Patrick R.
Nimax
a,
Florian
Zoller
a,
Tobias
Blockhaus
a,
Teresa
Küblböck
a,
Dina
Fattakhova-Rohlfing
b and
Karlheinz
Sünkel
*a
aDepartment Chemistry, Ludwig-Maximilians-University Munich, Butenandtstr. 9-11, 81377 Munich, Germany. E-mail: suenk@cup.uni-muenchen.de
bElectrochemical Storage Department, Institut für Energie- und Klimaforschung: Werkstoffsynthese und Herstellungsverfahren (IEK-1), Forschungszentrum Jülich GmbH Wilhelm-Johnen-Straße, 52425 Juelich, Germany
First published on 18th November 2019
Crystals of the aminotetracyanocyclopentadienyl radical were obtained from the reaction of CaCl2 with Ag[C5(CN)4(NH2)] and recrystallization in MeOH, performed in sunlight. The radical was characterized by X-ray diffraction, EPR and UV Vis spectroscopy as well as by cyclovoltammetry and DFT calculations.
Radical chemistry in organic synthesis and physical applications has been a growing field of research over the past few years. This is particularly due to the rapidly growing field of photoredox catalysis.16,17 Within this field, two major approaches have developed: (1) use of metal-based catalysts, either with “noble”18 or “earth-abundant” metals19 and (2) use of purely organic compounds, mostly organic dyes.20 Especially in the latter field, the importance of “push–pull-chromophores” has been stressed.21 Conceptually related, although not yet tested in the context of photocatalysis, are coordination compounds with “non-innocent ligands”22 and charge-transfer salts of radical cations with polycyano anions.23
Considering the fact that (un-)substituted cyclopentadienyl systems are arguably the most important ligands in transition-metal organometallics, we reasoned that joining all the above-mentioned concepts by preparing coordination compounds with donor–acceptor-substituted cyclopentadienyl anions or radicals might lead to a new class of photoactive compounds. Possible candidates would be cyclopentadienides with donor groups like OMe or NR2 and acceptors like COOMe or CN, some of which are known for quite a while (e.g. the potassium salt precursor of radical I in Chart 1).24,25 In the course of our studies on the coordination chemistry of polynitriles we already had looked at the crystal structures of the Na and Mn(II) salts of the [C5(CN)4(NH2)] (“ATCC”) anion.26 In both structures, the ATCC anion coordinates only via one nitrile function and shows a slight distortion towards an “allylene” form. We decided to have a closer look at the ATCC system.
Scheme 1 Synthesis of different salts of the aminotetracyanocyclopentadienide anion according to Webster. |
Fig. 1 and 2 show the molecular structures of both compounds.
The crystals of 1b show large voids (ca. 80 Å3 per unit cell, corresponding to 15% of the volume) and the final structure refinements were performed using the “SQUEEZE” routine in “PLATON”. The NH4+ cation in 1b might have been formed by Cu(II) assisted hydrolysis of a CN bond during recrystallization in MeOH.25,27 The molecular structure of 1b shows an asymmetric hydrogen bridge between both ammonium nitrogens. The N1–H10 bond is significantly longer than N1–H1 and N1–H2, but also significantly shorter than the N0–H10 bond (Fig. 1). The three hydrogen atoms around N1 could be localized in difference Fourier synthesis and were freely refined, while the remaining hydrogen atoms at N0 were positioned with a riding model and then only a common thermal parameter at fixed positions was refined. It should, however, always be kept in mind that localization of hydrogen atoms from the X-ray data has large uncertainties, particularly when the data set is influenced by large voids as in this case. N1 shows typical tetrahedral angles (99–121°), while in the structure of 2 (Fig. 2) the angles around N1 correspond more to sp2 hybridization (angle sum: 355°). This is supported by a significantly shorter ring carbon bond to the amino nitrogen in 2 than in 1b.
Both compounds show the same distortion to an “allylene” system with the amino group bonded to the central allyl carbon (Table S3, ESI†), as it was observed in the Na and Mn(II) complexes studied by us earlier. When looking at the molecular packing, the presence of large solvent-accessible voids (15% in 1b, 40% in 2, see also Fig. S1 and S2, ESI†) becomes obvious. In 1b ATCC anions form one-dimensional chains in the b direction using H-bridges N1–H1–N3, and these chains are connected to inversion related chains via H-bridges N1–H2–N2, thus forming 12-membered rings. These double-chains are connected in the a direction via the NH4 cations using H-bridges N0–H0B–N5 (see Fig. S3 and S4, ESI†). In 2 the ATCC molecules also form one-dimensional chains parallel to c (base vector: 1 0 −1), using two H-bridges N1–H11–N3 (see Fig. 3).
Fig. 3 Packing diagram of 2, including H-bonds, viewed down a. H1–N3: 2.36(5) Å, N1–N3: 3.24(5) Å, N1–H1–N3: 153(4)°. |
However, parallel chains stack in such a manner that the amino nitrogens are 3.345 Å above the ring plane of the neighbouring chain (see Fig. 4 and Fig. S5, ESI†). A similar stacking was also observed in the structure of the NaATCC salt, published earlier26a by us (see Fig. S6 and S7, ESI†). The only difference in the stacking is that in 2 the projection of the NH2 nitrogen on the ring plane nearly coincides with the centroid (distance N–Ct 3.379 Å), while in NaATCC the corresponding projection is closer to the CNH2 ring carbon (distance N–Ct 3.431 Å vs. a perpendicular distance to the ring plane of 3.342 Å). This arrangement is reminiscent of the π-dimers/π-stacks observed in chloranil and TCNQ anion radicals.28
ICP/AAS analysis of X shows only trace amounts of Ca2+ (0.37%) and no silver; the CHN elemental analysis corresponds to the composition [C5(CN)4(NH2)(O2)·3H2O]. This is similar to the findings in the sterically encumbered radicals A5 and D6 and shows that the compound is not air-stable. The oxygen does not show up in the crystal structure, however. There is a certain possibility that the oxidation product might be formulated as “[C5(CN)4(NH2)·2H2O2·H2O]”, which might contain H2O2 molecules together with water in the crystal voids, but this remains fully speculative.
We then looked at the mass spectra of 2. The EI spectrum shows besides the (M + H)+ peak also the (M + H + O)+, (M + H + CN)+, (M − HCN + O)+ and (M − HCN + O2)+ peaks, while the ESI(+) showed (M − H + O)+ signals as the base peak together with the (M − H + O2)+ peak and some unidentified higher-mass peaks. This is very similar to the mass spectra of A and D, showing again the oxidative instability of 2.5,6
Next we turned to the (solid-state) EPR spectra of 2 and – for comparison – also of the potassium and silver salts 1c and 1e (see Fig. 5). The crystals of 2 provide EPR signals at first, but they vanish over the course of hours. Very surprisingly, however, powdered samples of salts 1c and 1e, which had been exposed to sunlight, also showed EPR signals, with slightly lower g values. For comparison, the radicals A, C, D and G have g values in the range from 2.0013 to 2.0034.
As neither Ag+ nor [C5(CN)4(NH2)]− have unpaired electrons, the most probable explanation is that the anion acts as a “redox-non-innocent” ligand,22i.e. an equilibrium like:
Ag+ + [C5(CN)4(NH2)]− ↔ Ag0 + [C5(CN)4(NH2)]0 |
For further clarification we looked at the electrochemical properties of 1c and X by cyclovoltammetry (Fig. 6 and Fig. S8, ESI†). Both compounds show two pairs of peaks at potentials around E1/2 of 0.45 V and 0.91 V vs. Ag/0.01 M AgNO3, respectively, which could be assigned to the Cp−/Cp0 redox pair and its subsequent oxidation to the Cp0/Cp+ at more positive potentials according to the literature convention. The detection of the corresponding reduction peaks in the reverse scan together with the similar height of the peak currents for reduction and oxidation indicates chemically reversible behavior, implying the high stability of the formed species under the conditions of the electrochemical experiment. A rather large peak-to-peak separation (100 mV and 120 mV for the first and the second pairs of peaks, respectively, without any IR-correction), however, points to a sluggish kinetics (quasi-reversible behavior) of the corresponding redox transformations. For comparison, the CV data of B–F and of I24a,29 are listed in Table 1. As can be seen from these data, the alkyl substituted radicals C and D are formed at very negative potentials and are easily oxidized to the corresponding cations. Both cyano-substituted radicals F and I apparently are easily oxidized to the cyclopentadienyl cations when formed in the electrochemical experiment from the corresponding anions and do not show up in the cyclovoltammograms. Only the CF3 substituted radical E shows similar electrochemical behaviour to our compounds, although it is more easily oxidized to the cyclopentadienyl cation.
Fig. 6 Cyclovoltammogram of 1c and X. The voltammograms were taken in MeCN on the Pt disk electrode. Ag/0.01 M AgNO3 in MeCN was used as a reference electrode. |
Comp. | E (Cp−/Cp°)a | E (Cp°/Cp+)a | Ref. |
---|---|---|---|
a vs. FcH/FcH+; the data for D and I were reported related to SCE and were corrected by subtraction of 0.38 V. b In MeCN. c In CH2Cl2. d Irreversible. e Anodic peak potential (calc. vs. external Fc/Fc+). | |||
B | +0.15c | 7 | |
C | −1.78cd | +0.30c | 8 |
D | −1.91b | +0.58b | 6b |
E | +0.38 , | +0.52 , | 9 |
F | 1.0b,d | 10 | |
I | −0.92 , | 29 | |
X | 0.36b,e | 0.88b,e | This work |
1c | 0.36b,e | 0.89b,e |
Next, we had a look at the UV-Vis spectra of 1c. While in MeOH solution only a broad absorption ranging from 250 to 390 nm could be observed, the reflectance spectrum showed an even broader band ranging nearly over 300 nm, with weak maxima at 220, 277, 386 and 514 nm (Fig. 7, “1c_A”).
After 3 h irradiation at 254 nm, a strong band at 234 nm with a shoulder at 260 nm and a weaker band at 295 nm arise, together with two weak maxima at 380 and 511 nm (Fig. 7, “1c_B”). When the sample is stored in the dark for 8 hours, the absorption peaks at 234 and 295 nm disappear and a spectrum similar to the original one is obtained (Fig. 7, “1c_C”).
In an effort to find some explanations for the obtained results, we performed DFT calculations on the ATCC anion and the radical and zwitterion 1a. Based on recent DFT calculations on the [C5(CN)5] radical we decided to use the ωB97XD functional with a 6-311++G** basis set for calculations, both for the gas and solution phase (PCM model).30Fig. 8 shows a spin density plot calculated for the radical. Tables 2 and 3 show the most important results on the energetics of the anion and the radical.
Anion | Radical | |
---|---|---|
E tot (gas) [a.u.] | −617.891 | −617.738 |
E HOMO (gas) [a.u.] | −0.168 | −0.342 |
E LUMO (gas) [a.u.] | 0.120 | −0.047 |
ΔE (gas) [a.u.] | 0.287 | 0.295 |
ΔE (gas) [eV] | 7.82 | 8.02 |
EAadi,g (radical) [a.u.] | 0.153 | |
EAadi,g (radical) [eV] | 4.149 |
Anion | Radical | |
---|---|---|
E tot (solv) [a.u.] | −617.960 | −617.764 |
E HOMO (solv) [a.u.] | −0.277 | −0.320 |
E LUMO (solv) [a.u.] | 0.002 | −0.034 |
ΔE (solv) [a.u.] | 0.279 | 0.285 |
ΔE (solv) [eV] | 7.60 | 7.76 |
EAadi,solv (radical) [a.u.] | 0.196 | |
EAadi,solv (radical) [eV] | 5.333 |
The adiabatic electron affinity can be calculated from the difference of the total energies of the radical and the anion.30c,31 The obtained gas-phase value of ca. 4.15 eV is significantly lower than the reported calculated values for [C5(CN)5] (5.55 ± 0.03 eV30a,c,31) and [C5(CN)4H] (5.00 eV30c). Since 4.15 eV corresponds to a wavelength of ca. 299 nm, it appears plausible that radical 2 can be produced by irradiation of 1c with UV light at 256 nm. The electron affinity of the radical corresponds to the ionization potential of the anion, and is therefore correlated with the electrochemical oxidation potential of the anion. While exact calculations seem still impossible, most authors agree that a linear relationship of the type
E1/2 = EA − ΔΔGsolv − Eref |
Comparison of the calculated and measured UV-Vis spectra (Fig. S12–S14, ESI†) shows a moderate agreement, which is not unexpected considering that the measured spectra were solid-state spectra and the calculations were performed on solution species. Similarly, comparison of the calculated and observed bond lengths shows poor agreement for 2, and moderate agreement for 1b with the calculated geometry for 1a (Table S3, ESI†). Again, this is not too surprising, since the solid state crystal structures showed both hydrogen bridging and π-stacking, which do not occur in the gas phase used for the calculations.
MS: m/z (DEI+): 180.9 [HATCC+]. 13C{1H} NMR (DMSO-d6): δ = 148.04 (C5-NH3), 116.10 (C1-CN), 115.68 (C2-CN), 94.82 (C2), 79.53 (C1). Elemental analysis (C9N5H3·H2O·1.37HCl): calc. C 43.40, N 28.10, H 2.58, found: C 43.32, N 27.81, H 2.59.
A solution of 1e (164.5 mg, ca. 0.56 mmol) in MeOH (10 mL) was treated with NH4Cl (26.7 mg, 0.50 mmol) in one portion. The mixture was stirred for 8 h at room temperature. The solution was evaporated to a volume of 1 mL and filtered. Addition of CH2Cl2 precipitated 1b as a red powder. Yield 15 mg (15%).
MS: m/z (FAB−): 180.4 [ATCC−]. Elemental analysis (C9N6H6·0.19CH4O·0.11H2O): calc. C 53.54, N 40.75, H 3.40; found C 53.79, N 40.78, H 3.41.
MS: m/z (MALDI−) = 180.2 [ATCC−]. EA (K(C9N5H2)·0.75CH4O·H2O): calc. C 44.82, N 26.80, H 2.70, found C 44.99, N 26.90, H 2.45.
MS: (DEI+): m/z = 168.9 [100%, ATCC + O-HCN], 180.9 [15%, HATCC+]; 185.0 [5%, ATCC + O2-HCN]; 196.1 [1%, HATCC + O]; 207.0 [5%, ATCC + HCN]; (ESI+): m/z 181.1 [10%, HATCC+]; 195.1 [100%, C9N5OH+]; 211.1 [20%, C9N5O2H+]; MS (ESI−): m/z 179.0 [100%, C9N5H−]; 180.0 [95%, ATCC−]; 181.0 [10%, HATCC−]. ICP-AAS Ag 0.0%, Ca 0.35%. Elemental analysis (C9N5H2·O2·3H2O): calc. C 40.61, N 26.302, H 3.03, found: C 40.76, N 26.31, H 2.99.
Footnote |
† Electronic supplementary information (ESI) available. CCDC 1909579 and 1909580. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c9nj04354c |
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2020 |