Gold(iii) tetraarylporphyrin amino acid derivatives: ligand or metal centred redox chemistry?

EPR spectroscopy and DFT calculations show that the site of reduction of porphyrinato gold(iii) complexes depends on the counterions X, the meso substituents R and the solvent.


Introduction
Synthetic metallo porphyrins are of increasing interest due to their suitability as chromophores, as well as electron and hole acceptors in articial photosynthetic systems, 1 due to their catalytic and sensing properties, 2 due to their medical applications 3 as well as due to their propensity to stabilise unusual metal oxidation states. Specically, porphyrinato gold(III) complexes have evolved as efficient anticancer drugs. 4 Furthermore, they are catalysts for the cycloisomerization of allenones. 5 Recently, gold clusters with face-on coordinated free-base porphyrins have been reported. 6 Finally, gold(III) porphyrins are suitable ingredients in photoinduced electron transfer chains with the gold(III) porphyrin acting as electron acceptor. 7 The site of gold(III) porphyrin reduction, namely ligand or metal centred, has been discussed controversially. Based on early UV/Vis spectroscopic and theoretical studies the products of the reduction of gold(III) porphyrins had been described as porphyrin-centred p radical anions. 8 In a seminal paper, Kadish, Fukuzumi and Crossley provided compelling EPR spectroscopic evidence that the one-electron reduction of [A H ] + to A H is metal centred giving gold(II) porphyrins (Scheme 1). 9 Only a few ligand types, such as thiolates or thioethers, are capable to stabilise mononuclear gold in the oxidation state +II. 10 Further outstanding examples are the uorosulfate 11 and xenon complexes 12 of Au II . Nitrogen donor ligands such as porphyrinato ligands have been reported to stabilise Au II with respect to disproportionation and dimerization 9,13 to [Au II 2 ] species 14 as well (Scheme 1, A, B, [C + ]*). Disproportionation and dimerisation of [Au II (en) 2 ] 2+ D 2+ has been suppressed by encapsulation in the pores of a zeolite (en ¼ ethylenediamine). 15 The gold(II) porphyrin A H (Scheme 1) prepared by reduction of the corresponding gold(III) porphyrin cation [AH] + with the strongly reducing naphthalene radical anion yielded a broad EPR resonance centred at g av ¼ 2.06. 9 Hyperne coupling to 197 Au has been reported for the central g line [A( 197 Au) ¼ 27 G at 113 K; I( 197 Au) ¼ 3 2 , natural abundance 100%]. 9 An EPR resonance with a signicantly smaller peak-to-peak distance was observed for the gold(II) complex B of hematoporphyrin IX with g t ¼ 2.035, g k ¼ 1.970 and A t ( 197 Au) ¼ A k ( 197 Au) ¼ 15 G at 130 K suggesting a less pronounced metal character (Scheme 1). 13b,e The charge-shied state [C + ]* of a Zn II -Au III bis(tetraarylporphyrin) C + yielded an EPR resonance with g ¼ 2.182, 2.043, 1.979 and A( 197 Au) ¼ 180, 14, n.r. G in frozen toluene solution at 143 K for the gold(II) centre (Scheme 1). 13d The gold complex [Au(en) 2 ] 2+ D 2+ with the pure s-donor ligand ethylenediamine trapped in a zeolite shows g k ¼ 2.239, g t ¼ 2.051, A k ( 197 Au) ¼ 188 G and A t ( 197 Au) ¼ 22 G at room temperature (Scheme 1). 15 Gold(III)-centred reductions have been associated with a signicantly higher reorganization energy (ca. 1.25 eV) than porphyrin-based reductions (ca. 0.6 eV). 13c The large reorganization energy renders gold(III) porphyrins suitable electron acceptors in photoelectron transfer schemes. 7 Moreover, gold(III)-associated counterions should dissociate upon Au III to Au II reduction further retarding the back-electron transfer. For instance, chloride is associated to the Au III centre in solid AuCl(TPP) by electrostatic forces with a gold chloride distance of 3.01(1)Å. 16 Unfortunately, no solid structures of porphyrinato gold(II) complexes have been reported so far and further experimental or theoretical studies are lacking.
We had previously reported synthetically versatile mesosubstituted tetraaryl porphyrins with trans-AB 2 C substitution pattern including A ¼ nitrophenyl, aminophenyl or amidophenyl, C ¼ phenyl carboxylic acid or ester and B ¼ EWG or EDG substituted aryl groups. 17 These porphyrins can be metallated 17 and assembled to multiporphyrin amides, 17b,17c electron donor substituted amide-linked dyads 17a as well as electron donor (ferrocene) and electron acceptor (quinone) substituted amidelinked triads and tetrads 17b with well dened sequences from the N-terminus to the C-terminus. The different meso substituents of the porphyrin amino acids at the B position can be used to modulate the solubilities and to ne-tune the redox potentials which allows to design redox gradients. 17 With this family of porphyrins in hand, we disclose in this contribution the factors that control the relative stabilities of a gold(II) porphyrin and its valence isomeric gold(III) porphyrin radical anion. We report novel meso-substituted Au III porphyrin amino acid derivatives with trans-AB 2 C substitution pattern for potential incorporation into electron transfer chains via amide bonds. These gold(III) porphyrins were inspected by cyclic voltammetry, UV/Vis spectroelectrochemistry and by EPR spectroscopy upon selective one-electron reduction with cobaltocene. We provide strong EPR and UV/Vis spectroscopic evidence that all singly reduced gold(III) porphyrins are well described as gold(II) porphyrins essentially irrespective of the meso-substituents A, B and C and that the porphyrin radical anions are higher energy valence tautomers of the ground state Au II valence isomers. Detailed EPR parameters of the gold(II) porphyrinato complexes were obtained by spectral simulations of the experimental spectra (g tensors, (super)hyperne couplings, valence isomer ratios). The experimental data are corroborated and interpreted with the aid of density functional theory (DFT) calculations in the framework of electron transfer theory.

Results and discussion
Synthesis of free-base porphyrins and (porphyrinato)gold(III) complexes (series [1a] 6 ] in order to acquire NMR spectra with a satisfactory signalto-noise ratio. This shows that the counterion and the meso substituents determine the solubility. All gold(III) complexes were characterised by 1 H NMR, 13 C NMR, 31 P NMR and 2D NMR spectroscopy, IR spectroscopy and high-resolution mass spectrometry. The proton NMR spectra display the expected number and intensities of resonances. The chemical shis vary according to the substitution patterns paralleling the shis of the corresponding free-base porphyrins Ia-IIIa and IVa-IVc. The CH 3 -ester, NH 2 -amine and CH 3 -amide substituents display characteristic resonances at d ¼ 4.1, 4.7, 2.2 ppm, respectively. The [PF 6 ] À counterions show the characteristic septet at d ¼ À144 ppm in the 31 P NMR spectra. Upon auration the characteristic high-eld pyrrol NH resonances of the free-base porphyrin disappear. Furthermore, auration of the free-base porphyrins consistently shis the pyrrole CH proton resonances to lower eld by 0.5 ppm, in accordance with the positive charge of the metal centre. In the IR spectra, characteristic absorptions for group vibrations are found for the ester, amine, amide, nitro, triuoromethyl and acid substituents at around 1719, 1618, 1690, 1520/1346, 1324 and 1716 cm À1 , respectively. The [PF 6 ] À counterions display absorptions for the PF stretching and deformation modes at 835-843 and 556-558 cm À1 , respectively. ESI mass spectra fully conrm the integrity and stability of the complex cations displaying peaks at m/z values corresponding to the intact complex cation (see Exp. section). The differences between the rst and second reduction potentials amount to 0.60-0.68 V which corresponds to very high comproportionation constants of K C > 10 10 for the neutral complexes. 21 Hence, disproportionation of the neutral complexes into the corresponding cations and anions can be safely neglected and spectral signatures aer one-electron reduction will essentially be associated with the neutral complexes.
All gold(III) complexes were reduced electrochemically to the neutral species in an optically transparent thin layer electrochemical (OTTLE) cell using THF as solvent (MeOH for [4a] + ). In all cases, isosbestic points were observed corroborating the reversible nature of the rst reduction process ( Fig. 3 and ESI †). The shis of the Soret and Q bands as well as the observed isosbestic points closely resemble those found for the [Au(TPP)] + /Au(TPP) process in THF (ESI †) and in pyridine 13c or in PhCN. 9 In all cases, except for the [2a] (ref. 22)) in an EPR tube. The redox potential of CoCp 2 is perfectly in-between the rst and second reduction of the gold porphyrins ( Fig. 1 and 2) further avoiding over-reduction. The reaction mixture in the tube was immediately frozen by immersing into liquid nitrogen and subjected to X-band EPR spectroscopy. Hence, we obtained signicantly better resolved EPR spectra than previously reported for neutral porphyrinato gold complexes prepared by reduction of [A H ] + with the strongly reducing naphthalene radical anion in DMF (ca. À3 V vs. Fc/Fc + (ref. 22)). 9 In this case some over-reduction might have been occurred blurring the hyperne couplings to gold and nitrogen nuclei.
Indeed, Au(TPP) as prepared by reduction of [Au(TPP)] + by CoCp 2 in CH 2 Cl 2 shows a well-resolved EPR pattern which could be reasonably simulated by a rhombic g tensor with hyperne interaction to a single 197 Au nucleus (I ¼ 3 2 ; natural abundance 100%) and superhyperne coupling to four 14 N nuclei (I ¼ 1, natural abundance 99.6%). The high resolution allows a very good estimation of the high-eld parameters while the low-eld parameters are less well-resolved (Table 2, Fig. 4). Compared to the isoelectronic Cu(TPP) complex ( 63/65 Cu; I ¼ 3 2 ; combined natural abundance 100%; 17d)] is signicantly reduced in Au(TPP). This suggests a more covalent character of the Au II -N bonds compared to the Cu II -N bonds in their respective TPP 2À complexes. For [C + ]* with strongly electron donating meso substituents at the gold porphyrin a much larger hfc to 197 Au has been reported [A 1 ( 197 Au) ¼ 180 G]. 13d In accordance with the stronger nephelauxetic effect of porphyrins, complex D 2+ with the pure s donor ligand ethylenediamine features a signicantly larger hyperne coupling to 197 Au than Au(TPP) as well. 15 For complexes 1a-3a the broad EPR resonance corresponding to the Au II valence isomer is less well resolved due to the lower symmetry and hence different superhyperne interactions (Fig. 5). Furthermore, the broad Au II resonance is superimposed by a sharp slightly rhombic resonance around g ¼ 2.0. For 2a and 3a, this sharp resonance accounts for approximately 5-6% of the total EPR intensity. The pattern can be satisfactorily simulated by g 1,2,3 ¼ 2.018, 2.005, 1.994 and hyperne coupling to four nitrogen atoms (A 1,2,3 ¼ 1, 12, 1 G). These data t to Table 1 Redox potentials (peak potentials in parentheses) of porphyrinato gold(III) complexes measured 10 À3 M in 0.1 M [ n Bu 4 N][PF 6 ]/THF solution, potentials given relative to ferrocene/ferrocenium  gold(III) porphyrin radical anions 2a 0 and 3a 0 . For 4a-4c prepared in THF or MeOH, the corresponding gold(III) porphyrin radical anions 4a 0 , 4b 0 and 4c 0 are only present in negligible amounts (Fig. 6). Hence, in all these cases the equilibrium between the gold(II) valence isomers 2a-4c and their corresponding porphyrin radical anions 2a 0 -4c 0 is in favour of the gold(II) isomers. The very strong preference of 4a-4c over 4a 0 -4c 0 independent of the meso substituents might be due to a solvent effect overwhelming the substituent effects. Indeed, in THF or in MeOH solvent-separated ion-pairs [4a-4c] Fig. S29 and S30 †). The counterion location might affect the charge and spin distribution in the neutral species as well (vide infra).
The much more intense sharp EPR resonance present in the EPR spectrum of 1a obtained from [1a][PF 6 ] in CH 2 Cl 2 ( Fig. 5a) differs from the sharp resonances assigned to the porphyrin p radical anions 2a 0 and 3a 0 . Indeed, simulations of the resonance suggests the presence of a further radical species with g 1,2,3 ¼ 2.031, 2.005, 1.948 and hyperne coupling to a single nitrogen atom (A 1,2,3 ¼ 2, 17, 2 G). This is in good accordance with a nitroarene radical anion. 23 Hence, this distinct EPR resonance is assigned to a nitrophenyl radical anion valence isomer 1a 00 . The radical distribution 1a : 1a 0 : 1a 00 is estimated as 78 : 3 : 19. The decomposition into the component spectra is displayed in the ESI. † The effect of the type of counterions was probed by adding two equivalents of [ n Bu 4 N]Cl to the solution prior to reduction of the gold(III) porphyrin with CoCp 2 . No signicant changes are observed for Au(TPP), 3a (CH 2 Cl 2 ) or 4c (THF) in the presence of chloride. However, the presence of chloride transforms the 1a : 1a 0 : 1a 00 radical mixture almost completely into a 1a : 1a 0 mixture (65 : 35) as only the gold(II) resonance and the porphyrin radical anion resonance are observed under these conditions (ESI, † Table 2), similar to the 2a 0 and 3a 0 cases. Hence, for the nitro derivative 1a, three possible valence isomers are possible: the gold(II) radical (1a), the porphyrin based p radical (1a 0 ) and a further nitro group based p radical (1a 00 ). Assuming, that rapid freezing does not strongly affect the equilibria of valence isomers, we can conclude that the environment, namely anions and the solvent, appears to inuence these valence isomeric equilibria signicantly. The substituents inuence the equilibria as well, especially, when a strongly electron accepting nitro group is present. A conceivable intervalence transition between 1a and 1a 0 /1a 00 is not detected in the UV/Vis spectrum by comparison with the spectra of 2a and 3a (Fig. 3). This might be associated with the different orbital symmetries of 1a and 1a 0 /1a 00 .
The electronic structure of the gold(II) radicals 1a-4c, the valence isomeric equilibrium 1a/1a 00 and the effect of counterions will be addressed by theoretical methods in the next section.
DFT studies of (porphyrinato)gold(III) complexes (series [1a]    This journal is © The Royal Society of Chemistry 2016 Fig. 7 and ESI †). The most signicant differences between the cationic gold(III) complexes and their neutral congeners are found in the Au-N distances which increase by ca. 4% from 2.051 to 2.124Å in all cases (Table 3). The large changes of the Au-N distances (Table 3) contribute to the reorganisation energy of the reduction process. 13c The gold ions are located nearly perfectly in the centre of the four pyrrolic nitrogen atoms in all complexes. The macrocycle itself displays only minor distortions both in the cations as well as in the neutral complexes. A minor increase of the saddling (B 2u ) distortion is noted in the neutral complexes (Table 3). These metrical data of 1a-4c strongly suggest that the reduction of the metal centre to Au II is favoured in the electronic ground state. A reduction of the porphyrin to its radical anion 1a 0 -4c 0 should result in pronounced macrocycle distortions as well as in small Au III -N (radical anion) bond distances which is not observed. The calculated Mulliken spin densities are in full accordance with these structural parameters. In all neutral complexes the majority of the spin density is located at the metal centre (Mulliken spin density at Au: 0.44), especially in the 5d x 2 Ày 2 orbital ( Fig. 7 and ESI †). The remainder is distributed over the pyrrolic nitrogen atoms in the s-orbitals pointing towards the metal centre (Mulliken spin density at N: 0.14). This clearly advocates a gold-centred radical localised in the s-system of the    almost planar molecule rather than a porphyrin radical anion with the spin delocalised in the p-system of the porphyrin. The DFT determined Au II electronic ground states of 1a-4c perfectly match the experimentally derived ground states. The spin densities are also in full agreement with experimentally determined EPR parameters (g values, 197 Au hyperne coupling and 14 N superhyperne coupling). Compared with the isoelectronic Cu(TPP) [Mulliken spin density at Cu: 0.58; Mulliken spin density at N: 0.105] the spin densities are more delocalised onto the nitrogen atoms which is in agreement with the EPR results as well (ESI †). 17d The special case of the nitro derivative 1a which displays signicant amounts of the nitrobenzene p radical anion valence isomer 1a 00 in the EPR spectrum (Fig. 5a) was treated by DFT methods as well. However, all geometry optimisation attempts (with the employed functional, basis set and tight convergence criteria) converged to the stable Au II valence isomer 1a. In order to get an impression on the spin density distribution in valence isomer 1a 00 , the nitrobenzene radical anion [C 6 H 5 NO 2 ]c À was separately optimised by DFT methods giving NO distances of 1.349Å. These NO distances were then constrained to 1.349Å in geometry optimizations of 1a 00 giving the (constrained) optimised structure of 1a 00 as shown in Fig. 7. The Au-N bond lengths of 1a 00 are fully consistent with a gold(III) oxidation state (Table 3). Compared to [1a] + and 1a the C 6 H 4 NO 2 torsion angle with respect to the porphyrin plane C5-C12-C38-C43 is signicantly reduced from 66.1 and 62.4 to 50.8 suggesting a conjugative electron withdrawing effect of the gold(III) porphyrin as expected for a p-centred radical. The spin density is mainly located at the NO 2 substituent and partially delocalised over the p-system of the porphyrin. The Mulliken spin density at the gold atom in 1a 00 is essentially zero (Fig. 7).
As an unconstrained optimization of 1a 00 was unsuccessful, we investigated the effect of the counterion [PF 6 ] À on the charge and spin distribution in [1a/PF 6 ] À and [1a 00 /PF 6 ] À , respectively. Indeed, we succeeded in optimising both valence isomers [1a/PF 6 ] À and [1a 00 /PF 6 ] À without any constraints (Fig. 8). The Au II valence isomer [1a/PF 6 ] À is preferred by 12 kJ mol À1 . In this Au II isomer [1a/PF 6 ] À the [PF 6 ] À ion is not coordinated to the metal (Au/F 4.076Å) but only hydrogen-bonded to two CH groups of the aryl substituents (Fig. 8a). The spin density is again localised at the metal centre (Mulliken spin density at Au: 0.44) and the pyrrolic nitrogen atoms (Mulliken spin density at N: 0.14). In the nitro-based p radical [1a 00 /PF 6 ] À the [PF 6 ] À ion is much closer to the gold centre (Au/F 3.249Å, Fig. 8b). The presence of the negative charge close to the metal centre stabilises the Au III oxidation state and indeed the gold ion carries no spin density. Au-N distances (2.057Å; [1a 00 /PF 6 ] À ) fully agree with a gold(III) porphyrin but not with a gold(II) porphyrin (2.127Å; [1a/PF 6 ] À ). The N-O distances have increased from 1.284Å in [1a/PF 6 ] À to 1.315Å in [1a 00 /PF 6 ] À as expected for population of N-O antibonding orbitals. The spin density is largely conned to the NO 2 substituent and partially delocalized to the p-system of the porphyrin. The C5-C12-C38-C43 torsion angle of the nitrophenyl substituent decreases from 61.4 ([1a/ PF 6 ] À ) to 50.2 ([1a 00 /PF 6 ] À ) similar to the 1a/1a 00 (constrained) pair. In essence, the intramolecular electron transfer pathway between [1a/PF 6 ] À and [1a 00 /PF 6 ] À encompasses the Au-N and Au/F distances (totally symmetric stretching vibration of the gold coordination sphere), the symmetric NO 2 stretching mode and a phenyl torsional motion (Fig. 8).
With respect to photoinduced electron transfer reactions using porphyrinato gold(III) complexes as electron acceptors we suggest that the initial kinetic reduction product of a porphyrinato gold(III) complex should be a gold(III) porphyrin p radical anion (such as 1a 0 -4c 0 ) due to the smaller activation barrier and the better electronic coupling to electron donors (Scheme 3). In a following intramolecular valence isomerisation the electron shis to the central gold ion (s-system) with concomitant dissociation of the counterion giving the thermodynamic Au II product (such as 1a-4c) (Scheme 3). The latter chemical reaction will render the whole photoinduced ET process irreversible, which is advantageous for further reactivity of the redox sites. In the case of nitro substituted porphyrins a further valence isomer [1a 00 /PF 6 ] À with a nitrophenyl p radical anion is existent as well. Both the solvent, the present ions and the substituents determine the nal charge and spin distribution.

DFT calculations
Density functional calculations were carried out with the Gaussian09/DFT series 25 of programs. The B3LYP formulation of density functional theory was used employing the LANL2DZ basis set. To include solvent effects the integral equation formalism polarisable continuum model (IEFPCM CH 2 Cl 2 ) was employed. No (symmetry) constraints were imposed on the molecules, except for the NO distance constraint for 1a 0 . The presence of energy minima of the ground states was checked by analytical frequency calculations.

Conclusions
Auration of meso-tetraaryl substituted AB 2 C porphyrins with KAuCl 4 in the presence of HOAc/NaOAc cleanly gives the corresponding gold(III) porphyrinato complex cations. Aminosubstituted porphyrins are N-acetylated under these conditions and have to be prepared from the corresponding nitrosubstituted gold(III) porphyrins by reduction with SnCl 2 /HCl. The gold(III) complexes can be reduced at least three times. The potentials slightly depend on the electron withdrawing and donating nature of the substituents. The rst reduction is addressed by UV/Vis spectroelectrochemistry and by EPR spectroscopy. Upon one-electron reduction, the Soret band experiences a small bathochromic shi. The intensity of the Soret band of the electron rich complexes [2a] + (R 2 ¼ NH 2 ) and [4b] + (R 3 ¼ O n Bu) slightly increases upon reduction while all other neutral complexes feature less intense Soret bands as compared to their parent Au III complexes. These spectral data clearly suggest the presence of an unreduced porphyrinato ligand in all cases under these conditions. Chemical one-electron reduction of the porphyrinato gold(III) hexauorophosphate salts by cobaltocene yields the corresponding Au II porphyrin complexes with a characteristic EPR pattern displaying hyperne coupling to 197 Au and 14 N. The degree of 197 Au hfc and g anisotropy places the gold contribution to the spin density in (tetraphenylporphyrinato)gold(II) complexes in between that of [Au(en) 2 ] 2+ (ref. 15) and the neutral gold hematoporphyrin IX complex. 13e DFT calculations fully support the metal centred reduction in all cases, essentially irrespective of the substituent at the meso aryl groups. Only, the nitro substituent reduction competes signicantly with the Au III reduction and a valence isomeric equilibrium between the Au II valence isomer and the nitro p radical anion valence isomer is established. DFT calculations suggest that the position of the counterion triggers the position of the equilibria between the different valence isomers that interconvert by an intramolecular electron transfer process. These ndings allows the usage of meso-substituted Au III porphyrins as electron acceptors and electron storage materials in photo-induced redox processes, almost irrespective of the substitution pattern. Hence, the substituents ne-tune the redox potential or other properties such as solubility without compromising the thermodynamically preferred metal site of one-electron reduction. Combination of electronaccepting gold(III) porphyrins bearing carboxylic acid, amine and amide substituents, as introduced in this report, with lightharvesting porphyrins and electron donating porphyrins via amide connectivity 17 are currently explored in our laboratory and will be reported in due course.