Martin V.
Appleby
a,
Rory A.
Cowin
a,
Iona I.
Ivalo
a,
Samantha L.
Peralta-Arriaga
a,
Craig C.
Robertson
a,
Stuart
Bartlett
b,
Ann
Fitzpatrick
b,
Andrew
Dent
b,
Gabriel
Karras
b,
Sofia
Diaz-Moreno
b,
Dimitri
Chekulaev
a and
Julia. A.
Weinstein
*a
aDepartment of Chemistry, The University of Sheffield, Sheffield, S3 7HF, UK. E-mail: julia.weinstein@sheffield.ac.uk
bDiamond Light Source, Rutherford Appleton Laboratory, Harwell Science and Innovation Campus, Didcot, Oxfordshire OX11 0DE, UK
First published on 7th July 2023
The study aims to understand the role of the transient bonding in the interplay between the structural and electronic changes in heteroleptic Cu(I) diimine diphosphine complexes. This is an emerging class of photosensitisers which absorb in the red region of the spectrum, whilst retaining a sufficiently long excited state lifetime. Here, the dynamics of these complexes are explored by transient absorption (TA) and time-resolved infrared (TRIR) spectroscopy, which reveal ultrafast intersystem crossing and structural distortion occurring. Two potential mechanisms affecting excited state decay in these complexes involve a transient formation of a solvent adduct, made possible by the opening up of the Cu coordination centre in the excited state due to structural distortion, and by a transient coordination of the O-atom of the phosphine ligand to the copper center. X-ray absorption studies of the ground electronic state have been conducted as a prerequisite for the upcoming X-ray spectroscopy studies which will directly determine structural dynamics. The potential for these complexes to be used in bimolecular applications is confirmed by a significant yield of singlet oxygen production.
Homoleptic Cu(I) diimine complexes, [Cu(NN)2]+ possess sufficiently low-energy MLCT states, but MLCT states are usually too short-lived, in the sub-nanoseconds domain, unless bulky substituents are introduced next to coordinating N-atoms. The reason for the short lifetime is that excitation into an MLCT state changes the electronic configuration of Cu from nominally d10 to d9, causing large structural changes from a pseudo-tetrahedral geometry in the ground state to a pseudo-square-planar geometry, which provides yet another efficient deactivation pathway.3,4 Heteroleptic Cu(I) complexes, such as phosphine diimine complexes, [Cu(PP)(NN)]+, can benefit from the tuneable electronic properties of the diimine acceptor ligand while extending the lifetime of the MLCT state by inhibiting the excited state distortion using a bulky phosphine co-ligand.5 For example, Cu(I) xantphos diimine complexes have excited state lifetimes in the nano to microsecond range (10–100 times longer than the homoleptic diimine complexes) which make them suitable for applications relying on bimolecular processes.6,7 We have recently shown that Cu(I) heteroleptic complexes consisting of a phen based ligand (dmp) and phosphine ligand (xantphos = 4,5-bis(diphenylphosphino)-9,9-dimethyl-xanthene) produce singlet oxygen in high yields and are capable of killing both Gram positive and Gram negative bacteria.8 However, these complexes are limited in use due to limited UV-vis absorption which does not make the most of the solar spectrum, vital for sunlight-driven applications.
Whilst heteroleptic Cu(I) phenanthroline complexes have shown strong absorption of light below 400 nm, many applications require compounds and photosensitisers which absorb in the visible range. The lowest absorption band in these complexes is due to a metal-to-ligand charge transfer (MLCT) transition to the diimine ligand. Introducing a more electron-withdrawing diimine ligand such as 2,2′-biquinoline (biq) and its derivatives (Ered = −1.46,9vs. 2,9-dimethyl-1,10-phenanthroline (dmp), Ered = −2.11)7 may shift the MLCT absorption towards lower energies.
Herein, the excited state dynamics of a series of Cu(I) 4,4′-(CO(O)Et)2-2,2′-biquinoline (deebq) complexes are studied. The biquinoline ligand, deebq, has electron-withdrawing ester groups, which further shift the MLCT absorption band to lower energies, and enable future immobilisation of the complexes on surfaces using –COOH as an anchoring group, for which the ester is a suitable proxy for spectroscopic studies. Surface immobilisation improves reusability of the photosensitiser and prevents leaching of the complex from a support, both are of specific interest in applications for water purification and H2 production. Earth-abundant photosensitisers absorbing in the red region and achieving a good yield of singlet oxygen not only will lend themselves to sun-driven applications but also photodynamic therapy (PDT), and photo-driven synthetic applications.
The heteroleptic diphosphine diimine complexes discussed in this work (Fig. 1) include [Cu(xantphos)(deebq)]BF4 (D2) where xantphos = 4,5-bis(diphenylphosphino)-9,9-dimethyl-xanthene, and [Cu(DPEphos)(deebq)]BF4 (D3) where DPEphos = bis[(2-diphenyl-phosphino)phenyl] ether, a more flexible analog of xantphos. The homoleptic complex [Cu(deebq)2]BF4 (D1) was also studied for comparison. D2 and D3 have previously been reported for use in white-light emitting electrochemical cells.10
Fig. 1 Structures of the Cu(I) complexes studied: [Cu(deebq)2]BF4 (D1), [Cu(xantphos)(deebq)]BF4 (D2) and [Cu(DPEphos)(deebq)]BF4 (D3). |
We aim to further understand the interplay between the structural and electronic changes in heteroleptic Cu(I) diimine complexes that will inform future design of more efficient photosensitisers. Two potential mechanisms affecting MLCT excited state decay involve a transient formation of a solvent adduct, made possible by the opening up of the Cu coordination centre in the excited state due to structural distortion, and a coordination of the O-atom on the phosphine ligand to the copper center. Here, the dynamics of these complexes are explored by transient absorption (TA) and time-resolved infrared (TRIR) spectroscopy; X-ray absorption studies of the ground electronic state have been conducted as a prerequisite for the upcoming X-ray spectroscopy studies which will directly determine structural dynamics. The potential for these complexes to be used in bimolecular applications is evaluated by determining the yield of singlet oxygen production.
Samples were prepared in quartz cuvettes with a pathlength of 1 or 2 mm. The samples were stirred using a magnetic stirrer to minimise photodegradation. Steady-state UV-vis spectra were measured before and after TA measurements to confirm no photodegradation had occurred. The optical density of the sample at the excitation wavelength is specified for each experiment. Over the probe spectral region, the optical density of the sample was kept below 1. Experiments consisted of averaging 1 s at each time point for 5 scans. Temporal chirp was obtained from measuring and fitting the solvent response. This was applied to the raw data to remove contributions from the chirp and solvent response. Chirp correction and background subtraction was performed in Surface Xplorer (Ultrafast Systems). Global lifetime analysis was carried out in MATLAB R2022a.
The beams are split by an additional 50:50 Ge beam splitter. The split beams were focussed into two identical spectrometers (f = 320 mm, Horiba iHR320), with either 50 g mm−1 or 100 g mm−1 gratings which gave a spectral resolution of 6 and 3 cm−1 respectively. The probe and reference spectra were recorded employing two liquid nitrogen cooled HgCdTe (MCT) array detectors (Infrared Systems). Each detector had 128 pixels split into 2 lines; 96 pixels for the probe and 32 for the reference. Home-built software (LabVIEW) was used for data collection and processing the raw spectra to pump–probe TRIR spectra. A multichannel referencing scheme23,24 was used to improve the signal-to-noise ratio. Measurements were calculated from the average of 5 scans with 2 s averaged per time point. Samples were prepared in a Harrick Scientific IR cell with 2 mm CaF2 windows. The path length was adjusted using a PFTE spacer (650 μm). Samples were raster-scanned at the sample position using a motorised x–y stage to minimise photodegradation. In addition to the raster-scan, samples were also flowed through the Harrick cells using a peristaltic pump (ColePalmer, Teflon loop). The photostability of the sample was checked by recording the UV-vis spectra before and after laser exposure. Global Lifetime Analysis was performed using MATLAB R2022a. Additional TRIR studies were performed at the Laser for Science Facility, Rutherford Appleton Laboratory.
Fig. 2 UV-Vis absorption spectra of complexes D1–D3 in DCM, normalised to the maxima of the MLCT transition band. Emission recorded for D2 and D3 following excitation at the absorption maxima. |
The photophysical properties of the Cu(I) deebq complexes are summarised in Table 1. The homoleptic complex D1 is non-emissive in DCM Fig. 2. The deebq complexes, D2 and D3, show a main emission band at 720 and 725 nm respectively. The emission maximum in deebq complexes is red-shifted relative to the complexes with unsubstituted biq ligands by ∼25 nm (∼610 cm−1).9,29 Both D2 and D3 exhibit emission lifetimes of the order of tens of nanoseconds. A slightly longer excited state lifetime for D3 in comparison to D2 (82 ns vs. 52 ns, in aerated DCM) could be due to the increased flexibility of the DPEphos not only restricting the structural reorganisation in the excited state, but also inhibiting solvent interactions more effectively than xantphos.
Complex | λ abs /nm | λ em /nm | τ air , /ns | τ 1(TA) /ps | τ 2(TA)/ps | τ 3(TA)/ps | τ 1(TRIR)/ps | τ 2(TRIR)/ps | ϕ 1O2 |
---|---|---|---|---|---|---|---|---|---|
a In aerated solution. b Excitation wavelength 445 nm. c Quantum yield of singlet oxygen calculated with respect to perinapthenone (assumed ϕPN ≈ 1 in DCM).17 d A component with the lifetime significantly exceeding the instrument-limited time window of 7 ns. e Additional ultrafast component of <0.2 ps is present in all TA data. | |||||||||
D1 | 540, 580, 617 | — | — | 1.7 | 6.1 | 1070 | <0.2 | 1000 | — |
D2 | 500 | 720 (698) | 52 ± 7 | 0.9 | 15 | Infd | <0.2 | Infd | 0.13 ± 0.05 |
D3 | 495 | 725 (703) | 89 ± 5 | 0.7 | 14 | Infd | <0.2 | Infd | 0.09 ± 0.05 |
Complexes | D2 | D3 | D1 | |
---|---|---|---|---|
Bond lengths (Å) | ||||
Cu–N1 | 2.085(5) | 2.097(5) | 2.019(4) | |
Cu–N2 | 2.087(5) | 2.058(5) | 2.016(5) | |
Cu–P1 | 2.2664(17) | 2.2344(18) | Cu–N3 | 2.015(5) |
Cu–P2 | 2.2839(18) | 2.3222(18) | Cu–N4 | 2.040(5) |
Cu–O1 | 3.170(4) | 3.152(12) | — | |
Dihedral angle | ||||
N1–Cu–N2/P1–Cu–P2 | 92.94° | 92.81° | N1–Cu–N2/N3–Cu–N4 | 82.6° |
Bond angles (°) | ||||
N1–Cu–N2 | 79.7(2) | 79.2(2) | N1–Cu–N2 | 80.76(18) |
P1–Cu–P2 | 116.74(8) | 116.80(6) | N3–Cu–N4 | 81.66(19) |
P1–Cu–N1 | 114.45(14) | 124.95(15) | N1–Cu–N3 | 124.12(19) |
P1–Cu–N2 | 119.49(15) | 127.37(15) | N2–Cu–N3 | 132.49(19) |
P2–Cu–N1 | 111.63(14) | 97.59(14) | N1–Cu–N4 | 123.94(18) |
P2–Cu–N2 | 109.25(14) | 102.40(15) | N2–Cu–N4 | 120.00(19) |
Fig. 3 Crystal structures of complexes D1, D2 and D3 obtained by single-crystal X-ray crystallography. |
Fig. 4 (a) Experimental Cu K-edge XANES spectra for D1 (black), D2 (red) and D3 (blue) in MeCN solution at r.t. (b–d) Experimental XANES (blue) and results of preliminary calculations of XANES (black) of complexes D1–D3. The energy axis for the calculated spectra has been shifted upwards by 98.25 eV so that the edge corresponds to the edge of the experimental data. Experimental data are average of multiple runs, see Fig. S17† for the data from individual runs. |
Ground state XAFS spectra for each complex in the k-space and R-space are displayed in Fig. 5. The results from the best fits are shown in Table 3. The scattering paths described are the single scattering paths to the N, P and C atoms closest to the Cu centre. The fits show that the average Cu–N bond length is shorter (2.01 Å) for the homoleptic complex D1 than the two heteroleptic complexes, D2 and D3 (2.11 Å and 2.12 Å respectively). The obtained Cu–N and Cu–P distances for D1–D3 in MeCN are similar to those obtained in the single crystal (Table 2). The average Cu–N bond length for D1 is slightly larger than for the similar complex [Cu(biq(COOH)2)2]+, 1.98 Å.35 The average Cu–N and Cu–P bond lengths for D2 and D3 are also comparable to those reported for other heteroleptic Cu(I) complexes such as [Cu(xantphos)(2,9-(OMe)2-1,10-phenanthroline)], 2.08 Å (Cu–N) and 2.28 Å (Cu–P).30
Fig. 5 Fourier transform (left) and EXAFS extracted signal (right) for complexes D1 (A), D2 (B) and D3 (C) in MeCN solution (k weighting = 1) (blue line). The red lines represent the best fit to the data using the parameters included in Table 3. |
Complex | Paths | Fitted distance (Å) | Fitted Debye–Waller (Å2) |
---|---|---|---|
a D1: S02 = 0.8, E0 = 6(2) eV, k-weight = 1, k-range = 3–10 Å, R-range = 1–3.5 Å. R-Factor = 0.02. D2: S02 = 0.8, E0 = 7(3) eV, k-weight = 1, k-range = 2.7–12 Å, R-range = 1.35–3 Å. R-Factor = 0.06. D3. S02 = 0.8, E0 = 7(3) eV, k-weight = 1, k-range = 2.7–12 Å, R-range = 1.35–3 Å. R-Factor = 0.06. | |||
D1 | 4 Cu–N | 2.01(2) | 0.009(1) |
4 Cu–C | 3.10(2) | 0.005(2) | |
4 Cu–C | 3.32(2) | 0.005(2) | |
D2 | 2 Cu–N | 2.11(4) | 0.002(6) |
2 Cu–P | 2.26(3) | 0.014(11) | |
D3 | 2 Cu–N | 2.12(3) | 0.006(6) |
2 Cu–P | 2.27(2) | 0.013(7) |
For D1 the calculated XANES spectrum (Fig. 4b) only roughly reproduces the edge shoulder at 8985 eV. Importantly, although the calculated spectrum underestimates the strength of the edge feature at 8982 eV, it does predict the pre-edge shoulder at 8978 eV associated with the low-lying, mixed metal–ligand states. The XANES spectra of D2 and D3 are more accurately predicted, including the first prominent edge feature, but still fails to predict the absorption maxima. The modelling of the 8985 eV edge feature in D2 and D3 but not D1 suggests that these transitions are fully allowed in the heteroleptic complexes, but are only enabled by spin–orbit coupling in the homoleptic D1 complex. Similarly, the calculations suggest that the absorption maxima at ∼8994 eV is highly dependent on a strong spin–orbit coupling interaction between the copper centre and the ligand. This would increase the absorption cross section for the higher energy, more ligand-centred virtual states, creating the broad absorption pattern seen at these energies. The calculated bond length of the Cu–N bonds in D1 (Table S25†) correspond well with the experimental results obtained in solution by EXAFS and by X-ray diffraction in the solid state. For D2 and D3, however, the calculated bond lengths (Table S25†) for Cu–N are smaller, and the calculated Cu–P bond lengths are significantly larger (∼2.5 Å vs. ∼2.3 Å) than the experimental values. The calculated dihedral angle for complex D3 is in good agreement with the crystallographic measurements, however it is underestimated for complexes D1 and D2.
For D2 and D3, initially, transients at ∼420, 461, 560, and 627 nm are observed, with the latter becoming obscured by a broader peak, 560 nm, by 1 ps. The GSB is obscured by the pump scatter and electronic offset but can be partially observed at ∼520 nm. A dip in the spectra at ∼600 nm observed at early times (∼250 fs, which could be potentially assigned to stimulated emission) disappears by 1 ps. The transient at 420 nm shifts to ∼404 nm by ca. 100 ps, and remains unchanged within the temporal limit of the experiment (7 ns). The peak at 560 nm decays over this timescale and a new peak at 538 nm grows in, appearing as a blue shift in the TA spectra which may also be assigned to the decay of stimulated emission. An additional transient >650 nm grows in by 100 ps, and persists.
The results of the global analysis of the transient data for each complex and the corresponding time components are listed in Table 1. The dynamics of the homoleptic complex D1 requires four time constants to describe the fs-TA satisfactory, of sub-200 fs (IRF), 1.7 ps, 6.1 ps and 1070 ns, Fig. 6A. Structural distortion in a singlet state (S1) of [Cu(NN)2]+ is usually observed on the sub-ps timescale.36 The observed 1.7 ps process could relate to either geometry change in an Sn state, or to the flattening distortion in a T13MLCT state, populated via an ultrafast (not detected) intersystem crossing (ISC) from an initially populated S2 state. The decay component of 6.1 ps is assigned to ISC from the flattened S1 state to the flattened T1 state which is similar to the 10 ps values reported for homoleptic [Cu(dmp)2]+ complexes.37
Overall, it appears that there is a branched (rather than a consecutive) decay process of the initially populated S2 state in D1 that leads to the final T1, 3MLCT state. The T1 state has a lifetime of 1.1 ns in DCM, which is longer than [Cu(phen)2]+ (143 ps),36,37 but shorter than the 54 ns lifetime of the [Cu(dmp)2]+ complex.
For complexes D2 and D3, four components are required to fit the TA data (Fig. 6B and C). The sub-200 fs, 900 fs (D2) and 700 fs (D3) decay components, could be assigned to the structural distortion which is similar to the rate of structural distortion in [Cu(xantphos)(NN)]+ bearing bulky NN = derivatives of 2,9-dimethyl-1,10-phenanthroline.38 The slight decrease of the sub-ps component from D2 (900 fs) to D3 (700 fs), could be due to the increased flexibility of the DPEphos ligand compared to the xantphos ligand. The component, 14–15 ps for both complexes, is tentatively assigned to ISC corresponding to the blue shift of the 420 nm to 404 nm and the appearance of the peak at 538 nm in the TA spectra.
To further investigate the excited state dynamics of these complexes, TRIR was recorded Fig. 8, S7 to S10,† following excitation with 40 fs, 525 nm (D1) and 500 nm (D2 and D3) laser pulses in DCM. Upon excitation, bleach of the ground state absorbencies of several of the biq bands are observed, with intense transient bands formed across the entire region. The ν(CO) stretching vibration of the ester group at ca. 1728 cm−1 is bleached, and a transient at lower energies is observed, at ca. 1700 cm−1. This is consistent with electron density increase on the diimine ligand, thus confirming the MLCT nature of the lowest excited state.
Fig. 8 TRIR spectra of D1, D2 and D3 in DCM following excitation at 525, 500 and 500 nm respectively (2 mW, OD@λex = ∼0.6), collected with the pump set to magic angle (54.7°) with respect to the probe. The time delay between the pump and probe for each spectrum is shown in the legend. Kinetic traces are shown in the ESI.† |
For complex D1, global analysis revealed two decay components for the region between 1650 cm−1 and 1760 cm−1, corresponding to the (CO) of the ester group (Fig. 8) including a sub-ps component likely due to vibrational cooling, and 1 ns.
Global analysis of the data D2–D3 Fig. S8 and S11,† yields 2 components: a long component beyond the timescale of the experiment of 6 ns, and sub-ps component, likely vibrational cooling which would not be easily observed in the TA. This component corresponds to a small increase in the peak intensity at 1710 cm−1, the bleach also increases in intensity, meanwhile the peak of the transient shifts from 1699 cm−1 to 1702 cm−1 (the apparent position of the bleach shifts from 1733 cm−1 to 1728 cm−1). This is a little faster than the typical timescale of vibrational relaxation of the (CO) of ester groups.40,41
Overall, the dynamics of the excited states for D1–D3 elucidated from the transient absorption and TRIR data indicate at least three components of the MLCT excited state decay, including a structural change, an ISC, and the final decay to the ground state in the process of charge recombination.
For the homoleptic complex D1, a 200 fs component observed is comparable to the rate of the flattening distortion in [Cu(phen)2]+.36,37,42 A second decay component of 1.7 ps for D1 was tentatively assigned to IC; however this component was not observed in TRIR, and one may assume that this is a process that does not affect electron density on the ester groups of the diimine ligand and hence is associated only with the structural change around the metal center. A 6.1 ps component was tentatively assigned to a combination of intersystem crossing, ISC, and vibrational cooling due to the blue shift in the TA spectrum and due to the similar magnitude of the component to the rate of ISC observed for other homoleptic Cu(I) complexes (9–13 ps)−1.37 The final state, here assigned to 3MLCT T1, decays over ∼1–1.1 ns as observed in both TA and TRIR data.
For the heteroleptic complexes, D2 and D3 a structural distortion of the S1 state occurs over 700–900 fs which is followed by ISC to the flattened T1 state over 14–15 ps. The final T1 state is long-lived, exhibiting lifetimes of 52 and 89 ns (D2 and D3). The propensity of the compounds D2–D3 to bimolecular reactions is demonstrated by a high yield of singlet oxygen photosensitisation (9–13%).
One of the mechanisms of excited state quenching in the nascent MLCT excited state in Cu(I) complexes could be a non-covalent, transient interaction between the solvent molecules and the copper center due to structural distortion from pseudo-tetrahedral to pseudo-square planar configuration that makes the metal centre more accessible. A process that may influence and/or prevent solvent coordination may involve transient coordination of the –O-atom of the phosphine ligand to the copper center. Ground state X-ray absorption results from D1–D3 in solution reveal the local structure around the metal center. Preliminary DFT calculations identified the geometry of the ground states of D1–D3 which correspond well to the parameters obtained from the EXAFS modelling. Building on the ground state EXAFS data, and the excited state dynamics resolved by optical spectroscopies, the next step will be an investigation of structural changes and transient coordination of solvent and O-atom to the copper center in the excited state of these promising photosensitisers using excited state X-ray spectroscopies.
Footnote |
† Electronic supplementary information (ESI) available. CCDC 2239630–2239632. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d3fd00027c |
This journal is © The Royal Society of Chemistry 2023 |