Nanocrystals perovskites photocatalyzed singlet oxygen generation for light-driven organic reactions

Lead halide perovskite nanocrystals were prepared and used as photocatalysts for the in situ 1O2 generation to perform hetero Diels–Alder, ene and oxidation reactions with suitable dienes and alkenes. The methodology has been reasonably standardized and made applicable to a variety of olefinic substrates. The scope of the method is finely illustrated by the results in all the tested reactions, which allowed to obtain desymmetrized hydroxy-ketone derivatives, unsaturated ketones and epoxides. Some limitations were also observed especially in the case of the alkene oxidations as well as poor chemoselectivity was somewhere observed. 1O2 generated by lead halide perovskite nanocrystals as photocatalyst in organic reactions.


Introduction
Light-driven organic syntheses take an enormous advantage from the development of cheap, effective, and robust, made-easy and long-term stable photocatalysts. The use of external photocatalysts with light and redox properties is generally required for activating most of the organic molecules considering their poor light-harvesting ability in the visible light region. Transition metal complexes [1,2] and organic dyes [3] are commonly selected to achieve electron transfer (ET) or energy transfer to perform organic reactions. Nevertheless, these photocatalysts suffer from common disadvantages such as high cost, easy decomposition/degradation during the reactions, and poor recyclability. Inorganic semiconductor photocatalysts, such as TiO 2 [4], cadmium sulfide, and other transition metal sulfides and oxides [5] can be used as heterogeneous catalysts but the wide band gap (e.g. for TiO 2 ~ 3.2 eV) makes them only absorb UV or near UV light of the solar spectrum, with low photocatalytic efficiency and quantum yield. In the past 10 years, metal halide perovskites (MHPs) have emerged as one of the leading optoelectronic materials with attracting properties of long carrier lifetime, good band-gap tunability, excellent absorption coefficient, strong nonlinear response, unique carrier bipolar diffusion characteristics, high electron/hole mobility, high tolerance to defects, and facile processing methods [6].
Perovskite-based photocatalysts have fortified their applications in the field of photocatalytic organic reactions and found applications in alcohols and alkene oxidation reactions to aldehydes/ketones, aliphatic and aromatic C-H bond activation to afford alcohols or aldehydes/ ketones, aminomethylation of imidazo-fused heterocycles, α-alkylation of aldehyde, carbon-carbon coupling reactions, decarboxylation and dehydrogenation reactions and few others [6].
The ability of MHPs to drive oxygen to be inserted in different ways into an organic molecule, often characterized by a simple carbon skeleton, opens different applicative opportunities and the generation of Singlet Oxygen ( 1 O 2 ) is an interesting way to afford valuable synthons to be used in remarkable approaches to organic molecules. 1 O 2 is photochemically produced by the use of organic molecules (sensitizers) [7] or, alternatively, through classical hydrogen peroxide decomposition promoted by hypochlorite or hypobromite and ozonides decomposition. Moreover, 1 O 2 can also be obtained by thermal decomposition of unstable molecules, such as arene endoperoxides, or by photolysis of oxone [8]. These methodologies are not devoid of problems due essentially to the instability of the organic sensitizers or not completely applicable to the experimental conditions required by specific syntheses.
For these reasons modern chemical literature on 1 O 2 reports the photochemical properties of new catalytic systems that proved to be nicely applicable to the generation of such transient oxygen species [9][10][11][12]. 1 O 2 can undergo a variety of reactions; among them the hetero Diels-Alder (HDA) [4 + 2] cycloadditions, the [2 + 2] cycloadditions, the ene reactions and epoxidations are representative of the most relevant categories of methods for the preparation of valuable synthons [13][14][15][16].
Recently, we proposed a 1 O 2 generation made-easy method by the catalytic use of oxidized graphitic carbon nitride (g-C 3 N 4 ) under photochemical conditions. g-C 3 N 4 , a polymeric semiconductor consisting of carbon and nitrogen atoms, is prepared by bulk and hard template pyrolysis of melamine (MLM), or alternatively from other carbon/nitrogen-containing sources [17,18]. Some HDA and ene reactions were conducted on selected dienes and alkenes showing a general ability of oxidized g-C 3 N 4 to promote chemoselective and unselective oxidative processes in strong dependence from substrates nature. The results offered a good panorama of the ability of this type of catalysts to be employed in organic reactions to prepare valuable synthons that can be used in several value-added preparations [19].
Furthermore, triggered by the superior optical properties of MHPs discussed above, we devised novel composite photocatalysts with high visible-light absorption, namely the g-C 3 N 4 /MHPs systems, which were used for the first time as photocatalysts for the in situ 1 O 2 generation to perform HDA, ene and oxidation reactions with suitable dienes and alkenes. The standardized methodology was made applicable to a variety of olefinic substrates. The reactions afforded desymmetrized hydroxy-ketone derivatives, unsaturated ketones and epoxides. This is the first application of MHP-based composites for in situ 1 O 2 generation [20].
Starting from the encouraging results obtained by the application of bulk MHPs, and upon pursuing our interest in finding new and greener methods for 1 O 2 generation as well as sustainable applications in the synthesis of valuable organic intermediates, we further explored the possible use of MHPs nanocrystals (NCs) for the 1 O 2 generation, to investigate the scope of the method in HDA, ene and oxidation reactions with representative dienes and alkenes. We enlarged the number of substrates used to perform the reactions and found interesting as well as valuable limitations occurring with some of them. The obtained value-added compounds are commonly the synthons for a remarkably high range of synthetic transformations, widely applied in modern organic synthesis and here derive from a sustainable and greener 1 O 2 methodology in organic synthesis taking advantage of easy tunable catalysts.

3 2 Results
We have prepared lead halide perovskite NCs of general formula MPbX 3 (Table of Fig. 1) where the cation M can be either inorganic (Cesium, Cs) or organic (Formamidinium, FA) and the anion X = Cl, Br or I. Catalysts C1-4 were prepared according to the reported procedures avoiding the use of toluene that can interfere during the photochemical reactions and applying centrifugation immediately before suspension in chloroform [21,22]. Figure 1a reports the emission spectra where the four catalysts show maximum emission wavelengths in the range 450-6870 nm. Catalysts C1-4 are typical monodisperse nanocrystals and their crystal structure characterization, reported previously, agrees with the current literature with all the prepared NCs showing an orthorhombic unit cell [21]. Transmission electron microscopy (TEM)

3
analysis confirms that all the samples possess a nearly cubic shape with size in the range of 20-30 nm, with a representative example reported in Fig. 1b [22]. Finally, Fig. 1c shows the emission properties under UV illumination of the samples used.
The catalysts were always freshly prepared and used immediately to perform the planned organic reactions.
1 O 2 is known to be a valuable dienophile and enophile whose reactivity has been thoroughly investigated both experimentally [23,24] and theoretically [25,26] in the past; the choice of typical 1 O 2 HDA cycloaddition reactions with 1,3-cyclohexadiene or freshly distilled cyclopentadiene relies upon the need to optimize the reaction conditions in these catalyzed processes. This starting point aims to evaluate the solvent, reaction time and temperature as well as detection methods for the products to select the best performing catalyst from the table list. To start with, we conducted the first reactions according to the performed experiments, previously set-up in other works for sake of comparison [19,20]. We used acetonitrile as solvent conducting the irradiation reactions at room temperature for 24 h, starting with the reactions with 1,3-cyclohexadiene whose reduced reactivity in HDA reactions avoids competitive light-promoted polymerization reactions and, eventually, other detrimental oxidation processes.
We precise that, since the NCs are generated in suspensions with other solvents (see Table of Fig. 1), the photocatalyzed reactions were conducted in acetonitrile as the solvent that can be considered as the true reaction solvent, being in large excess. Scheme 1 reports the HDA cycloaddition reaction with 1,3-cyclohexadiene that was conducted with all the four catalysts C1-4.
Irradiation was conducted under simulated solar light for 24 h saturating the organic phase with pure oxygen, and leaving the reaction under oxygen atmosphere (using a balloon as the reservoir connected to the photochemical vessel). Table 1 reports the data relative to the different catalysts performances with the yields values. In all the reactions, we verified that the 1,3-cyclohexadiene conversions were quantitative by GC-MS analyses that served also for the detection of the reaction products, here and in the other reported experiments. The characterization of the final products was, in fact, secured by comparison of the mass spectra with data available from literature and from authentic sample analyses [19,20].
Intermediate 1 represents the primary HDA cycloadduct and was never detected in all the performed experiments. The 4-hydroxycyclohex-2-en-1-one (2) was the only reaction product in entries 1-5 and 7 of Table 1, a known compound derived from the reductive cleavage of the peroxide O-O bond and simultaneous oxidation of one of the hydroxy functionalities, simultaneously promoted by the catalyst; the structure was confirmed by comparison with authentic samples (standard) as well as NMR characterization [15].
Catalysts C1 and C3 gave from modest to good yields in compound 2. In the reaction conducted with catalyst C3, another product could be detected, the cyclohex-2-ene-1,4diol, as a mixture of stereoisomers, obtained in 27% yield. The best result was obtained with catalyst C2 that provided 65% yield of 2, ameliorating the results obtained with the oxidized g-C 3 N 4 (65% of compound 2) [19]. Catalysts C4, that contains an organic cation (FA), gave the hydroxylketone 2 in fair yields (58%).
Based on the above-illustrated results we also investigated the behavior of cyclopentadiene to probe the scope of the reaction with a highly reactive diene, testing all the prepared catalysts. Scheme 2 shows the structures of the products obtained from the reaction of photogenerated 1 O 2 with freshly distilled cyclopentadiene. Due to the high reactivity of this diene and the known tendency to dimerize, we verified that the reaction conversion was complete although affected by an important amount of cyclopentadiene Scheme 1 HDA cycloaddition reaction of 1,3-cyclohexadiene and 1 O 2 used as benchmark reaction to test the efficiency of all the catalysts polymerization material. The composition of the reaction mixtures is reported in Table 2 as it was found by GC-MS analyses. The experimental conditions clearly activate a very fast cyclopentadiene dimerization that is responsible for the reaction outcome as shown in Scheme 2. The primary HDA cycloadduct 3 was neither isolated nor detected in the final reaction mixtures from independent experiments. Compound 4a that derives from the peroxide bond reductive cleavage was detected only in the reaction with catalysts C3 just in 3% yield and C4 in 7% yield ( Table 2). For the first time in this type of experiments we could detect the presence of oxidation products directly obtainable from the HDA cycloaddition reaction of the 1 O 2 to cyclopentadiene. This is furtherly demonstrated by the presence of the oxidized diketone 4b (6% yield) in the reaction mixture obtained from catalyst C1. All the reported products are known in literature and the structures are consistent with the reported data [27][28][29][30][31][32][33][34][35][36][37]. Looking at the other products, catalyst C1 is the worst in terms of the capability to functionalize the cyclopentadiene (monomer or dimer) scaffold. Catalysts C2 and C3 showed somewhat a similar behavior; compound 5a was obtained in 22% yield; analogously compound 5f was obtained in 22% yield but only in the case of catalyst C2. Compounds 5b-d complete the composition of the reaction mixture with variable amounts.
Catalyst C4 displayed a variable composition of the reaction mixture regarding the functionalization of the cyclopentadiene dimer. Poor yields were generally obtained for compounds 5. The best result corresponds to product 5f, obtained in 13% yield. It must be emphasize that in all the experiments above reported the catalyst was simply filtered at the end of each reaction and disposed. Freshly prepared catalysts were employed for new reactions and repetitions.
The presence of large amounts of dicyclopentadiene oxidized compounds 5a-f in the reactions shown in Scheme 2, prompted us to perform control experiments trying to reduce as much as possible the competing cyclopentadiene dimerization/oxidation processes, thus orienting the catalyzed reactions on the cyclopentadiene monomer, only. Scheme 3 shows the results obtained by conducting the same reaction with freshly distilled cyclopentadiene at − 78 °C (dry ice bath) in the presence of catalyst C4 that gave the best results in terms of cyclopentadiene monomer oxidation.
The other experimental conditions, aside from the reaction temperature, were the same as before but the results are remarkably different. Besides the presence of 49% of cyclopentadiene dimer, the oxidation processes of this latter were reduced with respect to the reaction at room temperature, at least in term of a variety of compounds. In parallel, cyclopentadiene monomer undergoes the HDA cycloaddition  reaction; the detected products are the diketone 4b (7%) (see also entry 4 of Table 2) and the 4-hydroxy-cyclopentenone 4c, obtained in 18% yield, as a mixture of stereoisomers. This means that the natural orientation of cyclopentadiene to dimerize is hard to constrain; if this powerful side reaction could be forbidden, the reaction non the monomer would lead to better and valuable results. We pointed out that 1 O 2 is not only a powerful dienophile but it undergoes ene reactions with alkenes through a complex mechanism finely elucidated by Houk in his theoretical investigations [25]. In our previous works we also evidenced that simple g-C 3 N 4 catalyzed processes or its composites with MHPs promoted oxidation reactions with several olefins to afford in some cases valuable synthons for organic chemistry applications [19,20]. For these reasons oxidations of alkenes and ene reactions represent the second task of our investigations with the new catalysts of Fig. 1 that gave the best results from the HDA cycloaddition reaction with 1,3-cyclohexadiene, i.e. C2 and C4.
We started with cyclic alkenes that were found prone to give valuable oxidized products in the presence of in situ generated 1 O 2 and Scheme 4 reports the structures of the obtained compounds along with their yields, from the selected catalysts.
Both the catalysts C2 and C4 gave nice and in some cases comparable results. The reactions with cyclopentene and cycloheptene (the odd-membered rings) display a good chemoselectivity leading to the corresponding α,β-unsaturated ketones as major products. The reaction of cyclopentene with catalyst C2 affords the chloro-cyclopentanol 6b whose formation can be attributed to the presence of chloride anion as impurity deriving from the catalyst preparation; the ether 6c was also detected in the reaction with catalyst C4, only, presumably deriving from rearrangement of the corresponding peroxide derivative. In the reaction with cycloheptene, the epoxide 8a was also obtained in comparable amounts in both the catalyzed reactions.
Less chemoselectively, the reactions with cyclohexene and cyclooctene (the even-membered rings) gave a plethora of products the most important of which were surely the α,βunsaturated products 7a, b in the reaction with cyclohexene that represent more than 60% of the reaction mixture when catalyst C2 is used. Compounds 7c, d were also detected in the same reaction.
Catalyst C4 substantially broadens the variety of products but in modest yields. In the case of cyclooctene reactions, the epoxide 9a is the major product with both the catalysts (42%) and the reactions are accompanied by other product in a range of yields 7-17%, confirming a poor chemoselectivity, in general. All the detected products are known compounds whose characterization is reported in literature, taken as reference [38][39][40][41][42][43][44][45][46].
The investigations in this field continued with a variety of acyclic alkenes, ranging from the highly C=C double bond substituted, such as tetramethyl ethylene (TME) and trimethyl ethylene (tme), cis-and trans-disubstituted olefins up to a monosubstituted one (e.g. 1-octene) (Scheme 5).
The C2 and C4 catalyzed reactions were quite unsatisfactory. The ene adduct 10, quantitatively prepared in previous works [19], could not be obtained here and compound 11 was obtained in C2 catalyzed reaction in 16% yield; the presence of bromide in compound 11 is due to the catalyst structure. Only catalyst C4 allows for the introduction of the novelty regarding the disubstituted alkenes, such as cis-2-hexene and trans-4-nonene, that however did not give remarkable results both in terms Scheme 5 Oxidation products from acyclic alkenes in 1 O 2 photocatalyzed reactions. ND, not detected 1 3 of chemoselectivity and yields. The oxidation of the cis-2-hexene afforded mainly α,β-unsaturated carbonyl compounds (12a-c) while the epoxides 12d, e clearly derive from the epoxidation reactions of 12b, c. On the contrary, the monosubstitute 1-octene underwent oxidative pathways to afford in fair yields the diol 14a and the ketoacid 14b when the catalyst C2 was used. With catalyst C4 novel compounds were also formed, one of them 14f in 33% yield. The structures were confirmed upon comparison with known products reported in the literature [47][48][49][50][51][52][53][54][55][56][57][58].
The studies were extended to some aromatic alkenes, such as allyl benzene, indene and 9,10-dihydroanthracene. Similarly as before, the reactions were performed by using C2 and C4 catalysts that gave the best performances in HDA cycloadditions, under the typical set-up experimental conditions above reported (Scheme 6).
Indene gave a selective reaction outcome with catalyst C2, affording the diol 15b as single product in fair yield (39%). The reactions with catalyst C4 were quite unselective and afforded in fair yields a mixture of ketones and alcohols, one of them, 15a (7%) directly obtainable from the ene reaction of indene with 1 O 2 .
The dihydroanthracene afforded the anthraquinone 16a in excellent yields with both the catalytic systems. The C2-catalyzed reaction is accompanied by the formation of the anthracen-9(10H)-one (16b) in 7% yield and of the anthracene (16c) from complete reduction of the starting material in up to 28% yield. Catalyst C4 shows a brilliant 100% yield of compound 16a.
Finally, we investigated the oxidation of allylbenzene. A poor chemoselective process results from the yields of the obtained products 17a-g with both the catalytic systems. The primary ene adduct 17a was obtained using catalyst C2, only. The cinnamaldehyde (17b) and the corresponding alcohol 17c are the major products along with the hydroxylketone 17d in the same catalyzed reaction. Minor amounts of 17f, g complete the mix of the products. The reaction catalyzed by C4 did not give the same performaces since low yields and low chamoselectivity characterized the outcome. All the compounds reported in Scheme 6 are known in literature [19,20,[59][60][61][62][63][64][65][66].
In all the reported experiments, blank trials were performed to verify that the absence of the catalysts would not activate any of the oxidative reaction pathways shown in the present work.
To summarize, lead halide perovskite NCs (see Fig. 1) were prepared and used as photocatalysts in HDA, ene and general oxidation reactions of dienes and alkenes as well as representative aromatic systems. HDA cycloaddition reaction with 1,3-cyclohexadiene and cyclopentadiene were used to define the best catalysts to conduct photooxidative processes with the other substrates. Catalysts C2 and C4 were found to give the best results with HDA in the presence of 1,3-cyclohexadiene, affording compound 2 up to 68% yield. In spite of the new catalysts employed, cyclopentadiene suffers of its spontaneous tendency to dimerize and even polymerize under the experimental conditions, offering a variety of oxidized compounds of the dimer itself. Interesting results were obtained with cyclic alkenes of different ring size. The oxidation reactions are not completely chemoselective, especially in the case of cyclohexene and cyclooctene, but α,β-unsaturated ketones are often the main products of the reactions along with the epoxides. As already reported in previous studies [19,20], the results of acyclic alkenes do not offer significant chances of applicability of the methodology for this type of molecules; we have not found yet a catalyst structurally fitted to interact positively with linear olefins in order to give valuable oxidized compounds. Finally, aromatic compounds such as indene, dihydroanthracene and allyl-benzene were easily oxidized by the selected catalysts. In particular, catalysts C4 gave quantitatively the anthraquinone 16a. Indene is mainly oxidized to the corresponding diol and the allyl-benzene offer a wide selection of oxidized compounds.

Discussion and conclusions
Lead halide perovskite NCs proved to be valuable photocatalysts for in situ 1 O 2 generation; easy-made 1 O 2 can be more conveniently employed in C-O bond formation reactions in the presence of a variety of olefinic compounds. The catalysts used in the present work display remarkable fluorescence properties (Fig. 1) that make them fit to be used in solar-light-driven processes in a sustainable manner.
These features have been thoroughly investigated and their application in organic synthesis is constantly growing. These catalyst make easier the possibility to run photooxidation reactions of Type II according to the Scheme 7 that sketches the differences between photooxidations of Type I and Type II [67]. In the first case, radicals or radical ions are photogenerated by suitable triplet state sensitizers and oxygen is captured to give the products (Type I); for example, the lack of 1 O 2 ene reactions in some cases and other non 1 O 2 products such as in indene are candidates for Type I reactions derived from oxygen radicals and oxygen radical ions [68][69][70]. In the second case, the sensitizer promotes the 1 O 2 formation that undergoes oxidation reactions with variable substrates (Type II). In the present cases, sensitizers can be avoided and the low cost and easy procedures to prepare the tested catalysts offer a valuable and cheap alternative to perform photooxidation reactions.
The obtained results open, however, some questions and problems to be solved. First problem is the possibility to design, on the basis of the results obtained from these and previous investigations, more selective and in particular chemoselective catalysts. This problem is, in some cases, solved on the substrate side, i.e. the substrate structure is prone to be easily oxidized affording single products, such in the case 1,3-cyclohexadiane or 9,10-dihydroanthracene. In other, we are in the presence of a totally absence of selectivity or even any possibility to run a reaction. The design of such a catalysts implies also to face the problem of their stability and photochemical properties that must be suited for performing organic reactions. Second problem is the scale up in organic processes to make these methodologies applicable to the production of valuable synthons not easily obtainable with other approaches. The possibility to conduct photooxidations on a large scale could be extremely important to change the habit to use toxic reagents, such as heavy metals, to preform oxidations on total carbon organic molecules. Strictly linked to this problem is the third, which is essentially Scheme 7 Photooxidation reactions of Type I and Type II pathways technological, i.e. the possibility to industrialize the method making the catalyst recoverable in some way and/or linkable on a solid support that would make easier and easier their use in organic synthesis. Last point is the extension to other substrates that could be oxidized. At the moment, we have limited our reaction spectrum to the classical 1 O 2 processes, such as HDA and ene reactions. Within the oxidation reaction box, many other interesting processes can find a relevant place for disrupting traditional approaches into sustainable ones. On the mechanism side, Type II pathway reasonably fits with the nature of the obtained products. Competitive radical routes cannot be excluded in part and other investigations are required for a definitive answer. Moreover, organic reaction other than 1 O 2 involving processes awaits their contribution to novel photocatalytic methodologies.
In conclusion, we can affirm that lead halide perovskite NCs are interesting and valuable photocatalysts for the in situ generation 1 O 2 to perform HDA, ene and oxidation reaction with suitable dienes and alkenes. The methodology has been reasonably standardized and made applicable to a variety of olefinic substrates. The scope of the method has been finely illustrated by the results in all the tested reactions, which allowed for the obtaining of desymmetrized hydroxy-ketone derivatives, unsaturated ketones and epoxides. Notable limitations were also observed especially in the case of the alkene oxidation as well as the poor chemoselectivity somewhere observed. These tuned-up oxidative properties of suitably modified catalysts offer a remarkable improvement in determining a change in the approach to 1 O 2 generation methods that may open other ways to perform organic reactions through greener and sustainable ways.