Controlling the Two-Dimensional Self-Assembly of Functionalized Porphyrins via Adenine-Thymine Quartet Formation

The development of supramolecular synthons capable of driving hierarchical two-dimensional self-assembly is an important step towards the growth of complex and functional molecular surfaces. In this work the formation of nucleobase quartets consisting of adenine and thymine groups was used to control the 2D self-assembly of porphyrins. Tetra-(phenylthymine) zinc porphyrin (Zn-tetra-TP) and tetra-(phenyladenine) porphyrin (tetra-AP) were synthesised and scanning tunneling microscopy (STM) experiments performed to visualize their self-assembly at the liquid-solid interface between an organic solvent and a graphite surface. Mono-component solutions of both Zn-tetra-TP and tetra-AP form stable 2D structures with either thymine-thymine or adenine-adenine hydrogen bonding. Structural models based on STM data were validated using molecular mechanics (MM) simulations. In contrast, bi-component mixtures showed the formation of a structure with p 4 symmetry consisting of alternating Zn-tetra-TP and tetra-AP molecules in a chessboard type pattern. The relative positions of the porphyrin components were identified from STM images via contrast changes associated with the zinc atom present in Zn-tetra-TP. MM simulations suggest that hydrogen bonding inter-actions within these structures are based on the formation of adenine-thymine (ATAT) quartets with Watson-Crick base pairing between adenine and thymine groups.


INTRODUCTION
The use of recognition interactions to control the two-dimensional (2D) self-assembly of molecules provides a route to form complex and potentially functional nanostructured surfaces. [1][2][3] In combination with the molecular resolution of surfaces provided by scanning tunneling microscopy (STM) this goal has been the driving force behind a wealth of research activity. [3][4][5][6] These studies use concepts from reticular synthesis 7 and molecular tectonics. 8 Rigid and planar molecular building blocks known as tectons are decorated with functional groups at specific positions. Recognition interactions between functional groups on different tectons then form supramolecular synthons 9 linking the tectons together and driving self-assembly. Coordination bonds, [10][11] halogen bonds, [12][13] van der Waals interactions, [14][15] and even dynamic covalent bond formation 15 have all been employed as recognition interactions in the growth of complex 2D molecular networks. Hydrogen bonds in particular have been widely used as stabilizing interactions for 2D self-assembly. 1,[5][6]12 A key example of the use of hydrogen bonds as molecular recognition interactions in 2D self-assembly is the triple hydrogen bond interaction formed between perylenetetracarboxylic diimide (PTCDI) and melamine. [17][18][19] The archetypal set of supramolecular synthons based on hydrogen bonds are those formed between the nucleobases of DNA. The ability of these interactions to effect self-assembly of vastly complex biological systems makes them ideal candidates to control the formation of synthetic functional materials. [20][21][22] The planar structure of individual nucleobases, and of small groups of hydrogen-bonded nucleobases, means they are ideally suited to the formation of 2D molecular layers. This planarity, coupled with the strength and selectivity of their interactions has led to the 2D self-assembly of nucleobases being a widely investigated topic. [23][24][25][26][27][28][29][30][31][32][33][34][35][36][37][38][39] Selfassembly of the individual nucleobases adenine (A), [23][24][25][26] thymine (T), [27][28] guanine (G), [29][30] and 4 cytosine (C) 31 have been studied on different substrates including various coinage metals 23,25,[28][29]31 and graphite 24,27,30 and under a range of environmental conditions from ultra-high vacuum (UHV) 23,25,29 through to liquid-solid interfaces. 28 The 2D self-assembly of mixtures of different nucleobases has shown the formation of selective inter-nucleobase interactions but also highlighted the complexity of their hydrogen bonding landscape. 27,[32][33][34][35] The hydrogen bonding synthons observed for 2D self-assembly of nucleobases often go beyond standard Watson-Crick base pairing with a number of examples of the formation of nucleobase quartets. 33,[36][37][38][39] Guanine quartets (GGGG) occur naturally and play a key role in the structure of telomeric DNA. 40 Through their ability to coordinate to metal ions G quartets have been the basis of novel methods for metal detection based on DNA. 41 The formation of (GGGG), 37-39 (ATAT) 33 and (GCGC) 36 quartets have all been observed via 2D self-assembly on surfaces. Despite the wealth results on mono-component or mixed nucleobase systems comparatively few studies have taken the step of using the nucleobases as functional groups to drive the 2D self-assembly of larger molecular tectons. 39,[42][43][44][45][46] We have previously outlined the synthesis and 2D-self-assembly of thymine functionalized porphyrins. 46 Now in this study we extend the use of nucleobase functional groups to control self-assembly by investigating bi-component mixtures of both thymine and adenine functionalized porphyrins; (Chart 1). Porphyrins are a particularly attractive choice as the basis for more complex molecular tectons and their 2D self-assembly has been widely investigated. 47 Functional groups can be included on the meso positions of a porphyrin to create tetratopic tectons. In addition to their flexibility in terms of the inclusion of peripheral functional groups, the central porphyrin macrocycle can also bind a large variety of different metal ions. Inclusion of metal ions within the porphyrin core ultimately provides a route to modify the optical, electronic and catalytic properties 48 of the resulting tectons and offers the 5 opportunity for additional control over self-assembly through secondary coordination to the metal. 49 Finally, in terms of self-assembly the planar nature of porphyrins promotes adsorption and ordering on surfaces while their delocalized system of π electrons and distinct cross shape makes them easily discernable in STM images. 47 The self-assembly of these nucleobase functionalized porphyrins at the liquid-solid interface between highly oriented pyrolytic graphite (HOPG) and an organic solvent layer was investigated using scanning tunneling microscopy (STM). Molecular mechanics (MM) simulations were then used in conjunction with drift corrected STM images to produce molecular models for the observed structures. Bi-component mixtures of the porphyrins demonstrated selective molecular recognition interactions in the form of unusual adenine-thymine (ATAT) quartets containing Watson-Crick hydrogen bonding between adenine and thymine groups. The formation of these nucleobase quartets resulted in a 2D molecular network consisting of the two porphyrin species arranged in an alternating chessboard type pattern. Chart 1. Tetra-(phenylthymine) zinc porphyrin (Zn-tetra-TP) and tetra-(phenyladenine) porphyrin (tetra-AP) and model compound mono-(phenyladenine)-tri-(3,5-di-tert-butylphenyl)porphyrin (mono-A).

EXPERIMENTAL METHODS
All chemical reagents were used as-purchased from Alfa Aesar, Fisher Scientific, Sigma-Aldrich, or VWR International, unless stated otherwise. Anhydrous toluene was dried by passing through a column packed with 4 Å molecular sieves, degassed and stored over a potassium mirror in a nitrogen atmosphere. Anhydrous dichloromethane was purchased from Sigma-Aldrich (Fluka) and stored over 4 Å molecular sieves. Column chromatography was performed on Merck silica gel 60 (0.2-0.5 mm, 50 -130 mesh). Full details of the synthesis procedures and characterization details for each of the molecules discussed in the paper can be found in the ESI. 1 H and 13 C NMR spectra were recorded using Bruker spectrometers. EI M/S spectra were taken using a Bruker Apex IV 4.7 T mass spectrometer. MALDI-TOF M/S spectra were recorded with a Bruker Ultraflex III mass spectrometer using trans-2-[3-(4-tert-butylphenyl)-2methyl-2-propenylidene]-malononitrile (DCTB) as the matrix. ESI M/S spectra were recorded with a Bruker MicroTOF. Single crystal X-ray diffraction experiments for mono-A and 9-propyl adenine were performed on a Rigaku Saturn724+ diffractometer equipped with a rotating anode using monochromated Cu-Kα radiation (λ = 1.5418 Å) at 120 K. The structures were solved by direct methods using either SHELXS or SHELXT 50 and refined with SHELXL 51 using a least squares method. OLEX2 software was used as the solution, refinement and analysis program. 52 All hydrogen atoms were placed in geometrically calculated positions; non-hydrogen atoms were refined with anisotropic displacement parameters.
7 STM experiments were performed using a Keysight Technologies 5500 STM operating in constant current mode using mechanically cut Pt:Ir (90:10) tips. Highly oriented pyrolytic graphite (HOPG) substrates (Bruker) were freshly cleaved before each experiment. Prior to deposition both the substrate and the target porphyrin solution were pre-heated to 120°C for 5 mins. A liquid cell was used to contain 25 μl of solutions on the HOPG surface. The substrate was maintained at 120°C for 5 mins following deposition before controlled cooling at 2°C min -1 until the sample reached room temperature. The liquid cell was covered during the cooling process to prevent evaporation. STM images were then collected at the solid-liquid interface. All Dreiding force field 53 was used to assign atomic charges as well as to perform geometry optimization. A conjugate gradient algorithm was employed with an RMS force convergence parameter set to 5 × 10 -3 kcal mol -1 Å -1 . The non-bonded van der Waals, and electrostatic terms both described by a cubic spline function with a cut-off at 15.5 Å and a spline width of 3 Å.
Hydrogen bonding terms were described by a cubic spline function with a cut-off at 4.5 Å and a spline width of 0.5 Å. Full details of the simulations and the method used to calculate the network binding energies can be found in the ESI.

RESULTS AND DISCUSSION
The target nucleobase functionalized porphyrins were prepared by adapting our previously reported strategy for the synthesis of thymine-functionalized porphyrins. 46 The strategy requires the synthesis of a suitable aldehyde building-block which can be employed to synthesize the porphyrin. Developing our previous strategy, we prepared N-9-(4-formylphenyl)-C-6-(bis-bocamino)adenine. We found it necessary to protect the adenine group using a boc-functionality 54 prior to the porphyrin forming reactions. Thus, the synthesis of N-9-(4-formylphenyl)-C-6-(bisboc-amino)adenine was readily achieved from 4-formylphenylboronic acid and (bis-bocamino)adenine using a Cu(OAc)2-mediated Chan-Lam-Evans-modified Ullmann condensation reaction to facilitate the cross-coupling process. 55 The synthesis of the tetra-AP (Chart 1) was achieved by reaction of N-9-(4-formylphenyl)-C-6-(bis-boc-amino)adenine with a large excess of pyrrole using trifluoroacetic acid (TFA) to initiate the reaction, followed by oxidation with 2,3-dichloro-5,6-dicyano-1,4-benzoquinone(DDQ). Isolation of the pure target was challenging as the reaction yielded a mix of boc-protected species. TFA, required for the porphyrin synthesis, is known to lead to removal of boc groups and MALDI mass spectrometry confirmed the presence of products with loss of 0-8 boc-protecting groups. Thus, purification and isolation of a single product was difficult. To overcome this problem the range of boc-protected species were combined and the mixture treated with KOH in MeOH at 50°C. These conditions did lead to the desired product but deprotection was inefficient leading to an overall yield tetra-AP of just 2%.
Tetra-AP was found to be sparingly soluble in DMSO with lower solubility in other common organic solvents, behaviour similar to that observed for tetra-TP. 44 The synthesis of mono-AP (Chart 1) required the formation of a suitable dipyrromethene which was achieved by reaction of N-9-(4-formylphenyl)-C-6-(bis-boc-amino)adenine with a 9 large excess of pyrrole in the presence of InCl3. The bis-boc-adenine functionalized dipyrromethane was reacted with a suitable tert-butyl functionalized carbinol species, using Lindsay's approach, [56][57][58] in the presence of TFA with subsequent oxidation using DDQ. The reaction was slightly more efficient than the synthesis of tetra-AP with an 8% yield following purification. In addition, in order to aid identification of the two similarly sized porphyrin molecules in STM images, the zinc(II) complex of tetra-(phenylthymine) porphyrin (Zn-tetra-TP) was prepared by simple reaction of tetra-TP 46 with zinc acetate. Further detailed information concerning synthesis is presented in the ESI.  In addition to the structure of mono-AP we have investigated the single crystal X-ray structures of 9-propyladenine, N-1-propylthymine and a co-crystal of these two compounds (see ESI for details). As discussed above 9-propyladenine forms a hydrogen bonded arrangement which adopts an A2-A1 adenine...adenine interaction. N-1-propylthymine forms a dimeric arrangement with two thymine molecules adopting N-H…O hydrogen bonds in an R2 2 (8) double hydrogen-bonded arrangement, analogous to that observed for self-assembled arrays of tetra-TP. 46 Perhaps most pertinently to this study we isolated a 1:1 co-crystal of 9-propyladenine:N-1propylthymine which adopts a Hoogsteen hydrogen-bonding arrangement between the A1 face of the adenine moiety and thymine (see ESI).
In order to promote solubility of the nucleobase functionalized porphyrins a mixture of tetrahydrofuran (THF) and 1,2,4-trichlorobenzene (TCB) in a 1:9 volume ratio was used as a solvent. Mono-component solutions of both Zn-tetra-TP and tetra-AP were made with concentrations of ~ 4 × 10 -5 M. Mixed solutions of Zn-tetra-TP and tetra-AP consisted of a 1:1 volume ratio mixture of the above solutions.
STM investigations of the self-assembly of Zn-tetra-TP (Figure 2a) show that the metallated tecton forms an identical structure to its freebase counterpart. 46 Zn-tetra-TP forms an ordered 12 structure on a square lattice with p4 symmetry that is stabilized by hydrogen bond dimers between adjacent Zn-tetra-TP molecules. Bright features observed in Figure 2a correspond to the cores of the Zn-tetra-TP molecules. Additional structure observed in the STM image is difficult to assign to a specific molecular moiety and may be the results of an imaging artefact associated with the STM tip configuration. The unit cell for the Zn-tetra-TP structure, shown in the insert to   Similar to thymine groups on Zn-tetra-TP, the adenine groups on tetra-AP can also rotate with respect to the porphyrin core allowing them to adopt two distinct orientations when the molecule is adsorbed on a surface. In contrast to Zn-tetra-TP, the adenine groups on the tetra-AP molecules shown in the model presented in Figure 3b do not all display the same orientation with respect to the porphyrin core. Instead they adopt alternating orientations around the molecule.
This arrangement of the adenine groups is necessary in order to construct a molecular model that   Figure 4a shows the formation of domains of ordered network between 25 and 50 nm in size. Figure 4b shows a smaller scale drift corrected STM image that shows the structure in more detail. As with  Zn-tetra-TP or the tetra-AP mono-component networks it is also clear that the mixed system adopts a different structure to either of the molecules on their own. However, from the STM images shown in Figure 4b it is not possible to differentiate between the Zn-tetra-TP and the tetra-AP molecules. Extended STM scanning of the network did result in occasional changes in image contrast to that seen in Figure 4c. In these images the cross shape of the porphyrin core was no longer discernable but there was a notable difference in contrast between molecules with some porphyrins appearing consistently brighter than others.
Differences in STM contrast observed for the presence or type of centrally coordinated metal ion have been widely reported previously. 67,68 The bright appearance of the central metal ion in some metal containing porphyrins, e.g. cobalt tetraphenyl porphyrin, has been linked to an orbital mediated tunneling mechanism through the partially filled dz 2 orbital on the metal ion. 68 However, although this explanation is not expected to be valid for Zn tetraphenyl porphyrin, as the dz 2 orbital is completely filled, molecular contrast in STM images is also strongly linked to the chemical nature of the STM tip apex. 69,70 Chemically modified STM tips consisting of an Au tip functionalized with thiol molecules have been used to identify the location of specific functional groups within an organic monolayer through hydrogen bond interactions. 71 A similar approach has been used by Ohshiro et al. to distinguish between Zn porphyrin and freebase porphyrin through coordination bonding between Zn and a pyridine functionalized STM tip. 72 While the PtIr tips used in this work have not been specifically functionalized, both porphyrin molecules can coordinate to Zn either through a carbonyl oxygen present in thymine or a pyridine-like nitrogen in adenine. Given that the STM images shown in Figure 4 were collected at the interface between HOPG and an organic solvent layer containing dissolved Zn-tetra-TP and tetra-AP it is likely that one of these molecules could adsorb at the apex of the STM tip and

CONCLUSIONS
In conclusion we have shown that selective hydrogen boding interactions between complementary nucleobase functional groups can be used to drive the hierarchical 2D selfassembly of porphyrin molecular tectons at a surface. Porphyrin tectons functionalized in the meso-positions with either T or A groups were synthesized and their 2D self-assembly at the liquid-solid interface investigated using STM and MM simulations. When mixed solutions of both A and T functionalized porphyrins were deposited at the interface a 2D molecular network was observed that is different to those formed when either of the components was deposited on its own. By analysis of experimental data and its comparison to simulations a molecular model for the bi-component structure was suggested that consists of an alternating arrangement of A and T functionalized porphyrins in a 2D molecular networks with p4 symmetry. This network structure is stabilized via the formation of cyclic hydrogen bond configurations consisting of ATAT quartets. Formation of quartet structures is consistent both with the experimental data and represents a highly unusual observation of a nucleobase quartet structures at surfaces. 33 The use of selective hydrogen bonding between nucleobases attached to molecular tectons represents a powerful method for the control and design of complex 2D materials. Control over the self-assembly of porphyrins, which themselves play key roles in electro-catalytic 75-77 and optoelectronic 78,79 applications, demonstrates the efficacy and potential of this approach. The formation of nucleobase quartets as the stabilizing hydrogen bond motif in these structures is also an important result and has diverse implications for nucleobase driven 2D self-assembly.
These quartets share many similarities with naturally occurring "quadruplex" structures in DNA. 80 In addition to their biological functions quadruplexes also strongly coordinate to a variety of metals. This ability has seen them find application in schemes for the sensitive detection of metal-ions in solution. 41 Coordination of metal ions by surface supported G quartets on HOPG has previously been demonstrated. 81 Formation of nucleobase quartets during porphyrin self-assembly provides the possibility of multiple different metal centers, coordinated to porphyrins and quartets, arranged into hierarchically ordered surface structures. There is also the possibility that the addition of different metal ions in the presence of nucleobase functional groups could be used as a further control mechanism for the self-assembly process. Along with metal complexation by quartets, the natural extension of this work will be the synthesis and 2D self-assembly of porphyrins functionalized with G and C nucleobase groups followed by the synthesis of asymmetrically functionalized porphyrins 82 containing different combinations of nucleobase groups within a single tecton. Using STM to perform quantitative studies of the