DFT Studies of Dimethylaminophenyl-substituted Phthalocyanine and of Its Silver Complexes

1. Introduction

Phthalocyanine (C₈H₄N₂)₄H₂ (PcH₂) contains four isoindole units connected by nitrogen bridges. The so-obtained ring system has 18 delocalized π electrons responsible for intense absorption (the Soret band around 400 nm and the Q-band in the red/near infrared region). Therefore, phthalocyanines are used as dyes and pigments 1. Metal complexes derived from Pc^2- have applications in catalysis, solar cells, and photodynamic therapy [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17].

Very high absorption in the visible range and high chemical resistance implied the synthesis of complexes of various substituted phthalocyanines with central Al, Mg, and Zn atoms [18,19,20,21,22] to be used as photoinitiators for free-radical (FRP) and/or cationic (CP) polymerizations. Recently, Breloy et al. [23] have synthesized a complex of dimethylaminophenyl-substituted phthalocyaninewith silver in an unusual oxidation state +II, [dmaphPcAg] (see Figure 1). The species obtained by photoinduced electron transfer reactions of excited [dmaphPcAg]* in the presence and absence of bis(4-methylphenyl)iodonium hexafluorophosphate (Iod) initiated CP and FRP, respectively, both in laminate; as well as under air. Moreover, a polymer/silver nanocomposite with a homogenous narrow size distribution of spherical silver nanoparticles has been formed.

Our recent study will deal with the results related to [dmaphPcAg] and its reaction intermediates in more detail [23]. Its UV-vis absorption spectrum in CHCl₃ shows two intense absorption peaks at ca. 300 - 360 nm (Soret band) and at 700 nm (Q-band) (see Figure 2). According to TD-B3LYP/SMD (see below) calculations of its neutral molecule in doublet ground spin state), ²[dmaphPcAg]⁰, the Soret band corresponds most probably to π – π* ligand-metal electron transitions at 359 nm (oscillator strength f = 0.76). The Q-band consists of π – π* electron transitions at 723 nm (f = 0.9) and at 724 nm (f = 0.79). Analogously, in the neutral dmaphPcH₂ molecule in singlet ground state (Figure 3) the Soret band at ca 340 nm could be ascribed to π – π* electron transitions at 376 nm (f = 0.19) and at 404 nm (f = 0.23) and the Q-band at ca 740 nm is explained by π – π* electron transitions at 736 nm (f = 0.82) and at 713 nm (f = 0.22).

Under exposure to LED @ 385 nm of [dmaphPcAg] in CHCl₃ solutions with and without Iod, a sharp decrease in absorbance at 359 and 723 nm is observed simultaneously with splitting the Q-band and a rapid increase in a band at 450 nm indicating the formation of silver nanoparticles (see Figure 4) [23]. Well-dispersed spherical Ag nanoparticles in the solution (an average diameter of 9 nm) were confirmed by transmission electron microscopy. Fluorescence quenching increases with Iod concentration in a chloroform solution, which suggests an electron transfer reaction from the [dmaphPcAg] excited state to Iod.

A very weak EPR signal of in situ irradiated [dmaphPcAg] in CHCl₃ using LED@385 nm was detected at g = 2.0021 and its intensity was enhanced in the presence of Iod [23]. It can be assigned to an unstable π-radical cation which is probably transformed into a N-centered radical that was detected as an adduct with N-benzylidene-tert-butylamine N-oxide (PBN) or 2,2-dimethyl-3,4-dihydro-2H-pyrrole 1-oxide (DMPO). The carbon- and nitrogen-centered radicals were detected in similar EPR experiments with PBN and DMPO spin-traps in deoxygenated benzene under argon as well.

LED@385 nm irradiation of [dmaphPcAg] in CHCl₃ with Iod under argon formed a π-radical cation centered on [dmaphPcAg]. Stable (4-methyl)phenyl radical adducts with PBN, DMPO and nitrosodurene spin traps were detected in similar experiments in deoxygenated benzene under argon as well.

The Iod addition to [dmaphPcAg] photoinitiating systems caused higher conversions of epoxy and acrylate monomers [23]. Similar results can be obtained by cationic photopolymerization using the dmaphPcH₂/Iod system and dmaphPcH₂ alone. Therefore, the presence of Ag is not crucial for photopolymerization. Because radical species are strongly inhibited by oxygen, FRP does not proceed well in air. Unlike acrylate conversions, epoxy conversions are not affected by atmosphere conditions. Another interesting attribute of the above [dmaphPcAg] photoinitiating systems is the electron transfer reactions where the central Ag(II) atom is reduced to homogenously dispersed silver nanoparticles in the polymer matrix.

The following [dmaphPcAg] photoinitiation mechanism was proposed [23]:

The photoexcited [dmaphPcAg]* is formed by irradiation in the first step

[dmaphPcAg] + hυ → [dmaphPcAg]*

(1)

The reaction between photoexcited and ground-state species leads to the reduction of

Ag(II) to Ag(I) and generates the nitrogen-centered cation-radical in the second step

[dmaphPcAg]* + [dmaphPcAg] → [dmaphPcAg]^- + [dmaphPcAg]^+•

(2)

In the third step, silver nanoparticles and aromatic carbon-centered radicals are formed.

[dmaphPcAg]^- → Ag⁰ + [dmaphPc]^-•

(3)

Subsequently, a hydrogen-transfer reaction between [dmaphPcAg]^+• and [dmaphPcAg] leads to dehydrogenated aromatic-derived aminoalkyl radicals [dmaphPcAg - H]^• and Brønsted photoacids H⁺ according to (4)

[dmaphPcAg]^+• + [dmaphPcAg] → [dmaphPcAg] + [dmaphPcAg - H]^• + H⁺

(4)

The Brønsted acids subsequently initiate ring-opening reactions in epoxides. [dmaphPc]^-• and [dmaphPcAg - H]^• initiate the FRP of acrylates.

The addition to [dmaphPcAg] is the source of additional CP reactions. Upon irradiation, the exciplex [dmaphPcAg…(MePh)₂I⁺]* is formed

[dmaphPcAg] + (MePh)₂I⁺ + hυ → [dmaphPcAg…(MePh)₂I⁺]*

(5)

Subsequently, it undergoes electron transfer (6), where the [dmaphPcAg] ^+• cation-radical and a diaryliodo radical are formed

[dmaphPcAg…(MePh)₂I⁺]* → [dmaphPcAg] ^+• + (MePh)₂I^•

(6)

Similar electron transfer reactions as above within [dmaphPcAg]* cause the formation of silver nanoparticles, [dmaphPc]^-•, [dmaphPcAg-H]^•, and photoacids - see Eqs. (2) – (4). The intermediate diaryliodo radical is unstable and decomposes as follows

(MePh)₂I^• → MePh^• + MePhI

(7)

So formed phenyl radicals can abstract hydrogen and produce the derived aromatic aminoalkyl radical [dmaphPcAg-H] ^•

[dmaphPcAg] + MePh^• → [dmaphPcAg-H] ^• + MePhH

(8)

The generated [dmaphPc]^-• and [dmaphPcAg-H]^• radicals are able to initiate the FRP of acrylates.

Nevertheless, the above reaction mechanism (1) – (8) meets several problems. Whereas [dmaphPcAg], [dmaphPcAg]* and [dmaphPcAg…(MePh)₂I⁺]* have an odd number of electrons and their doublet ground spin state corresponds to natural radicals, [dmaphPcAg]^-, [dmaphPcAg]⁺, and the monodehydrogenated species [dmaphPcAg - H] have an even number of electrons and so they can form biradicals. These correspond to singlet or triplet spin states, but the singlet biradicals are not detectable by EPR measurements. Consequently, quantum-chemical calculations are necessary to describe their spin densities. In [23] quantum-chemical calculations of neutral [dmaphPcAg] in doublet spin state and neutral [dmaphPcH₂] in singlet spin state were performed only. In our more recent (TD-)B3LYP study [24] optimal geometries, electronic structure and electron transitions of ^m[dmaphPcAg]^q species in vacuum with charges q = +1 to -2 in the two lowest spin states m were investigated from the point of view of the Jahn-Teller effect. However, this study was not related to spin distribution in the ^m[dmaphPcAg]^q species.

The aim of our recent manuscript is to complete the previous studies by DFT calculations of ^m[dmaphPcAg]^q species in CHCl₃ with various charges q corresponding to silver oxidation states between +III and 0 and in two lowest spin states m. We want to describe their electron- and spin-density distributions. To complete the picture, we will also study the dehydrogenated/deprotonated ¹[dmaphPcH₂]⁰ species in the lowest spin states. We hope that the obtained results on electronic structure, energetics and electron transitions will contribute to verification of the above mentioned reaction mechanism (1) – (3) [23].

2. Results

Geometry optimization of ^m[dmaphPcAg]^q complexes in CHCl₃ in two lowest spin states without any symmetry restriction (see below) started from their optimized structures in vacuum obtained in [24]. After suitable modifications, these structures were used as starting ones for analogous geometry optimizations of dehydrogenated/deprotonated species [dmaphPcH₂] in the lowest two spin states as well. Although the singlet spin states of the systems under study were treated using an unrestricted formalism (the ‘broken symmetry’ treatment [25]), no spin-polarized solutions were obtained, i.e., their energies are identical as in the case of restricted DFT calculations. Gibbs free energies and relevant geometry parameters of the stable structures are presented in Table 1, Table 2 and Table 3.

2.1. Energetics

According to the Gibbs free energy data at room temperature (Table 1), the energies of the ^m[dmaphPcAg]^q complexes decrease upon reduction. However, even the structures of the [dmaphPcAg]^2- complexes corresponding to the formal oxidation state Ag(0) seem to be stable. Except [dmaphPcAg]⁺, the complexes in lower spin states are more stable. It implies that the ³[dmaphPcAg]⁺ biradical can be present in non-vanishing concentrations in the reaction system. Consequently, reactions (2) and (6) are correct. Despite the relative concentrations of the deexcitation products of the excited [dmaphPcAg]* and [dmaphPcAg…(MePh)₂I⁺]* species not necessarily satisfy the Boltzmann distribution law, our results indicate that only ¹[dmaphPcAg]^- and ³[dmaphPcAg]^- species can co-exist in comparable concentrations. In equilibria, the remaining ^m[dmaphPcAg]^q complexes with the same charges are present only in the more stable form because of the too large energy difference between their spin states. Therefore, if the LED@385 nm irradiation (corresponding to 310.7 kJ/mol energy) is fully absorbed by ²[dmaphPcAg]⁰ excitation, the reaction (2) is shifted to the right in agreement with [23] (the reaction Gibbs energy of -26.5 kJ/mol at 298 K).

According to the data in Table 1, deprotonated [dmaphPcH_n]^q species, n = 2 or 1, seem to be more stable than their dehydrogenated counterparts (i.e., with q preserved). However, their relative stability is also dependent on the reactions of their formation. The reaction Gibbs energy of ²Ag⁰ formation (with atomic Gibbs energy of -146.98864 Hartree) according to reaction (3) is highly positive (+235.0 kJ/mol for ³[dmaphPcAg]^- and +240.7 kJ/mol for ¹[dmaphPcAg]^-) and its equilibrium is shifted right due to subsequent formation and precipitation of Ag nanoparticles. Due to the lack of necessary data, we will not deal with the remaining (4) – (8) reaction equilibria.

2.2. Geometries

The DFT optimized geometries of the ^m[dmaphPcAg]^q complexes are very similar to the ²[dmaphPcAg]⁰ structure presented in Figure 1. Except for dimethylphenyl groups, all ^m[dmaphPcAg]^q structures are planar, only the central Ag atom might be slightly above the plane of four pyridine nitrogen atoms N_py (up to 0.019 Å in ³[dmaphPcAg]⁺, see Table 2). The values of the lengths of the Ag – N_py bonds, as well as of the N_py – Ag – N_py angles indicate that the C₄ symmetry axis is preserved in all the silver complexes except ⁴[dmaphPcAg]⁰, ¹[dmaphPcAg]^-, and ³[dmaphPcAg]^-. The Ag – N_py bond lengths increase with complex reduction up to ¹[dmaphPcAg]^- only. Therefore, the electron density transfer to Ag is not related to its out-of-plane movement.

The DFT optimized structures of ^m[dmaphPcH_n]^q, n = 2 → 0, are presented in Figure 3, Figure 5, Figure 6 and Figures S1 and S2 and in Table 3. Except for ¹[dmaphPcH]^-, their phthalocyanine cores are planar. Their H-N_py bond lengths increase during deprotonation/dehydrogenation of the central ring. Due to H-N_py bonds, the C₄ symmetry axis can be observed only in ²[dmaphPc]^- and ¹[dmaphPc]^2-.

2.3. Electronic Structure

The main features of the ^m[dmaphPcAg]^q complexes are presented in Figure 7, Figure 8, Figure 9 and Figures S3–S5 and Table 2. Their Ag – N_py bond orders and positive Ag charges decrease upon reduction up to [dmaphPcAg]^-. Except for complexes in singlet spin states, the d-electron populations of silver atoms are practically constant. The atomic charges of the pyrrole nitrogens N_py are more negative than that of the bridging nitrogens N_br, and the amine nitrogens N_amin have even less negative charges. During reduction of the complex, all N charges become even more negative. Positive charges of carbon atoms at the 4- and 7-positions of the isoindole units denoted as C_α decrease with reduction of the complex. Similarly, small negative charges of carbon atoms at isoindoles 5- and 6-positions denoted as C_β increase with reduction of the complex. Significantly more negative charges of the C_met methyl carbons do not depend on the charge q and spin state m of the ^m[dmaphPcAg]^q complexes. Only N_amin and C_met charges are not affected by the lower symmetry of the complexes studied.

The highest spin density at Ag atoms decreases only slightly with reduction. N_py atoms have ca two - three times lower spin density of the same sign, which rises with reduction. In both cases, the spin density increases with spin multiplicity. Except for ⁴[dmaphPcAg]^2-, the spin density of opposite sign at N_br atoms is about one order lower and of variable signs. The spin density at C_α atoms is relevant only in anionic complexes and in higher spin states. C_β atoms have even lower spin density. The spin density at N_amin and C_met atoms is vanishing. Only the N_br, N_amin and C_met charges are not affected by the lower symmetry of the complexes under study. Except for ³[dmaphPcAg]⁺, only vanishing spin density can be found in dimethylaminophenyl groups.

The main features of the ^m[dmaphPcH_n]^q species are presented in Figures S6 and S7, and Table 3. The H – N_py bond orders decrease with dehydrogenation/deprotonation. The same trend exhibits the most negative N_py charges, whereas the N_br ones exhibit a reverse trend. The even less negative N_amin charges increase with the negative charge of the whole species. The positive Cα charges decrease with the total charge of the species and increase with deprotonation/dehydrogenation. Only small changes in the very small C_β charges can be observed. Negative C_met charges are constant in all ^m[dmaphPcH_n]^q species under study and equal to those of ^m[dmaphPcAg]^q.

We have only two ²[dmaphPcH_n]^q species with non-zero spin. The small spin densities at the N_py and N_br atoms are of the same sign, unlike the higher ones at C_α atoms. The vanishing spin density is at C_β and N_amin atoms. No spin density at C_met atoms can be observed.

2.4. Electron Transitions

In this section we will compare the TD-DFT calculated electron transitions of ^m[dmaphPcAg]^q complexes in CHCl₃ (Figure 10, Figure 11, Figure 12, Figure 13 and Figures S8–S11) with UV-vis spectra of in [dmaphPcAg]⁰ in CHCl₃ before (Figure 2) and during (Figure 4) photolysis.

Very intense TD-DFT electron transitions of ²[dmaphPcAg]⁰ at 723 and 724 nm (Figure 10) can be assigned rather to the 730 nm shoulder than to the 700 nm peak in experimental spectra (Figure 2). Similarly, the intense electron transitions at 359 nm correspond more probably to the experimental peak at 360 nm than at 300 nm. It implies the existence of several species in the solution.

Intense electron transitions of ³[dmaphPcAg]⁺ calculated at 2390 nm, as well as less intense ones at 915 and 856 nm (Figure 11) are not observed in the experimental spectra of the reaction system in Figure 4. This excludes its presence at measurable concentrations. The same holds for ¹[dmaphPcAg]⁺ due to the very intense TD-DFT electron transition at 900 nm (Figure S8), which is absent in the experimental spectra (Figure 4). Analogously, the calculated relevant electron transitions of ⁴[dmaphPcAg]⁰ at 1276 and 1210 nm (Figure S9) missing in experimental spectra are in contradiction with the presence of this species in the reaction system (Figure 4).

On the other hand, the experimental peak at 680 nm can be assigned to an intense electron transition at 687 nm and the 660 nm peak to an intense 668 nm transition whereas the 593 nm electron transition might correspond to its shoulder. This supports the presence of ¹[dmaphPcAg]^- in the reaction system (Figure 4 and Figure 12). The presence of ³[dmaphPcAg]^- in the reaction system cannot be excluded because the experimental peak at 730 nm (Figure 4) can be explained as the very intense TD-DFT electron transition at 724 nm and the very intense electron transition at 619 nm (Figure S10) can be superimposed by the scarcely visible shoulder of the very intense peak at 700 nm.

The absence of ²[dmaphPcAg]^2- in the reaction system is due to the missing experimental peaks corresponding to the relevant electron transitions at 898 nm and 884 nm (Figure 4 and Figure 13). Analogously, the absence of experimental peaks corresponding to the TD-DFT electron transitions of ⁴[dmaphPcAg]^2- at 1153 nm and 812 nm (Figure 4 and Figure S11) is the reason for the exclusion of this species from the reaction system.

Very intense TD-DFT electron transitions of ¹[dmaphPcH₂]⁰ at 736 and 749 nm (Figure 14) can be assigned to the 750 nm peak in the experimental spectra (Figure 2). Similarly, the experimental peak at 340 nm can be explained by the calculated electron transitions at 358 nm and 364 nm. Therefore, the agreement between the experimental and calculated spectral parameters is worse than in the case of ²[dmaphPcAg]⁰.

Very intense TD-DFT electron transitions of ¹[dmaphPcH]^- at 675 nm and 671 nm (Figure 15) can be assigned to the experimental peak at 680 nm (Figure 4). The experimental peak at 380 nm can be explained by the very intense calculated electron transitions at 386 and 384 nm. Its neutral analogue ²[dmaphPcH]⁰ can be excluded from the reaction system due to missing experimental peaks corresponding to TD-DFT electron transitions at 1183 nm, 1154 nm, 832 nm, and 788 nm (Figure 4 and Figure S12).

The very intense calculated electron transitions of ¹[dmaphPc]^2- at 660 nm and 658 nm (Figure 16) can be assigned to the 660 nm or 680 nm shoulder (but not to both) of the 730 nm peak shoulder in the reaction spectra (Figure 4). TD-DFT electron transitions of ²[dmaphPc]^- at 778 nm and 777 nm (Figure S13) can also be superimposed by the intense 730 nm peak shoulder of the reaction system Figure 4) and this peak can be assigned to very intense 722 nm and 721 nm electron transitions, too.

We can conclude that our TD-DFT interpretation of the experimental UV-vis spectra indicates that the ²[dmaphPcAg]⁰, ¹[dmaphPcAg]^-, ³[dmaphPcAg]^-, ¹[dmaphPcH₂]⁰, ¹[dmaphPcH]^-_, ²[dmaphPc]^-, and ¹[dmaphPc]^2- species can be present in the reaction system under study.

3. Discussion

Two- or even three-electron reduction of the ³[dmaphPc]⁺ species causes only small changes in the electron and spin population at the central Ag atom, and the same holds for the Ag-N_py bond. The added electrons are prevailingly distributed on the dmaphPc^2- ligand only which is typical for ‘non-innocent’ ligands. This reduction does not cause a shift of Ag from the phthalocyanine plane. However, neutral Ag nanoparticles should be formed within the endothermic reaction (3) which is connected with ca. 0.8 electron addition to Ag simultaneously with its move from the ligand plane. The planarity of the phthalocyanine core of ²[dmaphPc]^- is preserved. These changes should proceed in several reaction steps including the aggregation of the [dmaphPcAg]^- species in the solution. However, this does not coincide with its negative charge.

The results of our ‘broken symmetry’ DFT calculations indicate the nonexistence of singlet biradicals. According to the energy data, the ³[dmaphPcAg]⁺ concentration should dominate over their singlet counterparts, but the UV-vis spectra [23] do not confirm its presence in reaction systems. Based on EPR measurements using spin traps [23], this can be explained by the very low stability of ³[dmaphPcAg]⁺. Moreover, only this species has non-vanishing spin density at N_amin atoms (see Figure 7) as supposed in [23]. The calculated electron transitions of the EPR silent ¹[dmaphPcAg]⁺ species do not agree with UV-vis spectra [23] and so it is not present in reaction solutions.

The presence of ²[dmaphPcAg]⁰ in reaction solutions is in agreement with the DFT and TD-DFT calculations as well as with the EPR and UV-vis measurements [23]. However, the calculated electron transitions cannot explain all the spectral peaks in Figure 2. This indicates the presence of other species in the CHCl₃ solution of ²[dmaphPcAg]⁰. The comparison of UV-vis spectra with the TD-DFT electron transition as well as the DFT energy data indicate the absence of ⁴[dmaphPcAg]⁰ in reaction solutions.

The coexistence of ¹[dmaphPcAg]^- and ³[dmaphPcAg]^- in reaction solutions is allowed by their energies and by the agreement of the calculated electron transitions with UV-vis spectra [23]. ¹[dmaphPcAg]^- is EPR silent.

Energy data indicate that the concentration of ²[dmaphPcAg]^2- should dominate over ⁴[dmaphPcAg]^2- but the calculated electron transitions of both species do not agree with UV-vis spectra [23], implying their absence in reaction solutions.

Based on TD-DFT electron transitions and UV-vis spectra [23], the EPR silent ¹[dmaphPcH₂]⁰ species and its deprotonated products ¹[dmaphPc]^- and ¹[dmaphPc]^2- can be present in reaction solutions. The same holds for the dehydrogenated product ²[dmaphPc]^- but not for ²[dmaphPcH]⁰.

4. Methods

Standard B3LYP [26] geometry optimization with Grimme’s GD3 dispersion correction [27] of ^m[dmaphPcAg]^q, q = -2 to +1, in two lowest spin states m, and ^m[dmaphPcH_n]^q , n = 2 to 0, q = 0, 1, or 2, in the lowest spin states m, in CHCl₃ solutions using the cc-pVDZ-PP pseudopotential and basis set for Ag [28] and cc-pVDZ basis sets for the remaining atoms [29] was performed. All calculations were carried out using an unrestricted formalism within ‘broken symmetry’ treatment [25]. Solvent effects were approximated by the SMD (Solvation Model based on solute electron Density) modification [30] of the integral equation formalism polarizable continuum model. The optimized structures were tested on the absence of imaginary vibrations by vibrational analysis. The excited state energies and the intensities of the corresponding electron transitions were evaluated using the time-dependent DFT method (TD-DFT) [31] for 90 - 120 states. The electronic structure was evaluated in terms of Natural Bond Orbital (NBO) population analysis [32]. All calculations were performed using the Gaussian16 [33] program package.

5. Conclusions

This study aims to supplement the quantum-chemical studies related to the [dmaphPcAg] photoinitiator of polymerization reactions in [23]. We tested the initial steps (1) – (3) of FRP proposed in [23] by (TD-)DFT interpretation of the EPR and UV-vis measurements by testing the presence of possible reaction intermediates ^m[dmaphPcAg]^q, q = 1 → -2, ¹[dmaphPcH₂]⁰, and its deprotonation/dehydrogenation products.

Our results suggest the presence of crucial unstable ³[dmaphPcAg]⁺ species which was deduced by EPR spin trap experiments [23], but its identification was complicated due to the presence of other similar compounds in reaction solutions.

In addition to ²[dmaphPcAg]⁰, our results suggest the coexistence of both ¹[dmaphPcAg]^- (EPR silent) and ³[dmaphPcAg]^- species in reaction solutions as well. Therefore, the supposed intermediates of the reactions (1) – (3) can be identified. The presence of the EPR silent species ¹[dmaphPcH₂]⁰, ¹[dmaphPc]^- and ¹[dmaphPc]^2- as well as of the radical ²[dmaphPc]^- in reaction solutions cannot be excluded.

However, the formation of Ag nanoparticles by reaction (3) may be a weak point of the proposed reaction mechanism from the energetic, stereochemistry, and electronic structure points of view. Unfortunately, we do not have sufficient information to propose any alternative reaction mechanisms. Further experimental and theoretical studies are desirable in this field.