Rinnakkaistallenteet Luonnontieteiden ja metsätieteiden tiedekunta
2018
Cyanide-assembled d10 coordination polymers and cycles: excited state
metallophilic modulation of solid-state luminescence
Koshevoy IO
Wiley-Blackwell
Tieteelliset aikakauslehtiartikkelit
© Authors
All rights reserved
http://dx.doi.org/10.1002/chem.201704642
https://erepo.uef.fi/handle/123456789/6303
Downloaded from University of Eastern Finland's eRepository
Supported by
Accepted Article
Title: Cyanide-assembled d10 coordination polymers and cycles:
excited state metallophilic modulation of solid-state luminescence Authors: Igor O. Koshevoy, Andrey Belyaev, Toni Eskelinen, Thuy
Minh Dau, Yana Yu. Ershova, Alexei S. Melnikov, Sergey P.
Tunik, and Pipsa Hirva
This manuscript has been accepted after peer review and appears as an Accepted Article online prior to editing, proofing, and formal publication of the final Version of Record (VoR). This work is currently citable by using the Digital Object Identifier (DOI) given below. The VoR will be published online in Early View as soon as possible and may be different to this Accepted Article as a result of editing. Readers should obtain the VoR from the journal website shown below when it is published to ensure accuracy of information. The authors are responsible for the content of this Accepted Article.
To be cited as: Chem. Eur. J. 10.1002/chem.201704642
Link to VoR: http://dx.doi.org/10.1002/chem.201704642
FULL PAPER
Cyanide-assembled d 10 coordination polymers and cycles:
excited state metallophilic modulation of solid-state luminescence
Andrey Belyaev,
[a]Toni Eskelinen,
[a]Thuy Minh Dau,
[a]Yana Yu. Ershova,
[a]Sergey P. Tunik,
[b]Alexei S.
Melnikov,
[c]Pipsa Hirva,
[a]and Igor O. Koshevoy*
[a]Abstract: The series of cyanide-bridged coordination polymers [(P2)CuCN]n (1), [(P2)Cu{M(CN)2}]n (M = Cu 3, Ag 4, Au 5) and molecular tetrametallic clusters [{(P4)MM´(CN)}2]2+ (MM´ = Cu26, Ag2
7, AgCu8, AuCu9, AuAg10) were obtained using the bidentateP2 and tetradentate P4 phosphane ligands. All title complexes were crystallographically characterized to reveal zig-zag chain arrangement for 1 and 3–5, while 6–10 possess metallocyclic frameworks with different degree of metal-metal bonding. The d10–d10 interactions were evaluated by QTAIM computational approach. The photophysical properties of1–10 were investigated in the solid state and supported by theoretical analysis. The emission of compounds1 and3–5, dominated by MLCT transitions located within {CuP2} motifs, is compatible with TADF behaviour and a small energy gap between theT1 andS1 excited states. The luminescence characteristics of6–
10 are strongly dependent on the composition of the metal core, the emission band maxima vary in the range 484–650 nm with quantum efficiency reaching 0.56 (6). The origin of the emission for6–8 and10 at room temperature is assigned to delayed fluorescence. Au–Cu cluster9, however, exhibits only phosphorescence that corresponds to theoretically predicted large value E(S1–T1). DFT simulation highlights a crucial impact of metallophilic bonding on the nature and energy of the observed emission, the effect being greatly enhanced in the excited state.
Introduction
Optical functionality of luminescent compounds and materials is determined by the characteristics of electronic states that are involved in radiative transitions.[1] The inorganic and organometallic complexes exhibit a selection of excited states, accessible via a number of charge transfer (CT) processes, which implicate an active participation of the metal ion valence shell (e.g.
metal-centered transitions, metal to ligand or ligand to metal CT).
Alternatively, the electronic transitions may occur mainly within a
coordinated organic motif (intraligand CT) though being substantially perturbed by the neighbouring metal center.[2]
Understanding the ways to control the nature and the composition of the frontier molecular orbitals provides a fundamental basis to influence both ground and excited states through the variation of their relative energies, multiplicity, rates of intersystem crossing and spin-orbit coupling. This foundational knowledge substantially facilitates rational molecular design of metal complexes that is aimed at the synthesis of functional systems with desirable photophysical properties, specifically required for such important applications like diverse photonic devices,[3] photocatalysis,[4]
imaging[5] and sensing techniques.[6]
The complexes of coinage metals, particularly those of copper, have been an active field of successful research as cheaper and environmentally friendlier substitutes of Os, Ir and Pt-based luminophores,[2g, 7] One of the stimuli that has largely driven the basic research of CuI and AgI light-emitting compounds, is the phenomenon of thermally activated delayed fluorescence (TADF),[8] which allows reaching exceptionally high efficiencies of electroluminescence through harvesting of all triplet and singlet electrogenerated excitons, without utilizing noble metal complexes as phosphorescent emitters. For the TADF materials, it is essential to have a small energy difference between the lowest singlet S1 and triplet T1 excited states, comparable to thermal energy kBT that ensures efficient reverse intersystem crossing (RISC) T1 S1 at ambient conditions, leading to a relatively long-lived fluorescence S1 S0. In order to achieve small splitting E(S1–T1), the overlap between frontier orbitals (HOMO and LUMO) has to be also small to diminish the exchange interaction of unpaired (excited and non-excited) electrons.[9] With respect to coordination complexes, this charge separation can be realized in the case of metal-to-ligand charge transfer (MLCT) states, frequently encountered for copper(I) compounds[7c, 8a, 10]
though more rarely observed for silver(I) congeners.[8b, 11]
Evidently, for tuning the emission parameters (wavelength, intensity, lifetime) one needs to alter rationally the ligand environment, paying a special attention to the electronic properties of the ligand, which accommodates the excited electron (i.e. where LUMO is mostly localized). Thus, a wide variation of luminescence colour often requires substantial synthetic efforts for the preparation of a library of suitable ligands.[12]
Another option to modulate the energy of excited states while retaining the coordination sphere is to employ metal–metal interactions, which are frequently found among the closed shell d10 ions of copper subgroup, and are capable to affect considerably the photophysical properties of the di- or polymetallic molecules.[2g, 13] These aggregates are typically phosphorescent due to the strong spin-orbit coupling (SOC), [a] A. Belyaev, T. Eskelinen, T. Minh Dau, Y. Yu. Ershova, Prof. P.
Hirva, Prof. I.O. Koshevoy
Department of Chemistry, University of Eastern Finland, Yliopistokatu 7, Joensuu, Finland
E-mail: igor.koshevoy@uef.fi [b] Prof. S.P. Tunik
Institute of Chemistry, St.-Petersburg State University 26 Universitetskiy pr., Petergof, St.-Petersburg, Russia [c] Dr A. Melnikov
Peter the Great St.-Petersburg Polytechnic University Polytechnicheskaya, 29, St.-Petersburg, Russia
Supporting information for this article is given via a link at the end of the document.
Accepted Manuscript
which is enhanced by metallophilic bonding. The latter was shown to have a crucial influence on the origin of emission and can be responsible for “switching” from TADF behaviour to pureT1 S0
triplet emission.[14] Notably, in the tetranuclear TADF luminophore [Cu4 2-dppm)4(CN)2]2+ the metallophilic contacts, even not operating in the ground state, may get activated upon photoexcitation as a result of molecular reorganization facilitated by a relative lability of the cyanide bridges.[15] This approach to influence the properties of the emissive states through changing non-covalent bonding in polymetallic systems seems to be an appealing strategy to design the emitters with adjustable optical behaviour that does not imply significant chemical modification of the ligand surrounding.
To investigate further this concept, we aimed at potentially dynamic cyanide-linked systems, which allow incorporating different d10 ions. For this purpose, utilization of multidentate phosphanes has proven to be a convenient tool to promote the formation of d10 polymetallic assemblies of different nuclearities.[2g,
13] In our previous work we employed chelating triphosphane bis(o-diphenylphosphinophenyl)phenylphosphane,P3 (Figure 1), for the preparation of intensely emissive cyanide compounds with MLCT-chromophore motifs M(P3)CN (M = Cu, Ag, Au).[16] Herein, we probe the related di- and tetraphosphane ligands (1,2- bis(diphenylphosphino)benzene P2, and tris(2- diphenylphosphinophenyl)phosphaneP4, Figure 1) to construct a series of coordination polymers and discrete metallocyclic cyanide compounds, in order to reveal a possible impact of the constituting metals and of the metal–metal interactions on the emission mechanism and photophysical performance of structurally similar polynuclear aggregates.
Figure 1.Chelating di-, tri-[16] and tetraphosphane ligandsP2–4 utilized for the synthesis of coinage metal cyanide complexes.
Results and Discussion
Synthesis and characterization. Diphosphane coordination polymers. Treatment a dimethylformamide solution of copper cyanide with equivalent amount of the diphosphane P2 under anaerobic conditions readily produces the copper polymer [(P2)CuCN]n (1) as green-yellow microcrystalline solid in good yield (Scheme 1). The analogous reaction of silver cyanide with P2 gives however only the ionic compound [(P2)2Ag][Ag(CN)2] (2, Figure S1 in the Supporting Information), the cation of which has been structurally investigated in the forms of PF6- and BF4-salts.[17]
Changing the molar ratio of the starting reagents CuCN:P2 to 2:1 allows for the isolation of another polymeric material formulated
Scheme 1.Synthesis of complexes1–5.
Figure 2.Fragments of the polymer chains formed by complexes1 (top) and5 (bottom). Thermal ellipsoids are shown at the 50% probability level. H atoms are omitted for clarity.
as [(P2)Cu{Cu(CN)2}]n (3), which, however, is formed in a mixture with1. Alternatively, when [Cu(NCMe)4]+ is first reacted with 0.5 eq of P2, followed by the addition of stoichiometric quantity of sodium cyanide, polymer3 is produced selectively. In an attempt to synthesize the heterometallic congeners of3, the Ag and Au cyanides were dissolved in the presence ofP2 and subsequently mixed with CuCN solution in DMF to afford crystalline complexes [(P2)Cu{M(CN)2}]n (M = Ag,4; Au,5).
The XRD structural analysis reveals zig-zag patterns of the polymer chains for1 and3–5, reminiscent of [(Cy3P)Cu(CN)]n[18]
and [(Ph3P)2M{M´(CN)2}]n[19] monophosphane congeners, respectively (Figure 2 and Figure S2; selected bond lengths
Accepted Manuscript
FULL PAPER
Figure 3.Molecular views of complexes6,8 and9. Thermal ellipsoids are shown at the 50% probability level. H atoms and counterions are omitted for clarity.
Symmetry transformations used to generate equivalent atoms (´): in6, –x, 1–y, –z; in8, 2–x, 1–y, 1–z.
and angles are listed in Table S2). In the case of1, the {(P2)Cu}
motifs are connected through the µ2 2-cyanide bridges, which complete pseudo-tetrahedral coordination geometry of the metal centers. The assignment of the cyanide orientation shown in Figure 1 is arbitrary, as the crystallographic data does not allow for clear distinguishing of the carbon and nitrogen atoms of the CN groups. However, different Cu–C/N bond lengths (1.935(3) and 2.000(3) Å) evidently point to the (P2)Cu(C,N) coordination environment of the copper ions. In general, the structural parameters (Cu–CN, Cu–P distances and angles) in1 agree with those for the related compounds bearing P2 or resembling ligands.[10c, 11a, 12d, 16, 18]
The polymers [(P2)Cu{M(CN)2}]n (3–5) are isomorphous (space groupP212121) and feature essentially isostructural chains, which are composed of copper-diphosphane blocks linked by means of nearly linear dicyanometallate [NC–M–CN]- fragments, similarly to the triphenylphosphane-stabilized coinage metal complexes [(Ph3P)2M{M´(CN)2}]n.[19] The {(P2)Cu} moieties are twisted with respect to the direction of the chain propagation that results in a helical arrangement of these assemblies (Figure 2).
Tetraphosphane clusters. It has been shown in our previous work that P4 tetraphosphane coordinates to CuX species via three phosphorus atoms leaving one non-bound –PPh2 group, which is potentially capable of additional interaction.[20] Following this concept, d10 cyanides can be easily depolymerized with 1 equivalent of P4 ligand to give presumably the mononuclear species (P2)M(CN), which efficiently assemble into tetranuclear clusters [{(P4)MM´(CN)}2]2+ (6–10) upon coupling with AgCF3SO3
or [Cu(NCMe)4]PF6 salts (Scheme 2). The attempts to synthesize the tetragold analogue using the AuCN and [Au(tmbn)2]SbF6
(tmbn = 2,4,6-trimethylmethoxybenzonitrile) were unsuccessful probably because of markedly lower coordination number of AuI ions compared to CuI and AgI ones, therefore an excessive amount of donor functions prevents cluster formation. Also, a relatively poor affinity of gold(I) to N-ligands might additionally inhibit formation of stable Au–C N–Au bridges.
Complexes 6–8 and 10 adopt 8-membered metal-containing cycles with clearly visible metal-metal bonding, supported by the
tetraphosphane (Figures 3 and S3, the structural data are listed in Table S3). The observed Cu–M (M = Cu,6; Ag,8) and Ag–M (M = Ag,7; Au,10) distances are shorter than the sums of the corresponding Van der Waals radii and correlate well with metal–
metal contacts reported earlier for the homo- and heteronuclear compounds of d10 Cu subgroup metal ions.[2g, 20-21] In this respect, the Cu–Au complex9 is structurally rather different from6–8, and 10 as it exhibits substantially longer M–M´ separations
Scheme 2.Synthesis of complexes6–10.
(the Cu…Au distances are 3.328 and 3.484 Å). The existence of metallophilic bonding was further evaluated by assessing the topological charge density of the computationally optimized structures. Quantum Theory of Atoms in Molecules (QTAIM) analyses show the relatively high values for the delocalization index, (A,B), between metal atoms in6,7 and8 that is indicative of appreciable metal-metal bonding.[22] For 9 and 10 the corresponding parameter is significantly lower, suggesting negligible interactions between Au and Cu or Ag. The following trend is observed for the value of the delocalization index between metal atoms (values of (A,B), which are defined as an average number of shared electrons between two bonding atoms, are given in parentheses):8 (0.23) >6(0.18) >7(0.16) >>10(0.09)
>9(0.07), which correlate well with the experimentally observed metal-metal distances, see Table S3.
Accepted Manuscript
The cluster frameworks are stabilized by the unsymmetrical bridging mode of the cyanide carbon atoms, which somewhat pull together the neighbouring metal ions. The {MM´(CN)}2
metallocyanide cores in compounds6,7 and10 represent nearly flat systems, while in the congener 8 the {CuAg(CN)}2 motif features a chair-like conformation. The latter configuration of the eight-membered metal-containing rings results in a large deviation of the Ag–N–C angles from linearity (137.4o) in comparison to6,7,10, in which the corresponding angles range from 159.8o to 167.7o. The absence of µ2 1-CN interactions in9 is likely because of the saturated coordination vacancies of both metal ions by -donation from P4 and cyanide ligands. QTAIM results (Table S4) support the stabilizing bridging effect of the CN groups on the {MM´(CN)}2motif. The interaction energies at the M´–C(1) bond critical points, calculated by the method proposed by Espinosa et al.,[23] range from –63 kJ/mol to –24 kJ/mol for complexes 6–8 and from –18 kJ/mol to –6 kJ/mol for9 and10, showing the same relative trend as the delocalization index between metal atoms for the respective compounds. Small negative values for H (total local energy) at the BCPs (bond critical points) and the above unity values for |V|/G at the BCPs in 6–8 point to a partially covalent character of a secondary M–µ2 1- CN bonding.[24] However, in complexes9and10, the positive H and the values for |V|/G < 1 are indicative of purely closed shell interactions.
The bonding mode of theP4 phosphane ligand in6–10 depends on the nature of the constituting metal centers. The pronounced tendency of CuI ions to reach a tetracoordinate environment makes all the phosphorus atoms to participate in metal binding in copper complexes, that is reflected by quite similar P–Cu distances in6 (2.263–2.297 Å),8 (2.289–2.319 Å) and9 (2.281–
2.337 Å). On the other hand, in Ag2 (7) and Ag–Au (10) compounds the P(1)–M interactions involving central phosphorus donor are essentially weaker that is indicated by P(1)–Ag (7, 2.578 Å) and P(1)–Au (10, 2.832 Å) contacts, which are visibly longer than typical P–Ag/Au bonds. Computational analysis confirms this trend: in complex10P(1)–Au interaction energy is only –13 kJ/mol, whereas for P(2)–Au bondEi(BCP) equals to – 158 kJ/mol.
According to the mass spectroscopic measurements performed in the ESI+ mode, no doubly charged molecular ions could be detected (Figure S4). Instead, only the signals of the dinuclear monocationic species [(P4)MM´(CN)]+ are observed, that is indicative of possible reversible dissociation of 6–10 in solution due to a relatively labile nature of the cyanide bridges.
The31P NMR spectra of 6–10 also point to stereochemical non- rigidity of these tetranuclear compounds in solution under ambient conditions. Thus, at 298 K clusters 6–8 show two broadened multiplets of 3:1 relative integral intensities (Figures 4 and S5).
This spectroscopic pattern corresponds to the equivalent PPh2
groups as a result of fast fluxional motion of the phosphane ligand within [(P4)MM´(CN)]+ motifs. The gold-containing complexes 9 and 10 display unresolved and poorly interpretable signals at room temperature (Figures S6 and S7) indicating somewhat slower intramolecular dynamic processes.
At the low temperature limit (193 K, Figures 4, S5–S7) all the clusters except8, which still remains non-rigid in the NMR time
Figure 4. VT 31P{1H} NMR spectra of complex 7. The inset shows the assignment scheme and spin–spin coupling network. The top spectrum (brown) shows the ABB´X spin system simulation (2J(P(3)–P(4)) = 76 Hz,3J(P(3/4)–
P(2)) = 159/161 Hz,3J(P(1)–P(2)) = 243 Hz,1J(107,109Ag–P(3/4)) = 445 and 475 Hz,1J(107,109Ag–P(2)) = 416 and 446 Hz,1J(107,109Ag–P(4/3)) = 363 and 385 Hz,
1J(107,109Ag–P(1)) = 158 and 168 Hz).
scale, display the 31P NMR spectroscopic patterns compatible with the solid state structures. The observed spectra show the ABB´X (6,7) or AQQ´X (9,10) spin systems, generated by the non-symmetrical {(P4)MM´} moieties with all inequivalent phosphorus atoms. The chemical shifts and the multiplicities of the signals, which reveal extensive 2,3J(P–P) and additionally
1,2J(107/109Ag–P) couplings (in 7 and 10), allow for a complete assignment of the spectroscopic patterns, which has been verified by the spin system simulation (Figures 4, S5, S7).
The bonding of the tetraphosphane ligand in most cases occurs via all the P-donor functions that is confirmed by the typical coordination shifts of the resonances with respect to the free ligand, and the presence of P–Ag magnetic coupling in silver- containing complexes. However, in the Ag–Ag (7) and Au–Ag (10) compounds the central P(1) atom apparently forms weaker bonds to the metal, likewise some dinuclear alkynyl complexes bearing the P4 ligand.[20] The conclusion is evidenced by significantly smaller1J(Ag–P(1)) coupling constants in7 (158 and 168 Hz,
Accepted Manuscript
FULL PAPER
Table 1. Solid-State Photophysical Properties of1–10.
298 K 77 K
em, nm
ema av, sb
kr,s-1c knr,s-1d em, nm
av, sb
1 528 0.31 4.3 7.2×104 1.6×105 538 546.3
2 453 0.26 5.6 4.6×104 1.3×105 455 908.2
3 480 0.31 3.7 8.4×104 1.9×105 490 1176.3
4 488 0.28 3.6 7.6×104 2.0×105 496 1440.0
5 482 0.15 2.4 6.1×104 3.5×105 488 727.0
6 533 0.56 4.4 1.3×105 1.0×105 553 610.4
7 484 0.52 5.4 9.5×104 0.9×105 497 300.1
8 544 0.17 13.6 1.2×104 0.6×105 620 779.9
9 650 0.10 9.3 1.1×104 1.0×105 662 89.3
10 585 0.45 3.9 1.1×105 1.4×105 588 48.9
[a] The uncertainty of the quantum yield measurement is in the range of
±5% (an average of three replica). [b] Average emission lifetime (for2,4,5 and9 at 298 K; for1–3,5,7–9 at 77 K) for the two-exponential decay determined by the equation av = (A1 12 + A2 22)/( A1 1 + A2 2);Ai = weight of thei exponent. [c]kr values were estimated by av.dknr values were estimated bykr (1 )/av.
Figure 4) compared to the values obtained for P(3) and P(4) resonances, which are in good agreement with the literature data for the other polynuclear silver-containing complexes[21d, 25], as well as by the virtual absence of1J(Ag–P(1)) in10, where the shift of P(1) signal ( = –13.4 ppm) is very close to that of the free ligand ( = –14.1 ppm).[20] This observation also correlates with crystallographic data, which show a substantial elongation of P(1)–Ag/Au distances in clusters7 and10 (see Table S3).
Photophysical properties and computational studies. The polymeric complexes1,3–5 are practically insoluble in common organic solvents, while the tetrametallic compounds 6–10 are of limited stability in solution upon exposure to light and do not show appreciable luminescence in fluid medium. This prompted us to focus on the photophysical measurements in the solid state, the corresponding data are summarized in Table 1. The optical properties of the analogues of2 with different counterions have been studied in details by other groups.[17, 26] The emission wavelength of crystalline2 (453 nm at 298 K) correlates well with that reported for [(P2)2Ag]NO3 under ambient conditions ( em = 445 nm).[26] For both species em is only slightly sensitive to the temperature variation and shows 2 nm bathochromic shift for2 in the temperature range 298–77 K (Table 1). This similarity indicates the same nature of the emissive excited state, which was originally assigned to the mixed (3IL+3MLCT) character at low temperature[26] and is expectedly independent on the counterion.
However, a dramatic variation in the emission decay time of 2 upon cooling from 298 K (av = 5.6 µs) to 77 K (av = 908.2 µs) found in the present study very likely evidences that different
Figure 5.Normalized solid-state excitation (dashed lines) and emission (solid lines) spectra of1–5 at 298 K.
excited states operate at low and room temperatures, thus apparently indicating of TADF behaviour, which has been already found for some mononuclear silver complexes.[8b, 11a]
Polymer1 with cyanide-only bridges between the {Cu(P2)} motifs shows bright green luminescence (Figure 5) with microsecond decay time ( = 4.3 µs). At 77 K the lifetime shows ~2 orders of magnitude growth and reaches the value of av = 546.3 µs, accompanied by 10 nm red shift of emission (Table 1 and Figure S8) that is assigned to increasing contribution of phosphorescence,T1 S0.
Most of the complexes exhibit two-exponential decays at both 298 and 77 K that is not exceptional.[10b, 15] A plausible explanation could be the presence of different emissive sites,[10b] e.g. formed due to molecular disorder found in the crystal structures.
Additionally, a possibility of two non-equilibrated states has been proposed for the systems at 77 K.[14] However, based on the available data it is hardly possible to make an unambiguous conclusion concerning the origin of two lifetimes, therefore the average values are used in the further discussion.
The temperature dependences of the lifetimes for polymers1,3–
5 in the range 77–298 K (Figure S9) did not reach the plateau region of a constant , which evidently lies below 77 K. Due to this reason, the experimental data do not not allow fitting the curve using a standard equation for the two-state model (eq. 1).
However, the detected changes of the physical parameters are typical for TADF phenomenon,[8a] observed for a number of copper complexes including those containing P2 and akin diphosphane ligands.[10c, 12d, 27]
TD-DFT calculations predict that the lowest energy electronic transitionS0 S1 for1 is of1MLCT/1IL type as they involve mainly the contribution of HOMO and LUMO (90%, Figures 6 and S10).
The HOMO is localized on the copper ion (36%) and theP2 ligand (55%), while the LUMO spans over theP2 ligand (99%). A rather low S0 S1 oscillator strength (f = 1.4.10-3) matches the distinct spatial separation of the frontier orbitals that reflects a small energy gap E between the T1 and S1.[8a] This minor energy splitting is anticipated to allow easy thermal population of a higher energy state (reverse ISCT1 S1) and subsequently results in a
Accepted Manuscript
Figure 6.The appearance of the frontier orbitals in the singlet ground state and the highest singly occupied molecular orbital of the first excited triplet state of polymers1 and4.
delayed fluorescence relaxationS1 S0 that is not frozen at 77 K.
The emission properties of1 ( em = 528 nm, = 4.3 µs at 298 K) are comparable to those of the mononuclear complex Cu(P2)(dipyrazolylborate) ( em = 535 nm, = 3.3 µs at 300 K),[27]
which has a nearly pure ML(P2)CT character of the excited states of both singlet and triplet multiplicities. On the other hand, halide dinuclear species [Cu(P2)(µ-Hal)]2 exhibit somewhat higher energy emissions ( em = 482–497 nm)[10c, 12d] presumably due to a substantial admixture of XLCT (halide P2) transitions.
The isostructural compounds 3–5, which possess dicyanometallate spacers between the chromophoric centers {Cu(P2)}, display alike emission profiles and excited state lifetimes at 298 K ( av = 2.4–3.7 µs). The decay times as a function of temperature show comparable dynamics to that of1 (Figure S9), with long-livedT1 state still being mixed with singlet character at 77 K. Very close emission energies of3–5 (480–488 nm) suggest
a negligible role of [M(CN)2]- linkers in the composition of lowest lying excited states that is confirmed by the computational analysis. For all these congener polymers the S1 state is associated exclusively with HOMO LUMO transition (99.9%) and therefore is predominantly formed in the course of electron density relocation from thed- orbitals of phosphane-chelated copper ion to the aromatic rings ofP2 ligand (i.e. MLCT/IL origin, Figures 6 and S10) with a small admixture (~6%) of L´LCT character ( -cyanide -P2). The highest singly occupied MO (HSOMO) of theT1 state resembles the ground state LUMO and is distributed over the phenylene rings of the diphosphane with significant share (15%) of phosphorus atoms.
The gold-copper polymer5 has the shortest emission decay time at 298 K among the title complexes ( av = 2.4 µs) that is also one of smallest values reported for TADF materials.[8b] Taking into account zero oscillator strength calculated for the S0 S1
transition that should prevent high rate of prompt fluorescence k(S1 S0), short might testify to a remarkably small energy separation E(T1–S1), which is also in line with TADF process observed even at 77 K.[8a]
The tetrametallic clusters 6–10 are moderately intense luminophores in neat powder with quantum yields reaching 0.56 (6) at 298 K (Table 1). The broad and structureless emission spectra at both 298 and 77 K (Figures 7 and S11) suggest that the lowest lying excited states have mainly a charge transfer character (see computational support below). The emission wavelengths of6–10 cover a wide range of the visible spectrum from blue (484 nm,7) to orange-red (650 nm,9) that stems from the different nature of metal ions incorporated into the cluster framework. The silver complex 7 exhibits the most blue shifted band within this series, while the Au–Cu congener9 shows the lowest energy luminescence that fits the earlier reports on structurally similar d10 species with metallophilic interactions.[28]
The temperature dependencies of emission lifetimes (Figure 7) for complexes 6–8 and 10 are compatible with TADF phenomenon that implies variable contribution of two distinct excited states. The decay times, which are drastically increased with respect to the ambient conditions, approximately come to the plateau region at around 77 K and thus correspond to the domination ofT1 state at low temperature.
Figure 7. (A) Normalized solid-state excitation (dashed lines) and emission (solid lines) spectra of6–10 at 298 K; (B) and (C) temperature dependencies of the average decay times of6–10 (the solid lines represent the fits according to eq. 1, for9 a linear fit was used).
Accepted Manuscript
FULL PAPER
Table 2. Excited states characteristics from the fit based on the two-state model (calculated parameter for9 is given for comparison).
6 7 8 9 10
E(S1–T1), cm-1 521 703 779 - 491
E(S1–T1)calc, cm-1 965 332 2257 4670 325
(T1), s 646 300 783 - 50
(S1), ns 129 53 105 - 118
The two-site model described by the eq. 1 to fit the experimental data (kB is the Boltzmann constant, (S1) and (T1) are the fluorescence and phosphorescence decay times):
) =
( )]
) [ ) ( )] (eq. 1)
provides the E(T1–S1) energy difference between theS1 andT1
states of 491–779 cm-1 (Table 2), which allows fast reverse ISC T1 S1 at 298 K and therefore results in relatively short lifetimes of 3.9–13.6 s upon warming. The prompt fluorescence lifetime derived for6 ( (S1) = 129 ns) is comparable to its relatives studied here and to other copper complexes with TADF property.[8a, 14-15]
Additionally, a relatively small gap E(T1–S1) = 521 cm1 and substantial SOC warrant efficient population of theS1 state to give TADF decay time of 4.4 s at 298 K. The corresponding radiative rate constant for6 (kr = 1.3×105 s-1, 298 K) is visibly greater than those found for di- and tetranuclear TADF Cu(I) species with metal-metal interactions.[14-15]
Ag–Cu complex8 has the longest (TADF) of 13.6 s among the studied species that reflects the increased T1–S1 energy difference and likely points to a less effective SOC than in the rest of the compounds. Additionally, 8 demonstrates bathochromic shift of emission maximum (76 nm or 2253 cm-1) with temperature decrease that is exceptionally large for the conventional TADF materials and exceeds the value of E(T1–S1) derived by eq. 1.
Hence, this luminescence thermochromism can be tentatively attributed to the temperature-variable ground state energy, which might be raised at 77 K probably due to the subtle changes of intermolecular interactions.[29] Comparison of the intramolecular structural parameters measured at 100 K and 250 K does not reveal noticeable deviations; e.g. the largest alteration of bond lengths is found in the elongation of C N distance for 0.01 Å upon lowering the temperature, while the Ag–Cu and M–P, M–C/N contacts shows less than 0.006 Å shortening.
Combining silver and gold ions in10 further enhances SOC that leads to a prominent acceleration of the triplet state radiative decay ( (T1) = 50 s). Such relatively fast phosphorescence is supposed to contribute to the emission observed at room temperature together with TADF, as has been described for the dicopper species with commensurately highT1 S0 rate ( (T1) = 43 s).[10d]
In this respect, it is important to note that AgI compounds showing TADF behaviour are not excessive[8b, 11] due to the lack of accessible MLCT states that typically results in the emission of the intraligand nature. Moreover, gold(I) ion with even higher
Table 3. Optimized metal–metal distances (Å) of the ground (S0), the lowest singlet (S1) and triplet (T1) excited states for6–10.
M…M X-ray S0 S1 T1
6 Cu–Cu 2.7222(3) 2.6710
2.6706
2.6398 2.4483
2.4470 2.6511
7 Ag–Ag 2.9925(3) 3.0179
3.0180
3.0075 2.7438
2.7382 2.9915
8 Ag–Cu 2.6926(2) 2.7219
2.7220
2.6932 2.7264
2.6645 2.7163
9 Au–Cu 3.3278(4)
3.4844(5)
3.4915 3.7135
3.5034 3.3843
2.7851 3.4574
10 Au–Ag 3.1306(3) 3.4185
3.4188
3.3553 2.7445
2.7456 3.3502
oxidation potential than those of copper(I) and silver(I) ones predominantly serves as an external heavy atom, which promotes
3 phosphorescence[7c, 30], dual1 /3 emission[31] or perturbs
1IL fluorescence.[32] On the other hand, intense phosphorescence emerges in the AuI aggregates with extensive metallophilic bonding from the variety of charge transfer transitions involving Au…Au interactions.[33] Therefore, gold(I) ions are considered as inappropriate blocks for the design of TADF materials, though delayed fluorescence has been proposed for [Au(P2)(PS)]
complex (PS = 2-diphenylphosphinobenzenethiolate) on the basis of emission characteristics measured at 298 and 77 K.[11a]
Thus, to the best of our knowledge, cluster 10 is the first AuI- containing compound, for which TADF phenomenon has been unambiguously established.
In contrast to6–8 and 10, Au–Cu complex 9 features a linear growth of av from 9.3 at 298 K to 89.3 µs upon cooling the sample to 77 K, therefore suggesting that only phosphorescence is detected in the given temperature window. This discrepancy in photophysical properties manifests very strong SOC that occurs in the excited state and thus favours fast ISC S1 T1 and subsequent radiative relaxation of the triplet stateT1 S0for9 with respect to the congeners6–8, which do not contain gold(I) ions.
Comparison with10 shows that not only SOC efficacy determines the mechanism of emission, but the singlet-triplet energy gap may play a key role. The av for10, which shows TADF behavior, is almost twice shorter than that for9 at 77 K, and it is reasonable to assume that 9 also has a substantially larger energy splitting E(S1–T1), which has been evaluated computationally (vide infra).
The optimized geometries of 6–10 are generally in a good agreement with crystallographic data, though the metal–metal distances in 9 and10 are clearly overestimated (Table 3). The major structural variations, which occur upon one-electron excitation, are associated with dramatic contraction of one intermetallic contact in each complex in both singlet and triplet excited states. The increase of the M–M bond order in the exciplex-like forms of d10 metal clusters has been highlighted in a number of earlier works.[15, 28a, 34] The most severe decrease of M…M separations has been predicted for complex9, which does not possess appreciable metallophilic bonds in the ground state (according to XRD and DFT analyses).
Accepted Manuscript
Figure 8.The appearance of the frontier orbitals in the singlet ground state and the highest singly occupied molecular orbital of the first excited triplet state of clusters6 and9.
Conversely, in the case of8, where rather short Ag–Cu spacings have been determined crystallographically, the expected deviation in excited states is rather minor.
The calculated electronic structures of6–10 reveal thatS1 andS2
states for all these clusters except9 are nearly degenerate with ca. 1 nm separation (Table S5). TheS0 Sn (n = 1, 2) excitations comprise the combination of electronic transitions HOMO/HOMO- LUMO/LUMO+1, which have the {dp(M,P), dp(M´,P)} -P4 nature mixed with a minor contribution (~3%) of L´( - cyanide) L( -P4) charge transfer, see Figures 8, S12 and S13.
The highest singly occupied MO (HSOMO) of the T1 state is unevenly distributed on the phenylene spacers of one of the P4 ligands in6,7 and10, meaning that the lowest lying triplet state also has mainly an M,M´LCT origin. In clusters8 and9, however, HSOMO is largely located on one of the metal–metal bonds (Ag–
Cu and Au–Cu, respectively). This significantly differentiates the parentage of theT1 state from M,M´LCT (in6,7 and10) to the metal-centered charge transfer in8 and9 and might explain the markedly larger values of E(S1–T1), Table 2. Particularly in the case of9, the predictedS1–T1 energy separation reaches 4670 cm-1 (0.58 eV) that far exceeds the thermal energy at room temperature (kBT is ca. 25.7 meV at 298 K) and lies above the E threshold anticipated for delayed fluorescence.[8a] Simultaneously, populating the M–M´ bonding orbital assumes stronger SOC that
favours fast radiative decayT1 S0 and in combination with large S1–T1 separation hinders TADF as observed for complex9. This contrasting behavior of9 shows that the metal–metal interactions, which strongly affect the SCO and are expected to be stronger for 10 than for 9, cannot be used as the only decisive factor to rationalize the emissive mechanism and should be considered together with the energies of the excited states.
The calculated energy gaps E(S1–T1), listed in Table 2, reasonably describe the experimentally found trend for clusters 6–10, which occurs on the variation of metal ions. The significant deviation of this parameter for complex 8, where the predicted value (2257 cm1) coincides with the energy difference of emission spectra at 298 and 77 K ( em = 76 nm or 2253 cm-1) might arise from the above mentioned lattice effects, which are not taken into account by the DFT method. Nevertheless, such generally adequate theoretical evaluation of the photophysical properties, in particular, identification of the borderline between delayed fluorescence and phosphorescence along with wide modulation of emission colour, offer crucially important guidelines for the rational molecular design of efficient and tunable inorganic luminescent materials.
Conclusions
Chelating bidentate phosphane ligand (P2) was used to generate a family of coordination polymers when coupled with coinage metal cyanides. The resulting infinite zig-zag chains comprise {CuP2} blocks, which are linked by the cyanide (1) or dicyanometallate spacers (3–5) and do not feature metal–metal interactions. Luminescence properties of these moderately intense emitters ( em = 0.15–0.31) are determined by the copper- diphosphane chromophore and therefore are virtually insensitive to the nature of the linking [M(CN)2]- units (M = Cu, Ag, Au). The invariance of the emission energies for isostructural complexes 3–5 is confirmed by computational analysis of their electronic structures, which indicate a prevailing MLCT d(Cu) (P2) character of the excited states. Radiative decay lifetimes for1,3–
5 as a function of temperature, monitored in the range 77–298 K, demonstrate a steep increase below 150 K upon cooling. The obtained profiles for (T) are in line with possible TADF behaviour, though proposed delayed fluorescence was not completely frozen at 77 K presumably due to the small energy separation of theT1
and S1 excited states, which are still equilibrated at this temperature.
The cationic clusters [{(P4)MM´(CN)}2]2+ (6–10) were obtained by treating the MCN/M+ mixtures with tetradentate phosphane (P4).
Based on crystallographic and QTAIM analyses, metallophilic interactions were identified in Cu (6), Ag (7) and Ag–Cu (8) complexes, while gold-copper/ and silver congeners9 and10 lack appreciable metal–metal bonding. Aggregates 6–10 are not emissive in solution probably due to stereochemical non-rigidity that was revealed by the NMR spectroscopic studies. In solid, the title compounds are luminescent both at 298 and 77 K with maximum quantum yield of 0.56 (6). Variable composition of the metal core in 6–10 causes large modulation of the emission energies, which span an interval from blue (484 nm,7) to orange-
Accepted Manuscript
FULL PAPER
red (650 nm,9). All clusters except 9 exhibit TADF property at room temperature. In contrast, Au-Cu complex 9 displays only phosphorescence in the investigated temperature window that is assigned to the strong SOC and, consequently, efficient radiative decay of the triplet state T1 S0. Additionally, theoretical evaluation of the energy gap E(S1–T1) for 6–10 shows that cluster9 has the largest splitting, which significantly exceeds the values, typical for the luminophores with delayed fluorescence.
Furthermore, computational studies corroborate a particularly important role of metallophilic interactions in the excited states, in which the contracted metal-metal distances are supposed to be primarily responsible for tuning the emission energies and governing the spin-orbit coupling that has a crucial influence on the origin of radiative processes. Thus, efficient TADF with adjustable emission colour can be attained using different combinations of copper subgroup metals including gold, which previously was not considered as a suitable metal to generate delayed fluorescence. This result, rationalized computationally, has been achieved without demanding alteration of the ligand environment that broadens and simplifies the approaches to novel inorganic/organometallic luminescent materials.
Experimental Section
General comments. Tris(2-diphenylphosphinophenyl)phosphane (P4) was synthesized according to the published procedure.[20] Other reagents and solvents were used as received. The solution1H,31P{1H} NMR spectra were measured on Bruker 500 MHz Avance III spectrometer. Mass spectra were recorded on a Bruker micrOTOF 10223 instrument in the ESI+ mode.
Microanalyses were carried out at the analytical laboratory of the University of Eastern Finland.
[(P2)CuCN]n (1). CuCN (10.0 mg, 0.111 mmol) was dissolved in dimethylformamide (10 mL) under a nitrogen atmosphere and a solution of 1,2-bis(diphenylphosphino)benzene (P2, 50.0 mg, 0.112 mmol) in chloroform (15 mL) was added. The resulting pale yellow reaction mixture was left undisturbed at room temperature for 12 h, while green yellow microcrystalline precipitate started to form within minutes. The solid was collected, washed with methanol, dichloromethane and diethyl ether to give a yellow powder of1 (49.0 mg, 82 %). IR [KBr; (CN); cm–1]: 2120.
Anal. Calcd for C31H24CuNP2: C, 69.46; H, 4.51; N, 2.61. Found: C, 69.05;
H, 4.13; N, 2.24.
[(P2)2Ag][Ag(CN)2] (2).P2 (100.0 mg, 0.224 mmol) and AgCN (30.0 mg, 0.227 mmol) and were mixed in dichloromethane (10 mL). The resulting colorless solution was stirred for 0.5 h in the absence of light. Then it was evaporated, the solid residue was dissolved in hot toluene, filtered and left for a slow evaporation at room temperature to give colorless crystalline material (98.0 mg, 75 %). IR [KBr; (CN); cm–1]: 2131. Anal. Calcd for C64H48Ag2N2P4: C, 64.16; H, 4.17; N, 2.41. Found: C, 64.21; H, 4.44; N, 2.31.31P{1H} NMR (CD2Cl2; 25oC; ): 0.8 (d,1J109AgP 264 Hz,1J107AgP 230 Hz).
[(P2)Cu{Cu(CN)2}]n (3). P2 (100.0 mg, 0.224 mmol) was dissolved in degassed dichloromethane (20 mL), [Cu(NCMe)4](PF6) (174.5 mg, 0.468 mmol) was added and the reaction mixture was stirred for 0.5 h at room temperature to give a clear solution. Then NaCN (23.0 mg, 0.469 mmol) in methanol (5 mL) was added. The mixture was stirred for additional 12 hours; the resulting white precipitate was collected, washed with methanol (50 mL) and vacuum dried (89.2 mg, 63 %). IR [KBr; (CN); cm–1]: 2122.
Anal. Calcd for C64H48Cu4N4P4: C, 61.44; H, 3.87; N, 4.48. Found: C, 61.66; H, 3.79; N, 4.61.
[(P2)Cu{Ag(CN)2}]n (4).P2 (100.0 mg, 0.224 mmol) and AgCN (30.0 mg, 0.224 mmol) and were mixed in chloroform (10 mL), followed by a solution of CuCN (20.5 mg, 0.228 mmol) in degassed dimethylformamide (5 mL).
The reaction mixture was stirred for 30 minutes in the absence of light. The resulting solution was filtered and left for a slow evaporation at room temperature to give colorless crystalline material (128.0 mg, 85 %). IR [KBr; (CN); cm–1]: 2130. Anal. Calcd for C64H48Ag2Cu2N4P4: C, 57.37; H, 3.61; N, 4.18. Found: C, 57.05; H, 3.66; N, 4.27 %.
[(P2)Cu{Au(CN)2}]n (5). Prepared analogously to4 using AuCN (50.0 mg, 0.224 mmol) instead of AgCN; colorless crystalline material (121.5 mg, 71 %). IR [KBr; (CN); cm–1]: 2164. Anal. Calcd for C64H48Au2Cu2N4P4: C, 50.64; H, 3.19; N, 3.69. Found: C, 50.24; H, 3.34; N, 3.95 %.
[(P4)Cu2CN]2(PF6)2 (6). P4 (50.0 mg, 0.061 mmol) was dissolved in dichloromethane (15 mL), CuCN (5.5 mg, 0.061 mmol) was added and the reaction mixture was stirred for 3 hours at room temperature to give a yellow solution. Then [Cu(NCMe)4](PF6) (22.5 mg, 0.060 mmol) was added.
The mixture was stirred for additional 1 hour, the resulting clear pale green solution was filtered through Celite and diluted with benzene (8 mL). Its slow evaporation at room temperature produced green block crystals (61.0 mg, 90 %). ESI MS (m/z): [0.5M CuCN]+877.15 (calcd 877.15), [0.5M]+ 968.09 (calcd 968.09). IR [KBr; (CN); cm–1]: 2114.1H NMR (CD2Cl2, 298 K; ): 6.86–7.41 (br unresolved m).31P{1H} NMR (CD2Cl2; 193 K; ): –9.7 (dt,3JPP 14 and 195 Hz, 1P, PPh2), –15.7 (dd br,2JPP 138,3JPP 14 and 58 Hz, 1P, PPh2), –18.8 (dd br,2JPP 138,3JPP 14 and 58 Hz, 1P, PPh2), –38.5 (dt,3JPP 58 and 195 Hz, 1P, P(C6H4PPh2)3) , -144.8 (sept, 1P, PF6). Anal.
Calcd for C110H84Cu4F12N2P10: C, 59.36; H, 3.80; N, 1.26. Found: C, 59.62;
H, 3.71; N, 1.20 %.
[(P4)Ag2CN]2(CF3SO3)2(7). P4(50.0 mg, 0.061 mmol) was dissolved in dichloromethane (15 mL), AgCN (8.0 mg, 0.060 mmol) was added and the reaction mixture was stirred for 4 hours at room temperature to give a pale yellow solution. Then AgCF3SO3 (15.5 mg, 0.060 mmol) in acetone (5 mL) was added. The mixture was stirred for additional 1 hour, the resulting clear pale green solution was filtered through Celite and evaporated. The obtained solid residue was recrystallized by a gas-phase diffusion of diethyl ether into dichloromethane/acetone solution of 7 at room temperature to afford nearly colorless block crystals (69.0 mg, 94 %). ESI MS (m/z): [0.5M]+1056.04 (calcd 1056.04). IR [KBr; (CN); cm–1]: 2129 cm-1.1H NMR (CD2Cl2, 298 K; ): 6.72 (s br, 6H), 7.01-7.20 (m, 36H), 7.25 (t,JHH 7.5 Hz, 6H), 7.31-7.35 (m, 18H), 7.43 (t,JHH 7.8 Hz, 12H), 7.52 (t, JHH 7.4 Hz, 6H).31P{1H} NMR (CD2Cl2; 193 K; ): ABB´X spin system; – 4.4 (m,2JPP 76 Hz,3JPP 6 and 159 Hz,1J109AgP 475 Hz,1J107AgP 445 Hz, 1P, PPh2), –5.0 (m,3JPP 6 and 243 Hz,1J109AgP 446 Hz,1J107AgP 416 Hz, 1P, PPh2), –13.3 (m,2JPP 76 Hz,3JPP 6 and 161 Hz,1J109AgP 385 Hz,1J107AgP
363 Hz, 1P, PPh2), –38.8 (m,3JPP 159, 161 and 243 Hz,1J109AgP 168 Hz,
1J107AgP 158 Hz, 2JAgP 16 Hz, 1P, P(C6H4PPh2)3). Anal. Calcd for C112H84Ag4F6N2O6P8S2: C, 55.79; H, 3.51; N, 1.16. Found: C, 55.93; H, 3.65; N, 1.11 %.
[(P4)CuAgCN]2(PF6)2 (8). Prepared analogously to6 fromP4 (50.0 mg, 0.061 mmol), AgCN (8.0 mg, 0.060 mmol) and [Cu(NCMe)4](PF6) (22.5 mg, 0.060 mmol). Recrystallized by slow evaporation of a dichloromethane/toluene solution at room temperature to give yellow crystalline material (59.0 mg, 84%). ESI MS (m/z): [0.5M]+ 1012.07 (calcd 1100.14). IR [KBr; (CN); cm–1]: 2090 cm-1.1H NMR (CD2Cl2, 293 K; ):
7.05–7.45 (br unresolved m).31P{1H} NMR (CD2Cl2; 293 K; ): –1.1 (dd br,
3JPP 134, 1JPAg 212 Hz, 1P, PPh2), –16.0 (q br, 3JPP 134 Hz, 1P, P(C6H4PPh2)3), –144.5 (sept, 1P, PF6). Anal. Calcd for C55H42AgCuF6NP5: C, 57.08; H, 3.66; N, 1.21. Found: C, 56.86; H, 3.77; N, 1.23.
