First π-linker featuring mercapto and isocyano anchoring groups within the same molecule: synthesis, heterobimetallic complexation and self-assembly on Au(111)

Azulene is a convenient platform for accessing heterobimetallic complexes and self-assembled monolayers of a π-linker with asymmetric junctions.


Introduction
Mercapto (-SH) and isocyano (-N^C) substituents are among particularly popular anchoring groups in coordination and surface chemistry as they are well-known to provide stable junctions at metal/organic interfaces. [1][2][3] Even though dimercapto-and diisocyano-functionalized molecular linkers have long been attracting interest of theorists [4][5][6][7][8][9] and experimentalists [10][11][12][13][14][15] in the quest for efficient organoelectronic materials, [16][17][18][19][20] species containing both -SH and -N^C functionalities in the same molecule are not presently known and constitute a formidable synthetic challenge. Indeed, a mercapto group is incompatible with reaction conditions commonly employed to form an isocyano substituent, 21 whereas free organic isocyanides are unlikely to tolerate chemical environments typically involved in the syntheses of mercaptans (thiols). [22][23][24] In the context of targeting isocyanothiols for bridging metal-based electron reservoirs, a potentially straightforward strategy to circumvent the above dilemma would be to anchor either the -N^C or the -SH terminus of such a hypothetical linker prior to forming and tethering its other end. There is only one related example in the literature, albeit not involving a mercapto group per se but rather its disulde surrogate. 25,26 In their elegant approach to covalently bind nickel clusters to a gold surface via the 4-isocyanophenylthiolate bridge, Kubiak and coworkers attached both -N^C ends of otherwise non-isolable 1,2-bis(4isocyanophenyl)disulde to trinuclear nickel clusters in the m 3 ,h 1 fashion. 25 The resulting salt, [{Ni 3 (m 3 -I)(m 2 -dppm) 3 (m 3 ,h 1 -C^NC 6 H 4 S-)} 2 ] 2+ (I À ) 2 (dppm ¼ bis(diphenylphosphino) methane), underwent homolysis of its S-S moiety upon exposure to a gold surface to give rectifying, presumably ionic, monolayer lms. 26 Earlier this year, Ratner and van Dyck proposed a new paradigm for the design of efficient molecular rectiers that involved two p-conjugated units asymmetrically anchored to metallic electrodes and separated by a decoupling bridge. 27 Their intriguing theoretical study suggested mercapto and cyano (-C^N) junctions for accommodating the asymmetric anchoring on the premises that the -SH and -C^N termini would facilitate alignments of a linker's HOMO (Highest Occupied Molecular Orbital) and LUMO (Lowest Unoccupied Molecular Orbital), respectively, through Fermi level pinning. 27 We note that, from a practical standpoint, mercapto/isocyano asymmetric anchoring would be worth considering as well, given that the isocyano group offers a substantially more stable junction within a wider range of organometallic platforms compared to its isomeric cyano congener. 2 Herein, we introduce chemistry of the rst, to the best of our knowledge, p-linker equipped with mercapto and isocyano anchoring groups. The linker's core is comprised of the nonalternant aromatic framework of azulene, a substitution-free molecular diode which has, among other unusual physicochemical characteristics, complementary orbital density distributions within its Frontier molecular orbitals (Fig. 1). 28
Our synthetic approach to constructing and metalating 3a is shown in Scheme 1. Treating pink 2-formamido-6-bromo-1,3diethoxycarbonylazulene 32,34 with ethyl 3-mercapto-propionate in reuxing pyridine afforded persimmon-coloured thioether 4 in a high yield. Dehydrating the 2-formamido group of 4 cleanly provided peach-red 2-isocyanoazulene derivative 5. Unlike 1,2bis(4-isocyanophenyl)disulphide (vide supra), 25 5 is thermally and air-stable for practical purposes and can be stored under ambient conditions for at least a few weeks without spectroscopically ( 1 H NMR, FTIR) detectable deterioration. Compound 5 reacted with Cr(CO) 5 (THF) via its 2-NC end to form orange Cr 0 adduct 6. No product featuring the thioether S / Cr(CO) 5 interaction 35 was documented in this reaction. The [(-NC) Cr(CO) 5 ] moiety of 6 tolerated the basic environment and subsequent acidication of the reaction mixture used to convert 6 into auburn organometallic thiol 7, which constitutes 3a with its 2-NC terminus anchored to the 16-e À [Cr(CO) 5 ] fragment. Metalation of the 6-SH end of 7 with PPh 3 AuCl under basic conditions yielded orange-red crystals of heterobimetallic Cr 0 / Au I complex 8 aer a simple workup.
The solid-state structure of 8$ 3 4 CH 2 Cl 2 features two very similar but crystallographically independent molecules of 8 in the asymmetric unit that are linked together via a weak Au/Au interaction 36 of 3.2102(4)Å ( Fig. 4, 5, S3 and S4 †). The partially positively charged 7-membered ring of the highly polarizable azulenic moiety in each of these molecules of 8 undergoes donor-acceptor face-centred stacking 37 with a Ph-ring of the other molecule's PPh 3 ligand giving the intercentroid distances 38 of 3.65 and 3.76Å. Heterobimetallic complex 8 may be viewed as a hybrid of our X-ray structurally characterized mononuclear Cr 0 and Au I adducts of 1 and 2, respectively, depicted in Fig    "aromatic donor-acceptor interactions". 37 The above structural perturbations do not signicantly affect the Au-S-C angle in 8 compared to that in 10, which are ca. 107.8 105.0 , respectively. Notably, the solid state structure of 10 exhibits neither aurophilic nor aromatic stacking interactions akin to those observed for 8. 29 The metric parameters for the octahedral [(-NC)Cr(CO) 5 ] core in 8 are quite similar to those observed for 9 (ref. 32) and many other complexes (ArylNC)Cr(CO) 5 . 39 Comparison of the Cr-CN and C^N bond distances 40 for 8 and 9 ( Table 1) may hint that the 2-isocyanoazulene ligand in 8 has a somewhat higher sdonor/p-acceptor ratio than that in 9, thereby reecting the difference in electron-donating/withdrawing characteristics of -SAuPPh 3 versus -N^C groups at position 6 of the azulenic scaffold. However, this suggestion should be taken cum grano salis as such subtle variations in d(Cr-CN) and d(C^N) are statistically ambiguous, especially under the 3s criterion. More drastic changes in the electronic nature of the isocyanide ligand's substituent do lead to signicant alterations in the Cr-CN and C^N bond lengths in (RNC)Cr(CO) 5 as illustrated in Table 1 for R ¼ t Bu (ref. 42) and FC^CF 2 . 43 Compounds 4-8 are highly coloured substances. The lowest energy electronic absorption band for 5 occurs at 484 nm (3 ¼ 1.55 Â 10 3 M À1 cm À1 ) and is 259 cm À1 red-shied compared to the S0 / S1 transition documented for 4 ( Fig. S1 †). This red shi arises from the greater electron-withdrawing inuence of the 2-isocyano group in 5 versus the 2-formamido group in 4 on the energy of the azulenic scaffold's LUMO (Fig. 1b). 31,44 The UVvis spectra of 6 and 7 are nearly identical and feature very intense absorption bands at 454 (3 ¼ 3.1 Â 10 4 M À1 cm À1 ) and 452 (3 ¼ 2.6 Â 10 4 M À1 cm À1 ), respectively, that have a substantial contribution from the dp(Cr) / pp*(CNAzulenyl) charge transfer ( Fig. 7 and S1 †). Our time-dependent DFT (TD-DFT) calculations for 7 suggest that the transition at 452 nm (TD-DFT: 416 nm) has 85% HOMO / LUMO character (Fig. 8a). Upon metalation of 7 to form 8, this band not only red-shis to 469 nm (TD-DFT: 463 nm for 8a, the truncated model of 8 featuring OMe and PMe 3 groups instead of OEt and PPh 3 , respectively, Fig. 8b) but also more than doubles in intensity (3 ¼ 5.4 Â 10 4 M À1 cm À1 ). This intensity gain is due to the addition of the n(S) / pp*(CNAzulenyl) character to the HOMO / LUMO transition observed for 8 (cf. the 445 nm band for 10 in Fig. 7). 29 As in the case of 9 and 10, 29,32 the LUMOs of 7 and 8a constitute the p*-system of the azulenic moiety with contributions from both anchoring groups while their HOMOs involve the entire 2-isocyano-6-azulenylthiolate motif (Fig. 8).
From Table 2 it is evident that as the net electron-releasing ability of X decreases (SAuPPh 3 > SCH 2 CH 2 CO 2 CH 2 CH 3 > SH > H > Br > N^C), the d( 13 CN) value for the isocyano carbon resonance increases in the range spanning ca. 8 ppm, thereby signifying gradual drop in the s-donor/p-acceptor ratio of the 2isocyano-6-X-azulene ligand. Concomitantly, both d( 13 CO trans ) and d( 13 CO cis ) values decrease, albeit in tighter chemical shi ranges ($1.0 and $0.5 ppm, respectively), indicating reduction in the electron richness of the Cr-centre. Even though the 13 C chemical shis of terminal CO and CNR ligands in low-valent complexes are inuenced considerably by the paramagnetic shielding term, s para , which reects the degree of p-backbonding, 40,45,46 it is more appropriate to interpret Dd( 13 CN) and Dd( 13 CO) as a combined s-donor/p-acceptor effect.
Closer examination of the 13 C NMR data in the top six rows of Table 2 unveiled remarkably consistent inverse-linear relationships d( 13 CO trans ) vs. d( 13 CN) and d( 13 CO cis ) vs. d( 13 CN), as illustrated in Fig. 9. This gure also conrms that remote modulation the Cr-centre's electron richness mediated by the 2,6-azulenic framework affects the trans-CO ligand to a greater extent than the cis-CO's of the [(NC)Cr(CO) 5 ] moiety. Would the trends depicted in Fig. 9 hold beyond the 2-isocyanoazulenic series? To address this question, we considered (RNC)Cr(CO) 5 species containing strongly electron-withdrawing substituents R, for which 13 C NMR data acquired in the same solvent (CDCl 3 ) were available (bottom four rows in Table 2). The expanded d( 13 CO trans ) vs. d( 13 CN) and d( 13 CO cis ) vs. d( 13 CN) plots that, in     Fig. 10, which again demonstrates excellent inverse-linear correlations now spanning substantially wider Dd( 13 CN) and Dd( 13 CO) windows. The above d( 13 CO)/d( 13 CN) NMR analysis serves as a convenient tool for quantifying even subtle electronic inuence of a CNR ligand's substituent R. In this regard, it offers a simple alternative to the well-established method involving correlation the carbonyl 13 C chemical shis with the corresponding CO force constants (k CO ) for complexes (RNC)Cr(CO) 5 . 40,49,50 Unfortunately, changes in k CO due to mild electronic perturbations of the R group are oen not clearly discernible. 40,51 Determining the values of k CO 's under the C 4v symmetry for complexes (RNC) 5 Cr(CO) 5 using the Cotton-Kraihanzel (C-K) approximation 52 is a straightforward but somewhat tedious task that carries fundamental limitations [52][53][54] and relies on the availability of the complete n C^O vibrational prole 52 (G nCO ¼ 2A 1 + B 1 + E, e.g., Fig. 11 (ref. 55 and 56) and Table S17 †). In the IR spectra of LM(CO) 5 species, the lower energy n C^O (A 1 ) band is oen obscured by the intense n C^O (E) band, 40,51-54 which compromises the accuracy of experimental determination of this n C^O (A 1 ) value (vide infra).
Similar to the trend in d( 13 CN) for the (RNC)Cr(CO) 5 adducts in Table 2, the 13 C NMR resonance for the terminal C-atom in the available uncoordinated 2-isocyanoazulenes moves upeld upon increasing electron-donating power of the substituent X at the azulenic 6-position (d ¼ 179.9, 32 178.0, 177.5, 176.3 ppm in CDCl 3 for X ¼^C, Br, H, SCH 2 CH 2 CO 2 CH 2 CH 3 , respectively). Yet, the n N^C stretching frequency for these free 2-isocyanoazulenes (2126 AE 1 cm À1 in CH 2 Cl 2 ) is insensitive to the nature of the group X. However, upon proceeding from 8 to (6, 7, 11) to 12 to 9, the n N^C band undergoes a small red shi (Table 3), thereby suggesting decrease in the s-donor/p-acid ratio of the isocyanide ligand, especially when 8 is compared to 9 and 12. Fig. 12a shows the FTIR spectrum of thiol 7 in CH 2 Cl 2 . In addition to the characteristic n S-H and n N^C bands at 2583 and 2140 cm À1 , respectively, it features a typical pattern in the n C^O stretching region for a LM(CO) 5 species. 57 The band at 2049 cm À1 corresponds to the n C^O mode A 1 (1) where all ve CO    ligands vibrate in-phase (cf. Fig. 11). The very weak band at 2000 cm À1 is due to the n C^O vibration of B 1 -symmetry, which is IRforbidden under the strict C 4v symmetry but gains slight intensity because of minor deviations of the structure from the idealized C 4v geometry. The intense n C^O band at 1958 cm À1 chiey represents the doubly degenerate vibration of Esymmetry. This n C^O (E) band obscures the remaining IR-active n C^O mode A 1 (2) . Interestingly, perturbations of the local C 4v symmetry in 7 through crystal packing interactions in the solid state are sufficient to split the E-mode into two separate n C^O peaks while unmasking the original A 1 (2) mode (Fig. 12b).
Exposing ca. 1 Â 1 cm 2 gold substrates to a 2 mM solution of 7 in CHCl 3 without protection from air and ambient lighting reproducibly afforded self-assembled monolayer (SAM) lms of 7 on the Au(111) surface. This chemisorption process is presumably accompanied by formation of the thiolate junction and the release of H 2 . 8,17,58 The reection absorption infrared (RAIR) spectrum of the SAM of 7 on Au(111) is shown in Fig. 13a. In addition to the n N^C absorption at 2135 cm À1 , it features two n C^O bands. The n C^O region in this RAIR spectrum, however, is quite different from that in Fig. 11a in terms of peak intensities and energies. The lowest energy intense n C^O band in the solution IR spectrum of 7, which is primarily attributed to the n C^O mode of E symmetry, practically vanishes upon the SAM formation, while simultaneously uncovering the hidden A 1 (2) band of much lower intensity. This observation implies approximately parallel orientation of the cis-CO ligands with respect to the gold surface. Indeed, surface IR selection rules 59 dictate that only vibrations contributing to dipole changes perpendicular to the surface are IR-active. Consequently, any vibrations occurring nearly parallel to the surface would have low IR intensity. Given that the C-N-C unit in 7 is expected to be essentially linear, the appearance of the RAIR spectrum in Fig. 13a suggests upright orientation (i.e., straight C-S-Au surface angle) of the molecules in the SAMs of 7.
The "hollow-linear" coordination of organic thiolates in their SAMs on Au(111), akin to that depicted in Fig. 13b, has been predicted to accommodate the strongest S-Au interaction and induce S / Au(111) charge transfer via S(3p)-Au p-bonding. 60,61 In the context of the chemistry presented herein, this means that the gold surface would effectively function as an electronwithdrawing "substituent", thus, enhancing p-acidity of the 2isocyanoazulene ligand and, in turn, decreasing electron richness of the [Cr(CO) 5 ] unit. The A 1 (1) and A 1 (2) n C^O bands at 2058 and 1995 cm À1 in the RAIR spectrum in Fig. 11 both exhibit signicant blue shis compared to the corresponding n C^O peaks in the solution FTIR spectrum of 7 (2049 and 1958 cm À1 , respectively, Fig. 10a). The magnitudes of these shis appear to be too high, especially in the case of the A 1 (2) mode, to be attributed solely to differences in intermolecular interactions within the SAM vs. solution of 7. The larger change in energy of the n C^O A 1 (2) mode compared to that of the A 1 (1) mode upon chemisorption of 7 stems from the greater contribution of the trans-CO stretch to the former. 62 The tilt angle of the aromatic moiety in SAMs of benzenoid mercaptoarenes on Au(111) can be highly variable. 17 We have recently shown that 2-mercaptoazulene and several of its derivatives form monolayer lms on Au(111) with approximately upright assembly of the azulenylthiolate constituents. 63 Our optical ellipsometry measurements on multiple SAM samples of 7 provided consistent SAM thickness values that nicely corroborate the monolayer nature of these lms and upright orientation of the molecules on the gold surface (Table  4). In terms of their composition, the SAMs of 7 and 9 on Au(111) differ only in the surface anchoring group (thiolate vs. isocyanide) and appear to exhibit essentially identical thicknesses. 64 Notably, neither RAIR spectroscopic nor ellipsometric data collected for the SAMs of 7 on Au(111) would be consistent   with the "on-top-bent" 59 or any other adsorption models of 7 invoking a bent C-S-Au surface geometry. The ellipsometric measurements on SAM lms formed from our recently reported 29 6-mercapto-1,3-diethoxycarbonyl-azulene and 6-mercapto-2-chloro-1,3-diethoxycarbonylazulene also corroborate that these 6-mercaptoazulenes self-assemble on Au(111) surfaces in the upright fashion (Table 4).

Conclusions
The asymmetric nonbenzenoid aromatic framework of azulene proved to be a convenient platform for accessing the rst plinker terminated with both mercapto and isocyano junction moieties. Anchoring the 2-isocyano end of this linker was an important prerequisite to successfully installing its 6-mercapto terminus. The 13 C NMR signatures of the octahedral [(-NC) Cr(CO) 5 ] core in related complexes 6,7,8,9,11, and 12 provided a sensitive spectroscopic handle for tuning electron richness of the Cr 0 -centre through mediation by the 2,6-azulenic framework. Moreover, the remarkably consistent inverse-linear trends d( 13 CO trans )/d( 13 CN) and d( 13 CO cis )/d( 13 CN) for a wide spectrum of complexes (RNC)Cr(CO) 5 offer a simple and more accurate alternative to the d( 13 CO)/k CO strategy in quantifying electronic inuence of the substituent R in isocyanide ligands. This 13 C NMR approach utilizes feedback from the entire [(-NC)Cr(CO) 5 ] unit rather than focusing on the [Cr(CO) 5 ] fragment in the d( 13 CO)/k CO method. In addition, the C 4v -symmetric [(-CN) Cr(CO) 5 ] moiety served as a distinctly informative n N^C /n C^O infrared reporter for probing self-assembly of the 6-mercaptoazulenic motif on the Au(111) surface. We hope that the chemistry of the 2-isocyano-6-mercaptoazulenic platform introduced herein will facilitate further development and experimental validation of the emerging concept of asymmetric anchoring relevant to the design of organic electronics materials. Efforts to access and isolate completely free (i.e., unmetalated) 3a are currently in progress. a Average of ve measurements at different spots on multiple SAM samples. b Calculated from the X-ray structural data for 8, 6-mercapto-1,3-diethoxycarbonylazulene (ref. 29), and 6-mercapto-2-chloro-1,3diethoxy-carbonylazulene (ref. 29), as well as by assuming straight C-S-Au surface angle and the Au(111)-S distance of 2.45Å (ref. 58).