1,2,4,5‐Tetrakis(tetramethylguanidino)‐3,6‐diethynyl‐benzenes: Fluorescent Probes, Redox‐Active Ligands and Strong Organic Electron Donors

Abstract In this work, the change of reactivity induced by the introduction of two para‐ethynyl substituents (CCSi(iPr)3 or CCH) to the organic electron‐donor 1,2,4,5‐tetrakis(tetramethylguanidino)‐benzene is evaluated. The redox‐properties and redox‐state dependent fluorescence are evaluated, and dinuclear CuI and CuII complexes synthesized. The Lewis‐acidic B(C6F5)3 substitutes the proton of the ethynyl −CCH groups to give new anionic −CCB(C6F5)3 − substituents, leading eventually to a novel dianionic strong electron donor in its diprotonated form. Its two‐electron oxidation with dioxygen in the presence of a copper catalyst yields the first redox‐active guanidine that is neutral (instead of cationic) in its oxidized form.

In several works we described intramolecular (reversible) electron-transfer processes between GFAl igands and metal atomsi nm ono-a nd dinuclear copper complexes, [11,[45][46][47] including the first dinuclear copper complexess howingr eversible, thermally stimulated redox isomerism (also denoted valence tautomerism). [47] Starting with the archetypical compounds 1a and 1b,s everal derivatives were obtained by substituting the two remaining aromatic protons (e.g. by halides, [26,48] nitro [48] or even additionalg uanidinog roups [27] ), or by modifying the guanidino groups. [49,50] These substitutions affect the redox properties as well as the optical properties. [24] Herein we report on the synthesis and the chemistry of compounds, in which the two remaininga romatic protons of 1a/ 1b are substituted by ethynyl groups.F igure 1s hows the Lewis structures of the three compounds 2a, 2b and 3 studied in this work. The synthesis of 2a was described in ap reliminary work. [51] As detailed in the following, the peculiarities of these three compounds are the redox-state dependentf luorescence, and the additional reactivity inscribed by the ethynyl groups (especially for compound 3). Moreover,t he first dinuclear metal complexes of 2a and 3 are synthesized and analysed.

Results and Discussion
Synthesis and characterization of 1,2,4,5-tetrakis(tetramethylguanidino)-3,6-diethynyl-benzenes The synthesis of the three compounds (see Scheme 2) commencesw ith 4,7-dibromo-2,1,3-benzothiadiazole. Conversion to 5,6-dinitro-4,7-bis[2-[tris(1-methylethyl)silyl]ethynyl]-2,1,3benzothiadiazole is followed by reduction to give 1,2,4,5-tetra(amino)-3,6-bis-[(triisopropylsilyl)ethynyl]benzene. Reaction with chloro-N,N,N',N'-tetramethyl-formamidinium-chloride leads to 2a (43% isolated yield) [51] and reactionw ith 2-chloro-1,3-dimethyl-4,5-dihydro-1H-imidazolium-chloride leads to 2b (14 %i solated yield). The low isolated yield of 2b is due to its relativelyh igh solubility in organic solvents that hampers its isolation by precipitation. Removal of the two silyl groups from compound 2a is achieved with tetrabutylammonium fluoride in THF,y ieldingp ure compound 3 in good yield (78 %). The addition of an extra protons ource is not required.T he presence of terminal alkynes was evidenced by NMR and IR spectroscopy.H ence, the two protons of the alkyne groups show at d = 3.05 ppm in the 1 HNMR spectrum. In the IR spectrum, sharp absorptions at 3260 and at 2084 cm À1 are assigned to the alkyne stretching modes n(CÀH) and n(CC), respectively. Interestingly, the three compounds differ distinctly in their solubility.C ompound 2a is soluble in THF,b ut much less soluble in Et 2 Oo rt oluene. It is completely insoluble in more polar solvents sucha sM e 2 CO or CH 3 CN. It is highly soluble in CH 2 Cl 2 , but decomposes in this solvent within hours to unknown products.B yc ontrast, 2b is much more soluble in CH 3 CN or toluene. Compound 3 is generally barely soluble in standard organic solvents, and seems to decompose within hours in tol-uene and especially in CH 2 Cl 2 solution.P lease note that cyclic voltammetry studies in CH 2 Cl 2 are still possible (see below), but no reactions of thesec ompounds in this solvent could be carried out. For comparison, compounds 1a and 1b are stable and soluble in CH 3 CN andC H 2 Cl 2 solutions.T he differences in solubility and stability limit ac omparison of the reactivity of the three compounds.
Ta ble 1c ompares some bond parameters for 2a, [51] 2b and 3,a nd the solid-states tructures of 2b and 3 are visualized in Figure 2. In similarity to the structures of other GFAs, the CN 3 planes of the guanidino groups are highly twisted with respect to the central aromatic C 6 ring plane (see the analysiso ft his issue in ref. 52). Due to this preferredc onformation,t here is no steric strain in the molecule. The imino N=Cb ond lengths( N1-C4/N4-C9i n2a/2b and N1-C7/N10-C22 in 3)a re similar for all compounds (shortest and longest bonds of 1.283(3) and 1.294(4) ,r espectively), and fall in at ypical range for N=C double bonds in neutral guanidines. [24] These bonds are very sensitivet oc hanges in the electronic structure, and are elongated significantly upon protonation, metal coordination or oxidation (see discussion below).
Scheme2.Synthesis of the 2a, [51] 2b and 3.The yieldsrefer to the isolated, pure compounds.  Next, we inspected the opticalp roperties of the three compounds.D ue to the huge differencei ns olubility,t he spectra had to be recorded in different solvents. The optical properties of all discussed compounds are collected in Ta ble 2. In the electronic absorption spectra,a ll three compounds 2a, 2b and 3 display one band in the visible region,w ith maximao fa b-sorptionat433 (2a in THF), 429 (2b in THF) and 420 (3 in toluene) nm (see Figure 3f or compound 3 in toluene). The extinction coefficient is only slightly higherf or 2a than for 2b (by ca. 10 %), but significantly highert han for 3.A ll three compounds show relatively strong fluorescence (maximum of emission at 502 (2a), 504 (2b)a nd 500 (3) nm, see Figure 3), in differencet ot he fluorescent-silent compounds 1a and 1b.T he quantum yields increase in the row 3 (F = 12 %) < 2a (F = 18 %) < 2b (F = 31 %). The more rigid guanidino groups in 2b might be responsible for the remarkable difference in the quantum yield between 2a and 2b (both in THF solution). In this context it is worth noting that the quantum yield of fluorescence of 2a in solution massively increases upon decrease of the temperature. [51] Quantum-chemical calculations (B3LYP/def2-TZVP) were carried out to get information about the nature of the electronic transition. The calculatedl owest-energy electronic transition (TD-DFT calculation) is in excellent agreementw ith the experi-mentalr esults( observed: 433 nm for 2a and 420 nm for 3; calcd 429 nm for 2a [51] and 403 nm for 3), and can safely be assignedt ot he HOMO!LUMO transition (see SupportingI nformation, Figures S47 and S48). The C 6 ring and the guanidino groups,b ut not the ethynyl groups contribute to the HOMO orbital. By contrast, the LUMO is localized on the C 6 ring and the ethynyl groups, and the guanidino groups contribute only marginally (see Figure 4). Hence,i nt he HOMO!LUMO transition an electron is excited from one p-system to an orthogonal p-system, like in typical cross-conjugated cruciform fluorophores. [53] Redoxp roperties The redox properties are first analysed in electrochemical studies. In Ta ble3,t he redox potentials obtainedf rom cyclic voltammetry (CV) for the compounds 2a, 2b and 3 are compared with those of 1a and 1b.I na ll cases aq uasi-reversible twoelectron redox process is observed. At high potentials, ao neelectron redox process follows,l eading eventually to the GFA trication. The alkynyl groups shift the redox potential to slightly higher values. This shift is larger for the CCH groups than for the CCSi(iPr) 3 groups. Compound 3 in CH 2 Cl 2 solution (see Figure 5) shows the quasi-reversible two-electron redoxp rocess, assigned to the redox couple 3 2 + /3,w ith the highest potential (E 1/2 = À0.61 V, E ox = À0.49 V) of the tetrakis-guanidine compounds studied herein. Another reversible one-electron process whichi su sually observedf or GFAs (GFA 2 + /GFA ·3 + ), is also expected for 3.H owever,t he potential window in dichloromethane and the one-electron process seem to be in Table 2. Comparison of the optical properties for several compounds:    Motivated by the resultso ft he cyclic voltammetry measurements,w er eacted compound 3 with oxidizing reagents. Reaction of 3 with two equivalents of ferrocenium hexafluorophosphate in acetonitrile at room temperature indeed leads to clean two-electron oxidation( Scheme3). The product salt 3 2 + (PF 6 À ) 2 ,o btained in 89 %i solated yield, can be re-crystallized by slow diffusion of diethyl ether into an acetonitrile solution.
In the UV-vis spectrum, as mall bathochromic shift of the lowest-energy absorption from l max = 420 nm to 433 nm upon oxidation is measured (see Ta ble 2). Interestingly,t his small shift is accompanied by am assive increaseo ft he extinction coefficient (by af actor of 5.2). Moreover,o xidation completely extinguishes the fluorescences ignal.
As already mentioned, compound 3 consists of two crossconjugated p-systems. The donor p-system, being the HOMO of the neutral compound,i nvolves the aromatic ring and the guanidino groups. The acceptor p-system, being the LUMO of the neutral compound, involves the aromatic ring and the alkynyl groups. Hence the compound could be described as a cross-conjugated cruciformc hromophore. For the dication 3 2 + , the LUMO is localized on the central C 6 ring and the guanidino groups (see Supporting Information, Figure S48), in similarity to the HOMO of the neutralc ompound. However,t he HOMO (a g symmetric) and HOMO-1 (a u symmetric)o f3 2 + are centred on the alkynyl groups, the central C 6 ring, and the guanidino groups.F or 3,t he lowest-energetic electronic excitation (calculated at 402.7 nm) is ap ure HOMO!LUMO transition. According to TD-DFT (B3LYP/def2-TZVP), the HOMO!LUMO transition of 3 2 + (calculated at 675.9 nm) is symmetry forbidden, since both orbitals exhibit a g symmetry.T hus, an electronic excitation with high HOMO-1 ! LUMO character (77.5 %), calculated at 439.2 nm, is assigned to the observed band at 433 nm (see Supporting Information, Figure S47). The distinct changes of the electronic excitationsa re responsible for the extinction of fluorescence upon oxidation of 3.
Hence compound 3 shows distinct redox-state dependent fluorescence, meaning that the fluorescences ignal could be used as aprobe for its redox state.
In another experiment, we reacted 2a with CuI. This reaction leads to the dinuclear Cu I complex [2a(CuI) 2 ]i n8 2% isolated yield. Figure 8d isplays the structure of the complexi nt he solid state. As expected, the imino N=Cd ouble bond lengths increaseu pon coordination, from 1.294(4)/1.284(4) in 2a . Interestingly, the fluorescence is completely extinguished, in line with the results obtained for coordination of Cu I to tetrakisguanidinophenazine ligands. [55] TD-DFT calculations (B3LYP/def2-TZVP) found ar elatively strong electronic excitation (HOMO-2! LUMO) at 453.5 nm and aw eak excitation (HOMO!LUMO) at 486.4 nm (see SupportingI nformation,F igures S51 and S52). While the LUMO is centred predominantly at the C 6 ring and the ethynyl groups, the HOMO and HOMO-1 are located on the C 6 ring, the guanidino groups and the CuI groups.H ence the orbitals involved in the electronic excitations are significantly different to those involved for free 2a.
Then, we reactedt he complex [2a(CuI) 2 ]w ith an excesso fI 2 (3 equivalents) in an attempt to isolate ac omplex with an oxidized guanidinel igand unit. However,t he metal-free salt (2a) 2 + (I 3 À ) 2 is isolatedi np ure form in 55 %y ield (see Scheme 5a). This result indicates that the metal-ligandb onds break upon ligand oxidation. In the case of the analoguec omplex [1a(CuI) 2 ], reaction with I 2 gives ad iamagneticc oordination polymer{ [ 1a(CuI) 2 ](I 3 ) 2 } n with twofold oxidized bridging guanidine ligand units(see Scheme5b). [56] Interestingly,t his chain polymer is found to be an electric semiconductor with a relative small band gap of 1.05 eV (as estimated from an Arrhenius plot of the temperature dependence of the electrical conductivity). Hence all attempts to obtain ad inuclear copper complex with the oxidized, dicationic form 2a 2 + as ligand, failed. The reason for the distinctly different ligand behaviours of 2a and 1a is not yet clear,but it might arise from the slightly higherr edox potential of 2a (see Ta ble 3) and probably also from the differences in solubility and applieds olvents that might shift possible equilibria to other sides.
We also studied the coordination chemistry of compound 3 (Scheme 6). Complexation with CuI gave [3(CuI) 2 ]i n6 3% yield. Reactiono f3 with Cu(OAc) 2 resulted in the formation of the complex [3{Cu(OAc) 2 } 2 ]i n5 5% yield. Hence, dinuclear Cu I as well as Cu II complexes of the neutral ligand could be synthesized. Thes olid-state structures of both complexes are illustrated in Figure 9, and selected bond lengths are compiled in Ta ble 4. In both cases, the fluorescencei sc ompletely extinguished upon copper coordination (see the analysiso ft he electronic excitations for [3(CuI) 2 ]w ith TD-DFT in the Supporting Information, Figures S49 and S50).
Cyclic voltammograms of [3{Cu(OAc) 2 } 2 ]i nC H 2 Cl 2 solution show only irreversible redox processes (see Supporting Information, Figure S33), for example, two oxidation waves at  À0.30 Vand À0.05 V, as well as ab road shoulder at À0.45 V. A sharp reduction wave is detected at À0.45 V. The irreversibility of the redoxe vents might point againt ot he cleavageo ft he metal-ligand bonds upon ligand oxidation. The complex [3(CuI) 2 ]i ss table in solution under inert-gas,b ut is rapidly transformed to other products upon contact to air (see Supporting Information for ap reliminary UV-vis spectroscopic study on this issue, Figure S30). In this case, CÀCc oupling reactions might take place. The product is not soluble in standard organic solvents, in line with ap olymerics tructure. The rational synthesis of such coupling products is an attractive goal, which is however clearly outside the scopeo ft his work.
The synthesis of [3(CuCl 2 ) 2 ]w as attempted but the complex could not be isolated, suggesting as imilar reactivitya sc ompound 2a.A gain we observe ad ifferent behaviour to that of 1a,f or which the dinuclearc opper complex [1a(CuCl 2 ) 2 ]i s formed.
The results of this study show that compounds 2a and 3 could be used for the synthesis of dinuclear Cu II and Cu I complexes. However,w ith the oxidized form of the ligands, the complexes are not stable and the metal-ligandb ond is cleaved. This is in marked contrast to the properties of 1a, that formss table complexes in the neutral and in the oxidized form. The differences are most likely caused by the higher redox potentials of 2a and 3 compared with 1a,a nd to some extend maybe also by the differences in the applied solvents (which are necessary due to the large differences in solubility) that might affect the positiono fe quilibria. For 3,f urther reactivity arises from the terminal alkynylg roups,a nd is currently studied in our group.
Reactivity at the terminal alkynyl hydrogens of compound 3 Next, we tested the possibility to replace the protons from the two terminal alkynyl groups by reactionw ithaL ewis acid. Indeed, reactiono fc ompound 3 with two equivalents of tris-(pentafluorophenyl)boranei ntoluene at 60 8Cgives the neutral zwitterionic bis-alkynylboronate compound 4 in 65 %i solated yield (Scheme 7). In this reaction, the proton of each terminal alkyne group is replaced by the borane, and the released proton captured by one of the guanidino groups. The reaction is an example of terminal alkynea ctivation by frustrated Lewis pairs. The previously reported reactions of at erminal alkyne RCCH (various rests R, for example, Ph or H, were tested) with the frustrated Lewis pair combination B(C 6 F 5 ) 3 and ab ulky Lewis base LB (e.g. tBu 3 P) yield salts [LBH] + [RCCB(C 6 F 5 ) 3 ] À that could react furtherwith the Lewis base or acid. [57,58,59] In our reaction, the terminala lkyne and the basic guanidino groups are assembled in one molecule, and therefore an overall neutral compound is obtained.
The compound is soluble in CH 2 Cl 2 (in contrast to 2a/2b or 3 withouts igns of decomposition) and acetone, but insoluble in most other solvents (including CH 3 CN). It can be crystallized from as aturated dichloromethane solution. Compound 4 is only weakly fluorescent,w ith the maximum of emission at 445 nm (l ex = 315 nm), showingasignificant shift compared to 500 nm for compound 3 (l ex = 420 nm) The fluorescences ignal is extremely temperature-sensitive. It rises at lower temperatures and decreasesath ighertemperatures (see Supporting Information,F igure S37). Twofoldd eprotonation of this compound would result in ad ianionic, extremely electron-rich GFA. Unfortunately,a ll attempts to deprotonate this compound (using triethylamine, butyllithium or sodium amide) failed and resultedi nt he recovery of unreacted 4.Ont he other hand, oxidation of 4 coupled with deprotonation using catalytic amountso fc opper salts with O 2 is successful, giving the zwitterionic compound 5 in 44 %i solated yield. The catalysti s equal to that previously used for oxidation of protonated 1a with O 2 (see also Scheme 1). [30] The new compound 5 is quite soluble in Me 2 CO or THF,b ut to our surprise almost insoluble in CH 2 Cl 2 .T he solid-state structures of 4 and 5 are shown in Figure 10, and selected structural parameters are compiled in Ta ble 5. In 4,t he CÀCb ond distances in the central C 6 ring vary only slightly (1.402(3)/1.407(3) and 1.415(3) for C1-C2/ C1-C3 and C2-C3). By contrast, they vary much in 5 (1.497(2)/ 1.382(3)/1.449(3) for C1ÀC2/C1ÀC3 and C2ÀC3), indicating loss of aromaticity.M oreover,t he N1ÀC1 and N4ÀC2 bond lengths are considerably shorteri n5 compared with 4.O nt he other hand, the effect of oxidation on the bond lengthsw ithin the alkynyl groups is miniscule. Hence the structural comparison between 4 and 5 is in line with the Lewis structures in Scheme7.
As observed upon oxidation of 2a or 3,t he electronic absorptiono f4 in the visible region experiences ab athochromic shift upon oxidation (accompanied in this case by deprotonation), from 401 nm in 4 to 455 nm in 5,a nd also am assive increase in its extinction coefficient (by af actor of 3.6). The fluo-rescence, being already small in 4 at room temperature, is extinguished in 5.
Cyclic voltammetryw as used to obtain information about the reduction potential (see Figure 11). In DMF solution, ar eversible two-electron redox process, assigned to the redox couple 5/5 2À ,i sd etected at E 1/2 = À0.83 Vv s. Fc + /Fc (E ox = À0.75 V). Ao ne-electronp rocess, assigned to the redoxc ouple 5 · + /5,o ccurs at E 1/2 =+0.91 Vv s. Fc + /Fc (E ox =+0.97 V). Another oxidation wave at E ox =+1.26 Vv s. Fc + /Fc clearly belongs to an irreversible redox event, presumably leading to degradation. Moreover,t he voltammogram shows weaker waves (at À0.33 Vi nd irection of oxidation and À1.40 Vi nd irection of reduction). These waves are presumably caused by the extremelyh igh reactivity of 5 2À ,t hat quicklyu ndergoes reactions with dioxygen or other oxidizing impurities. Hence the redox potential of the reduced, dianionic form 5 2À is significantly lower than those of 1a or 2a.I nf act, 5 2À has the lowest redox potential of all tetrakisguanidines. On the other hand, its Figure 10. Illustration of the solid-state structures of 4 and 5 in the solid state.D isplacemente llipsoids drawn at the 50 %p robability level. Hydrogen atoms bound to nitrogen (green colour)were locatedi ndifference Fourier syntheses and refined, either fully or with appropriate distance and/or symmetry.Methyl hydrogen atoms omitted. redox potential is still slightly higher than that of the strongest guanidinee lectron donor, hexakis(N,N'-dimethyl-N,N'-ethyleneguanidino)-benzene, for which an E 1/2 value of À0.96 Vv s. Fc + /Fc was obtained. [27] So far,i tw as not possible to isolate a salt of the dianion 5 2À ,w hich appearst ob ee xtremelyr eactive and sensitivetodioxygen.

Conclusions
In this work the chemistry of redox-active1 ,2,4,5-tetrakis(tetramethylguanidino)-3,6-diethynyl-benzenes (compounds 2a, 2b and 3)a re studied. Substitution of the remainingt wo hydrogens of the redox-activeg uanidine1 ,2,4,5-tetrakis(guanidino)benzene by ethynylg roups leads to redox-activec ompounds with redox-state dependentf luorescence. The fluorescenceo f the neutralr educed forms is extinguished upon oxidation. The four guanidino groups allow the use of the compounds as redox-active bridging ligandsi ns everald inuclear Cu I and Cu II complexes.I nc ontrastt o1 ,2,4,5-tetrakis(guanidino)benzene, the guanidine-metal bond is cleaved upon ligand oxidation. One of the new compounds synthesized in this work has two terminal alkynyl groups (3). Reaction of this compound with two equivalents of the Lewis acid B(C 6 F 5 ) 3 leads to migration of the two CÀHp rotons to the guanidinog roups andf ormation of two new CÀBb onds by addition of two equivalents of the borane (4). Hence the combination 3/B (C 6 F 5 ) 3 acts as a frustrated Lewis pair that activates the terminal alkyne groups. The catalytic oxidation/deprotonation of 4 with dioxygen leads to the first redox-activeg uanidine thati sn eutral( instead of dicationic)i ni ts twofold oxidized state (5). Consequently,i ts reductiono ccurs at the lowest reduction potential ever measured forredox-active tetrakis-guanidine compounds.
The results of this study show that redox-active1 ,2,4,5-tetrakis(tetramethylguanidino)-3,6-diethynyl-benzenesd isplay ad iverse chemistry.T he topic of ongoing research in our group is their use (after substitution of the protons in 3 by organic groups with suitable functionalities) as building blocks for the construction of metal-organic frameworks. We are also systematically studying how substituents at the alkynyl groups affect the redox-state dependent fluorescence properties.