Pyrazinacenes exhibit on-surface oxidation-state-dependent conformational and self-assembly behaviours

Acenes and azaacenes lie at the core of molecular materials’ applications due to their important optical and electronic features. A critical aspect is provided by their heteroatom multiplicity, which can strongly affect their properties. Here we report pyrazinacenes containing the dihydro-decaazapentacene and dihydro-octaazatetracene chromophores and compare their properties/functions as a model case at an oxidizing metal substrate. We find a distinguished, oxidation-state-dependent conformational adaptation and self-assembly behaviour and discuss the analogies and differences of planar benzo-substituted decaazapentacene and octaazatetracene forms. Our broad experimental and theoretical study reveals that decaazapentacene is stable against oxidation but unstable against reduction, which is in contrast to pentacene, its C–H only analogue. Decaazapentacenes studied here combine a planar molecular backbone with conformationally flexible substituents. They provide a rich model case to understand the properties of a redox-switchable π-electronic system in solution and at interfaces. Pyrazinacenes represent an unusual class of redox-active chromophores.


Supplementary Note 1.
Variation in the height profiles of molecules contained within the line structures can be assigned to conformational variation of the phenyl substituents of the compounds, i.e. different dihedral angles between the planes of phenyl groups and the plane of the pyrazinacene backbone. For 2, variation in the phenyl group conformation is relatively rare and appears as 'defects' in the line structures.
The observed defects within the arrays correspond to different conformers being included in the arrays; the STM profile height is lower due to rotation of the phenyl groups of one molecule. Possibly in order to reduce stresses within the array, the molecules adjacent to this defect rotate their phenyl groups and are of brighter STM contrast and greater height, as shown in Supplementary Fig. S13b (Conformations B2 and C2). Therefore, there exist three different conformations of 2 (indicated in the STM image overlay of Supplementary Fig. S13b) although that closely resembling the crystal structure conformation (Conformation A2) dominates. In contrast, phenyl group dihedral angle variation in 1 (Figure Supplementary S13a) is significantly more common although there appears again to be three possible S17 conformations. The contrast of STM images of the line structure of 1 suggests that conformations with larger dihedral angles between phenyl groups and pyrazinacene backbone are preferred. It is difficult to specify these angles (suggested values are given in Supplementary Figure S13) but again rotation of the phenyl rings leads to their different heights in STM and allowed us to identify three distinct conformations, A1 (similar conformation to energy minimized structure), B1 (with a single phenyl group with a lower dihedral angle with the pyrazinacene) and C1 (where two phenyl groups at the same end of the molecule have lower dihedral angles). It is likely that this difference in conformational preference between 1 and 2 (i.e., 1 commonly adopts a range of conformations while 2 largely adopts a conformation similar to its crystallographic form plus occasional defects) originates in interactions with the substrate lattice so that 2 is more easily accommodated in its preferred conformation while 1 is required to adapt (through dihedral angle variation) in order to form the line structures. Therefore, the line structures may be the lower energy state favored by both molecules over a dispersed state at the expense of phenyl group dihedral angle variation.

Supplementary Note 2.
To discuss the energies of the lines in the XPS N1s spectra we have analyzed the electronic populations of the adsorbed molecules. The Voronoi charges S1,S2 gained/lost for each atom are given in Supplementary Fig S16 above. In particular, it can be seen that for 2-ox and 2-ox2 the Voronoi populations are not related to the average distance between the nitrogens and the surface. Indeed, the geometry of the 2-ox system indicates that the nitrogens on one side of the central part of the molecule are closer to the surface than those on the other side. The average value for the Voronoi population, however, is -0.16 in both cases (i.e. left and right side of the central π system). For the 2-ox2 system, the average value is very close to that obtained in the case of 2-ox, i.e. -0.17 e. While the average values are the same in both cases, differences up to 0.04 e may occur between specific atoms, most probably caused by the local geometry: the distance between a given nitrogen and the closest Cu atom will specifically influence its electronic population. Indeed, in vacuum these differences amount to less than 0.01 e in most cases. For 2. we see that N bonded to H atoms exhibits population around -0.04 e, while for the other nitrogens exhibiting a free electron pair, the average population is -0.16 e. This agrees favorably with the observation that only the native molecules 1 and 2 exhibit two peaks. The first lower intensity signal is due to nitrogens bonded to N-H (corresponding to a population of -0.04 e) with a second one specific to other pyridine-type N. This peak is rather similar in the Voronoi charge argument regardless of the position of the N atom within the molecule and whether it is dehydrogenated or cyclodehydrogenated. This peak corresponds to an average electron population of -0.16 to -0.17 e. All this reasoning supports the experimental assignment that the molecules 1 and 2 are physisorbed, not chemisorbed, to the substrate. (Supplementary Information Pages S19-S20).
Supplementary Figure 17. Photoreduction of 2. a, Electronic absorption spectra of 2 (in dichloromethane) taken at 10 second intervals during irradiation with a high intensity UV lamp. At 0 s and 10 s, 2-ox is largely present with 2 predominating after 20 s. Beyond 30 s a shift in absorbance maximum may be due to decomposition to an unidentified product. Reducing intensity of the absorption maximum also suggests decomposition, which is in contrast to 1 where isobesticity is observed for this process. b, Photographs of solutions of 2-ox stored in the dark (upper) or stored under room light (lower) taken at 1 day intervals revealing a gradual reappearance of the orange hue due to 2 in solution stored in light. Figure 18. Thermal analyses of 1. a, Thermogravimetric analysis. Weight loss of 7% from room temperature corresponds well with loss of 2.5 mols of water (expected from elemental analysis -see synthesis section). Peaks in the DTA trace correspond to a phase transition (301 °C, probably melting) and commencement of decomposition or dihydropyrazine dehydrogenation (440 °C). The latter is signified by appearance of the deep blue colour due to the decaazapentacene chromophore -see b. b, Solid state electronic absorption spectra of 1 at room temperature (orange line) and after heating in air at 420 °C for 10 minutes.

Supplementary
Supplementary Figure 19. Thermal analyses of 2. a, Thermogravimetric analysis. Total loss of volatiles at low temperatures corresponds well with that found by elemental analysis (⁓25 %). Decomposition commences at ⁓400 °C. The gradual loss of weight up to decomposition of this sample is due to reactions involving decomposition of tetrahydrofuran solvent of crystallization. b, Solid state electronic absorption spectra of 2 at room temperature (orange line) and after heating in air at 400 °C for 10 minutes. There is a ⁓50 nm red shift in the absorption maximum of 2 in the solid state. Figure 20. Additional thermogravimetric analysis (TGA). a, TGA of 1. Weight loss of 7% from room temperature corresponds well with loss of 2.5 mols of water (expected from elemental analysissee synthesis section). Peaks in the DTA trace correspond to a phase transition (301 °C, probably melting) and commencement of decomposition or dihydropyrazine dehydrogenation (440 °C). b, TGA of 2. Total loss of volatiles at low temperatures corresponds well with that found by elemental analysis (⁓25 %). Decomposition commences at ⁓400 °C. The gradual loss of weight up to decomposition of this sample is due to reactions involving decomposition of tetrahydrofuran solvent of crystallization. Note that 2 also sublimes close to its decomposition temperature during TGA measurements so that the remnant mass is low.

Supplementary Figure 21.
Cyclic voltammetry for 2-ox revealing a first reduction at -0.293 V (in CHCl3/0.2 M tetrabutylammonium perchlorate; scan rate: 100 mV s -1 ). 1-ox could not be obtained in a form suitable for electrochemical measurements due to its poor solubility. Supplementary Scheme S1 (also Supplementary Table S1 -see below). The reactions were run for 24 hours with 1 mol% of the catalyst by irradiating the reaction mixture with light emitted by a Royal Blue LED under air at 25 °C. Both pyrazinacenes were capable of causing C-C bond formation between THIQ and nitromethane. The catalytic performances of 1 and 2, however, differ slightly (see Supplementary   Table S1). According to the observed conversions and isolated yields, octaazatetracene 2 proved to be a more efficient catalyst than decaazapentacene 1. Due to the greater compatibility of the absorption maximum of 2 (⁓450 nm) with the high energy emission band of the Royal Blue LED (⁓450 nm), 2 afforded better conversion (72 %) and isolated yield (70 %) of the CDC product than 1 (67% and 63 %, resp.). Note that 1 compared to 2 possesses an absorption maximum red-shifted by ⁓50 nm. (Fig. 5a