Tris(2-Pyridyl)Arsine as a New Platform for Design of Luminescent Cu(I) and Ag(I) Complexes

The coordination behavior of tris(2-pyridyl)arsine (Py3As) has been studied for the first time on the example of the reactions with CuI, CuBr and AgClO4. When treated with CuI in CH2Cl2 medium, Py3As unexpectedly affords the scorpionate complex [Cu(Py3As)I]∙CH2Cl2 only, while this reaction in MeCN selectively leads to the dimer [Cu2(Py3As)2I2]. At the same time, the interaction of CuBr with Py3As exclusively gives the dimer [Cu2(Py3As)2Br2]. It is interesting to note that the scorpionate [Cu(Py3As)I]∙CH2Cl2, upon fuming with a MeCN vapor (r.t., 1 h), undergoes quantitative dimerization into the dimer [Cu2(Py3As)2I2]. The reaction of Py3As with AgClO4 produces complex [Ag@Ag4(Py3As)4](CIO4)5 featuring a Ag-centered Ag4 tetrahedral kernel. At ambient temperature, the obtained Cu(I) complexes exhibit an unusually short-lived photoluminescence, which can be tentatively assigned to the thermally activated delayed fluorescence of (M + X) LCT type (M = Cu, L = Py3As; X = halogen). For the title Ag(I) complexes, QTAIM calculations reveal the pronounced argentophilic interactions for all short Ag∙∙∙Ag contacts (3.209–3.313 Å).

Working in this field, we have paid attention to pyridylarsines, "heavy pnictogen" analogs of pyridylphosphines. The latter are a famous family of ligands which are widely used for the design of Cu(I) and Ag(I) complexes showing very efficient TADF or/and phosphorescence [40][41][42][43][44]. The survey of the literature reveals, however, that the azaheterocyclic-substituted arsines are very limited in number [45][46][47], and their Cu(I)/Ag(I) derivatives are even scarcer [38,46,47]. Herein, for the first time, we have employed tris(2pyridyl)arsine as a stabilizing platform for Cu(I) and Ag(I) complexes. The latter have been studied in terms of solid-state luminescence, thermal stability and electronic structure.

Synthesis and Characterization
It was reported that tris(2-pyridyl)phosphine (Py3P) reacted with Cu(I) halides to irreversibly give air-stable dimeric complexes [Cu2(Py3P)2X2] (X = Cl, Br, I) in high yields [48]. Therefore, it would be reasonable to expect that Py3As could also give the related complexes. Meanwhile, we have found that the reactivity of Py3As differs from that of Py3P. Namely, CuI easily reacts with Py3As in CH2Cl2 to furnish the unexpected scorpionate [Cu(Py3As)I] (1) only, isolated as a solvate 1•CH2Cl2 in 89% yield (Scheme 1). By contrast, CuBr under similar conditions affords a dimeric complex [Cu2(Py3As)2Br2] (2) in 92% yield. Our numerous attempts to synthesize CuBr-based scorpionate under the varied conditions (different Cu/Py3As ratios, diverse solvents) failed: complex 2 was already formed in all the experiments. Of note, CuCl also easily reacts with Py3As, but a product formed was found to be easily oxidized in air to produce unidentified green Cu(II) complexes.

Scheme 1. Reactions of Py3As with CuBr and CuI.
To explain the different reactivity of Py3As towards CuBr and CuI, the dimerization energies for the equilibriums 2[Cu(Py3As)X] ↔ [Cu2(Py3As)2X2] (X = Br, I) have been assessed at the PBE0/def2-TZVP level of the theory. The performed calculations reveal that the formation of a dimer is indeed thermodynamically more preferable for X = Br by 2.34 kcal•mol -1 , whereas scorpionate is favorable for X = I by 0.97 kcal•mol -1 (Supplementary Materials, Figure S10). It should be noted, however, that an additional stabilization of scorpionate [Cu(Py3As)I] in 1•CH2Cl2 is possible through the solvate molecules (vide infra).
Our attempts to transform 2 into scorpionate [Cu(Py3As)Br] using recrystallization from different solvents failed. Meanwhile, we have found that the fuming of 1•CH2Cl2 with MeCN vapor (23 °C, 1 h) results in quantitative and irreversible dimerization into [Cu2(Py3As)2I2] (Figure 1a). The reaction is accompanied by the changing of the parent emission color of 1 (vide infra) to green, which is specific for complexes of [Cu2(Py3As)2X2] Scheme 1. Reactions of Py 3 As with CuBr and CuI.
To explain the different reactivity of Py 3 As towards CuBr and CuI, the dimerization energies for the equilibriums 2[Cu(Py 3 As)X] ↔ [Cu 2 (Py 3 As) 2 X 2 ] (X = Br, I) have been assessed at the PBE0/def2-TZVP level of the theory. The performed calculations reveal that the formation of a dimer is indeed thermodynamically more preferable for X = Br by 2.34 kcal·mol -1 , whereas scorpionate is favorable for X = I by 0.97 kcal·mol -1 (Supplementary Materials, Figure S10). It should be noted, however, that an additional stabilization of scorpionate [Cu(Py 3 As)I] in 1·CH 2 Cl 2 is possible through the solvate molecules (vide infra).
Our attempts to transform 2 into scorpionate [Cu(Py 3 As)Br] using recrystallization from different solvents failed. Meanwhile, we have found that the fuming of 1·CH 2 Cl 2 with MeCN vapor (23 • C, 1 h) results in quantitative and irreversible dimerization into [Cu 2 (Py 3 As) 2 I 2 ] (Figure 1a). The reaction is accompanied by the changing of the parent emission color of 1 (vide infra) to green, which is specific for complexes of [Cu 2 (Py 3 As) 2 X 2 ] type (X = P or As). The powder X-ray diffraction (PXRD) pattern of the product [Cu 2 (Py 3 As) 2 I 2 ], differs from that of [Cu 2 (Py 3 As) 2 Br 2 ] (2), but closely resembles that of a similar phosphine derivative, [Cu 2 (Py 3 P) 2 Br 2 ] (Figure 1b). Since the attempts to grow X-ray quality crystals of 1a met with no success, its structure has been proved by PXRD, mid-IR and 1 H NMR data only. Eventually, we have found that complex 1a can be successfully synthesized in an almost quantitative yield by the straightforward reaction of Py 3 As with CuI in MeCN (r.t., 10 min) (Figure 1a). Thus, a noticeable effect of solvents on the reaction outcome has been established for the reaction of Py 3 As with CuI.
type (X = P or As). The powder X-ray diffraction (PXRD) pattern of the product [Cu2(Py3As)2I2], differs from that of [Cu2(Py3As)2Br2] (2), but closely resembles that of a similar phosphine derivative, [Cu2(Py3P)2Br2] (Figure 1b). Since the attempts to grow Xray quality crystals of 1a met with no success, its structure has been proved by PXRD, mid-IR and 1 H NMR data only. Eventually, we have found that complex 1a can be successfully synthesized in an almost quantitative yield by the straightforward reaction of Py3As with CuI in MeCN (r.t., 10 min) (Figure 1a). Thus, a noticeable effect of solvents on the reaction outcome has been established for the reaction of Py3As with CuI. In the next step, we have examined the complexation of Py3As with Ag(I) using AgClO4 as a precursor. The reaction easily occurs in MeCN medium, and the following recrystallization of the product in water produces a fournuclear cluster [Ag@Ag4(Py3As)4](CIO4)5•3H2O (3•3H2O) in high yield (Scheme 2). Of note, the same product is also formed even when the AgClO4/Py3As molar ratio deviates from the stoichiometric one (5:4) being 1:1, 2:1, or 1:2. For comparison, the reaction of bis(2-pyridyl)phenylarsine (Py2PhAs) with AgClO4 under similar conditions has also been implemented to deliver a dinuclear complex [Ag2(Py2PhAs)2(MeCN)2](CIO4)2•CH3CN (4•CH3CN), the structure of which resembles that of 2. Again, the AgClO4/Py2PhAs ratio does not affect the reaction outcome: only complex 2 is formed. The X-ray derived structures of the prepared complexes are displayed in Figure 2, and the selected interatomic distances are listed in Table 1. In the structure of 1•CH2Cl2 (Figure 2), the Cu atom is coordinated by Py3As in a N,N′,N″-tripodal manner, and the In the next step, we have examined the complexation of Py 3 As with Ag(I) using AgClO 4 as a precursor. The reaction easily occurs in MeCN medium, and the following recrystallization of the product in water produces a fournuclear cluster [Ag@Ag 4 (Py 3 As) 4 ](CIO 4 ) 5 ·3H 2 O (3·3H 2 O) in high yield (Scheme 2). Of note, the same product is also formed even when the AgClO 4 /Py 3 As molar ratio deviates from the stoichiometric one (5:4) being 1:1, 2:1, or 1:2. For comparison, the reaction of bis(2-pyridyl)phenylarsine (Py 2 PhAs) with AgClO 4 under similar conditions has also been implemented to deliver a dinuclear complex [Ag 2 (Py 2 PhAs) 2 (MeCN) 2 ](CIO 4 ) 2 ·CH 3 CN (4·CH 3 CN), the structure of which resembles that of 2. Again, the AgClO 4 /Py 2 PhAs ratio does not affect the reaction outcome: only complex 2 is formed.
type (X = P or As). The powder X-ray diffraction (PXRD) pattern of the product [Cu2(Py3As)2I2], differs from that of [Cu2(Py3As)2Br2] (2), but closely resembles that of a similar phosphine derivative, [Cu2(Py3P)2Br2] (Figure 1b). Since the attempts to grow Xray quality crystals of 1a met with no success, its structure has been proved by PXRD, mid-IR and 1 H NMR data only. Eventually, we have found that complex 1a can be successfully synthesized in an almost quantitative yield by the straightforward reaction of Py3As with CuI in MeCN (r.t., 10 min) ( Figure 1a). Thus, a noticeable effect of solvents on the reaction outcome has been established for the reaction of Py3As with CuI. In the next step, we have examined the complexation of Py3As with Ag(I) using AgClO4 as a precursor. The reaction easily occurs in MeCN medium, and the following recrystallization of the product in water produces a fournuclear cluster [Ag@Ag4(Py3As)4](CIO4)5•3H2O (3•3H2O) in high yield (Scheme 2). Of note, the same product is also formed even when the AgClO4/Py3As molar ratio deviates from the stoichiometric one (5:4) being 1:1, 2:1, or 1:2. For comparison, the reaction of bis(2-pyridyl)phenylarsine (Py2PhAs) with AgClO4 under similar conditions has also been implemented to deliver a dinuclear complex [Ag2(Py2PhAs)2(MeCN)2](CIO4)2•CH3CN (4•CH3CN), the structure of which resembles that of 2. Again, the AgClO4/Py2PhAs ratio does not affect the reaction outcome: only complex 2 is formed. The X-ray derived structures of the prepared complexes are displayed in Figure 2, and the selected interatomic distances are listed in Table 1. In the structure of 1•CH2Cl2 (Figure 2), the Cu atom is coordinated by Py3As in a N,N′,N″-tripodal manner, and the The X-ray derived structures of the prepared complexes are displayed in Figure 2, and the selected interatomic distances are listed in Table 1. In the structure of 1·CH 2 Cl 2 ( Figure 2), the Cu atom is coordinated by Py 3 As in a N,N ,N"-tripodal manner, and the iodine atom completes a distorted tetrahedral environment of the metal (τ 4 = 0.85) [49]. Note that the iodine atom of 1 deviates from the axis passing from As and Cu atoms by only 0. could occur (vide infra). To sum up, Py3As ligands in 3 exhibit an As,N,N′,N″-coordination manner that is similar to that of Py3P in [Ag@Ag4(Py3P)4] 5+ complexes [52].
The structure of the cationic part of 4•CH3CN, [Ag2(Py2PhAs)2(MeCN)2] 2+ , is formed by two Ag(I) cations bridged by two Py2AsPh ligands in a head-to-tail manner. Both metal cations are also ligated by MeCN, thereby adopting a distorted trigonal pyramidal geometry. Again, the Ag…Ag distance in 4 being 3.2206(4) Å implies argentophilic interaction, which is actually taking pace according to the theoretical calculations (vide infra).   Complex 2 consists of two CuBr units bridged by two Py 3 As ligands (µ 2 -As,N,N') in a head-to-tail manner so that an inversion center is located in the middle of the molecule ( Figure 2). Therefore, each Cu atom adopts a distorted tetrahedral [Cu@AsN 2 Br] environment (τ 4 = 0.90). The intramolecular Cu···Cu distances (ca. 3.94 Å) are too long for the appearance of metallophilic interactions (cf. twice van der Waals radius of Cu is 2.80 Å [51]). The dihedral angle between planes of the coordinated pyridine rings of 2 is about 124.8 • . Overall, the structure of 2 closely resembles that of similar Py 3 P-based complex [Cu 2 (Py 3 P) 2  independent [Ag@Ag 4 (Py 3 As) 4 ] 5+ cations have a similar structure, which consists of a C 3 -symmetrical Ag@Ag 4 tetrahedral kernel supported by four Py 3 As ligands. The central Ag atom adopts a slightly distorted tetrahedral geometry (τ 4 = 0.98) constituted of four As atoms. Each vertex Ag atom of the Ag@Ag 4 kernel is coordinated by three N atoms of three neighboring Py 3 As ligands, thus accepting a trigonal pyramidal geometry [3 + 1].
Considering that the average distance between central and vertex Ag atoms of Ag@Ag 4 kernel (3.25 Å) is shorter than twice van der Waals Ag radius (3.44 Å) [51], argentophilic interactions could occur (vide infra). To sum up, Py 3 As ligands in 3 exhibit an As,N,N ,N"coordination manner that is similar to that of Py 3 P in [Ag@Ag 4 (Py 3 P) 4 ] 5+ complexes [52]. The structure of the cationic part of 4·CH 3 CN, [Ag 2 (Py 2 PhAs) 2 (MeCN) 2 ] 2+ , is formed by two Ag(I) cations bridged by two Py 2 AsPh ligands in a head-to-tail manner. Both metal cations are also ligated by MeCN, thereby adopting a distorted trigonal pyramidal geometry. Again, the Ag . . . Ag distance in 4 being 3.2206(4) Å implies argentophilic interaction, which is actually taking pace according to the theoretical calculations (vide infra).
The synthesized compounds are moderately soluble in dichloromethane and acetonitrile (1·CH 2 Cl 2 and 2), or in water (3·3H 2 O and 4·CH 3 CN). All of them are air-stable, except for 4·CH 3  The thermal stability of the above complexes has been studied by TGA, DTG and DTA techniques under argon atmosphere ( Figure S9). The solvate molecules of 1·CH 2 Cl 2 are lost at the range of 100-127 • C, but complex 1 itself is stable up to ≈200 • C. A higher stability is inherent in 2, which begins to decompose at about 240 • C. Compounds 3·3H 2 O and 4·CH 3 CN lose the solvate molecules at 120-140 • C and 135-155 • C, respectively, after which they remain stable at least up to 270 • C.

Theoretical Consideration
The electronic structure of luminescent Cu(I) complexes 1 and 2 has been investigated at the PBE0/def2TZVP level of the theory (for details, see §6 in ESI) to understand electronic transitions responsible for the excitation. For both complexes, the highest occupied molecular orbital (HOMO) and nearby HOMO-n are largely contributed by metal' d-orbitals and the lone pairs in halogen atoms ( Figure 3, Figures S11, S12; Tables S2, S3). The lowest unoccupied molecular orbital (LUMO) and nearby LUMO+n are mainly π-orbitals on the pyridine rings ( Figure 3, Figures S11, S12 ; Tables S2, S3). Interestingly, a lone pair in As atom does not contribute to the highest MOs (HOMO-HOMO-6). The fact that the frontier MOs of 1 and 2 are well separated in the space indicates a quite small energy gap between the lowest singlet (S 1 ) and triplet (T 1 ) exited states. Overall, the predicted HOMO/LUMO distribution is very typical for emitting TADF Cu(I) halide complexes.
A detailed consideration of HOMOs of 1 and 2 reveals small energy gaps separation between the HOMO and HOMO-1 levels which are, moreover, populated by d-orbitals with different spatial orientations (Figures S11 and S12). In particular, the HOMO-1/HOMO separation is 140 cm −1 for 2, and it is just 2 cm −1 for 1. According to the literature [12,13], such a scenario demonstrates a strong SOC mixing of the S 1 and T 1 states, which are originated from HOMO-1 → LUMO and HOMO → LUMO transitions, respectively. This, in turn, accelerates the rates of both S 1 → T 1 and T 1 → S 1 spin-forbidden processes, and hence, increases total rate of the luminescence. The experimental results fully confirm these predictions (vide infra). TD-DFT computations of 1 and 2 testify to the (M + X) LCT character (X = Br or I) of the low-energy absorptions (Tables S4, S5) that is specific for the related complexes. Furthermore, the (M + X) LCT nature of the lowest excited states is confirmed by a specific spin distribution in the computed T 1 state of 1 ( Figure S15). The lowest (LSOMO) and highest (HSOMO) single occupied molecular orbitals of the T 1 state of 1 ( Figure S15) closely resemble HOMO and LUMO in its S 0 state ( Figure 3). The calculated ∆E(T 1 -S 0 ) energy gap for the optimized states of 1 being 1.81 eV reasonably agrees with the emission energy of 1·CH 2 Cl 2 at a pure phosphorescence regime (77 K), i.e., 1.98 eV or 625 nm (vide infra). Therefore, one can expect that the emitting excited states of the compounds discussed should be of the (M + X) LCT type.
are originated from HOMO-1 → LUMO and HOMO → LUMO transitions, respectively. This, in turn, accelerates the rates of both S1 → T1 and T1 → S1 spin-forbidden processes, and hence, increases total rate of the luminescence. The experimental results fully confirm these predictions (vide infra). TD-DFT computations of 1 and 2 testify to the (M + X) LCT character (X = Br or I) of the low-energy absorptions (Tables S4, S5) that is specific for the related complexes. Furthermore, the (M + X) LCT nature of the lowest excited states is confirmed by a specific spin distribution in the computed T1 state of 1 ( Figure S15). The lowest (LSOMO) and highest (HSOMO) single occupied molecular orbitals of the T1 state of 1 ( Figure S15) closely resemble HOMO and LUMO in its S0 state (Figure 3). The calculated ∆E(T1-S0) energy gap for the optimized states of 1 being 1.81 eV reasonably agrees with the emission energy of 1•CH2Cl2 at a pure phosphorescence regime (77 K), i.e., 1.98 eV or 625 nm (vide infra). Therefore, one can expect that the emitting excited states of the compounds discussed should be of the (M + X) LCT type.  [53][54][55][56][57][58][59]. The balance between the Lagrangian kinetic energy G(r) and potential energy density V(r) at the bond critical points (3, -1) for Ag(I)•••Ag(I) interactions in 3 and 4 [viz. -G(r)/V(r) < 1] shows some covalent contribution in these short contacts [60]. The Laplacian of electron density is typically decomposed into the sum of contributions along the three principal axes of maximal variation, giving three eigenvalues of the Hessian matrix (λ1, λ2 and λ3), and the sign of λ2 can be utilized to distinguish the bonding (attractive, λ2 < 0) weak interactions from the non-bonding ones (repulsive, λ2 > 0) [61,62]. Thus, the Ag(I)•••Ag(I) interactions in 3 and 4 have an attractive character. For illustration, the calculated electron density distribution at the metal atoms of 4 is plotted in Figure 4.   [53][54][55][56][57][58][59]. The balance between the Lagrangian kinetic energy G(r) and potential energy density V(r) at the bond critical points (3, -1) for Ag(I)···Ag(I) interactions in 3 and 4 [viz. -G(r)/V(r) < 1] shows some covalent contribution in these short contacts [60]. The Laplacian of electron density is typically decomposed into the sum of contributions along the three principal axes of maximal variation, giving three eigenvalues of the Hessian matrix (λ 1 , λ 2 and λ 3 ), and the sign of λ 2 can be utilized to distinguish the bonding (attractive, λ 2 < 0) weak interactions from the non-bonding ones (repulsive, λ 2 > 0) [61,62]. Thus, the Ag(I)···Ag(I) interactions in 3 and 4 have an attractive character. For illustration, the calculated electron density distribution at the metal atoms of 4 is plotted in Figure 4. Table 2. Values of the density of all electrons -ρ(r), Laplacian of electron density -∇ 2 ρ(r) and appropriate λ 2 eigenvalues, energy density -H b , potential energy density -V(r), Lagrangian kinetic energy -G(r), and electron localization function -ELF at the bond critical points (3, −1), corresponding to Ag(I)···Ag(I) interactions in 3 and 4.

Photoluminescence of 1 and 2
At ambient temperature, Cu(I) complexes 1, 1a and 2 emit pronounced solid state photoluminescence (PL), whereas Ag(I) derivatives 3 and 4 appear to be almost non-emissive. The recorded emission and excitation spectra of 1, 1a and 2 are shown in Figure 5, and the measured PL properties are given in Table 3. As follows from these data, scorpionate 1 manifests an orange PL, while dimers 1a and 2 emit in a green region. In the terms of PL quantum yields (PLQYs), the emission of the studied compounds is moderate at 298 K (10-14%). All the emission profiles are of broad and structureless shape that is inherent in PL of the charge transfer origin [14]. The corresponding excitation curves are displayed by typical bands extending from the UV-edge and are sharply falling close at ≈450 nm (1a, 2) or 590 nm (1). The fact that the emission and emission profiles of iodide 1a and bromide 2 are almost superimposable is not confusing; previously, the similar cases were documented for other halide complexes, including the related ones [38,48,63]. The PL lifetimes

Photoluminescence of 1 and 2
At ambient temperature, Cu(I) complexes 1, 1a and 2 emit pronounced solid state photoluminescence (PL), whereas Ag(I) derivatives 3 and 4 appear to be almost nonemissive. The recorded emission and excitation spectra of 1, 1a and 2 are shown in Figure 5, and the measured PL properties are given in Table 3. As follows from these data, scorpionate 1 manifests an orange PL, while dimers 1a and 2 emit in a green region. In the terms of PL quantum yields (PLQYs), the emission of the studied compounds is moderate at 298 K (10-14%). All the emission profiles are of broad and structureless shape that is inherent in PL of the charge transfer origin [14]. The corresponding excitation curves are displayed by typical bands extending from the UV-edge and are sharply falling close at ≈450 nm (1a, 2) or 590 nm (1). The fact that the emission and emission profiles of iodide 1a and bromide 2 are almost superimposable is not confusing; previously, the similar cases were documented for other halide complexes, including the related ones [38,48,63]. The PL lifetimes of 1, 1a and 2 at 298 K are remarkably short (0.8-1.9 µs) compared to the most known Cu(I) complexes. Accordingly, the radiative constants (k r = PLQY/τ) being (0.53-1.75)·10 5 s -1 are relatively high, thereby indicating a strong SOC effect that is also predicted by DFT calculations (vide supra). For comparison, radiative rates (k r ) of the similar Py 3 P-based complexes [Cu 2 (Py 3 P) 2 X 2 ] at 298 K are much lower: 2.9·10 4 and 2.6·10 4 s -1 for X = Br and I, respectively. The acceleration of the emission rates observed in the arsine complexes is obviously attributed to much stronger SOC effect of arsenic compared to that of phosphorus. Previously, this effect was already demonstrated for Cu(I)-arsine complexes [38].   510 520 0.9 23 12 b a λex = 500 nm, b λex = 390 nm.
Temperature-dependent PL spectra of 1 and 2 ( Figure 6) demonstrate the significant enhancement of PL intensity upon cooling that is accompanied by a slight bathochromic shift of the bands by 10-20 nm (Table 3). When passing from 298 to 77 K, the PL lifetimes of 1 and 2 increase by 13 and 25 times, thus amounting 25 and 23 μs at 77 K, respectively. Taken together, these observations suggest the TADF manifestation at ambient temperature and phosphorescence at 77 K. According to the DFT and TD-DFT computations, the 1 (M + X) LCT emissive state is responsible for TADF (298 K), and 3 (M + X) LCT state is active at the phosphorescence regime (77 K). It should be underlined that the same emission scheme was previously proved for the related dimeric complexes based on Py3P and PhPy2As ligands. The ∆EST energy gaps between the 1 (M + X) LCT and 3 (M + X) LCT states of 1 and 2 can be roughly estimated by the red-shifting of the left flank of emission bands at their half height ( Figure S17). The estimated ∆EST gaps, being 340 and 890 cm -1 for 1 and 2, respectively, fall close to the range of such values (∆EST < 1500 cm -1 ) for TADF-active Cu(I) complexes [12][13][14].  (for 1a, 2) and 500 nm (1·CH 2 Cl 2 ). Temperature-dependent PL spectra of 1 and 2 ( Figure 6) demonstrate the significant enhancement of PL intensity upon cooling that is accompanied by a slight bathochromic shift of the bands by 10-20 nm (Table 3). When passing from 298 to 77 K, the PL lifetimes of 1 and 2 increase by 13 and 25 times, thus amounting 25 and 23 µs at 77 K, respectively. Taken together, these observations suggest the TADF manifestation at ambient temperature and phosphorescence at 77 K. According to the DFT and TD-DFT computations, the 1 (M + X) LCT emissive state is responsible for TADF (298 K), and 3 (M + X) LCT state is active at the phosphorescence regime (77 K). It should be underlined that the same emission scheme was previously proved for the related dimeric complexes based on Py 3 P and PhPy 2 As ligands. The ∆E ST energy gaps between the 1 (M + X) LCT and 3 (M + X) LCT states of 1 and 2 can be roughly estimated by the red-shifting of the left flank of emission bands at their half height ( Figure S17). The estimated ∆E ST gaps, being 340 and 890 cm -1 for 1 and 2, respectively, fall close to the range of such values (∆E ST < 1500 cm -1 ) for TADF-active Cu(I) complexes [12][13][14].
Emission and excitation spectra were recorded on a Fluorolog 3 spectrometer (Horiba Jobin Yvon) equipped with a cooled PC177CE-010 photon detection module and an R2658 photomultiplier. The absolute PLQYs were determined at 298 K using a Fluorolog 3 Quanta-phi integrating sphere. Temperature-dependent excitation and emission spectra as well as emission decays were recorded using an Optistat DN optical cryostat (Oxford Instruments) integrated with the above spectrometer.
Emission and excitation spectra were recorded on a Fluorolog 3 spectrometer (Horiba Jobin Yvon) equipped with a cooled PC177CE-010 photon detection module and an R2658 photomultiplier. The absolute PLQYs were determined at 298 K using a Fluorolog 3 Quantaphi integrating sphere. Temperature-dependent excitation and emission spectra as well as emission decays were recorded using an Optistat DN optical cryostat (Oxford Instruments) integrated with the above spectrometer.

X-ray Crystallography
The data were collected on a Bruker Kappa Apex II CCD diffractometer using ϕ, ω-scans of narrow (0.5 • ) frames with Mo Kα radiation (λ = 0.71073 Å) and a graphite monochromator. The structures were solved by direct methods SHELXL97 and refined by a full matrix least-squares anisotropic-isotropic (for H atoms) procedure using the SHELXL-2018/3 programs set [64]. Absorption corrections were applied using the empirical multiscan method with the SADABS program [65]. The positions of the hydrogen atoms were calculated with the riding model. Free solvent accessible volume in compound 3 derived from PLATON routine analysis was found to be 8.5% (1001.0 Å 3 ). This volume is occupied by H 2 O molecules, but this structure is based on very weak data (a better dataset cannot be obtained). Therefore, we employed the PLATON/SQUEEZE procedure to calculate the contribution to the diffraction from H 2 O molecules and thereby produced a set of H 2 O-free diffraction intensities. The final formula of 3, C 60 H 48 Ag 5 As 4 N 12 Cl 5 O 10 , was derived from the SQUEEZE results. The crystallographic data and details of the structure refinements are summarized in Table S1. CCDC 2090740, 2090741, 2126017 and 2126016 contain the supplementary crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Center at https://www.ccdc.cam.ac.uk/structures/ (accessed on 1 July 2022).

Conclusions
In conclusion, the reactions of tris(2-pyridyl)arsine (Py 3 As), an earlier unexplored ligand, with CuI, CuBr, and AgClO 4 have been studied. The interaction with CuI features a remarkable solvent-directing effect on the product structure. This reaction in CH 2 Cl 2 results in the crystallization of mononuclear scorpionate [Cu(Py 3 As)I]·CH 2 Cl 2 , whilst MeCN favors the selective formation of the dimer [Cu 2 (Py 3 As) 2 I 2 ]. Noteworthy, scorpionate [Cu(Py 3 As)I]·CH 2 Cl 2 , when fumed with MeCN vapor, easily (r.t., 1 h) and quantitatively dimerizes into [Cu 2 (Py 3 As) 2 I 2 ]. On the contrary, the treatment of Py 3 As with CuBr, regardless of solvent nature, affords the dimer [Cu 2 (Py 3 As) 2 Br 2 ] only. At ambient temperature, the above Cu(I) complexes manifest visible photoluminescence with noticeably short decay times (0.8-1.9 µs) and moderate quantum yields (10-14%). Taking into account the results of DFT computations and temperature-dependent photophysical measurements, the emission observed has been tentatively assigned to the thermally activated delayed fluorescence of (M + X) LCT kind (M = Cu, L = Py 3 As; X = halogen).
Complexation of Py 3 As with AgClO 4 results in the assembly of complex [Ag@Ag 4 (Py 3 As) 4 ](CIO 4 ) 5 , containing Ag-centered tetrahedron Ag@Ag 4 supported by four Py 3 As ligands. According to QTAIM analysis of this cluster, argentophilic interactions between its central and peripheral Ag(I) cations are observed.
To sum up, our findings reveal that the coordination chemistry of Py 3 As in some cases, e.g., in the reactions with Cu halides, may differ from that of Py 3 P. Moreover, the Py 3 As-derived Cu(I) complexes demonstrate higher emission rates (k r ) at 298 K compared to the similar Py 3 P-based analogs, thus highlighting the prospects of employing the arsine ligands for the design of short-lived TADF materials.