Copper(I) Phosphinooxazoline Complexes: Impact of the Ligand Substitution and Steric Demand on the Electrochemical and Photophysical Properties.

A series of seven homoleptic CuI complexes based on hetero-bidentate P^N ligands was synthesized and comprehensively characterized. In order to study structure-property relationships, the type, size, number and configuration of substituents at the phosphinooxazoline (phox) ligands were systematically varied. To this end, a combination of X-ray diffraction, NMR spectroscopy, steady-state absorption and emission spectroscopy, time-resolved emission spectroscopy, quenching experiments and cyclic voltammetry was used to assess the photophysical and electrochemical properties. Furthermore, time-dependent density functional theory calculations were applied to also analyze the excited state structures and characteristics. Surprisingly, a strong dependency on the chirality of the respective P^N ligand was found, whereas the specific kind and size of the different substituents has only a minor impact on the properties in solution. Most importantly, all complexes except C3 are photostable in solution and show fully reversible redox processes. Sacrificial reductants were applied to demonstrate a successful electron transfer upon light irradiation. These properties render this class of photosensitizers as potential candidates for solar energy conversion issues.

Introduction

Photoactive CuI complexes are considered as a highly promising alternative to traditional systems based on noble metals such as ruthenium, iridium, rhenium or platinum.1, 2, 3, 4, 5 Indeed, CuI compounds were already successfully applied as photosensitizers in the light‐driven reduction of protons to H2,6, 7, 8, 9, 10, 11 as photoredoxcatalysts for organic transformations12, 13, 14, 15, 16 or in devices such as organic light‐emitting diodes (OLEDs),17, 18, 19, 20, 21, 22 dye‐sensitized solar cells (DSSCs)23, 24, 25, 26 and light‐emitting electrochemical cells (LECs).27, 28 Unfortunately, a limited stability under operating conditions still hampers their large‐scale application in molecular solar energy conversion schemes.21, 29, 30, 31, 32, 33, 34 It is known, that in particular heteroleptic CuI complexes of the type [(P^P)Cu(N^N)]+ (with P^P representing a diphosphine and N^N a diimine ligand) can undergo ligand exchange reactions in solution upon light irradiation.29, 31, 33, 34, 35 This is mainly caused by the formation of thermodynamically more favored homoleptic bisdiimine complex [Cu(N^N)2]+.29, 31, 33, 34, 35 Hence, this drawback drove the search and development of novel CuI complexes with an increased stability, but also having other desired properties like a broad absorption in the visible and a reversible redoxchemistry.4, 36, 37, 38, 39, 40

One possible option to achieve this aim is the replacement of the original P^P ligand by a heterobidentate P^N‐ligand.17, 41 Particularly, the combination of a N‐heteroaryl moiety, which possesses a wide range of tunable electronic properties and a soft phosphine donor seems promising.41, 42 Several CuI complexes based on different types of P^N ligands, mainly as multinuclear cuprous halide complexes, have already been prepared and investigated.43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53 In these examples the phosphine unit is either directly bound to the N‐heteroaryl moiety (e.g., 2‐(diphenylphosphino)‐pyridine46, 49 or 8‐(diphenylphosphino)‐quinoline44, 45) or connected via an aliphatic spacer (e.g. 2‐[2‐(diphenylphosphino)‐ethyl]‐pyridine51, 53). Consequently, the previous examples are typically bridging or only monodentate ligands due to the small bite angle.44, 45, 46, 49, 51, 53, 54 Moreover, these systems still suffer from a limited stability.21, 55, 56

In contrast, Zeng et al. found that 1,2‐phenyl‐bridged P^N‐ligands can form phosphorescent and stable CuI complexes in solution and in the solid state.57 Nevertheless, the impact of different substituents, steric effects and chirality on the electrochemistry and photophysics of copper P^N complexes has not been studied in detail yet.

In a previous study we showed for the first time, that a phosphinooxazoline (phox) based P^N ligand enables stable mononuclear CuI complexes with interesting photophysical properties.41 Moreover, also the ability of these complexes to act as photosensitizers for the light‐driven production of H2 was demonstrated.41 Hence, the impact of the spatial arrangement and steric demand in such phox ligands on the properties of the resulting CuI complexes is of high interest. In consequence, a systematic series of seven homoleptic CuI complexes (Figure 1) with different size, type and number of substituents at the oxazoline moiety was prepared.

Figure 1

General structure of the ligands L1 to L7 and the resulting homoleptic CuI complexes C1 to C7.

Following, a combination of NMR spectroscopy, X‐ray analysis, cyclic voltammetry, absorption and emission spectroscopy as well as density functional theory (DFT) calculations was used to identify structure–property relationships. The presence of single crystals of all compounds enabled a detailed discussion of their solid state structures. In addition, time‐resolved emission spectroscopy and time‐dependent DFT (TD‐DFT) calculations were applied to also examine excited state properties. Most remarkably a dependence of the complex properties on the chirality of the respective P^N ligand was found. Finally, measurements with sacrificial reductants on representative complexes were performed to demonstrate successful electron transfer upon light irradiation. All in all, the gained knowledge paves the way to improved photoactive CuI complexes, which might be used in solar energy conversion schemes in the future.

Results and Discussion

Synthesis and Structural Characterization

A broad variety of phox ligands were prepared (see Figure 1) to allow for a comprehensive evaluation of the ligand impact on the ground and excited state properties of the resulting CuI complexes. Main attention was given to the steric influence of the different substituents at the 4‐position of the 2‐oxazoline (4,5‐dihydrooxazole) moiety. This position was chosen for modification due to its proximity to the copper center, because this likely has a strong impact on the geometry as well as on excited state relaxation processes. A special distinction has been made between derivatives without any substituents (L1), with only one substituent (L4–L7) and two substituents (L2–L3) at the 2‐oxazoline (Figure 1). Furthermore, phox ligands with aliphatic (Me, iPr, iBu) and aromatic substituents (Ph, Bn) were prepared. While L1, L2 and L4–7 are literature known, L3 was specifically designed for this study. It is also worth noting, that the ligands L4–L7 possess a chirality center at the 4‐position of the 2‐oxazoline. The synthesis of the ligands L1, L2, L5 and L6 was performed following a synthesis procedure from literature,41, 58, 59 starting from the corresponding and commercially available chiral amino alcohols (see Scheme 1). These amino alcohols were first reacted in a Witte–Seeliger reaction58 to the aryl‐bromide precursor and then converted to the corresponding phox ligands via an Ullman‐type coupling (Scheme 1).59 In contrast, the preparation of L4 and L7 was done using the easily accessible amino acids (i.e., d,l‐valine, (S)‐phenylalanine) as natural feedstock, while for L3 a commercially available amino acid (2,2‐diphenylglycine) was utilized. For these three ligands the required aryl‐bromide precursors were obtained from the respective amino acids in a two‐step reaction using LiAlH4 for reduction60 followed by the acid catalyzed reaction with bromo‐benzoylchloride (Scheme 1).59 In the case of L4 a racemic mixture of the R and S configured isomer was obtained, while the ligands L5–L7 are chiral and the stereochemical information of the educts was preserved This means that L6 is R configured and L5 as well as L7 possess a S configuration at the stereocenter.

Scheme 1

Overview of the common synthesis procedure of the ligands L1 to L7 either starting from the respective amino acids or the amino alcohols. i) Reduction: LiAlH4, THF,60 ii.a) Witte–Seeliger reaction: ZnCl2, PhCl,58 ii.b) step 1: NaHCO3, H2O/DCM; step 2: TsCl, TEA, DCM,59 iii) Ullman‐type coupling: CuI, DMEDA, Cs2CO3, toluene.59

The ligands L1–L7 were subsequently coordinated to a CuI center by a ligand substitution reaction using the [Cu(MeCN)4]PF6 adduct (MeCN=acetonitrile) as a precursor. Following a general synthesis procedure [Cu(MeCN)4]PF6 (1 equiv) and the respective phox ligand (L1–L7, 2 equiv) were suspended in dry dichloromethane and heated to reflux for 5 h under argon atmosphere. After isolation and purification, the homoleptic CuI complexes C1–C7 of the type [Cu(phox)2]PF6 were obtained as yellow to orange crystals. The obtained yields differ in the range from 11 % for C3 to 60 % for C2 (for further details see the Supporting Information). The much lower yield of C3 significantly differs from all other complexes and might be caused by steric constraints (two adjacent phenyl groups) and hence kinetic instability in solution. Such instability, for example, substitution reactions in solution, is well known from sterically congested homo‐ and heteroleptic CuI complexes.29, 31, 33, 34, 35

In a next step the molecular composition and structures of all complexes were confirmed by standard analytical techniques. In the 31P{1H} NMR spectra all complexes display one single signal in the range between −8.5 (C2) and −5.3 ppm (C6) (Figure S2), indicating a weak influence of the different substituents. A relatively small and broadened 31P{1H} NMR signal of C3 is another indication for a limited stability of this particular complex in solution.

Single crystals suitable for X‐ray crystallography of all complexes C1–C7 were either obtained by common diffusion techniques or directly from recrystallization in methanol. The crystallographic data and depiction of all solid‐state structures can be found in the Supporting Information (Chapter 4), while only a selection is presented here. For instance, complex C4 crystallizes in the centrosymmetric space group P21/n. Consequently, C4 is therefore invariant under the parity operation P (inversion through a point), which prevents the specification of an absolute structure (i.e., whether it is the R or the S enantiomer). This is in line with the preparation conditions, where a racemic mixture of L4 was used for complexation. In contrast, the crystal structures of C5–C7 (Figures 4 and S4), which contain the chiral ligands L5–L7, have a valid absolute‐structure determination (see Flack parameters, Table 1). Structural analysis revealed that C5 and C7 possess a S configuration, whereas C6 has an R configuration on the stereogenic center, as depicted in Figure 1. Interestingly, complex C7 crystallizes in the monoclinic crystal lattice in the space group C2 and therefore possesses symmetry operations like a two‐fold rotation axis (see Figure 2).

Figure 4

Simplified solid‐state structures (ORTEP representation) of C4–C7 highlighting the ligand orientation in the respective CuI complexes. The substituents in C4 and C6 point away from each other, whereas in C5 and C7 they point towards each other. Thermal ellipsoids are at a probability level of 50 %. The hydrogen atoms, the phenyl groups of the PPh2 moiety, counter anions and solvent molecules are omitted for clarity.

Table 1

Selected crystallographic bond lengths (pm) and angles (°) of the complexes C1–C7. For atom labelling also compare with Figure 2 and the Supporting Information. The respective CCDC reference numbers are given in the supporting information. These data are provided free of charge by the Cambridge Crystallographic Data Centre.

	C1	C2	C3	C4	C5	C6	C7
space group	P21/n	P21/n	P4cc	P21/n	P21	P212121	C2
Flack parameter	–	–	–	n.a.[a]	0.004(4)	0.012(2)	0.007(6)
Cu−P1	2.2230(8)	2.2640(8)	2.2143(10)	2.2428(4)	2.2428(4)	2.2263(8)	2.2628(15)
Cu−P2	2.1860(8)	2.2726(7)	2.2160(10)	2.2502(4)	2.2502(4)	2.2186(7)	2.2629(15)
Cu−N1	2.047(2)	2.094(2)	2.070(3)	2.0488(12)	2.0488(12)	2.071(2)	2.128(5)
Cu−N2	2.007(2)	2.070(2)	2.061(3)	2.0712(12)	2.0712(12)	2.066(2)	2.128(5)
P1‐Cu‐N1	91.20(8)	86.38(6)	88.39(9)	91.54(4)	91.54(4)	86.88(7)	89.24(12)
P2‐Cu‐N2	97.25(7)	86.98(7)	88.56(9)	87.86(4)	87.86(4)	87.81(7)	89.24(12)
plane angle[b]	85.88	71.09	73.93	77.56	78.48	68.01	88.31

[a] Not applicable, because of a centrosymmetric point group. [b] Determined ligand plane intersection angle (lpia) between the two ligand planes.

Figure 2

Solid state structures (ORTEP representation with thermal ellipsoids at a probability level of 50 %) of complex C1 with atom labeling (top) and of complex C7 highlighting the two‐fold rotation symmetry (bottom). The hydrogen atoms, counter anions and solvent molecules are omitted for clarity.

–

–

–

Furthermore, all CuI complexes display a strongly distorted tetrahedral geometry around the copper center as a common feature for this class of compounds.4, 61 In the solid state structure of C1 the two phenyl‐(2‐oxazoline) moieties are almost perpendicular oriented towards each other, without any perturbation in the ligand backbone. This can be visualized by the ligand plane intersection angle (lpia), which describes the angle between the two ligand planes, that are spanned through the chelating P^N‐heteroatoms and the copper center (see Table 1).4, 61 This angle is 85.88° for complex C1 without any substituents at the 4‐position of the 2‐oxazoline (Figure 3). The observed lpia is also well represented by the DFT calculations (BP86‐D3(BJ)/def2‐TZVP), which provide a value of 81.9° for C1 (Table 3). For C2–C7 the lpia significantly differs from the ideal 90° arrangement, for example, 71.09° (DFT, 75.4°) for C2 or 68.01° (DFT, 68.6°) for C6, and is strongly influenced by the various substituents. However, a clear trend concerning the kind and size of substituents cannot be observed. Instead, the individual packing in the solid state seems to superimpose any general trend.

Figure 3

Space‐filling representation of C1 (left) and simplified solid state structure (right) showing only the copper center and the chelating heteroatoms (with thermal ellipsoids at a probability level of 90 %). The simplified structure of C1 displays the two ligand planes, that is, the plane through the atoms P1‐Cu1‐N1 (yellow) and P2‐Cu1‐N2 (blue).

Table 3

Compilation of the ligand plane intersection angle (lpia, °), reorganization energies and energy differences between the S1 and T1 state of the complexes C1–C7 assessed from their calculated structures.

	C1	C2	C3	C4	C5	C6	C7
crystal	85.88	71.09	73.93	77.56	78.48	68.01	88.31
S0 [a]	81.88	75.35	76.03	71.89	79.91	68.59	84.10
S0 [b]	83.50	73.13	75.86	72.24	82.82	69.12	83.22
S1 [b]	45.72	56.98	59.41	41.03	66.09	38.53	68.24
ΔlpiaS_1-S_0b	37.78	16.15	16.45	31.21	16.73	30.59	15.00
ΔES_1-S_0b	84.45	68.85	58.85	93.94	64.84	95.95	58.41
ΔET_1-S_1b	0.15	0.17	0.13	0.17	0.16	0.17	0.15

[a] BP86‐D3(BJ)//def2‐TZVP. [b] B3LYP‐D3(BJ)//def2‐SVP.

Additionally, in C1 the copper center is asymmetrically chelated by the two L1 ligands. One ligand is more loosely bound and exhibits a bite angle of 91.20(8)°, while for the second ligand the Cu−P bond length is significantly shortened (Cu−P1: 2.2230(8) pm vs. Cu−P2: 2.1860(8) pm) and the bite angle is 97.25(7)°. A similar coordination behavior is also found for C4 and C5. The detailed inspection of the extended crystal structures and crystal packing effects revealed pairwise intermolecular π‐stacking interactions (Figure S4) for C1, C4, C5 and C7. In C4 and C5, one aryl ring of adjacent PPh2 groups participates in a perpendicular T‐shaped stacking interaction, whereas in C1 the interaction takes place between one PPh2 unit and the central aryl moiety in 2‐position of the oxazoline. A distance of about 500 pm between the centroids of the aryl groups clearly indicates such interactions,62, 63, 64 which are predominantly responsible for the asymmetric coordination environment. The highly symmetrical structure of C7 also exhibits pairwise π‐stacking interactions, but these are located between the substituents. Two neighboring benzyl units form an intermolecular parallel face‐centered π‐stacking with a centroid distance of 470 pm. Moreover, the DFT calculated ground state structures (Table 3 and Supporting Information) show a good agreement to the measured ones.

Although significantly different it is difficult to rank the substituents in terms of their steric requirements.65 To have a more profound definition and classification of the “size of the substituent” the molecular surfaces were calculated with the GEPOL algorithm.66 By calculating a smaller model system, that is, only considering the 4‐substituted 2‐oxazoline moiety (Figure S5, Table S4), the surface size follows the order L1C2

To verify these findings, an analysis of the buried volume (%V bur) using the SambVca 2.0 online software67 was carried out. This method developed by Cavallo et al. describes the space filling of the first coordination sphere in transition metal complexes.68, 69, 70 The results (Table S5) are in line with the surface sizes obtained by the GEPOL algorithm. This means, that the complexes C4 and C6 employing R configured ligands are much more densely packed around the copper center than the complexes C5 and C7 with S configured ligands.

Electrochemical Properties

Cyclic voltammograms (CVs) of the complexes C1–C7 were recorded in acetonitrile solution. Apart from C3 all compounds display a fully reversible reduction and oxidation event each (Figure 4, Table 3). It needs to be mentioned, that the reduction in C3 is only reversible at scan rates above 500 mV s−1, whereas the oxidation stays irreversible (Figure S6). The reduction of C1–C6 occurs at potentials between −2.25 to −2.21 V and corresponds to a one‐electron reduction of the P^N‐ligand (Figure 5).41, 44, 71 DFT calculations confirmed this assignment and revealed that the spin density of the reduced complex is only located at the coordinating atoms and the central aryl unit (Figure S8). Therefore, this process is largely invariant from the different substituents at the 2‐oxazoline moiety, because they are not part of the conjugated system of the phox ligand. Solely in complex C7 the reduction is slightly shifted by about 50 mV to more positive potentials.

Figure 5

Cyclic voltammograms of C1–C7 in acetonitrile solution referenced vs. the ferrocene/ferricenium (Fc/Fc+) couple. Left: reductive scans. Right: oxidative scans. C1/C2 (red, solid/dashed), C4/C6 (blue, solid/dashed), C5/C7 (green, solid/dashed), Conditions: scan rate of 100 mVs−1, with 0.1 m nBu4NPF6 as supporting electrolyte.

In contrast, the reversible oxidation potentials, which formally can be assigned to a metal‐centered Cu+/Cu2+ oxidation process, are influenced by the substituents. The oxidation potential of C2 (0.58 V) is shifted by 300 mV more anodic compared to that of C1 (0.29 V). Apparently, the CuII center ([Ar] 3d9) in the oxidized species favors a square planar coordination environment. This flattening seems to be hampered in C2 through the steric impact of the two methyl groups at the oxazoline moiety. This is in line with findings on CuI complexes with 2,9‐substituted phenanthroline ligands, for example, in [Cu(N^N)(POP)]+ (POP=bis[2‐(diphenylphosphino)‐phenyl]ether), where the complex with N^N=2,9‐dimethyl‐phenanthroline shows a shift of 150 mV towards more positive potentials compared to the complex without substituents at the diimine ligand.72 Moreover, there is a direct correlation between the oxidation potentials and the different orientations of the substituents in the chiral ligands. The complexes with a larger expansion of the molecular volumes C4 and C6 (0.49 V and 0.47 V) are oxidized at higher potentials compared to C5 and C7 (0.41 V and 0.42 V). All in all, these observations seem to be in relation to the entatic state principle,73 which describes a pre‐organization of the ligand sphere in order to stabilize a certain coordination mode. This structural pre‐distortion is known for CuI complexes in a protein matrix,74, 75, 76 as well as for other synthetically CuI complexes,77, 78, 79, 80, 81 for example, CuI complexes with guanidine ligands80, 81 or heteroleptic CuI complexes with sterically demanding phenanthroline ligands.78

The diffusion coefficients (, Table 2) were obtained from the scan rate‐dependent CVs and the baseline corrected forward‐scan peak potentials (i ) by using the Randles‐Sevcik equation (Figure S7). As a result, there seems no direct correlation between the diffusion constants and the chirality or the steric demand, because is always about 1.13×10−5 cm2 s−1. This is most likely due to a similar globular shape in solution for C1–C7.

Table 2

Summary of the electrochemical properties of the complexes C1–C7 in acetonitrile solution at room temperature. Potentials are referenced to the ferrocene/ferricenium (Fc/Fc+) redox couple. Diffusion coefficients D are determined using the Randles–Sevcik equation (see Supporting Information).

	Ered1/2	Eox1/2	D [cm2 s<M‐1]
C1	−2.25	0.29	1.15×10−15
C2	−2.25	0.58	1.14×10−5
C3	−2.14[a,b]	0.76 [a,c]	n.d.[d]
C4	−2.23	0.49	1.13×10−5
C5	−2.25	0.41	1.06×10−5
C6	−2.21	0.47	1.15×10−5
C7	−2.19	0.42	1.14×10−5

[a] Irreversible. [b] Anodic peak. [c] Cathodic peak. [d] Not determined.

D [cm2 s

Absorption and Emission Properties

The absorption spectra of C1–C7 in dichloromethane (Figure 6) are dominated by strong ligand centered (LC) transitions (π‐π* and n‐π*) in the wavelength range below 300 nm.41 Notably, the different substituents and substitution pattern have only a marginal influence on the band shape and energy. Solely in C6 and C7 additional LC transitions below 270 nm, caused by the aromatic substituents, increase the extinction coefficient to some extend (ϵ 250 nm=30 000 m −1 cm−1). In comparison, the complexes C4 and C5 with aliphatic chains in 4‐position at the 2‐oxazoline possess an extinction coefficient of ϵ 250 nm=25 000 m −1 cm−1. The comparably weak (ϵ 400 nm=1500–2300 m −1 cm−1) and broad tails for C1–C5 and C7 in the range between 350 and 450 nm can be attributed to metal‐to‐ligand charge transfer (MLCT) processes.41 The respective TD‐DFT calculations of C1 also suggest that the lowest‐lying excitations are mainly involving the frontier orbitals HOMO‐n (n=0, 1, 2) and LUMO+m (m=0, 1, 2). The HOMO (as well as the HOMO‐1 and HOMO‐2) of both complexes corresponds to a CuI d orbital. The LUMO and LUMO+1 are ligand‐centered orbitals, where the electron density is distributed over the coordinating nitrogen atoms and the joint of the 2‐oxazoline and the central aryl ring itself. The LUMO+2 is mostly localized on one of the aryl substituents of the PPh2 moiety (Figure S11b, Table S6). Consequently, the most dominant electronic transition (HOMO→LUMO) can be assigned to a charge‐separation from the copper center to the π*‐orbitals of the ligand sphere.41 This is also the case for C6, resulting in a far more pronounced and clearly separated MLCT band from the other optical transitions (Figure S11a, Tables S6 and S7). Compared to the unsubstituted parent complex C1, the low‐energy transitions are bathochromically shifted with a simultaneous increase in extinction coefficients (Figure 6). This observation is also reflected in the TD‐DFT calculations, where the S0→S1 and S0→S2 transitions are shifted to lower energy compared to C1.

Figure 6

UV/vis absorption spectra of C1/C2/C3 (red, solid/dashed/dotted), C4/C6 (blue, solid/dashed) and C5/C7 (green, solid/dashed) in dichloromethane under inert conditions. The inset is an enlargement of the MLCT region.

In comparison to structurally related mono‐ and multinuclear CuI complexes bearing P^N, P^N^P, P^N^N^P or N^P^N ligands, the low‐energy bands of C1–C7 are generally bathochromically shifted, for example, [(POP)Cu(P^N)]+, where P^N=8‐diphenyl‐phosphanylquinoline only exhibits an absorption maxima (shoulder) at 360 nm,44 and [Cu2(PNNP)Br2], with PNNP=1,3‐bis(1‐(2‐(diphenylphospanyl)phenyl)‐1H‐pyrazol‐3‐yl)benzene has also only weak absorption bands in the 325–375 nm region.82 This renders the present CuI phosphinooxazoline complexes more attractive for light‐harvesting applications. Nevertheless, the ability to harvest visible (sun)light is still inferior compared to benchmark photosensitizers like [Ru(bpy)3]2+ (λ max,MLCT=452 nm, ϵ=13.000 m −1 cm−1) or [fac‐Ir(ppy)3] (λ max=375 nm, ϵ=7.200 m −1 cm−1).83, 84 The light absorption of [Cu(P^N)2]+ complexes could possibly further improved by the use of larger, stiffer and sterically more demanding hetero‐bidentate P^N ligands.

Only complexes C2 and C4–C7 show emission (Figure S9, Table 4) in deaerated dichloromethane solution at room temperature (293 K), but of very low intensity so that reliable values for the quantum yields could not be obtained. The spectra are broad and structureless with a maximum at about 600 nm, which is indicative for 3MLCT states.

Table 4

Photophysical data of C1–C7 in dichloromethane solution and in the solid state at room temperature under argon. Relative photoluminescence quantum yields in solution (φ PLQY,l) were determined using [Ru(bpy)3]PF6 as standard (φ R=0.095 in MeCN95, 96). Absolute photoluminescence quantum yields in the solid (φ PLQY,s) were determined using an integrating sphere with an experimental error of 5 % of the obtained values. The excitation wavelength in all experiments was λ=355 nm.

	λemDCM	FWHM[d]	Φ PLQY,l	λemsolid	FWHM[d]	Φ PLQY,s	τemsolid
	[nm]	[nm]	[%]	[nm]	[nm]	[%]	[ns]
C1	‐[a]	–[c]	–[c]	–[a]	–[c]	0.3	–[c]
C2	603	122	1.4	549	99	8.5	1329±5
C3	507, 590[b]	–[c]	–[c]	585	110	0.6	666±207
C4	592	146	0.18	550	95	3.6	1457±12
C5	592	130	0.59	557	102	3.4	819±3
C6	609	136	0.34	574	99	2.4	488±6
C7	591	137	0.26	519	104	3.8	2781±7

[a] The emission intensity was below the detection limit. [b] Shoulder. [c] Not determined. [d] Full width at half maximum of the emission. [e] The long‐lived major lifetime component is given. For further details, see Supporting Information and ref. 97.

Excited State Structure

After excitation with light that corresponds to the wavelength of the MLCT region CuI is formally oxidized to CuII and the electron configuration changes from d10 to d9. In solution, this charge transfer induces a structural reorganization from a tetrahedral geometry in the ground state (S0) to a distorted square‐planar ligand field in the singlet excited state (S1). The related triplet excited state (T1) exhibits a similar geometry.41, 85 To get a deeper insight into the influence of the different substituents on this flattening process, the geometries of S0 and S1 of C1–C7 (Figures S10) were optimized at the B3LYP‐ D3(BJ)//def2‐SVP level of theory. As described in the discussion of the crystal structures above, in the ground state the lpia of C1–C7 is correlated to the steric information of the substituents, but an impact of π‐stacking effects and distortions within the ligands is also present. All in all, the difference in lpia between S0 and S1 geometries is directly associated with the steric encumbrance of the ligand within the complex. For C1, which does not contain substituents at the 2‐oxazoline moiety, a difference in lpia (Δlpia) between the S0 (83.50°) and the S1 state (45.72°) of 37.78° is the highest one of all complexes (Table 3). Instead, the Δlpias are significantly lower for C2 (4,4′‐dimethyl, 16.15°) and C3 (4,4′‐diphenyl, 16.45°), but practically unaltered in these two complexes containing two substituents each. However, the energy difference between the ground and the excited state is higher in the case of C3. This is due to an increased steric repulsion, which limits the structural changes in the ligand sphere.

Interestingly, for the unsymmetrically and singly substituted complexes C4–C7 the estimated reorganization energy is almost independent of the mass or size of the substituents, but strongly depends on the R and S configuration. TD‐DFT calculations of the complexes with R configured ligands (i.e., C4 and C6) exhibit a Δlpia of approx. 31° (Table 3). This is much larger compared to the S configured complexes C5 and C7 with a Δlpia of approx. 16°. Hence, only in C5 and C7, where the substituents at adjacent ligands are pointing towards each other, the flattening distortion is hampered (for S0 and S1 structures see Figure S10). Another promising alternative to prevent unwanted flattening upon photoexcitation is the design of linear CuI complexes based on for example, cyclic alkyl(amino)carbenes, N‐heterocyclic carbenes and different pyridine or amide ligands.86, 87, 88, 89 The coplanar arrangement of the ligands can suppress non‐radiative decay and reduce structural reorganization resulting in highly efficient CuI emitters.

Solid‐State Emission and Lifetime

Unlike their behavior in solution the complexes C2 and C4–C7 are clearly luminescent in the solid state (Figure S12b) with luminescence quantum yields Φ between 0.3 to 8.5 % (Table 4). Upon excitation in the MLCT regime the crystalline solids of C2 and C4–C7 exhibit structureless emission bands with a full width at half maximum (FWHM) of about 100 nm (Table 4). Interestingly, the emission maxima are significantly hypsochromically shifted of up to 72 nm for C7 compared to the maxima in solution. This shift can be explained by the luminescence rigodochromic effect90 as well as by the smaller changes of the molecular geometry in the solid state upon excitation.91 Linfoot et al. showed, that even in the solid state it is highly important to maximize the steric repulsion between the ligands and the metal center in order to increase the photoluminescence quantum yield of CuI complexes.92 They studied the solid state emission of [(POP)Cu(N^N)]+ complexes (with POP=bis[2‐(diphenyl‐phosphino)‐phenyl ether) and N^N = 4,4′‐dimethyl‐ or 4,4′,6,6′‐tetramethyl‐2,2′‐bipyridine) and found that already a methyl group largely hinders the flattening in the solid state and therefore increases the emission.86 The same behavior is observed in our case, when changing from C1 (no substituents) to C2 (two methyl groups). Furthermore, the complexes C1–C7 display an unexpected strong dependence of the solid‐state emission color from the different substituents (Figure S12a). The emission covers a spectral region ranging from 519 nm in C7 up to 585 nm for C3. However, there seems no clear correlation between the size or kind of the substituents and the emission maxima. In addition, emission lifetime measurements of crystalline samples of the complexes C1–C7 were performed at room temperature. After excitation at 355 nm, C2–C7 possess luminescence lifetimes in the sub‐microsecond timescale (Table 4). C7 exhibits the longest emission lifetime, with a long‐lived component of about 2.8 μs (Table 4). The energy separation between the S1 and T1 state in C1–C7 is estimated to be around 130–170 meV (Table S8). Thermally activated delayed fluorescence (TADF), however, is therefore expected to be less dominant and the emission lifetime is most likely only associated with the decay of a 3MLCT state.93, 94

[%]

[%]

Photostability and Photoreactivity

With respect to possible applications in solar energy conversion schemes a high photostability in solution is essential. Unfortunately, heteroleptic diimine‐diphosphine CuI complexes frequently suffer from a limited stability in solution under irradiation or catalytic conditions.31, 32, 33, 34, 98 Irradiation of the prototype photosensitizer [(P^P)Cu(N^N)]PF6 (P^P=xantphos and N^N=bathocuproine) in acetonitrile with a solar light source (i.e., a 150 W Xe arc lamp) leads to ligand dissociation and the formation of the respective homoleptic complexes (Figure S14). In contrast, a ligand exchange reaction in the related heteroleptic complex [(L2)Cu(bathocuproine)]PF6 does not occur (Figure S14). Furthermore, the homoleptic complexes C1–C7 are not prone to changes in their molecular composition, which results in more photostable (Figure S15) complexes compared to heteroleptic CuI complexes bearing a P^P ligand. The photostability measurements in acetonitrile solution have also shown, that the S configured complexes display a slightly higher stability than the R configured, which can be explained by the reduced steric constraints.

In a next step, the electron transfer properties of these novel photosensitizers upon light irradiation were tested by the interaction with commonly used sacrificial reductants as well as with a water reduction catalyst.99 For this purpose, an acetonitrile solution containing C2 (0.1 mm) and either dimethylphenylbenzimidazoline (BIH, 0.5 mm) or triethylamine (TEA, 100 mm) was irradiated. Simultaneous UV/vis measurements revealed the constant build‐up of a new band at 570 nm (Figure S16). The speed and magnitude of interaction using TEA, however, is substantially slower than with BIH, which is in accordance with the lower oxidation potential of BIH.99 The interplay of C2 with a water reduction catalyst was probed with the commonly used iron carbonyl complex [Fe3(CO)12]6, 7, 35, 41 (Figure S17). From the Stern–Volmer experiment, an apparent emission quenching can be seen (K SV = 1.99×103  m −1). These spectroscopic observations are in line with our previous results, where complex C2 was used for the light‐driven reduction of protons to hydrogen within a fully noble‐metal‐free system composed of [Fe3(CO)12] as water reduction catalyst and triethylamine as sacrificial reductant.41 There, C2 showed a low, but fairly constant production of H2 with a turnover number of 53 within 24 h.41

Conclusions

Based on the phosphinooxazoline ligand a systematic series of sterically modified P^N ligands L1–L7, without any substituents (L1), with only one substituent (L4–L7) and two substituents (L2–L3) at the 2‐oxazoline moiety, were prepared. These heterobidentate P^N ligands were then used to design a new class of homoleptic CuI photosensitizers C1–C7 and to study their impact on the photophysical and electrochemical properties. The different ligands are either available by a two‐step procedure starting from the commercially available amino alcohols or via a three‐step synthesis using natural amino acids. The complexes C1–C7 (yellow to orange solids) were mostly obtained in good yields with a remarkably photostability in solution. A comprehensive X‐ray analysis revealed a uniform coordination behavior around the copper center and exposed major differences of the ligand arrangement in S and R substituted complexes.

The redox processes of all complexes (except C3) are fully reversible, which is in strong contrast to most of the conventional heteroleptic diamine‐diphosphine CuI photosensitizers. While the reduction potentials are not affected by the different substituents or substitution pattern, the oxidation of C2–C7 occurs at higher potentials compared to C1, which does not bear any substituents. The absorption spectra are largely unaffected by the different substituents, with only complex C6 having a phenyl substituent exhibiting a pronounced bathochromic shift of the MLCT band. (TD‐)DFT calculations corroborate the findings and were used to determine the excited state structures. The respective MLCT state undergoes a strong flattening distortion, which is independent of the specific kind, size or number of substituents. Instead, the specific properties are dictated by the chirality of the ligands. Furthermore, the flattening distortion is the main reason for the weak emission in solution. In contrast, all complexes show a clear emission in the solid state with a dependence of the emission color on the steric information. Additionally, the successful interaction of C2 with sacrificial reductants as well as with an iron carbonyl water reduction catalyst was demonstrated. As another key advantage compared to heteroleptic diimine‐diphosphine CuI photosensitizers these novel [Cu(N^P)2]+ complexes were found to be quite photorobust and do not suffer from photoinduced ligand exchange reactions in solution.

All in all, this renders these class of compounds as suitable for various applications within solar energy conversion schemes. At the same time, this study highlights the importance of having control over chirality and steric information in CuI complexes. In the future, multidentate N^N^P or P^N^N^P ligands100 as well as macrocyclic phenanthroline ligands36 might be used as suitable alternatives to P^N ligands to further improve CuI based photosensitizers.

Experimental Section

CCDC https://www.ccdc.cam.ac.uk/services/strctures?id=doi:10.1002/chem.201904379 contain the supplementary crystallographic data for this paper. These data are provided free of charge by http://www.ccdc.cam.ac.uk/.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

Click here for additional data file.