Spontaneous Water-Promoted Self-Aggregation of a Hydrophilic Gold(I) Complex Due to Ligand Sphere Rearrangement

Aggregating gold(I) complexes in solution through short aurophilic contacts promotes new photoluminescent deactivation pathways (aggregation-induced emission, AIE). The time dependence of spontaneous AIE is seldom studied. We examine the behavior of complex [Au(N9-hypoxanthinate)(PTA)] (1) in an aqueous solution with the aid of variable-temperature NMR, time-resolved UV–Vis and photoluminescence spectroscopy, and PGSE NMR. The studies suggest that partial ligand scrambling in favor of the ionic [Au(PTA)2][Au(N9-hypoxanthinate)2] pair followed by anion oligomerization takes place. The results are rationalized with the aid of computational calculations at the TD-DFT level of theory and IRI analysis of the electron density.


Introduction
The aqueous chemistry of linear gold(I) ([Xe] 5d 10 ) is far less developed than that of its square-planar congener, gold(III) ([Xe] 5d 8 ). The higher Pearson softness of gold(I) relative to gold(III) makes it less suitable for water coordination and thus disproportionate to give metallic gold(0) and gold(III). However, if the unstable gold(I) center is protected with suitable ligands, the complexes are stable in aqueous media and water-enriched mixtures [1]. This allows the exploitation of the interesting and characteristic properties of gold(I), e.g., catalytical [2][3][4][5][6][7][8] and biological [9][10][11] activity in these appealing media.
As an instance, gold(I) complexes have been popularized during the last years as Aggregation-Induced Emission Luminogens (AIEgens) [12] thanks to the ability of gold(I) to self-aggregate through short aurophilic contacts of <3.2 Å [13][14][15]. Gold(I) AIEgens essentially comprise highly hydrophobic complexes, e.g., pentafluorophenylgold(I) units bound to alkyl-or arylisocyanide ligands featuring long alkyl spacers or chains [16]. Aggregation is commonly forced by adding water to a neat solution of AIEgen. The spatial proximity of the gold(I) atoms in the nanosized aggregates gives rise to new excited states accessible by optical excitation, switching on new photoluminescent deactivation pathways. The degree of aggregation and the photoluminescence intensity are controlled by a balance between hydrophobicity and hydrophilicity.
There are fewer examples where AIE is achieved in gold(I) chemistry, not by modifying the solvent composition but spontaneously to gain thermodynamic stability. These include spontaneous self-assembly, as in the case of metallogelation [15,17,18], and ligand rearrangement conducive to diauracycles for releasing strain [19] or forming stabler species [20].
Molecules 2023, 28, x FOR PEER REVIEW 2 of 15 In this context, the water-soluble phosphine 1,3,5-triaza-7-phosphaadamantane (PTA; see Figure 1, left) [21] is widely employed as a ligand because of its small cone angle of 103°, the absence of π-delocalized electron density, and its extremely rich chemistry [22][23][24]. These properties allow gold(I)···gold(I) aggregation and the observation of AIE unencumbered by intraligand transitions. There are numerous studies on the self-aggregation of discrete gold(I) complexes of PTA and DAPTA (3,7-diacetyl-1,3,7-triaza-5-fosfabiciclo[3.3.1]nonane) [25,26] by Rodríguez's group; see [27][28][29] as examples. However, the time dimension of the process is seldom studied. We reported in 2021 that complex [Au(N 9 -adeninate)(PTA)], an auration product of the natural purinic nucleobase adenine (see Figure 1, center), experiences a spontaneous supramolecular rearrangement in water solution leading to broad red phosphorescence centered at 700 nm [30]. Aurophilic dimers further stabilized by C-H···π interactions were proposed as a source of phosphorescence, in contrast to the initial blue-fluorescent hydrogen-bonded dimers. The rearrangement process occurs even within the semirigid fibrous matrix of a hydrometallogel. We herein expand our previous study, considering the natural oxypurinic nucleobase hypoxanthine (6-hydroxypurine; see Figure 1, right). Hypoxanthine is an intermediate in adenine biosynthesis and the major product of the C 2 -deamination of guanine. The substitution of the C 6 -amino group of adenine by a C 6 -hydroxy one gives rise to a different balance and pattern of supramolecular hydrogen bond interactions. There are fewer examples of metal complexes of hypoxanthine, or oxypurines in general, than of adenine [31,32]. The spontaneous rearrangement process of complex [Au(N 9 -hypoxanthinate)(PTA)] (1) taking place exclusively in water solution is studied using time-resolved spectroscopical techniques, TD-DFT computational calculations, and IRI electron density topological analysis.

Synthesis and Characterization of Complex [Au(N 9 -Hypoxanthinate)(PTA)] (1)
Complex [Au(N 9 -hypoxanthinate)(PTA)] (1) was prepared in a single step by coordinating a [Au(PTA)] + unit to the hypoxanthinate anion formed in situ by deprotonation of hypoxanthine with the acetylacetonate [(acac) -] ligand of [Au(acac)(PTA)] (see Figure 2). A N 9 coordination mode of the [Au(PTA)] + fragment to hypoxanthinate is proposed for complex 1 by analogy with the X-ray structure of other reported gold(I) complexes of the adeninate anion [18,30,[33][34][35]. N 9 is also the most basic position of purines. Complex 1 is freely soluble in water, where it experiences a spontaneous color change at room temperature associated with a nascent red photoluminescent emission (see below). It is also soluble in methanol, sparingly soluble in ethanol, and barely soluble in organic solvents such as dichloromethane, chloroform, diethylether, and n-hexane.  We herein expand our previous study, considering the natural oxypurinic nucleobase hypoxanthine (6-hydroxypurine; see Figure 1, right). Hypoxanthine is an intermediate in adenine biosynthesis and the major product of the C 2 -deamination of guanine. The substitution of the C 6 -amino group of adenine by a C 6 -hydroxy one gives rise to a different balance and pattern of supramolecular hydrogen bond interactions. There are fewer examples of metal complexes of hypoxanthine, or oxypurines in general, than of adenine [31,32]. The spontaneous rearrangement process of complex [Au(N 9 -hypoxanthinate)(PTA)] (1) taking place exclusively in water solution is studied using time-resolved spectroscopical techniques, TD-DFT computational calculations, and IRI electron density topological analysis.

Synthesis and Characterization of Complex [Au(N 9 -Hypoxanthinate)(PTA)] (1)
Complex [Au(N 9 -hypoxanthinate)(PTA)] (1) was prepared in a single step by coordinating a [Au(PTA)] + unit to the hypoxanthinate anion formed in situ by deprotonation of hypoxanthine with the acetylacetonate [(acac) − ] ligand of [Au(acac)(PTA)] (see Figure 2). A N 9 coordination mode of the [Au(PTA)] + fragment to hypoxanthinate is proposed for complex 1 by analogy with the X-ray structure of other reported gold(I) complexes of the adeninate anion [18,30,[33][34][35]. N 9 is also the most basic position of purines. Complex 1 is freely soluble in water, where it experiences a spontaneous color change at room temperature associated with a nascent red photoluminescent emission (see below). It is also soluble in methanol, sparingly soluble in ethanol, and barely soluble in organic solvents such as dichloromethane, chloroform, diethylether, and n-hexane. In this context, the water-soluble phosphine 1,3,5-triaza-7-phosphaadamantane (PTA; see Figure 1, left) [21] is widely employed as a ligand because of its small cone angle of 103°, the absence of π-delocalized electron density, and its extremely rich chemistry [22][23][24]. These properties allow gold(I)···gold(I) aggregation and the observation of AIE unencumbered by intraligand transitions. There are numerous studies on the self-aggregation of discrete gold(I) complexes of PTA and DAPTA (3,7-diacetyl-1,3,7-triaza-5-fosfabiciclo[3.3.1]nonane) [25,26] by Rodríguez's group; see [27][28][29] as examples. However, the time dimension of the process is seldom studied. We reported in 2021 that complex [Au(N 9 -adeninate)(PTA)], an auration product of the natural purinic nucleobase adenine (see Figure 1, center), experiences a spontaneous supramolecular rearrangement in water solution leading to broad red phosphorescence centered at 700 nm [30]. Aurophilic dimers further stabilized by C-H···π interactions were proposed as a source of phosphorescence, in contrast to the initial blue-fluorescent hydrogen-bonded dimers. The rearrangement process occurs even within the semirigid fibrous matrix of a hydrometallogel. We herein expand our previous study, considering the natural oxypurinic nucleobase hypoxanthine (6-hydroxypurine; see Figure 1, right). Hypoxanthine is an intermediate in adenine biosynthesis and the major product of the C 2 -deamination of guanine. The substitution of the C 6 -amino group of adenine by a C 6 -hydroxy one gives rise to a different balance and pattern of supramolecular hydrogen bond interactions. There are fewer examples of metal complexes of hypoxanthine, or oxypurines in general, than of adenine [31,32]. The spontaneous rearrangement process of complex [Au(N 9 -hypoxanthinate)(PTA)] (1) taking place exclusively in water solution is studied using time-resolved spectroscopical techniques, TD-DFT computational calculations, and IRI electron density topological analysis.

Synthesis and Characterization of Complex [Au(N 9 -Hypoxanthinate)(PTA)] (1)
Complex [Au(N 9 -hypoxanthinate)(PTA)] (1) was prepared in a single step by coordinating a [Au(PTA)] + unit to the hypoxanthinate anion formed in situ by deprotonation of hypoxanthine with the acetylacetonate [(acac) -] ligand of [Au(acac)(PTA)] (see Figure 2). A N 9 coordination mode of the [Au(PTA)] + fragment to hypoxanthinate is proposed for complex 1 by analogy with the X-ray structure of other reported gold(I) complexes of the adeninate anion [18,30,[33][34][35]. N 9 is also the most basic position of purines. Complex 1 is freely soluble in water, where it experiences a spontaneous color change at room temperature associated with a nascent red photoluminescent emission (see below). It is also soluble in methanol, sparingly soluble in ethanol, and barely soluble in organic solvents such as dichloromethane, chloroform, diethylether, and n-hexane.   Figure S3] nucleobase resonances are observed as broadened singlets at 7.99 (C 8 H) and 7.85 (C 2 H) ppm, deshielded with respect to free hypoxanthine [8.21 ppm (C 8 H), 8.19 ppm (C 2 H); see Figure S4]. Also, the 1 H NMR phosphine signals appear as an AB system (lower rim, NCH 2 N) in the interval 4.46-4.56 ppm and as a virtual singlet (upper rim, PCH 2 N) centered at 4.30 ppm. Curiously, no effective 2 J PH coupling is detected, which seems to be characteristic of (PTA)gold(I) complexes [22]. The 31 P{ 1 H} NMR spectrum (ν 0 of 162 MHz, in D 2 O; see Figure S5) is complicated with additional unexpected signals, but a major singlet at −52.63 ppm suggests one type of [Au(PTA)] + coordination mode for the hypoxanthinate ligand. These spectra were recorded at temperatures ranging from 10 • C to 70 • C (see Figures 3 and S6). The low freezing point of D 2 O restricted the temperature interval, precluding collection at lower temperatures. Up to three signals are observed in the 31 P{ 1 H} NMR spectrum at 10 • C. The downfield signal at −39.98 ppm is isochronous with that of the [Au(PTA) 2 ] + cation (−39.77 ppm, see below). In contrast, the upfield one at −55.56 ppm is assigned to coordination products of [Au(PTA)] + to hypoxanthinate different from that of N 9 . The presence of [Au(PTA) 2 ] + can only be explained if the ligand scrambling process of Figure 4 is considered. In this process, two neutral molecules of complex 1 interchange their ligands. This renders the homoleptic ions [Au(PTA) 2 ] + and [Au(N 9 -hypoxanthinate) 2 ] − . Indeed, the high polarity of water favors the partial formation of an ionic species. When the temperature is increased to 70 • C, the signals coalesce, averaging the chemical shifts to −51.21 ppm. This suggests that the different species are related by a thermal equilibrium.
The presence of hypoxanthine as a ligand in complex 1 is suggested by the observed characteristic ν(C=O) (1682 cm −1 ) and ν(C=N) (1621 cm −1 ) stretching vibrations of the nucleobase and the loss of a broad ν(N-H) absorption centered at 2750 cm −1 in the UATR-FTIR spectrum (see Figures S1 and S2). The 1 H NMR [ν0 of 400 MHz, in deuterium oxide (D2O); see Figure S3] nucleobase resonances are observed as broadened singlets at 7.99 (C 8 H) and 7.85 (C 2 H) ppm, deshielded with respect to free hypoxanthine [8.21 ppm (C 8 H), 8.19 ppm (C 2 H); see Figure S4]. Also, the 1 H NMR phosphine signals appear as an AB system (lower rim, NCH2N) in the interval 4.46-4.56 ppm and as a virtual singlet (upper rim, PCH2N) centered at 4.30 ppm. Curiously, no effective 2 JPH coupling is detected, which seems to be characteristic of (PTA)gold(I) complexes [22]. The 31 P{ 1 H} NMR spectrum (ν0 of 162 MHz, in D2O; see Figure S5) is complicated with additional unexpected signals, but a major singlet at −52.63 ppm suggests one type of [Au(PTA)] + coordination mode for the hypoxanthinate ligand. These spectra were recorded at temperatures ranging from 10 °C to 70 °C (see Figures 3 and S6). The low freezing point of D2O restricted the temperature interval, precluding collection at lower temperatures. Up to three signals are observed in the 31 P{ 1 H} NMR spectrum at 10 °C. The downfield signal at −39.98 ppm is isochronous with that of the [Au(PTA)2] + cation (−39.77 ppm, see below). In contrast, the upfield one at −55.56 ppm is assigned to coordination products of [Au(PTA)] + to hypoxanthinate different from that of N 9 . The presence of [Au(PTA)2] + can only be explained if the ligand scrambling process of Figure 4 is considered. In this process, two neutral molecules of complex 1 interchange their ligands. This renders the homoleptic ions [Au(PTA)2] + and [Au(N 9 -hypoxanthinate)2] -. Indeed, the high polarity of water favors the partial formation of an ionic species. When the temperature is increased to 70 °C, the signals coalesce, averaging the chemical shifts to −51.21 ppm. This suggests that the different species are related by a thermal equilibrium.    The freshly prepared water solutions of complex 1 behave as an expected nonelectrolyte. However, its molar conductivity experiences a slight increase from 24.06 to 34.34 cm 2 Ω −1 mol −1 during a 54-h period (see Figure S10).
These experimental findings, for example, the presence of [Au(PTA) 2 ] + in the 31 P{ 1 H} NMR spectrum of complex 1, the detection of high-nuclearity species in the MALDI-MS(+) spectrum, the increase in the molar conductivity, and particularly the spontaneous color change of the solution (see below), reveal that complex 1 is not static in water solution.
These experimental findings, for example, the presence of [Au(PTA)2] + in the 31 P{ 1 H} NMR spectrum of complex 1, the detection of high-nuclearity species in the MALDI-MS(+) spectrum, the increase in the molar conductivity, and particularly the spontaneous color change of the solution (see below), reveal that complex 1 is not static in water solution.
To confirm whether the proposed equilibrium depicted in Figure 4 is indeed an equilibrium and if it can also be reached from the right side of the equation, an equimolecular mixture of complexes 2 and 3 in D 2 O has been characterized by variable-temperature 1 H and 31 P{ 1 H} NMR (see Figure S20 and Figure 6, respectively). It can be observed that the 31  ions are observed in their respective MALDI-MS spectra: {511 Da for [Au(PTA)2] + and 98 Da for (ClO4) -} (see Figures S13 and S14). The presence of bound hypoxanthinate ligands in complex 3 is suggested by the ν(C=O) (1663 cm −1 ) and (C=N) (1610 cm −1 ) stretching vibrations of the nucleobase in the UATR-FTIR spectrum of 3 (see Figure S16).
To confirm whether the proposed equilibrium depicted in Figure 4 is indeed an equilibrium and if it can also be reached from the right side of the equation, an equimolecular mixture of complexes 2 and 3 in D2O has been characterized by variable-temperature 1 H and 31 P{ 1 H} NMR (see Figures S20 and 6, respectively). It can be observed that the 31

Photophysical Studies
As was advanced, the initially colorless water solutions of complex 1 turn to an amber color and emit red light (λex of 365 nm) as a spontaneous result of aging (see Figure 7).

Photophysical Studies
As was advanced, the initially colorless water solutions of complex 1 turn to an amber color and emit red light (λ ex of 365 nm) as a spontaneous result of aging (see Figure 7).
To confirm whether the proposed equilibrium depicted in Figure 4 is indeed an equilibrium and if it can also be reached from the right side of the equation, an equimolecular mixture of complexes 2 and 3 in D2O has been characterized by variable-temperature 1 H and 31 P{ 1 H} NMR (see Figures S20 and 6, respectively). It can be observed that the 31

Photophysical Studies
As was advanced, the initially colorless water solutions of complex 1 turn to an amber color and emit red light (λex of 365 nm) as a spontaneous result of aging (see Figure 7).  The UV-Vis spectra of freshly prepared solutions of complexes 1-3 (50 µM in water solution) are plotted in Figure 8a,b. The spectrum of hypoxanthine is also included for comparative purposes. The spectrum of complex 1 features two distinct asymmetric absorptions at 245 and 200 nm. The energy of the bands is coincident with those of free hypoxanthine, suggesting an intraligand origin for them, with a slight perturbation arising from the [Au(PTA)] + fragment. This is confirmed by computational calculations at the TD-DFT level of theory (see below). The coordination of two hypoxanthinate ligands to a single gold(I) center has a more significant impact on the spectrum. The spectral profile of complex 3 still features two absorptions, but these are broader and less defined. Notably, the band edge is redshifted. The spectrum of complex 2 shows two structured absorptions at 242 and 209 nm, with shoulders at 257 and 214 nm. The aqueous solutions of PTA do not show absorptions at wavelengths longer than 210 nm [36], so a metal-perturbed intraligand origin is proposed.
sorptions at 245 and 200 nm. The energy of the bands is coincident with those of free hypoxanthine, suggesting an intraligand origin for them, with a slight perturbation arising from the [Au(PTA)] + fragment. This is confirmed by computational calculations at the TD-DFT level of theory (see below). The coordination of two hypoxanthinate ligands to a single gold(I) center has a more significant impact on the spectrum. The spectral profile of complex 3 still features two absorptions, but these are broader and less defined. Notably, the band edge is redshifted. The spectrum of complex 2 shows two structured absorptions at 242 and 209 nm, with shoulders at 257 and 214 nm. The aqueous solutions of PTA do not show absorptions at wavelengths longer than 210 nm [36], so a metal-perturbed intraligand origin is proposed.  The photoemission spectrum of complex 1 (2 mM in water solution) has been recorded at different times for 54 h (see Figure 9). Initially, the emission spectrum (λ ex of 320 nm) featured a broad asymmetric band with a maximum of 445 nm and an emission lifetime of 73 ns. As time passed, a new lower-energy band (620 nm) with a longer lifetime of 628 ns as a shoulder of the former appeared. The excitation spectra of both bands overlap (see Figure S21); however, the low-energy band can be better resolved if the excitation wavelength is moved to 340 nm (see Figure 9b). During the 54-h period, an intensity increase in both bands is observed. orded at different times for 54 h (see Figure 9). Initially, the emission spectrum (λex of 320 nm) featured a broad asymmetric band with a maximum of 445 nm and an emission lifetime of 73 ns. As time passed, a new lower-energy band (620 nm) with a longer lifetime of 628 ns as a shoulder of the former appeared. The excitation spectra of both bands overlap (see Figure S21); however, the low-energy band can be better resolved if the excitation wavelength is moved to 340 nm (see Figure 9b). During the 54-h period, an intensity increase in both bands is observed. This time-dependent variation of the photoemission in aqueous solution is ascribed to the formation of fixed-size small aurophilic oligomers (see below), which may be formed from discrete neutral molecules of complex 1 or the ions originated from the partial ligand rearrangement proposed in Figure 4.

Pulsed-Field Gradient Spin Echo NMR Studies
The translational diffusion coefficients (Dt) of the ligands conforming to complex 1 were measured by pulsed-field gradient spin-echo (PGSE) 1 H NMR (ν0 of 400 MHz, 25 mM in D2O) [37,38]. The hydrodynamic radius (rH) and the molecular size of the diffusing species are inversely proportional to Dt, as expressed by the Stokes-Einstein equation. The Dt and derived rH of characteristic 1 H NMR signals of the PTA and hypoxanthinate ligands measured from a freshly prepared sample and after 48 h are compared in Table 1.  Table 1 shows that both ligands have different Dt irrespective of the sample preparation time, suggesting they do not belong to the same molecule. Instead, the ligands may be bound to different gold(I) atoms, diffusing differently. This fact agrees with the formation of the ionic pair due to the redistribution of ligands proposed to occur in an aqueous solution. Although the time variations of Dt and rH are small in both cases, the ones of PTA are almost negligible. Thus, the PTA-containing species, namely [Au(N 9 - This time-dependent variation of the photoemission in aqueous solution is ascribed to the formation of fixed-size small aurophilic oligomers (see below), which may be formed from discrete neutral molecules of complex 1 or the ions originated from the partial ligand rearrangement proposed in Figure 4.

Pulsed-Field Gradient Spin Echo NMR Studies
The translational diffusion coefficients (D t ) of the ligands conforming to complex 1 were measured by pulsed-field gradient spin-echo (PGSE) 1 H NMR (ν 0 of 400 MHz, 25 mM in D 2 O) [37,38]. The hydrodynamic radius (r H ) and the molecular size of the diffusing species are inversely proportional to D t, as expressed by the Stokes-Einstein equation. The D t and derived r H of characteristic 1 H NMR signals of the PTA and hypoxanthinate ligands measured from a freshly prepared sample and after 48 h are compared in Table 1.  Table 1 shows that both ligands have different D t irrespective of the sample preparation time, suggesting they do not belong to the same molecule. Instead, the ligands may be bound to different gold(I) atoms, diffusing differently. This fact agrees with the formation of the ionic pair due to the redistribution of ligands proposed to occur in an aqueous solution. Although the time variations of D t and r H are small in both cases, the ones of PTA are almost negligible. Thus, the PTA-containing species, namely [Au(N 9 -hypoxanthinate)(PTA)] (1) and [Au(PTA) 2 ] + , do not seem to participate in the oligomerization processes.

Computational Studies
The time-dependent optical properties of 1 in aqueous solution are explained with the aid of three computational models. On the one hand, there is the [Au(N 9 -hypoxanthinate)(PTA)] neutral monomer, model 1a. On the other hand, there are the [Au(N 9 -hypoxanthinate)(PTA)] 2 neutral and [Au(PTA) 2 ][Au(N 9 -hypoxanthinate) 2 ] ionic dimers, models 1b and 1c, respectively (see Figure 10). An aurophilically bound dimer for 1b instead of a hydrogen-bonded one was chosen to allow a more reasonable comparison with 1c. The structures of models 1a-1c were optimized at the RI-DFT/PBE0-D3(BJ)/def2-TZVP level of theory with a continuum solvation model representing water. A TD-DFT calculation of the first singlet and triplet vertical excitations was carried out. The energies, oscillator strengths, and orbital contributions of a selection of these excitations are collected in Table 2. An overlay of the experimental UV-Vis spectrum of complex 1 in water with the calculated singlet excitations of models 1a-1c is plotted in Figure 11. The molecular orbitals contributing to the excitations are collected in the Supplementary Materials.

Computational Studies
The time-dependent optical properties of 1 in aqueous solution are explained with the aid of three computational models. On the one hand, there is the [Au(N 9 -hypoxanthinate)(PTA)] neutral monomer, model 1a. On the other hand, there are the [Au(N 9 -hypoxanthinate)(PTA)]2 neutral and [Au(PTA)2][Au(N 9 -hypoxanthinate)2] ionic dimers, models 1b and 1c, respectively (see Figure 10). An aurophilically bound dimer for 1b instead of a hydrogen-bonded one was chosen to allow a more reasonable comparison with 1c. The structures of models 1a-1c were optimized at the RI-DFT/PBE0-D3(BJ)/def2-TZVP level of theory with a continuum solvation model representing water. A TD-DFT calculation of the first singlet and triplet vertical excitations was carried out. The energies, oscillator strengths, and orbital contributions of a selection of these excitations are collected in Table 2. An overlay of the experimental UV-Vis spectrum of complex 1 in water with the calculated singlet excitations of models 1a-1c is plotted in Figure 11. The molecular orbitals contributing to the excitations are collected in the Supplementary Materials.     The TD-DFT calculation on model 1a predicts the two most intense transitions at 247 and 205 nm, corresponding to the two absorption bands observed in the UV-Vis spectrum of complex 1. Thus, the low-energy band is assigned to a mixture of HOMO to LUMO and HOMO-1 to LUMO+1 transitions consisting of charge transfers from the PTA ligand to the gold(I) center. In contrast, the high-energy band is assigned to a HOMO-4 to LUMO metalperturbed intraligand transition within the hypoxanthine ligand, with a second arising from a HOMO-1 to LUMO+4 ligand-to-metal charge transfer.
The TD-DFT predictions for models 1b and 1c are more complicated, showing increased transitions covering the whole UV-Vis spectrum of complex 1. It is noteworthy that the transitions to the first excited singlet state are less energetic for the dimerized models (272 nm, 1b; 288 nm, 1c) than for the monomer one (253 nm). This explains the increased absorbance of the low-energy tail of the UV-Vis spectrum of 1 upon oligomerization. The most intense transition among the lower-energy ones of model 1b consists of a mixture of up to three charge transfers from the PTA-centered HOMO-1, HOMO-2, and HOMO to LUMO, which has a major contribution on the intermetallic axis. The analogous transitions for model 1c, which appear almost degenerate due to the C2 symmetry of the The TD-DFT calculation on model 1a predicts the two most intense transitions at 247 and 205 nm, corresponding to the two absorption bands observed in the UV-Vis spectrum of complex 1. Thus, the low-energy band is assigned to a mixture of HOMO to LUMO and HOMO-1 to LUMO+1 transitions consisting of charge transfers from the PTA ligand to the gold(I) center. In contrast, the high-energy band is assigned to a HOMO-4 to LUMO metal-perturbed intraligand transition within the hypoxanthine ligand, with a second arising from a HOMO-1 to LUMO+4 ligand-to-metal charge transfer.
The TD-DFT predictions for models 1b and 1c are more complicated, showing increased transitions covering the whole UV-Vis spectrum of complex 1. It is noteworthy that the transitions to the first excited singlet state are less energetic for the dimerized models (272 nm, 1b; 288 nm, 1c) than for the monomer one (253 nm). This explains the increased absorbance of the low-energy tail of the UV-Vis spectrum of 1 upon oligomerization. The most intense transition among the lower-energy ones of model 1b consists of a mixture of up to three charge transfers from the PTA-centered HOMO-1, HOMO-2, and HOMO to LUMO, which has a major contribution on the intermetallic axis. The analogous transitions for model 1c, which appear almost degenerate due to the C 2 symmetry of the model, feature a similar origin to that of 1b. However, in this case, the gold(I) atom bound to the PTA ligands has more weight in the transition.
Finally, the optimized electron density was studied using the interaction region indicator (IRI). The RGB-scale mapping of the IRI isosurfaces with the electron density (ρ) weighted by the sign of the second largest eigenvalue of the Hessian (λ 2 ) is a powerful tool to simultaneously reveal and visualize both covalent and non-covalent interactions at a cheap computational cost. The IRI isosurfaces (isovalue of 1.1) of models 1b and 1c are plotted in Figure 12. The sign of λ 2 distinguishes between attractive (blue), van der Waals (green), and repulsive (red) interactions. model, feature a similar origin to that of 1b. However, in this case, the gold(I) atom bound to the PTA ligands has more weight in the transition.
Finally, the optimized electron density was studied using the interaction region indicator (IRI). The RGB-scale mapping of the IRI isosurfaces with the electron density (ρ) weighted by the sign of the second largest eigenvalue of the Hessian (λ2) is a powerful tool to simultaneously reveal and visualize both covalent and non-covalent interactions at a cheap computational cost. The IRI isosurfaces (isovalue of 1.1) of models 1b and 1c are plotted in Figure 12. The sign of λ2 distinguishes between attractive (blue), van der Waals (green), and repulsive (red) interactions. A strong aurophilic interaction as a green-to-blue disk between the gold(I) atoms and several C-H···π interactions as a green surface between the C-H bonds of PTA and the πelectron density of the hypoxanthinate ligands keep the fragments bound in both cases.

General Procedures
All reactions were carried out at room temperature in an open-air atmosphere. Complexes [Au(acac)(PTA)] [35] and [Au(tht)2](ClO4) [39] were prepared as described in the bibliography, whereas (NBu4)[Au(acac)2] was prepared as described in [40] but employing (NBu4)[AuCl2] instead of [N(PPh3)2][AuCl2]. Hypoxanthine and 1,3,5-triazaphosphaadamantane were purchased from Sigma-Aldrich (Madrid, Spain) and employed as received. The Milli-Q water employed in photoluminescence measurements was saturated with nitrogen gas by continuous bubbling for 10 min.  A strong aurophilic interaction as a green-to-blue disk between the gold(I) atoms and several C-H· · · π interactions as a green surface between the C-H bonds of PTA and the π-electron density of the hypoxanthinate ligands keep the fragments bound in both cases.

Conclusions
The efficient dipole solvation properties of water prompt the partial redistribution of the asymmetrically coordinated [Au(N 9 -hypoxanthinate)(PTA)] into the stable charged species [Au(N 9 -hypoxanthinate) 2 ] − and [Au(PTA) 2 ] + . Afterwards, [Au(N 9 -hypoxanthinate) 2 ] − experiences an aggregation process as reflected by its decreasing D t upon aging. [Au(PTA) 2 ] + remains undisturbed in solution, equilibrating the negative electric charge of the supposed [Au n (hypoxanthinate) 2n ] n− aggregate and stabilizing it by C-H· · · π interactions. The latter species is proposed to be the source of phosphorescence. However, an unequivocal assignment of the optical properties cannot be provided without conclusive structural data for complex 1 and the emissive products. In contrast to regular AIE, no poor solvent addition is needed to force gold(I)· · · gold(I) clustering and achieve photoemission. A tentative scheme of the processes occurring after dissolving complex 1 in water is given in Figure 13.

Conclusions
The efficient dipole solvation properties of water prompt the partial redistribution of the asymmetrically coordinated [Au(N 9 -hypoxanthinate)(PTA)] into the stable charged species [Au(N 9 -hypoxanthinate)2] -and [Au(PTA)2] + . Afterwards, [Au(N 9hypoxanthinate)2] -experiences an aggregation process as reflected by its decreasing Dt upon aging. [Au(PTA)2] + remains undisturbed in solution, equilibrating the negative electric charge of the supposed [Aun(hypoxanthinate)2n] n-aggregate and stabilizing it by C-H···π interactions. The latter species is proposed to be the source of phosphorescence. However, an unequivocal assignment of the optical properties cannot be provided without conclusive structural data for complex 1 and the emissive products. In contrast to regular AIE, no poor solvent addition is needed to force gold(I)···gold(I) clustering and achieve photoemission. A tentative scheme of the processes occurring after dissolving complex 1 in water is given in Figure 13. Supplementary Materials: The following supporting information can be downloaded at: www.mdpi.com/xxx/s1, Figure S1: UATR-FTIR spectrum of complex 1. Figure S2: UATR-FTIR spectrum of hypoxanthine. Figure S3: 1 H NMR (400 MHz, D2O) spectrum of complex 1. Figure S4: 1 H NMR (400 MHz, D2O) spectrum of hypoxanthine. Figure S5: 31 P{ 1 H} NMR (162 MHz, D2O) spectrum of complex 1. Figure S6: Collection of 1 H NMR spectra of complex 1 in D2O (25 mM) at different temperatures. Figure S7: MALDI-MS(+) spectrum of complex 1. Figure S8: MALDI-MS(+) spectrum of an aged solution of complex 1. Figure S9: MALDI-MS(-) spectrum of complex 1. Figure  S10: Molar conductivity in aqueous solution of complex 1 at different times. Figure S11 Figure S13: MALDI-MS(+) spectrum of complex 2. Figure S14: MALDI-MS(-) spectrum of complex 2. Figure S15: UATR-FTIR spectrum of complex 2. Figure S16: UATR-FTIR spectrum of complex 3. Figure S17: MALDI-MS(+) spectrum of complex 3. Figure S18: MALDI-MS(-) spectrum of complex 3. Figure S19: 1 H NMR (400 MHz, D2O) spectrum of complex 3. Figure S20: 1 H NMR (400 MHz, D2O) spectrum of the equimolecular mixture of complexes 2 and 3. Figure S21: Excitation spectra of complex 1 in aqueous solution with an emission wavelength of 450 nm (black line) or 650 nm (blue line). Figure S22: Molecular orbitals involved in the S1 transition of model 1a. Figure S23: Molecular orbitals involved in the S5 transition of model 1a. Figure S24: Molecular orbitals involved in the S20 transition of model 1a. Figure S25: Molecular orbitals involved in the T1 transition of model 1a. Figure S26: Molecular orbitals involved in the S1 transition of model 1b. Figure S27: Molecular orbitals involved in the S3 transition of model 1b. Figure S28: Molecular orbitals involved in the S12 transition of model 1b. Figure S29: Molecular orbitals involved in the S25 transition of model 1b. Figure S30: Molecular orbitals involved in the T1 transition of model 1b. Figure S31: Molecular orbitals involved in the S1 transition of model 1c. Figure S32: Molecular orbitals involved in the S4 transition of model 1c. Figure S33: Molecular orbitals involved in the S5 transition of model 1c. Figure  S34: Molecular orbitals involved in the S14 transition of model 1c. Figure S35: Molecular orbitals involved in the T1 transition of model 1c. Table S1. Cartesian coordinates of the optimized structure Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/molecules28155680/s1, Figure S1: UATR-FTIR spectrum of complex 1. Figure S2: UATR-FTIR spectrum of hypoxanthine. Figure S3 Figure S13: MALDI-MS(+) spectrum of complex 2. Figure S14: MALDI-MS(-) spectrum of complex 2. Figure S15: UATR-FTIR spectrum of complex 2. Figure S16: UATR-FTIR spectrum of complex 3. Figure S17: MALDI-MS(+) spectrum of complex 3. Figure S18: MALDI-MS(-) spectrum of complex 3. Figure S19: 1 H NMR (400 MHz, D 2 O) spectrum of complex 3. Figure S20: 1 H NMR (400 MHz, D 2 O) spectrum of the equimolecular mixture of complexes 2 and 3. Figure S21: Excitation spectra of complex 1 in aqueous solution with an emission wavelength of 450 nm (black line) or 650 nm (blue line). Figure S22: Molecular orbitals involved in the S 1 transition of model 1a. Figure S23: Molecular orbitals involved in the S 5 transition of model 1a. Figure S24: Molecular orbitals involved in the S 20 transition of model 1a. Figure S25: Molecular orbitals involved in the T 1 transition of model 1a. Figure S26: Molecular orbitals involved in the S 1 transition of model 1b. Figure S27: Molecular orbitals involved in the S 3 transition of model 1b. Figure S28: Molecular orbitals involved in the S 12 transition of model 1b. Figure S29: Molecular orbitals involved in the S 25 transition of model 1b. Figure S30: Molecular orbitals involved in the T 1 transition of model 1b. Figure S31: Molecular orbitals involved in the S 1 transition of model 1c. Figure S32: Molecular orbitals involved in the S 4 transition of model 1c. Figure S33: Molecular orbitals involved in the S 5 transition of model 1c. Figure S34: Molecular orbitals involved in the S 14 transition of model 1c. Figure S35: Molecular orbitals involved in the T 1 transition of model 1c. Table S1. Cartesian coordinates of the optimized structure [RI-DFT/PBE0-D3(BJ)/COSMO(ε = 80.1)/def2-TZVP/def2-ECP(Au)] of model 1a. Table S2