Fluorescence vs. Phosphorescence: Which Scenario Is Preferable in Au(I) Complexes with Benzothiadiazoles?

The photoluminescence of Au(I) complexes is generally characterized by long radiative lifetimes owing to the large spin-orbital coupling constant of the Au(I) ion. Herein, we report three brightly emissive Au(I) coordination compounds, 1, 2a, and 2b, that reveal unexpectedly short emission lifetimes of 10–20 ns. Polymorphs 2a and 2b exclusively exhibit fluorescence, which is quite rare for Au(I) compounds, while compound 1 reveals fluorescence as the major radiative pathway, and a minor contribution of a microsecond-scale component. The fluorescent behaviour for 1–2 is rationalized by means of quantum chemical (TD)-DFT calculations, which reveal the following: (1) S0–S1 and S0–T1 transitions mainly exhibit an intraligand nature. (2) The calculated spin-orbital coupling (SOC) between the states is small, which is a consequence of overall small metal contribution to the frontier orbitals. (3) The T1 state features much lower energy than the S1 state (by ca. 7000 cm−1), which hinders the SOC between the states. Thus, the S1 state decays in the form of fluorescence, rather than couples with T1. In the specific case of complex 1, the potential energy surfaces for the S1 and T2 states intersect, while the vibrationally resolved S1–S0 and T2–S0 calculated radiative transitions show substantial overlap. Thus, the microsecond-scale component for complex 1 can stem from the coupling between the S1 and T2 states.


Introduction
Au(I) is a heavy d 10 ion, whose coordination compounds have widespread applications, particularly in areas associated with light emission [1][2][3][4]. For these applications, understanding the origin of photophysical processes and the structure-property relationships is essential, as they play a decisive role in the characteristics of functional materials [5][6][7]. Au coordination compounds generally reveal a phosphorescent nature of luminescence (both conventional phosphorescence and delayed fluorescence) [8][9][10][11], which is tailored by the heavy atom effect. The latter promotes spin-orbit coupling (SOC) between singlet and triplet states, thus facilitating intersystem crossing (ISC) between them, which relaxes the spin 'forbidden' radiative transitions [12]. In general, the heavier the metal atom of a coordination compound, the stronger the SOC [13][14][15]. However, recent studies revealed that the presence of heavy atoms does not guarantee efficient ISC, while the nature of ligands may play a critical role in determining the radiation pathway [16][17][18]. To this end, the theoretical understanding of the relaxation processes is crucial, especially since conventional ISC between the S 1 and T 1 states is not necessarily the only pathway [19,20]. For instance, in a recently reported Au(III) complex, the S 1 state couples with T 4 and T 5 , resulting in phosphorescence as a major radiation pathway, while the relative Au(I) complex mainly demonstrates fluorescence due to the absence of triplet states energetically close to S 1 [21]. The study of two Pt(II) complexes revealed the existence of two channels, S 1 -T 1 and S 1 -T 2 ; structural differences in the complexes result in a dominance of either one or the other channel, which induces their different luminescent behaviour [22]. Recently reported benzothiadiazole-based Au(I) compounds feature high quantum yields, reaching 97% in various organic matrices [23]. Notably, in the latter complexes, only a prompt fluorescence was observed.
Inspired by these encouraging examples, we study the luminescent properties of novel Au(I) complexes 1, 2a, and 2b with two closely related ligands bearing benzothiadiazole (btd) moiety. In this context, amino-substituted btds open up a great scope for the design of photoactive materials: a slight variation of the composition and structure of btds results in a wide variation of energetic characteristics of electronic transitions, which is the key to fine-tuning the photophysical properties [24][25][26][27]. Herein, we chose two phosphorusnitrogen ligands, comprising the NH-btd moiety as a chromophore, and PPh 2 moiety as a readily accessible unit for the metal coordination. Synthesized compounds 1-2 exhibit prompt fluorescence, unusual for Au(I) complexes. We provide an interpretation of this phenomenon using quantum chemical DFT calculations.
Molecules 2022, 27, x FOR PEER REVIEW 2 of 13 channels, S1-T1 and S1-T2; structural differences in the complexes result in a dominance of either one or the other channel, which induces their different luminescent behaviour [22]. Recently reported benzothiadiazole-based Au(I) compounds feature high quantum yields, reaching 97% in various organic matrices [23]. Notably, in the latter complexes, only a prompt fluorescence was observed. Inspired by these encouraging examples, we study the luminescent properties of novel Au(I) complexes 1, 2a, and 2b with two closely related ligands bearing benzothiadiazole (btd) moiety. In this context, amino-substituted btds open up a great scope for the design of photoactive materials: a slight variation of the composition and structure of btds results in a wide variation of energetic characteristics of electronic transitions, which is the key to fine-tuning the photophysical properties [24][25][26][27]. Herein, we chose two phosphorus-nitrogen ligands, comprising the NH-btd moiety as a chromophore, and PPh2 moiety as a readily accessible unit for the metal coordination. Synthesized compounds 1-2 exhibit prompt fluorescence, unusual for Au(I) complexes. We provide an interpretation of this phenomenon using quantum chemical DFT calculations.

Synthesis and Crystal Structure of the Compounds
The new compound N-(diphenylphosphino)-2,1,3-benzothiadiazole-4-amine (PN; Scheme 1) was prepared by the reaction of 4-amino-2,1,3-benzothiadiazole and chlorodiphenylphosphine, in the presence of triethylamine as a base. The corresponding 1,3-phosphinoamine, PCN, was prepared by the one-pot condensation of diphenylphospine, benzaldehyde, and 4-amino-2,1,3-benzothaidiazole as reported recently [28]. Reactions of equimolar amounts of (THT)AuCl (THT-tetrahydrothiophene) with PN and PCN yielded complexes [Au(PN)Cl] (1) and [Au(PCN)Cl] (2), respectively. Changing the molar ratio of the phosphinoamines and Au to 1:2 produced the same complexes. For complex 2, we obtained two polymorphs, 2a and 2b. The first one (2a) precipitated by slow concentration of a tetrahydrofuran (THF) solution under vacuum, while the second polymorph (2b) was formed from oversaturated oily THF solution. Both polymorphs can be obtained as a single phase, as evidenced by powder X-ray diffraction (XRD) analysis (Figures S1-S3). According to single-crystal XRD analysis, compounds 1, 2a, and 2b exhibit a linear coordination environment of the metal by the Cl and P atoms ( Figure 1). Except for these, Au(I) reveals neither intra-nor intermolecular contacts. No other specific intermolecular According to single-crystal XRD analysis, compounds 1, 2a, and 2b exhibit a linear coordination environment of the metal by the Cl and P atoms ( Figure 1). Except for these, Au(I) reveals neither intra-nor intermolecular contacts. No other specific intermolecular interactions were found. The geometry of molecule 2a mainly differs from 2b in the torsion angle {Au-P-C-N} (56.2 • in 2a versus 165.7 • in 2b). According to DFT calculations, the single molecules have similar Gibbs energy (the energy difference is less than 1 kJ·mol −1 ), while the calculated energy barrier (both in vacuum and in THF), corresponding to the transition from 2a to 2b, amounts to 30 kJ·mol −1 (Figure 2). This implies that the conformers 2a and 2b can easily transform to each other in the solution, which is corroborated by the presence of one set of signals in the corresponding 31 P and 1 H NMR spectra. Upon crystallization, either one or the other conformer is implemented, depending on a slight variation in the conditions; specifically, almost the entire amount of the dissolved complex can precipitate as pure phase 2a. interactions were found. The geometry of molecule 2a mainly differs from 2b in the torsion angle {Au-P-C-N} (56.2° in 2a versus 165.7° in 2b). According to DFT calculations, the single molecules have similar Gibbs energy (the energy difference is less than 1 kJ·mol -1 ), while the calculated energy barrier (both in vacuum and in THF), corresponding to the transition from 2a to 2b, amounts to 30 kJ·mol -1 ( Figure 2). This implies that the conformers 2a and 2b can easily transform to each other in the solution, which is corroborated by the presence of one set of signals in the corresponding 31 P and 1 H NMR spectra. Upon crystallization, either one or the other conformer is implemented, depending on a slight variation in the conditions; specifically, almost the entire amount of the dissolved complex can precipitate as pure phase 2a.

Photophysical Properties and TD-DFT Calculations
The UV-Vis spectra were measured for 1-2 in the solid state ( Figure 3, Table 1). The shape of the spectra is quite similar, with the exception that the long wavelength band for compound 1 is hypsochromically shifted as compared to 2a and 2b. According to TD-DFT interactions were found. The geometry of molecule 2a mainly differs from 2b in the torsion angle {Au-P-C-N} (56.2° in 2a versus 165.7° in 2b). According to DFT calculations, the single molecules have similar Gibbs energy (the energy difference is less than 1 kJ·mol -1 ), while the calculated energy barrier (both in vacuum and in THF), corresponding to the transition from 2a to 2b, amounts to 30 kJ·mol -1 ( Figure 2). This implies that the conformers 2a and 2b can easily transform to each other in the solution, which is corroborated by the presence of one set of signals in the corresponding 31 P and 1 H NMR spectra. Upon crystallization, either one or the other conformer is implemented, depending on a slight variation in the conditions; specifically, almost the entire amount of the dissolved complex can precipitate as pure phase 2a.

Photophysical Properties and TD-DFT Calculations
The UV-Vis spectra were measured for 1-2 in the solid state ( Figure 3, Table 1). The shape of the spectra is quite similar, with the exception that the long wavelength band for compound 1 is hypsochromically shifted as compared to 2a and 2b. According to TD-DFT

Photophysical Properties and TD-DFT Calculations
The UV-Vis spectra were measured for 1-2 in the solid state ( Figure 3, Table 1). The shape of the spectra is quite similar, with the exception that the long wavelength band for compound 1 is hypsochromically shifted as compared to 2a and 2b. According to TD-DFT calculations for the single molecules at their ground-state optimised geometry (Tables S1-S6), the long wavelength band corresponds to the S 0 -S 1 transition between HOMO and LUMO orbitals (Figure 4, Figures S4 and S5). The transition has a locally excited (LE) character with a notable contribution of an intraligand charge transfer (ILCT) from the periphery to the thiadiazole moiety, which is common for btd derivatives ( Figure 4) [28][29][30][31]. In addition, compound 1 features a minor contribution of Au and Cl orbitals to the HOMO, which implies the presence of (M+X)LCT (metal-and halogen-to-ligand charge transfer). Upon photoexcitation, solids 1-2 exhibit a bright emission, with the absolute quantum yields of 30-34% (Table 1). The position of the bands does not depend on the excitation wavelength in the range of 300-400 nm ( Figure S6). The luminescence spectra of compounds 2a and 2b reveal a hypsochromic shift of the emission band by ca. 450 and 100 cm −1 , respectively, compared with free PCN. Compound 1 exhibits a more pronounced hypsochromic shift by ca. 2750 cm −1 compared with PN. The absorption and emission spectra of 2a and 2b in THF solution ( Figure S7) are identical, suggesting that there is one emitting species in the solution. calculations for the single molecules at their ground-state optimised geometry (Tables S1-S6), the long wavelength band corresponds to the S0-S1 transition between HOMO and LUMO orbitals (Figures 4, S4 and S5). The transition has a locally excited (LE) character with a notable contribution of an intraligand charge transfer (ILCT) from the periphery to the thiadiazole moiety, which is common for btd derivatives ( Figure 4) [28][29][30][31]. In addition, compound 1 features a minor contribution of Au and Cl orbitals to the HOMO, which implies the presence of (M+X)LCT (metal-and halogen-to-ligand charge transfer). Upon photoexcitation, solids 1-2 exhibit a bright emission, with the absolute quantum yields of 30-34% (Table 1). The position of the bands does not depend on the excitation wavelength in the range of 300-400 nm ( Figure S6). The luminescence spectra of compounds 2a and 2b reveal a hypsochromic shift of the emission band by ca. 450 and 100 cm -1 , respectively, compared with free PCN. Compound 1 exhibits a more pronounced hypsochromic shift by ca. 2750 cm -1 compared with PN. The absorption and emission spectra of 2a and 2b in THF solution ( Figure S7) are identical, suggesting that there is one emitting species in the solution.
(a) (b)    Unexpectedly, the emission decay curves of solids 1-2 exhibit kinetics on the nanosecond scale, which yields the emission lifetimes of 10-20 ns (Figures S8-S10). This is not typical for Au complexes, which usually reveal microsecond-scale kinetics owing to efficient SOC between the singlet and triplet states. The emission bands change negligibly upon cooling from room temperature to 77 K ( Figure S11). Compounds 2a and 2b show exclusively prompt luminescence, while 1 features a weak afterglow signal after a time delay of 300 µs; the position of the corresponding band approximately coincides with that in the steady-state spectrum (Figure 5a). Thus, we can speculate that both prompt and delayed emission result from the same excited state. Since the microsecond-scale component (with an estimated lifetime of 100 µs) is not detected against the background  Unexpectedly, the emission decay curves of solids 1-2 exhibit kinetics on the nanosecond scale, which yields the emission lifetimes of 10-20 ns (Figures S8-S10). This is not typical for Au complexes, which usually reveal microsecond-scale kinetics owing to efficient SOC between the singlet and triplet states. The emission bands change negligibly upon cooling from room temperature to 77 K ( Figure S11). Compounds 2a and 2b show exclusively prompt luminescence, while 1 features a weak afterglow signal after a time delay of 300 μs; the position of the corresponding band approximately coincides with that in the steady-state spectrum (Figure 5a). Thus, we can speculate that both prompt and delayed emission result from the same excited state. Since the microsecond-scale component (with an estimated lifetime of 100 μs) is not detected against the background of the nanosecond-scale kinetics, we assume it has a contribution of less than 1% of the overall emission. One should note that the afterglow intensity decreases under constant irradiation of a sample of 1, even in deaerated conditions. Within an hour, the signal drops by 10 times under a 450 W source of white light. Under the same conditions, the overall steady-state emission spectrum changes negligibly. This implies that the triplet states decompose due to processes in the solid, and are less photostable compared to the singlet excited states. The afterglow signal and its behaviour persist for samples from several parallel syntheses.
To gain insights into the peculiar fluorescent behaviour of compounds 1-2, we per- One should note that the afterglow intensity decreases under constant irradiation of a sample of 1, even in deaerated conditions. Within an hour, the signal drops by 10 times under a 450 W source of white light. Under the same conditions, the overall steady-state emission spectrum changes negligibly. This implies that the triplet states decompose due to processes in the solid, and are less photostable compared to the singlet excited states. The afterglow signal and its behaviour persist for samples from several parallel syntheses.
To gain insights into the peculiar fluorescent behaviour of compounds 1-2, we performed TD-DFT calculations for the emission transitions from the singlet and triplet excited states (Table 1), as well as the SOC matrix elements between these states (Tables S7 and S8). For complexes 2a and 2b, the calculated SOC is weak (the coefficients are 10 cm −1 and less), which is in line with the observed absence of a phosphorescence. The calculated SOC for 1 is several times higher, which is still quite weak, but it enables the mixing of the singlet and triplet states. For the complexes, the brightest excitation transition in the long wavelength region is S 0 -S 1 , while the others (including S 2 and S 3 ) feature a much shorter wavelength and a 10 times lower magnitude of oscillator strength (Table 2). Therefore, we consider the S 1 -T n channels as the most probable pathways. The calculated T 1 energy level is widely separated from the S 1 level (the difference between vertical S 0 -S 1 and S 0 -T 1 transitions of ca. 7000 cm −1 ). This results in a very slow ISC from the S 1 sate and thus enables fluorescence from this state. A similar observation was observed for Pt complexes that also feature high ∆E(S 1 −T 1 ) of 6000 cm −1 resulting in the fluorescence along with conventional phosphorescence [32]. For complex 1, the T 2 level locates energetically much closer to S 1 than does T 1 . Vibrationally resolved calculated T 2 -S 0 and S 1 -S 0 emission spectra strongly overlap owing to the presence of a large number of vibrational sublevels of the S 1 and T 2 manifolds (Figure 5b). The comparison of energies at different geometry states ( Figure 6) revealed that the potential energy surface (PES) for the S 0 state has a local minimum at the T 2 equilibrium geometry, while no imaginary frequencies were found for this geometry. As evidenced by the energy diagram, the PESs for the S 1 and T 2 states intersect: for the equilibrium T 2 -geometry, the energies for the S 1 and T 2 states show reverse order as compared to the equilibrium of the S 1 geometry. Note also that the HOMO orbital for the relaxed T 2 state differs from that for S 1 state by a higher contribution of Au and Cl atoms ( Figure S12). In addition, the SOC matrix elements between S 1 and T 2 at the T 2 equilibrium geometry are almost twice as large as those at the S 0 and S 1 geometries (Table S8). Therefore, we assume that the vibronic SOC [33] between the S 1 and T 2 states is a process responsible for ISC in complex 1, and, consequently, for the minor microsecond-scale emission in the visible region. This process is much less favourable with respect to the S 1 -S 0 radiation relaxation, thus resulting in the fluorescence as the major pathway. A portion of absorbed energy could release as a phosphorescence from the T 1 state, but our calculations predict its manifestation in the infrared region. Thus, it is not detectable under the measurement conditions. that the vibronic SOC [33] between the S1 and T2 states is a process responsible for ISC in complex 1, and, consequently, for the minor microsecond-scale emission in the visible region. This process is much less favourable with respect to the S1-S0 radiation relaxation, thus resulting in the fluorescence as the major pathway. A portion of absorbed energy could release as a phosphorescence from the T1 state, but our calculations predict its manifestation in the infrared region. Thus, it is not detectable under the measurement conditions.
(a) (b) Figure 6. Representation of energy levels for different geometry states (So, S1, T1, and T2) of compounds 1 (a) and 2a,b regarding the example of 2a (b) in the energy range from 0 to 30 × 10 3 cm -1 .
Complexes 2a and 2b exhibit a larger ΔE(S1−T2) compared to complex 1, and the vibrational sublevels for the S1 and T2 manifolds show almost null overlap. In addition, the potential energy surfaces for S1 and T2 do not intersect. This may be the reason why 2a and 2b do not show an afterglow signal. Figure 6 represents energy diagrams that summarize the different behaviours arising from the differences in the excited states of complexes 1 and 2a. For 2b, ( Figure S13) the diagram is similar to that for 2a.

General
All manipulations for the syntheses were performed using the Schlenk technique and a glovebox. Solvents were purified using the standard technique and stored under an argon atmosphere. The 1,3-aminophosphine was synthesized as described recently [28]. Elemental analyses were performed on various MICRO cube instruments (Langenselbold, Germany) for C, H, N, and S elements. IR-spectra were recorded on a Fourier IR spectrometer FT-801 (Simex, Novosibirsk, Russia) in KBr pellets. 1 H NMR spectra (500.13 Complexes 2a and 2b exhibit a larger ∆E(S 1 −T 2 ) compared to complex 1, and the vibrational sublevels for the S 1 and T 2 manifolds show almost null overlap. In addition, the potential energy surfaces for S 1 and T 2 do not intersect. This may be the reason why 2a and 2b do not show an afterglow signal. Figure 6 represents energy diagrams that summarize the different behaviours arising from the differences in the excited states of complexes 1 and 2a. For 2b, ( Figure S13) the diagram is similar to that for 2a.

General
All manipulations for the syntheses were performed using the Schlenk technique and a glovebox. Solvents were purified using the standard technique and stored under an argon atmosphere. The 1,3-aminophosphine was synthesized as described recently [28]. Elemental analyses were performed on various MICRO cube instruments (Langenselbold, Germany) for C, H, N, and S elements. IR-spectra were recorded on a Fourier IR spectrometer FT-801 (Simex, Novosibirsk, Russia) in KBr pellets. 1 H NMR spectra (500.13 MHz) and 31 P NMR spectra (202.45 MHz) were obtained with a Bruker DRX-500 spectrometer (Madison, Wisconsin, USA) in dry CDCl 3 at room temperature (RT); the solvent peak was used as an internal reference.
The diffuse reflectance UV-Vis spectra of the solid samples ( Figure S14) were obtained with a Shimadzu UV-3101 spectrophotometer (Kyoto, Japan) at RT. Samples for the measurements were prepared by a thorough grinding of a mixture of the compounds under study (about 0.01 mol fraction) with BaSO 4 , which was also used as a standard. Spectral dependences of the diffuse reflectance were converted into spectra of a Kubelka-Munk function. Emission and excitation spectra of solids were recorded at RT on a Fluorolog 3 spectrometer (Horiba Jobin Yvon, Edison, NJ, USA) equipped with a cooled PC177CE-010 photon detection module (Horiba Jobin Yvon, Edison, NJ, USA) with a PMT R2658 photomultiplier (Horiba Jobin Yvon, Edison, NJ, USA). Excitation and emission spectra were corrected for source intensity (lamp and grating) and emission spectral response (detector and grating) by standard correction curves. For the measurements, powdered samples were placed between two nonfluorescent quartz plates. The emission decay curves were recorded on the same instrument using the TCSPC (time-correlated single photon counting) technique. A Horiba Nanoled laser diode (350 nm) was used as an excitation source. The curves were fitted by either one or two exponential decays; the fit yielded the emission lifetime values. Absolute PL quantum yields (QY) for the solids were recorded using the Quanta-ϕ device of Fluorolog 3. For the photostability experiment, solid sample 1 was constantly irradiated under Ar conditions with a full-range spectrum Xe lamp ushio uxl450s-0 (450 W; Horiba Jobin Yvon, Edison, NJ, USA) for 1 h.
The UV-Vis spectra of the solutions in 1 cm quartz cuvettes in an argon atmosphere were obtained with Agilent Cary 60 spectrometer (Santa Clara, California, USA). The emission and excitation spectra of the same solutions were recorded on an Agilent Cary Eclipse spectrometer.

Theoretical Calculations
The ground state geometries were optimized in a vacuum, without any constraints, at the DFT level using the PBE0 functional [34], def2-TZVP(-f) basis set, def2-ECP effective core potential (60 core electrons) for the Au atoms and D3(BJ) dispersion correction. The equilibrium geometries of the S 1 (for 2a) and T 2 states (for comparison between the geometries, see Figure S15) were optimized by TD(TDA)-PBE0-D3BJ/def2-TZVP(-f), and the T 1 state was optimized by UPBE0-D3BJ/def2-TZVP(-f) in the gaseous state, without any constraints. The equilibrium geometry of the S 1 state for 2b was optimized by TD-BHandHLYP-D3BJ rather than TD-PBE0-D3BJ, because the optimization with the PBE0 functional yielded a twisted structure ( Figure S16), with an underestimated energy of fluorescence. Optimization at the TD-BHandHLYP-D3BJ/def2-TZVP(-f) level of theory provided more reliable geometry and emission energy. The Frank-Condon and emission energies for all complexes were then calculated at the TD(A)-PBE0/def2-TZVP(-f) level. Regular TD-DFT was used for the singlet states, and the Tamm-Dancoff approximation of TD-DFT was used for the triplet states due to triplet instability problems in TD-DFT [35]. For the acceleration of calculations resolution of identity for the Coulomb part (RI) and the chain of spheres for the Fock exchange (COSX), approximations were used with the corresponding auxiliary basis [36,37]. Correspondence to the minima of optimized geometries were verified by analytical harmonic vibrational frequency calculations for the ground states and T 1 (UKS), and by numerical harmonic vibrational frequency calculations for the excited states.
The scalar-relativistic ZORA method was used with the SARC-ZORA-TZVP basis set for Au atoms to take into account relativistic effects in the photophysical properties; the ZORA-def2-TZVP basis set was used for the other atoms with SARC/J and AutoAux auxiliary basis sets for acceleration via the RIJCOSX algorithm. The quasi-degenerate perturbation theory was used for the calculation of spin-orbital coupling regarding the ZORA-TDA-DFT results [38]. The SOC integrals were calculated by the RI-SOMF(1X) method [39]. The calculations of SOC between states do not include vibronic coupling and non-adiabatic couplings. All geometry optimizations and TD(TDA)-DFT calculations were conducted using hte Orca 4.2.1 program. Molecular orbitals were visualized using ChemCraft 1.8.
Vibronically resolved fluorescence (from S 1 state) and phosphorescence (from T 1 and T 2 states) spectra were calculated by the path integral approach implemented in the ESD subprogram of Orca 5.0.3. Low-energy normal modes were cut due to their anharmonicity (TCutFreq value is 100 cm −1 ). The line shape was set by VOIGT option, which is the mean product of the Gaussian and Lorentzian curves (Linew parameter is 20 cm −1 and Inlinew parameter is 100 cm −1 ).

Synthesis of PN
At 0 • C, a solution of Ph 2 PCl (0.594 mL, 3.31 mmol) in toluene (5 mL) was added dropwise to a solution of NH 2 -btd (0.500 g, 3.31 mmol) and Et 3 N (0.922 mL, 6.62 mmol) in toluene (5 mL). The resulting mixture was allowed to warm up to RT and stirred overnight. The precipitate of Et 3 N·HCl was removed by filtration and rinsed four times with 3 mL of toluene. The solutions were combined, and all volatiles were evaporated. The residue was purified by column chromatography (silica gel, ethyl acetate:hexane = 1:10), followed by recrystallization from diethyl ether. Yield 0.876 g (79%). Calc. for C 18

X-ray Data
Single-crystal XRD data for compounds 1-2 (Tables S9-S11) were collected at 150 K with a Bruker D8 Venture diffractometer (Madison, Wisconsin, USA) with a CMOS PHO-TON III detector (Bruker, Madison, Wisconsin, USA) and IµS 3.0 microfocus source (MoK α radiation (λ = 0.71073 Å), collimating Montel mirrors; Incoatec GmbH, Geesthacht, Germany). The crystal structures were solved using the SHELXT [40] and were refined using the SHELXL [41] programs with OLEX2 GUI [42]. Atomic displacement parameters for non-hydrogen atoms were refined anisotropically. Hydrogen atoms were placed geometrically, with the exception of those in the amino groups, which were localized from the residual electron density map and refined with the restraining of the N-H bond (0.88 Å). The structures of 1-2 were deposited to the Cambridge Crystallographic Data Centre (CCDC) as a supplementary publication, No. 2217671-2217673.
Powder XRD data for the compounds (Figures S1-S3) were collected at RT with a Bruker D8 Advance diffractometer in Bragg-Brentano geometry with an energy discriminating Eyger XE T detector (CuK α radiation).

Conclusions
In summary, we synthesized Au(I) compounds 1-2 comprising closely related PN and PCN ligands. The latter comprise NH-btd moiety as a chromophore, and PPh 2 moiety as a readily accessible unit for the metal coordination. The main difference between the ligands is that one of them (PN) contains a P-N bond, while in the other (PCN), the P and N atoms are separated by a methylene bridge (Scheme 1). Such differences alter the nature of the luminescence of the corresponding complexes: polymorphs 2a and 2b exclusively exhibit fluorescence, which is not typical of Au(I) compounds, while compound 1 reveals fluorescence as the major radiative pathway, and a minor contribution of the microsecond-scale component.
We interpreted the unusual fluorescent behaviour using quantum chemical TD(A)-DFT calculations. In the complexes, the frontier molecular orbitals, responsible for the S 0 -S 1 and S 0 -T 1 transitions, are located almost exclusively on the PN and PCN ligands, i.e., the transitions mainly have an intraligand nature. Owing to the overall small metal contribution to the orbitals, the SOC is small. In addition, the T 1 state has much lower energy than does the S 1 (by ca. 7000 cm −1 ) state, which further hinders the ISC from S 1 . It thus enables the latter state to decay in the form of fluorescence as a dominant radiative process. In the specific case of complex 1, the emission origin is not only limited to the prompt S 1 decay, but also occurs with a mixing of singlet and triplet states, resulting in the presence of a minor microsecond-scale component. This behaviour is probably governed by a small contribution of (M+X)LCT, in the case of 1, which is absent in 2a and 2b. As a possible channel, the SOC between the S 1 and T 2 states can contribute to ISC, owing to rich vibronic structure of the corresponding singlet and triplet manifolds.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/molecules27238162/s1, Figures S1-S3: Simulated and experimental powder XRD patterns; Tables S1-S6: DFT calculation of Frank-Condon excitation processes; Figures S4 and S5: Frontier molecular orbitals of 1-2 in their S 0 optimized geometry; Figure S6: UV-Vis, emission and excitation spectra of the solids; Figure S7: UV-Vis, emission, and excitation spectra of 2 in THF; Figures S8-S10: Emission decay kinetics for the compounds; Figure S11: Emission of the compounds at 77-300 K; Tables S7 and S8: Calculated values of selected spin-orbit coupling matrix elements for the complexes; Figure S12: Comparison of frontier orbitals of complex 1 in different equilibrium geometries; Figure S13: Representation of energy levels for different geometry states of the complexes; Figure S14: IR-spectra of the compounds; Figure S15: Comparison of equilibrium geometries of the complexes in different states; Figure S16. Overlaid geometries for S 1 equilibrium state of 2b obtained by two different DFT methods. Table S9: Crystal data and structure refinement for the compounds; Table S10: Selected bond lengths for the structures; Table S11: Selected bond angles for the structures.