A new class of luminescent Cu( I ) complexes with tripodal ligands – TADF emitters for the yellow to red color range †

A new class of emissive and neutral Cu( I ) compounds with tripodal ligands is presented. The complexes were characterized chemically, computationally, and photophysically. Under ambient conditions, the powders of the compounds exhibit yellow to red emission with quantum yields ranging from about 5% to 35%. The emission represents a thermally activated delayed ﬂ uorescence (TADF) combined with a short-lived phosphorescence which represents a rare situation and is a consequence of high spin – orbit coupling (SOC). In the series of the investigated compounds the non-radiative rates increase with decreasing emission energy according to the energy gap law while the radiative rate is almost constant. Furthermore, a well- ﬁ t linear dependence between the experimental emission energies and the transition energies calculated by DFT and TD-DFT methods could be established, thus supporting the applicability of these computational methods also to Cu( I ) complexes.


Synthesis of the ligands
The three ligands tpypo L1, tpyps L2, and tpypse L3 were prepared in a two-step synthesis (Scheme 1). Tris(2-pyridyl)phosphine (tpyp) was synthesized from 2-bromopyridine and phosphorus trichloride via the classical method. 82,83 For this purpose 2-lithiopyridine was generated with nBuLi at −78°C. Further reaction with phosphorus trichloride under salt elimination at −100°C gave the desired product. Meanwhile, other synthetic methods are known to prepare tpyp, for example, with 2-bromopyridine and red phosphorus. [84][85][86] In the second step, the oxidized ligand tpypo L1 was synthesized according to a previously published method, which was slightly modified. In the literature, [87][88][89][90] hydrogen peroxide solution was used to oxidize tpyp, whereas in our procedure tBuOOH (80% in DTBP) was used. The other two ligands tpyps L2 and tpypse L3 could be easily prepared by refluxing tpyp with elemental sulfur or gray selenium in toluene. 87,88 The progress of the reaction could be monitored via 31 P-NMR spectroscopy. An excess of sulfur or selenium in the reaction mixture was separated by filtration via a syringe filter. Yields between 45% and 94% were reached in the oxidation reactions.
The synthesis of the arsine ligand tpyaso L4 was carried out analogously to tpypo L1. First, the arsine compound tpyas is formed followed by oxidation with tBuOOH solution to tpyaso L4 (Scheme 2). In contrast to tpyp, the arsine compound tpyas is stable in air.
The methane ligand tpym L5 was prepared by a literatureknown method. 91 First, 2-picoline is lithiated with nBuLi followed by reaction with 2-fluoropyridine under salt elimination (Scheme 3). The obtained mixture of the desired tpym L5 and the by-product di(2-pyridyl)methane dpym was separated by distillation.

Synthesis of the copper(I) complexes
In this section the syntheses of nine new copper(I) complexes are presented (Scheme 4). To obtain the complexes, the ligand was dissolved in acetonitrile and reacted with a copper(I) halide (CuCl, CuBr, and CuI) at ambient temperature. A colored precipitate was formed, which was collected by centrifugation. The yields of the different Cu(I) complexes varied significantly from 17% to 75%, depending on the solubility of the respective complex in the solvent.
We have also studied the coordination compounds of the phosphine tpyp with Cu(I) halide. Since the compounds are insoluble in common organic solvents, they could not be adequately characterized. It is assumed that the compounds are coordination polymers involving P-Cu interactions.

X-ray crystal structures
Crystal structures of tpyas and of the copper(I) complexes C1-C9 have been determined using X-ray diffraction measurements. The crystallographic data and structure refinement details are summarized in Tables 1 and 2. Selected bond distances and angles are listed for ligand tpyas in Fig. S1 † and for complexes C1-C9 in Tables 3 and 4. The crystal structure of tpyas is discussed only in the ESI, † since no complexes have been made with this ligand.
Suitable single crystals of copper complexes C1-C9 could be obtained from a saturated acetonitrile solution (at ambient temperature) after storing for a few days at 4°C. The complexes crystallize either in the monoclinic or orthorhombic crystal system. As a representative example of all the copper compounds investigated here, the molecular structures of the three complexes [CuItpypo] C3, [CuItpyaso] C8, and [CuItpym] C9 are shown in Fig. 1. The copper centers are coordinated by the halide anion and the three N atoms of the tripodal ligand in a distorted tetrahedral configuration. In each case, the halide atom deviates slightly from the Y-Cu axis (Y = P, As, C), which is due to packing effects that result from interactions with neighboring molecules (see Fig. S2-S4 † for the three complexes C1-C3 exhibiting the most pronounced bending of the halide). This is supported by quantum chemical calculations performed on the complex [CuCltpypo] C1 which predict that the energy of the molecule is minimal if the chloride is lying on the P-Cu axis corresponding to an angle of P1-Cu1-Cl1 = 180°. However, bending the chloride away from this axis by 10°, corresponding to an angle of P1-Cu1-Cl1 = 170°, results only in a minor energy increase of 1 kJ mol −1 . This indicates that the potential energy surface describing the halide bending is very flat. For the three complexes [CuCltpypo] C1, [CuBrtpypo] C2, and [CuItpypo] C3 the halide bending is most pronounced, with a bending angle of about 10°(P1-Cu1-X1 ≈ 170°) away from the P-Cu axis (see the angle Y(P, As, C)-Cu-X) in Tables 3 and 4 The Cu-X bond length increases from chloride to iodide due to the increasing atomic size of the halide, for example, in the series [CuXtpypo] from Cu1-Cl1 = 2.212(1) Å to Cu1-Br1 = 2.344(1) Å to Cu1-I1 = 2.499(1) Å. The bond length of the P-O bond is about 1.48 Å and that of the P-S bond is about 1.94 Å. Thereby they are in the range of the corresponding bond lengths of triphenylphosphine oxide (P-O 1.479(2) Å) and triphenylphosphine sulfide (P-S 1.950(3) Å). 92,93 Computational studies To learn more about the tripodally coordinated copper complexes, quantum chemical calculations have been carried out for complexes C1-C3 and C6-C9, using density functional theory (DFT) and time-dependent density functional theory (TD-DFT) with the functional B3LYP and the basis set def2-TZVPP. It has been shown that this method gives good results for Cu(I) complexes, especially for a description of the transition energies. 94 We want to focus the discussion on [CuItpypo] C3. For this compound, the optimized ground state geometry and the first excited triplet state geometry were calculated. The results are displayed in Fig. 2. In the ground state, the halide is lying on the axis that is defined by the P1-Cu1 atoms (Fig. 2a). However, in the triplet state geometry, the halide is bent away from this axis (Fig. 2b). This is clearly displayed by the three I-Cu-N angles. In the ground state geometry, all three angles I-Cu-N are nearly equivalent to values of about 123°. In the triplet state, the angle I1-Cu1-N1, amounting to 110.3°, is significantly smaller while the two other angles, I1-Cu1-N2 amounting to 127.4°and I1-Cu1-N3 amounting to 128.3°, are slightly larger than in the ground state (compare with Table S1 †). Moreover, it can be seen that in the triplet state geometry two Cu-N bonds are distinctly shortened, i.e. from the ground state value of 2.145 Å to 1.986 Å (Cu1-N2) and from 2.144 Å to 1.985 Å (Cu1-N3), respectively, while the Cu1-N1 bond length increases from 2.146 Å to 2.265 Å. This shows that two Cu-N bonds are strengthened and one Cu-N bond is weakened in the triplet state.
The occurrence of such a distortion in the excited state can be understood when the composition of the frontier orbitals is analyzed (Fig. 3). For [CuItpypo] C3, the HOMO is mainly localized on the copper atom and the halide, whereas the LUMO is mainly distributed over two of the three pyridine moieties of the tripod ligand. TD-DFT calculations reveal that transitions between these frontier orbitals largely determine the first excited singlet state S 1 and triplet state T 1 , especially in the triplet state geometry. Therefore, these states can be classified to be of (metal + halide)-to-ligand charge transfer ( 1,3 (M + X) LCT) character. To quantify the amount of charge that is transferred on excitation the natural charges for the copper and iodine atoms were calculated for the ground and the first excited triplet state. They amount to +0.321 (Cu) and −0.562 (I) for the ground state (S 0 geometry) and to +0.665 (Cu) and −0.324 (I) for the first excited triplet state (T 1 geometry). This nicely confirms the occurrence of a pronounced charge trans-fer away from the metal and halide upon excitation. Especially, the transfer of charge away from the copper center, in a rather crude approximation representing a formal oxidation of Cu(I)    to Cu(II), has an important consequence. As Cu(I) prefers a tetrahedral coordination environment, whereas Cu(II) prefers a planar one, significant structural reorganizations occur upon excitation. In the case of [CuItpypo] C3 these reorganizations are mainly represented by the bending of the halide away from the Cu-P axis and by changes of the Cu-N binding distances as described above.
Interestingly, a comparison of the two tripodal copper complexes, K[Cu(SC 6 F 5 )(HB(3,5-i Pr 2 pz) 3 )], representing a Cu(I) complex, and [Cu(SC 6 F 5 )(HB(3,5-i Pr 2 pz) 3 )], representing a Cu(II) complex, supports the results that are predicted from our calculations. 95,96 Compared to the Cu(I) complex, two of the three Cu-N bonds are shorter for the Cu(II) homologue (Cu(I): Cu-N11 = 2.188(7) Å, Cu-N31 = 2.062 (7)  For completeness it is remarked that similar geometry distortions for the triplet state geometry were also found for all other complexes presented in this study. This is shown in Table S1 † also for the two complexes [CuItpyaso] C8 and [CuItpym] C9.
In the scope of our calculations, the energy gaps between HOMO and LUMO show only slight variations between about 2.5 and 2.7 eV (data for all complexes are summarized in Table S2 †). For an estimate of the transition energies, a comparison of the HOMO-LUMO gaps is, however, not sufficient. Instead, TD-DFT calculations have been performed.
The TD-DFT transition energies of the different compounds, as calculated for the optimized T 1 state geometry, differ significantly from the experimental emission energies (see Table S3 †). However, it is not expected that these calculated gas phase data can exactly reproduce the experimental data, since (i) TD-DFT results have the tendency to underestimate the transition energies that correspond to charge-transfer states 94,97,98 and since (ii) especially for Cu(I) complexes, the experimental emission energies depend strongly on the environment and, in particular, on its rigidity, 28,31,32 which is not taken into account in our calculations. Nevertheless, the calculated transition energies nicely reproduce the trend of the emission energies as determined for the Cu(I) complexes doped into amorphous PMMA ( polymethylmethacrylate) (Fig. 4). It is remarked that the experimental trend is not repro-   duced by TD-DFT calculations performed for the ground state geometry. Furthermore, our calculations suggest that the energy separations ΔE(S 1 -T 1 ) between the T 1 and S 1 states are rather small, lying between about 1000 and 1350 cm −1 (compare Table S3 †). For such small energy splittings the occurrence of a thermally activated delayed fluorescence is expected. A more detailed discussion of this aspect is presented in the next section.

Photophysical studies
Again, the discussion will focus on [CuItpypo] C3. Fig. 5 shows the absorption spectrum of this compound and of the tpypo L1 ligand recorded in dichloromethane. The absorption bands observed in the range between about 230 nm and 300 nm are present in both the complex and the ligand. Therefore, they are assigned to result from ligand centered transitions. In contrast, in the range between about 300 and 500 nm, the ligand does not absorb, but for the complex two separate absorption bands are observed. This allows us to assign these bands to charge transfer transitions in agreement with the results obtained from DFT and TD-DFT calculations that predict lowlying (M + X)LCT states.
In fluid solution, an emission of [CuItpypo] C3 is not observed at ambient temperature even after oxygen was carefully removed by at least five freeze-pump-thaw cycles. In contrast, when doped into a PMMA matrix, the complex exhibits a red emission with a maximum at λ max = 635 nm and an emission quantum yield of Φ PL = 16%. As powder, the emission is orange with a maximum at λ max = 600 nm and a quantum yield of Φ PL = 20% (Fig. 5).
The emission of [CuItpypo] C3 in PMMA compared to the neat powder is red shifted by 35 nm. This can be attributed to differences in matrix rigidity. With decreasing rigidity the complex can undergo more distinct distortions upon (M + X) LCT excitation resulting in an energy stabilization of the emitting state(s). In addition to this red shift, the distortions cause an increase of non-radiative deactivations to the ground state due to increased Franck Condon factors that couple the excited state to the ground state. 99,100 As a consequence, the quantum yield decreases with decreasing rigidity of the environment from 20% ( powder) to 16% (PMMA) to ≪1% (solution).
The emission decay time and its temperature dependence show further important information. As for [CuItpypo] C3 doped into PMMA the decay strongly deviates from a monoexponential behavior (due to distinct inhomogeneity effects), 101 we focus on the emission properties of the powders. 102 At ambient temperature, an emission decay time of τ(300 K) = 4 µs and an emission quantum yield of Φ PL (300 K) = 20% are found for [CuItpypo] C3. When the powder sample is cooled to T = 77 K, the decay time increases to τ(77 K) = 24 µs and the emission quantum yield to Φ PL (77 K) = 63%. Calculation of the radiative and non-radiative rates according to k r = Φ PL τ −1 and k nr = (1 − Φ PL )τ −1 , respectively, reveals that the radiative rate increases by a factor of about two from k r (77 K) = 3 × 10 4 s −1 to k r (300 K) = 5 × 10 4 s −1 , whereas the non-radiative rate increases tenfold from k nr (77 K) = 2 × 10 4 s −1 to k nr (300 K) = 2 × 10 5 s −1 . Furthermore, the powder sample shows a slight blue-shift of the emission maximum from λ max (77 K) = 610 nm to λ max (300 K) = 600 nm.
An increase of the radiative rate and a blue-shift of the emission on increasing the temperature from T = 77 K to 300 K has often been reported for Cu(I) complexes and can be explained by the occurrence of a thermally activated delayed fluorescence (TADF). [28][29][30][31][32]34,35 Accordingly, at low temperature, only emission from the lowest excited triplet state T 1 occurs. With temperature increase, a thermal population of the energetically only slightly higher lying first excited singlet state S 1 becomes possible. As the S 1 →S 0 transition is significantly more allowed than the spin-forbidden T 1 →S 0 transition, an effective reduction of the emission decay time results. Moreover, as the S 1 state lies energetically higher than the T 1 state, a blue shift of the emission is expected to occur with increasing temperature.
However, for the compound [CuItpypo] C3, only a moderate increase of the radiative rate by a factor of about 2 was found to occur on heating. This is much less than what has been reported for other Cu(I) complexes. For example, the compounds presented in ref. 29 (Cu(I) halide dimers with different chelating aminophosphane ligands) experience a radiative rate increase by a factor of 40 to 150 and the compound discussed in ref. 36 (Cu(I) chloride dimer with a chelating diphosphane ligand) even shows a much higher factor of 490. The small radiative rate increase found for [CuItpypo] C3 can be rationalized by taking into account an additional and efficient emission decay path from the triplet state to the singlet ground state. The compounds discussed in ref. 29 and 36 exhibit long triplet decay times between 250 µs and 2200 µs. In contrast, the triplet decay time of [CuItpypo] C3 amounts to only 24 µs (T = 77 K value). As a consequence, a further shortening of the emission decay time by involving the TADF mechanism at higher temperature is much less effective. Further support for this interpretation is given in ref. 37 (three-coordinate Cu(I) carbene complex) where a Cu(I) complex with a similarly short triplet decay time of 34 µs is investigated. This compound also displays only a moderate increase of the radiative rate by a factor of about 3 when the TADF process is activated.
Obviously, the only moderate increase of the radiative rate upon heating can be rationalized by a significant contribution of the triplet state emission even at ambient temperature. Accordingly, both the singlet state S 1 and the triplet state T 1 contribute to the overall emission at ambient temperature. Consequently, the expected blue-shift of the emission maximum with increasing temperature is not as clearly displayed for the studied complexes as for TADF-only emitters (compare ref. [28][29][30][31][32][34][35][36]. However, the corresponding blueshift is displayed on the high-energy flanks of the spectra measured at T = 77 K and 300 K, respectively (compare with Table 5). Fig. 6 shows the emission spectra of the compounds C1-C3 and C6-C9.

The complexes [CuCltpypo] C1 and [CuBrtpypo]
C2 exhibit the same trends in the emission behavior as those described for [CuItpypo] C3. For all the other investigated compounds corresponding trends are expected to occur. An overview of the respective emission parameters is given in Table 5 An analysis of the radiative and non-radiative rates for the studied series of complexes reveals an interesting trend. At ambient temperature, powders of all compounds show similar radiative rates k r ranging from 3 × 10 4 to 6 × 10 4 s −1 . In contrast, the non-radiative rates k nr increase drastically with increasing emission wavelength. Such a trend is not unexpected and can be described by the energy gap law. In its simplest form it can be written as 103 wherein ΔE represents the energy separation between the states involved in the transition, γ a molecular coupling parameter, ħ the Planck constant, and ω n the dominant vibrational frequency that induces the non-radiative process. Accordingly, an exponential dependence of the non-radiative rate k nr on the emission energy ΔE is expected to occur. Indeed, this is experimentally observed. A graphical representation of this correlation is given in Fig. 7. These results demonstrate that the realization of compounds showing efficient red light emission remains a challenging task.

Conclusion and outlook
Emitter materials that exhibit a thermally activated delayed fluorescence (TADF) or a phosphorescence are highly attractive for use in OLEDs as they can convert all injected excitons into light. For applications, the emitter should also exhibit a short (radiative) decay time to minimize saturation effects. 104 The complexes investigated in this study are interesting in this regard; when compared to other Cu(I) complexes they exhibit relatively short phosphorescence (T 1 →S 0 emission) decay times, for example, amounting to only about 24 µs (radiative decay time 38 µs) in the case of [CuItpypo] C3. Thus, it can be concluded that spin-orbit coupling is particularly effective when compared to other Cu(I) complexes for which triplet decay times of several hundred microseconds or longer are not unusual. 29,32,36 Besides this already effective radiative deactivation process via phosphorescence, an additional radiative TADF path becomes important at ambient temperature for the investigated compounds. The combined emission paths of phosphorescence and TADF result in a distinct increase of the overall radiative rate when compared to TADF-only emitters. This property is highly attractive when the compounds are applied as emitters in OLEDs, in particular, to reduce saturation effects. Another interesting observation was made for the investigated compounds. In this series, the non-radiative rate increases strongly with decreasing emission energy following the energy gap law. As a consequence, the emission quantum yields decrease towards the red range of the spectrum from 28% for [CuItpym] C3 to 4% for [CuItpypse] C7. Accordingly, shifting the emission to the blue using methods of chemical engineering will result in a significant reduction of the nonradiative rates and will therefore lead to an increase of the emission quantum yield. An extrapolation of the data presented to shorter wavelengths reveals that for an emission wavelength of 460 nm, an emission quantum yield greater than 70% would be expected.

General remarks
The syntheses and handling of air-and moisture-sensitive substances were carried out using standard Schlenk and glovebox techniques. Solvents were dried using standard procedures 105 and stored over Al 2 O 3 /molecular sieves 3 Å/R3-11G catalyst (BASF).
The numbering of the hydrogen and carbon atoms is shown for the three ligands tpyp, tpyas, and tpym in Fig. 8.
Electrospray ionization (ESI) mass spectra were recorded on a Thermo Fisher Scientific LTQ FT Ultra using methanol, acetonitrile, or dichloromethane as the solvent. IR spectra were recorded on a Bruker Alpha FT-IR spectrometer using powder samples at ambient temperature. Elemental analysis was done using an Elementar vario MICRO cube. UV-Vis absorption measurements were carried out using a Varian Cary 300 double beam spectrometer. Emission spectra were recorded with a Fluorolog 3-22 (Horiba Jobin Yvon) spectrophotometer which was equipped with a cooled photo-multiplier (RCA C7164R). For the decay time measurements, the same photomultiplier was used in combination with a FAST ComTec multichannel scaler PCI card with a time resolution of 250 ps. As the excitation source for the decay time measurements, a pulsed diode laser (Picobrite PB-375L) with an excitation wavelength of λ exc = 378 nm and a pulse width <100 ps was used. For absolute measurements of photoluminescence quantum yields at ambient temperature and at 77 K, a Hamamatsu Photonics (C9920-02) system was applied. Doping of polymethylmethacrylate (PMMA) films was performed by dissolving the respective complex (<1 wt%) and the polymer in dichloromethane. After this, the solution was spin-coated onto a quartz-glass plate. All calculations were carried out with Gaussian09. 108 As the functional, B3LYP was used and as the basis set, def2-TZVPP was used. As the starting geometry, the structures obtained from X-ray measurements were used. No symmetry constraints were applied.
The data collection for the single crystal structure determinations was performed on a Stoe IPDS-II or IIT or a Bruker D8 QUEST diffractometer by the X-ray service department of the Fachbereich Chemie, University of Marburg. The Stoe IPDS-II and IIT devices are equipped with a Mo-K α X-ray source (λ = 0.71073 Å), a graphite mono-chromator and an active imaging plate. Stoe IPDS software (X-AREA) was used for data collection, cell refinement and data reduction, respectively. 109 The D8-QUEST is equipped with a Mo-K α X-ray micro source (Incotec), a fixed chi goniometer and a PHOTON 100 CMOS detector. Bruker software (Bruker Instrument Service, APEX2, SAINT) was used for data collection, cell refinement and data reduction. 110 The structures were solved with SIR-97 111 or SHELXS-97, 112 refined with SHELXL-2014 112 and finally validated using PLATON 113 software, all within the WinGX 114 software bundle. Absorption corrections were either applied with WinGX (multi-scan 115 or analytical 116 ) or beforehand with the APEX2 software (multi-scan). 117 Compound C8 was described as a non-merohedral twin. An HKLF5 absorption correction using Stoe X-Red32 (TwinAbs) 109 was applied. Graphic representations were created using Diamond 3. 118 C-bound Hatoms were constrained to the parent site. In all graphics the displacement ellipsoids are shown for the 50% probability level, and hydrogen atoms are shown with arbitrary radius. CCDC 1021437-1021446 contain the supplementary crystallographic data (excluding structure factors) for the structures reported in this paper.

Synthesis of the ligands
Preparation of tris(2-pyridyl)phosphine oxide (tpypo) L1. tpyp (375 mg, 1.4 mmol, 1.0 eq.) was dissolved in toluene (30 mL). After the addition of tBuOOH solution (80% in DTBP, 2.0 mL, 16 mmol, 11 eq.), the reaction solution was stirred at room temperature for 5 d and the progress was monitored via 31 P-NMR spectroscopy. After completion of the reaction, the solvent was removed in vacuo and the crude product was triturated with isopropanol (10 mL). After drying in vacuo, a colorless powder was obtained. Yield: 372 mg (1.3 mmol, 94%). Anal. Calc. for C 15  Preparation of tris(2-pyridyl)phosphine sulfide (tpyps) L2. tpyps was prepared by a modified literature method. 87,88 Tpyp (1.01 g, 3.80 mmol, 1.0 eq.) was dissolved in degassed toluene (20 mL), and sulfur (134 mg, 4.17 mmol, 1.1 eq.) was added. The suspension was refluxed for 24 h and the completion of the reaction was monitored via 31 P-NMR spectroscopy. After cooling to room temperature the suspension was filtered via a syringe filter and the filtrate was evaporated to dryness in vacuo. The crude product was recrystallized from ethanol to give a beige-colored powder. Yield: 517 mg (1. 74 31  Preparation of tris(2-pyridyl)phosphine selenide (tpypse) L3. tpypse was synthesized using a modified literature method. 88 Tpyp (300 mg, 1.14 mmol, 1.0 eq.) was dissolved in degassed toluene (15 mL) and gray selenium (90 mg, 1.14 mmol, 1.0 eq.) was added. The suspension was refluxed for 4 d and the completion of the reaction was monitored via 31 P-NMR spectroscopy. After cooling to room temperature the suspension was filtered via a syringe filter and the yelloworange filtrate was evaporated to dryness under vacuum. The crude product was triturated with diethyl ether (15 mL  Preparation of tris(2-pyridyl)arsine (tpyas). tpyas was prepared analogously using the following literature procedure. 82 2-Bromopyridine (10.93 g, 69.2 mmol, 3.9 eq.) was dissolved in diethyl ether (100 mL). After cooling to −78°C nBuLi (2.55 M in n-hexane, 28 mL, 71.4 mmol, 4.0 eq.) was added quickly and the solution turned dark red. The solution was first stirred for 1 h at −78°C followed by stirring for 1 h without cooling bath. Then the reaction solution was cooled to −100°C and a solution of arsenic trichloride (1.5 mL, 17.9 mmol, 1.0 eq.) in diethyl ether (50 mL) was added dropwise (5-10 mmol h −1 ) at a constant temperature before warming slowly to room temperature overnight. The reaction solution turned dark green and a colorless precipitate was formed. After the addition of 2 M HCl (25 mL), the colorless precipitate dissolved. The aqueous phase was isolated and the organic phase was extracted three times with 2 M HCl (25 mL). The aqueous phases were combined and alkalized ( pH = 8-10) with a saturated NaOH solution. The formed precipitate was filtered off and washed two times with cold water (10 mL). After drying in vacuo, a colorless powder was obtained. Yield: 1.13 g (3.7 mmol, 21%). Anal.
Calc. for C 15   Preparation of tris(2-pyridyl)arsine oxide (tpyaso) L4. tpyas (57 mg, 0.18 mmol, 1.0 eq.) was dissolved in toluene (15 mL) and tBuOOH solution (80% in DTBP, 0.12 mL, 0.96 mmol, 5.3 eq.) was added. The reaction solution was stirred at room temperature for 5 d. The solvent was removed under vacuum and the crude product was washed with n-pentane. After drying in vacuo, a colorless powder was obtained. Yield: 51 mg (0.16 mmol, 89%). Anal. Calc. for C 15  General procedure for the copper complexes The ligand tpypo, tpyps, tpypse, tpyaso or tpym was dissolved in a minimal amount of acetonitrile and CuCl, CuBr or CuI was added. After stirring for 1 d at room temperature the precipitated product was isolated by centrifugation. The supernatant liquid was used for the preparation of single crystals. After a few days of storage at 4°C, single crystals were obtained.