275 Å) of indole-2-carboxylic acid. Three main structural motifs are present: catemer rows, pentamer clusters, and hexamer clusters. Some of these species are sufficiently mobile to prevent the sharp resolution of their features. Adapted with permission from ref 30. Copyright 2015 American Chemical Society.

mol−1 of the dimer on a per-molecule basis. We assume that the pentamers, while metastable, coexist with dimer species at the surface based on this relatively small difference in cohesion energy. However, the most stable species on this surface is the catemer rows, which do not have a dimer configuration. DFT calculations comparing individual clusters to catemer chains of 2−12 molecules addressed the question of the relative stability of these species (Figure 4c). The calculations of the catemer structures employed the bulk crystal catemer structure and then optimized the geometry, and the trend of cohesive energy per molecule vs chain length scales as 1/n. This trend is consistent with a fixed energy penalty of the terminal molecules, due to lost hydrogen bonds, and it asymptotically approaches a value almost 10 kJ mol−1 lower in energy than the molecular clusters. However, for chains of equivalent size to the cyclic clusters, the difference in energy is well within 2kBT at room temperature (4.96 kJ mol−1). In a nonequilibrium growth environment, in which the structures initially formed can become kinetically trapped, this small difference in cohesive energy between small catemers and pentamer clusters could lead to the coexistence of these two species at the surface. This work confirms the bonding model for the FcCOOH pentamers, and demonstrates that this bonding motif might be more common for metastable species with the appropriate hydrogen bonding elements and geometry, but it does not address where or how these clusters form. The indole-2-COOH work was notable in that vapor deposition experiments only produced the crystalline catemer structures,31 while pulsedeposition produced a surface with a mixture of catemers and metastable species.

3. A MODEL FOR CYCLIC PENTAMER FORMATION In order to test the bonding model of the cyclic pentamers observed for FcCOOH, the self-organization of indole-2carboxylic acid on Au(111) was studied.30 The presence of a hydrogen bonding donor (NH) adjacent to a carboxylic acid on a five member ring should result in an energetic minimum for metastable pentamer formation. The NH group is a stronger Hbond donor than the CH of FcCOOH, and as a result the bulk crystal structure is comprised of rows of catemers, rather than dimers. If the bonding model for cyclic FcCOOH pentamers reflects a more general behavior of systems with this arrangement of hydrogen bonding donors and acceptors, then there should be a strong propensity for the formation of cyclic pentamers. Figure 3 shows a typical surface after pulse-deposition of indole-2-COOH on Au(111). There are three major bonding motifs for the supramolecular structures on this surface. The structures consisting of aligned chevrons are the catemer rows seen in the bulk crystal structure, and these are most likely the most stable structure on this interface. There are also two types of clusters present on this surface: cyclic pentamers and rectangular hexamers. The cyclic pentamers have no analogous structure in the bulk crystal, and appear to have the same bonding motif as the cyclic FcCOOH pentamers. In order to confirm the structural similarity of these cyclic pentamers to those of FcCOOH, DFT calculations were performed on a strained dimer system. The DFT calculations of the strained dimer geometries indicated the presence of a well for geometries present in cyclic pentamers and hexamers, Figure 4a, which is similar to the case of FcCOOH. The optimized geometry of the indole pentamer is stabilized by contributions from the NH donor of the pyrrole ring, Figure 4b, which is similar to the role of the CH donors in FcCOOH pentamers. Calculations of the cohesive energy of cyclic clusters of two to six molecules indicates that a dimer is most stable, but the pentamer and hexamer are within 3 kJ

4. METASTABLE CLUSTER FORMATION IN SOLUTION A rapidly evaporating droplet will likely have temperature and concentration gradients not achievable in solution or on a surface. A supercooled droplet might allow the molecules in solution to sample parts of the supramolecular configuration potential energy surface not normally accessible through conventional growth techniques, and thus not observed. The first observed evidence for metastable cluster formation in solution came from the FcCOOH monolayer on Au(111).32 Pulse-deposition generally produces a heterogeneous surface, and for FcCOOH this surface contained three main regions: D

DOI: 10.1021/acs.accounts.7b00522 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research

Figure 4. (a) Calculated energies of a strained dimer system as a function of the CO···H angle. A shallow well exists for clusters of five or six molecules. (b) Optimized geometry, obtained from DFT calculations, of the cyclic pentamer which shows contributions from O···HN bonding. (c) Calculated cohesive energy per molecule for clusters of 2−6 molecules (open squares) are compared to the calculated cohesive energy per molecule of chains of 2−12 molecules with a catemer binding motif (black circles). The black circles are fit to E0 − Ep/n, red line, with the horizontal dashed line representing the asymptote, E0, which is the per-molecule cohesive energy of an infinite chain. Adapted with permission from ref 30. Copyright 2015 American Chemical Society.

arrays of ordered dimer structures, quasicrystalline domains of cyclic pentamers and dimers, and regions of disordered clusters, all shown in Figure 5a−d. The disordered regions tend to have lower coverage than the ordered dimer and quasicrystalline domains, and contain clusters of varying size and internal structure. The most common cluster structures that are not cyclic pentamers have a double-row structure, similar to the

FcCOOH dimers. The distribution of cluster sizes for these double-row clusters was determined from 14 STM images covering 2.85 × 106 Å2 of low coverage, disordered regions. If in forming these clusters the addition of each molecule is an independent event, then the size distribution should be binomial. Figure 5f displays the overlay of the best binomial fit to the actual distribution of clusters in the observed areas. The fit is quite poor, even when only accounting for even-numbered clusters, with large excesses of clusters of 5, 6, and 8 molecules, and a distinct lack of clusters larger than 8 molecules. It is possible that 5, 6, and 8 are magic number clusters,33 but the apparent lack of a strong substrate interaction, and the wide range of cluster sizes and structures, make this a poor explanation for this nonstatistical distribution. Instead, we proposed that clusters were formed in solution and then adsorbed onto the surface, as a result of the nonequilibrium conditions present in the droplet or thin film during deposition. A study of 1,1′-ferrocenedicarboxyic acid, Fc(COOH)2 gave more conclusive evidence that metastable clusters were forming in solution.34 The structures produced after pulse deposition could be broadly characterized into three classes: ordered dimers, rectangular hexamers, and disordered clusters. The ordered dimers consisted of packed arrays of dimers and dimerrow clusters, both of which are similar to those observed for FcCOOH. Surprisingly, there was a large excess of rectangular hexamer clusters, with some regions apparently containing almost entirely disordered, jammed packing of this cluster (see Figure 6a). The inability of these hexamers to influence the packing of neighboring clusters, except for essentially steric constraints, implies that the carboxylic acid groups all orient toward the center of the cluster. The disordered packing, in which neighboring clusters are apparently randomly oriented, and the seemingly nonstatistical prevalence of hexamers can give clues as to how these clusters formed. A fairly complex set of assumptions must hold for a nucleation and growth model to apply to these rectangular hexamers. First, the Fc(COOH)2 molecules must adsorb on the surface, either as single molecules or dimers, then diffuse (either in the thin evaporating solvent film or spontaneously on the surface), eventually forming clusters. Additionally, there must be a deep free-energy minimum, or a magic number configuration,33 for these chiral species on Au(111). Finally, the clusters must be highly mobile, to allow for smaller clusters or orphan molecules to diffuse out of apparently jammed regions

Figure 5. (a) A 620 Å × 600 Å STM image (+1.0 V, 10 pA) containing FcCOOH quasicrystalline domains (red), ordered dimers (green), and disordered clusters (blue). Panels (b)−(d) display a 105 Å × 100 Å area of each type of feature on this surface. (e) A 185 Å × 190 Å STM image (+1.0 V, 10 pA) with examples of clusters of various sizes in red boxes, with the white box highlighting a cyclic pentamer. (f) Histogram of clusters, by size, observed in the disordered molecular cluster regions (red), with a binomial distribution fit (blue). Adapted with permission from ref 32. Copyright 2014 Royal Society of Chemistry. E

DOI: 10.1021/acs.accounts.7b00522 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research

Figure 6. (a) STM image of high-coverage chiral hexamers, outlined in rectangles for coverage analysis. (b) Simulated arrangement of rectangles, with an aspect ratio of 3:2.2, determined by the jamming limit of a random sequential adsorption model. (c) Distance/angle correlation function of the RSA simulation (black) and experimental (red) rectangular clusters. The integrated distance (bottom) and angular (right) correlation functions are also compared. Panels (d) and (e) are composite images produced for the “s” and “z” enantiomers, respectively. Adapted with permission from ref 34. Copyright 2015 American Institute of Physics.

Figure 7. (a) Negative ion mode ESI-MS spectrum of 1.7 mM Fc(COOH)2 in methanol containing the trimer to hexamer family of peaks. The peaks indicated by the red arrow represent the [Mn − H]− peaks for n = 3−6. Additional peaks present represent sodiated clusters and adduct peaks. The inset is the full spectrum, with only the anionic monomer and dimer peaks visible. (b) Semilog plot of the natural log of the integrated intensity of the [Mn − H]− peaks for n = 1−7 (open circles) vs cluster size, with a linear fit omitting the n = 6 peak. (c) Residuals of the semilog plot and fit for each peak. Adapted with permission from ref 38. Copyright 2017 American Chemical Society.

aspect ratio, which is in agreement with the experimentally observed coverage of 0.55 monolayers (Figure 6a and b). Additionally, there is strong agreement between the distance/ angular correlation functions for the experimental data and RSA prediction, with the only substantive difference being a slight preference for the experimental clusters to align with one of the six high-symmetry axes of the Au(111) surface, as seen in Figure 6c. This analysis gives strong evidence for cluster formation in solution. Electrospray ionization mass spectrometry (ESI-MS) can directly measure hydrogen-bonded clusters in aerosol droplets. A previous ESI-MS study of Fc(COOH)2 showed no clusters other than dimers without the addition of alkali cations, but these studies were performed at substantially lower concentrations than those used during pulse deposition (0.01 to 1 mM for Kubota et al. vs 7 mM in our experiment).37 We performed our own measurements using a 1.7 mM solution of Fc(COOH)2 in methanol and observed an anionic hexameric cluster peak.38

of packing, so that they can form hexameric clusters. The disorder of jammed regions would then be attributed to the cooling of the sample, but in this case there is still no alignment due to van der Waals interactions. A simpler model describing this kind of packing is random sequential adsorption.35 In this model, the hexamers must exist in solution at some point during the deposition process (i.e., the rapidly evaporating droplet or an evaporating thin film). The clusters need not be a majority species, but rather can exist as a metastable species, which then can precipitate from solution as predicted by Ostwald’s rule.36 The precipitated cluster then can adsorb, possibly changing supramolecular configuration but remaining intact, and does not undergo any ripening process or further diffusion or reorientation. In this model, the relative distances and orientations to neighboring clusters is statistical, with cluster geometry as a constraining factor. Our analysis involves using the methodology of Vigil and Ziff35 with rectangles with a 3:2.2 aspect ratio. RSA calculations predict a jammed coverage of 0.54 for rectangles with this F

DOI: 10.1021/acs.accounts.7b00522 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research Anionic cluster peaks ([Mn − H]−) for clusters of n = 1−7 were observed, with the pentamer and heptamer peaks barely visible above the baseline noise (Figure 7a). When comparing relative intensities of the observed peaks, adduct and sodiated peaks were not included in the analysis due to the propensity of these species to act as nucleating sites for clusters.37 Clusters having “magic number” sizes will deviate from the general trend of peak intensity with cluster size. All clusters except the n = 6 cluster exhibited an exponential decay in intensity as cluster size increased. A semilog plot of the integrated intensity of each anionic peak vs cluster size shows that the hexameric peak is almost 40 times larger than expected. The exceptional intensity of the hexamer peak indicates that this is a magic number peak for a rapidly evaporating spray. While the conditions in an electrospray ionization aerosol are not identical to a rapidly evaporating droplet of neutral clusters, the combination of the STM and ESI-MS results strongly support the mechanism of chiral hexamers forming in solution. Annealing experiments can resolve which species adsorbed on a surface are metastable and which is the most stable species. Such experiments were performed on two surfaces prepared by pulse deposition of a 6 mM solution of Fc(COOH)2 in methanol on Au(111).38 The representative images are presented in Figure 8, and they show an evolution of the initial metastable structures into ordered forms consisting of dimers with increasing annealing temperatures and times. This behavior is the hallmark of a system evolving under kinetic control, as increasing annealing temperature leads to the

formation of different structures as local kinetic barriers are overcome, allowing the system to access new local minima in the potential energy surface.

5. CONTROLLING METASTABLE CLUSTER STRUCTURE Our previous studies have produced a model for the formation of metastable cyclic pentamers in nonequilibrium growth conditions. Using this knowledge, we chose to deposit 1-Hindole-2,3-dione (isatin) on Au(111) with the expectation that it would form pentamers.39 After both pulse and vapor deposition, this molecule almost exclusively formed cyclic pentamers (Figure 9a). Metastable species can form during

Figure 9. (a) A 250 Å × 250 Å image of vapor deposited isatin on Au(111). Representative cyclic pentamers are highlighted in green. (b) A 125 Å × 195 Å image of the isatin decorated surface after a 20 min 40 °C anneal. An ordered row structure is highlighted in blue. (c) A 30 Å × 25 Å composite image of one chirality of isatin pentamers in panel (a). (d) Optimized geometry from DFT calculations of the isatin pentamer, with the critical intermolecular hydrogen bonding circled. Adapted with permission from ref 39. Copyright 2017 American Chemical Society.

vapor deposition, if the mobility of the species is low enough, or the flux high enough, to prevent substantial reorganization.21 Annealing of the isatin pentamers produces a feature resembling the catemer rows of indole-2-COOH (Figure 9b), indicating that the pentamer is metastable. Density functional theory calculations, in concert with the STM observations, produce a bonding model with NH···O between the carbonyl at the 2 position with the amine of an adjacent ring, and CH···O between the carbonyl at the 3 position with the hydrogen at the 7 position of the aromatic ring, as seen in Figure 9d. Given this bonding model, substitutions of the carbonyl at the 3 position and the hydrogen at the 7 position should disrupt this pentamer feature. The metastable clusters formed by two different derivatives of isatin were investigated to test this model. This approach is similar to the comparison of FcCH2COOH and FcCOOH, in which the larger separation of the acetic acid group from the CH donors of the Cp ring suppressed pentamer formation. The two molecules investigated were 3-methyl-2-oxindole, which substitutes a methyl and hydrogen group for the carbonyl at the

Figure 8. (a) A 321 Å × 315 Å image of a 6 mM solution of Fc(COOH)2 as deposited. Dimer rows (green), dimer arrays (blue), square tetramers (yellow to orange), and chiral hexamers (pink to purple) are highlighted. (b) A 400 Å × 400 Å image obtained after a 1 h 50 °C anneal, consisting mainly of disordered tetramers and ordered tetramer rows. (c) A 300 Å × 287 Å image obtained after a total annealing time of 3 h at 50 °C and 1 h at 65 °C. There are two chiral domains of dimers, and the insets on the left and right correspond to composite images of the left and right chiral domains. (d) A 450 Å × 450 Å image after a 1 h 75 °C anneal. The inset is a composite of the ordered linear dimer rows present after the anneal. Adapted with permission from ref 38. Copyright 2017 American Chemical Society. G

DOI: 10.1021/acs.accounts.7b00522 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research

CH···O bonding between the 3 and 7 positions of adjacent molecules, while the NH···O bond is elongated by almost 0.5 Å in 7-fluoroisatin pentamers, possibly due to destabilizing F−O repulsion. This work demonstrates that the indole heterocycle with adjacent hydrogen bonding acceptors and donors on the fivemember ring of the heterocycle has a strong affinity for metastable pentamer formation. Additionally, it is possible to modify the internal structure of the cyclic pentamer or fully suppress pentamer formation through substitution at the sites stabilizing the cyclic pentamer. The combination of experimental observations with complementary electronic structure calculations allows for the development of detailed bonding models for metastable clusters observed on the surface, which can then be exploited to alter the internal structure of these supramolecular features.

3 position, and 7-fluoroisatin, which changes the weak hydrogen bonding donor CH to a fluorine at the 7 position. The results are presented in Figure 10. The 3-methyl-2-

6. CONCLUSIONS This Account gives a detailed summary of our initial investigations into the production of metastable molecular clusters at the vacuum/surface interface. Growth in conditions far from equilibrium has allowed us to produce supramolecular features with symmetries not observed in bulk crystal structures, resulting in the first report of a molecular quasicrystal. The combination of local structural probes, electronic structure calculations, and ensemble measurements of clustering in aerosols allowed us to determine the nature and origin of metastable clusters observed after pulse-deposition. An improved understanding of the intermolecular interactions leading to metastable cluster formation resulted in our ability to manipulate the structures produced under nonequilibrium growth conditions. These are just the initial steps in understanding and exploiting nonequilibrium growth phenomena. Specifically, for the direct injection of a solution in to vacuum, the role of factors such as solute concentration, solvent polarity, and droplet size are poorly understood, though possibly important. Future work should involve characterizing the impact of these factors, in addition to exploratory work to identify new chemical structure motifs which produce metastable clusters with novel supramolecular configurations.

Figure 10. (a) A 200 Å × 200 Å image of vapor deposited 3-methyl-2oxindole on Au(111). Some representative cyclic pentamers are highlighted in green, and catemer-like rows in blue. (b) A 158 Å × 158 Å image of vapor deposited 7-fluoroisatin on Au(111), with an ordered domain highlighted in blue and representative metastable hexamers in pink. 20 Å × 20 Å composite images of a cyclic isatin pentamer (c), a cyclic 3-methyl-2-oxindole pentamer (d), and a 7-fluoroisatin hexamer (e). Adapted with permission from ref 39. Copyright 2017 American Chemical Society.

oxindole monolayer contained several species, with the two dominant features being cyclic catemers, with a different supramolecular configuration than isatin, and rows resembling catemers. The substitution at the 7-position in 7-fluoroisatin completely suppressed the formation of cyclic pentamers, and only ordered arrays of this molecule and metastable hexamers are present after deposition. The substitution at the 3 position in 3-methyl-2-oxindole results in an apparent qualitative change in the cyclic pentamer structure, with the molecules oriented slightly farther away from the cluster center than in isatin, but is insufficient to fully suppress cyclic pentamer formation. We assign the slightly darker feature to the heterocycle, and the bright feature to the methyl group, which would project away from the gold substrate. This substitution did not stop pentamer formation, but it does subtly alter the structure of the metastable cluster. The fluorine substitution in 7-fluoroisatin blocks hydrogen bonding between the 3 and 7 positions. DFT calculations of the optimized geometries of the dimer and cyclic pentamer for all three molecule estimated the relative cohesive energy, on the per molecule basis, for each species. The isatin pentamer was more stable than the dimer by approximately 12 kJ mol−1, which is not surprising given that CH···O bonding is important in the bulk crystal structure. The 3-methyl-2-oxindole and 7-fluoroisatin pentamers were less stable than dimers by 6 and 12 kJ mol−1, respectively. All three species have similar NH···O bond lengths in their bulk crystal structures (and these values are in close agreement with the calculated dimer bond lengths), but this value deviates significantly for isatin and 7-fluoroisatin pentamers. The calculated NH···O bond length in isatin pentamers is 0.07 Å shorter than the dimer, likely due to stabilization from the



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Steven A. Corcelli: 0000-0001-6451-4447 S. Alex Kandel: 0000-0001-8191-1073 Notes

The authors declare no competing financial interest. Biographies Ryan D. Brown graduated from Kenyon College in 2004 with a B.A. in Chemistry. He performed his doctoral studies at the University of Chicago under the guidance of Steven J. Sibener, and graduated in 2014. He has performed postdoctoral research with S. Alex Kandel at the University of Notre Dame since 2014. His research interests concern molecular organization and growth processes at interfaces. Steven A. Corcelli was awarded a Sc.B. in Chemistry from Brown University in 1997. He then matriculated at Yale University where he received his Ph.D. in 2002 working with John C. Tully. He then transitioned to the University of Wisconsin at Madison for a postdoc H

DOI: 10.1021/acs.accounts.7b00522 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research

V. Self-organisation of dodeca-dendronized fullerene into supramolecular discs and helical columns containing a nanowire-like core. Chem. Sci. 2015, 6, 3393−3401. (18) Bzdek, B. R.; DePalma, J. W.; Johnston, M. V. Mechanisms of Atmospherically Relevant Cluster Growth. Acc. Chem. Res. 2017, 50, 1965−1975. (19) Gebauer, D.; Kellermeier, M.; Gale, J. D.; Bergstrom, L.; Coelfen, H. Pre-nucleation clusters as solute precursors in crystallisation. Chem. Soc. Rev. 2014, 43, 2348−2371. (20) Steiner, T. The hydrogen bond in the solid state. Angew. Chem., Int. Ed. 2002, 41, 48−76. (21) Barth, J.; Costantini, G.; Kern, K. Engineering atomic and molecular nanostructures at surfaces. Nature 2005, 437, 671−679. (22) Elemans, J. A. A. W.; Lei, S.; De Feyter, S. Molecular and supramolecular networks on surfaces: From two-dimensional crystal engineering to reactivity. Angew. Chem., Int. Ed. 2009, 48, 7298−7332. (23) Slater, A. G.; Perdigao, L. M. A.; Beton, P. H.; Champness, N. R. Surface-based supramolecular chemistry using hydrogen bonds. Acc. Chem. Res. 2014, 47, 3417−3427. (24) Mali, K. S.; Pearce, N.; De Feyter, S.; Champness, N. R. Frontiers of supramolecular chemistry at solid surfaces. Chem. Soc. Rev. 2017, 46, 2520−2542. (25) Smith, J. D.; Cappa, C. D.; Drisdell, W. S.; Cohen, R. C.; Saykally, R. J. Raman thermometry measurements of free evaporation from liquid water droplets. J. Am. Chem. Soc. 2006, 128, 12892−12898. (26) Rabani, E.; Reichman, D.; Geissler, P.; Brus, L. Drying-mediated self-assembly of nanoparticles. Nature 2003, 426, 271−274. (27) Wasio, N. A.; Quardokus, R. C.; Forrest, R. P.; Lent, C. S.; Corcelli, S. A.; Christie, J. A.; Henderson, K. W.; Kandel, S. A. Selfassembly of hydrogen-bonded two-dimensional quasicrystals. Nature 2014, 507, 86−89. (28) Bauert, T.; Merz, L.; Bandera, D.; Parschau, M.; Siegel, J. S.; Ernst, K.-H. Building 2D crystals from 5-fold-symmetric molecules. J. Am. Chem. Soc. 2009, 131, 3460−3461. (29) Levine, D.; Steinhardt, P. Quasi-crystals 0.1. Definition and structure. Phys. Rev. B: Condens. Matter Mater. Phys. 1986, 34, 596− 616. (30) Wasio, N. A.; Quardokus, R. C.; Brown, R. D.; Forrest, R. P.; Lent, C. S.; Corcelli, S. A.; Christie, J. A.; Henderson, K. W.; Kandel, S. A. Cyclic hydrogen bonding in indole carboxylic acid clusters. J. Phys. Chem. C 2015, 119, 21011−21017. (31) De Marchi, F.; Cui, D.; Lipton-Duffin, J.; Santato, C.; MacLeod, J. M.; Rosei, F. Self-assembly of indole-2-carboxylic acid at graphite and gold surfaces. J. Chem. Phys. 2015, 142, 101923. (32) Quardokus, R. C.; Wasio, N. A.; Christie, J. A.; Henderson, K. W.; Forrest, R. P.; Lent, C. S.; Corcelli, S. A.; Kandel, S. A. Hydrogenbonded clusters of ferrocenecarboxylic acid on Au(111). Chem. Commun. 2014, 50, 10229−10232. (33) Bohringer, M.; Morgenstern, K.; Schneider, W.; Berndt, R.; Mauri, F.; De Vita, A.; Car, R. Two-dimensional self-assembly of supramolecular clusters and chains. Phys. Rev. Lett. 1999, 83, 324−327. (34) Quardokus, R. C.; Wasio, N. A.; Brown, R. D.; Christie, J. A.; Henderson, K. W.; Forrest, R. P.; Lent, C. S.; Corcelli, S. A.; Kandel, S. A. Hydrogen-bonded clusters of 1,1′-ferrocenedicarboxylic acid on Au(111) are initially formed in solution. J. Chem. Phys. 2015, 142, 101927. (35) Vigil, R.; Ziff, R. Random sequential adsorption of unoriented rectangles onto a plane. J. Chem. Phys. 1989, 91, 2599−2602. (36) Threlfall, T. Structural and thermodynamic explanations of Ostwald’s rule. Org. Process Res. Dev. 2003, 7, 1017−1027. (37) Kubota, N.; Fukuo, T.; Akawa, R. Electrospray ionization mass spectrometric analysis of self-assembled 1,1′-ferrocenedicarboxylic acid. J. Am. Soc. Mass Spectrom. 1999, 10, 557−560. (38) Brown, R. D.; Coman, J. M.; Christie, J. A.; Forrest, R. P.; Lent, C. S.; Corcelli, S. A.; Henderson, K. W.; Kandel, S. A. Evolution of metastable clusters into ordered structures for 1,1′-ferrocenedicarboxylic acid on the Au(111) surface. J. Phys. Chem. C 2017, 121, 6191− 6198.

with James L. Skinner. In 2005, he joined the faculty in the Department of Chemistry and Biochemistry at the University of Notre Dame. His research interests involve the use of theory and computation to understand structure and dynamics in the condensed phase. S. Alex Kandel received a B.S. in Chemistry from Yale University in 1993. He completed his doctoral studies with Richard N. Zare at Stanford in 1999, and went on to do postdoctoral research with Paul S. Weiss at the Pennsylvania State University. He joined the department of Chemistry and Biochemistry at the University of Notre Dame in 2001. His research interests include self-assembly and selforganization, molecular electronics, and gas-surface chemical reactions.



ACKNOWLEDGMENTS Support for this work was provided by the National Science Foundation (NSF CHE-1507213).



REFERENCES

(1) Desiraju, G. Supramolecular synthons in crystal engineering - a new organic-synthesis. Angew. Chem., Int. Ed. Engl. 1995, 34, 2311− 2327. (2) Chen, B.; Xiang, S.; Qian, G. Metal-organic frameworks with functional pores for recognition of small molecules. Acc. Chem. Res. 2010, 43, 1115−1124. (3) Northrop, B. H.; Zheng, Y.-R.; Chi, K.-W.; Stang, P. J. Selforganization in coordination-driven self-assembly. Acc. Chem. Res. 2009, 42, 1554−1563. (4) Talapin, D. V.; Shevchenko, E. V.; Bodnarchuk, M. I.; Ye, X.; Chen, J.; Murray, C. B. Quasicrystalline order in self-assembled binary nanoparticle superlattices. Nature 2009, 461, 964−967. (5) Zhang, J.-P.; Zhang, Y.-B.; Lin, J.-B.; Chen, X.-M. Metal azolate frameworks: From crystal engineering to functional materials. Chem. Rev. 2012, 112, 1001−1033. (6) Prins, L.; Reinhoudt, D.; Timmerman, P. Noncovalent synthesis using hydrogen bonding. Angew. Chem., Int. Ed. 2001, 40, 2382−2426. (7) Zang, L.; Che, Y.; Moore, J. S. One-dimensional self-assembly of planar pi-conjugated molecules: Adaptable building blocks for organic nanodevices. Acc. Chem. Res. 2008, 41, 1596−1608. (8) Claridge, S. A.; Liao, W.-S.; Thomas, J. C.; Zhao, Y.; Cao, H. H.; Cheunkar, S.; Serino, A. C.; Andrews, A. M.; Weiss, P. S. From the bottom up: dimensional control and characterization in molecular monolayers. Chem. Soc. Rev. 2013, 42, 2725−2745. (9) Grzelczak, M.; Vermant, J.; Furst, E. M.; Liz-Marzan, L. M. Directed self-assembly of nanoparticles. ACS Nano 2010, 4, 3591− 3605. (10) Moulton, B.; Zaworotko, M. From molecules to crystal engineering: Supramolecular isomerism and polymorphism in network solids. Chem. Rev. 2001, 101, 1629−1658. (11) Desiraju, G. R. Crystal engineering: From molecule to crystal. J. Am. Chem. Soc. 2013, 135, 9952−9967. (12) Barth, J. V. Molecular architectonic on metal surfaces. Annu. Rev. Phys. Chem. 2007, 58, 375−407. (13) Desiraju, G. R. Crystal engineering: A holistic view. Angew. Chem., Int. Ed. 2007, 46, 8342−8356. (14) Stannard, A.; Russell, J. C.; Blunt, M. O.; Salesiotis, C.; del Carmen Gimenez-Lopez, M.; Taleb, N.; Schroeder, M.; Champness, N. R.; Garrahan, J. P.; Beton, P. H. Broken symmetry and the variation of critical properties in the phase behaviour of supramolecular rhombus tilings. Nat. Chem. 2012, 4, 112−117. (15) Mattia, E.; Otto, S. Supramolecular systems chemistry. Nat. Nanotechnol. 2015, 10, 111−119. (16) Maggini, L.; Bonifazi, D. Hierarchised luminescent organic architectures: design, synthesis, self-assembly, self-organisation and functions. Chem. Soc. Rev. 2012, 41, 211−241. (17) Guerra, S.; Iehl, J.; Holler, M.; Peterca, M.; Wilson, D. A.; Partridge, B. E.; Zhang, S.; Deschenaux, R.; Nierengarten, J.-F.; Percec, I

DOI: 10.1021/acs.accounts.7b00522 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research (39) Silski, A. M.; Brown, R. D.; Petersen, J. P.; Coman, J. M.; Turner, D. A.; Smith, Z. M.; Corcelli, S. A.; Poutsma, J. C.; Kandel, S. A. C−HO hydrogen bonding in pentamers of isatin. J. Phys. Chem. C 2017, 121, 21520−21526.

J

DOI: 10.1021/acs.accounts.7b00522 Acc. Chem. Res. XXXX, XXX, XXX−XXX