An unknown component of a selective and mild oxidant: structure and oxidative ability of a double salt-type complex having κ1O-coordinated permanganate anions and three- and four-fold coordinated silver cations

Compounds containing redox active permanganate anions and complexed silver cations with reducing pyridine ligands are used not only as selective and mild oxidants in organic chemistry but as precursors for nanocatalyst synthesis in low-temperature solid-phase quasi-intramolecular redox reactions. Here we show a novel compound (4Agpy2MnO4·Agpy4MnO4) that has unique structural features including (1) four coordinated and one non-coordinated permanganate anion, (2) κ1O-permanganate coordinated Ag, (3) chain-like [Ag(py)2]+ units, (4) non-coordinated ionic permanganate ions and an [Ag(py)4]+ tetrahedra as well as (5) unsymmetrical hydrogen bonds between pyridine α-CHs and a permanganate oxygen. As a result of the oxidizing permanganate anion and reducing pyridine ligand, a highly exothermic reaction occurs at 85 °C. If the decomposition heat is absorbed by alumina or oxidation-resistant organic solvents (the solvent absorbs the heat to evaporate), the decomposition reaction proceeds smoothly and safely. During heating of the solid material, pyridine is partly oxidized into carbon dioxide and water; the solid phase decomposition end product contains mainly metallic Ag, Mn3O4 and some encapsulated carbon dioxide. Surprisingly, the enigmatic carbon-dioxide is an intercalated gas instead of the expected chemisorbed carbonate form. The title compound is proved to be a mild and efficient oxidant toward benzyl alcohols with an almost quantitative yield of benzaldehydes.


Introduction
As a useful mild oxidant in organic chemistry, a 'compound' prepared by Firouzabadi and reported as '[Ag(py) 2 ]MnO 4 ' (Firouzabadi's compound, FC) has been known for a long time. 1 According to literature data, it is an important reagent in a range of oxidation reactions including the preparation of various oxocompounds and sulfones. 1-5 It can be used as a catalyst for the CN coupling reactions of aromatic hydrocarbons and primary amines including the synthesis of the pharmaceutically important quinazoline heterocycles3 such as Gefitinib, the drug used for treating breast and lung cancers.5 Furthermore, FC decomposes into Ag/manganese oxide particles, which are candidates for catalyzing the selective oxidation of 3-picoline into niacin, the key compound of vitamin B3 synthesis. 6 The successful application of FC in organic oxidation reactions requires detailed knowledge about the structure, properties as well as the catalytically active chemical sites and the reactivity of the compound. However, the chemical identity of FC is questionable 7 as its structural data are missing. According to Sajó et al. 8 it is indeed a ~1:1 mixture of the 4[Ag(py) 2 MnO 4 ]·[Ag(py) 4 ]MnO 4 double salt (compound 1) and the [Ag(py) 2 ]MnO 4 (compound 2). 9 Therefore, the structural characterization and properties identification are imperative for both compound 1 and 2 in order to appreciate the mild and selective oxidation capacity of FC. In this work we study the structural and vibrational spectroscopic characteristics, as well as thermal properties and reactivity with selected organic materials of compound 1 and compare its structure and properties with the known perchlorate analogues (listed in Table 1) and selected pyridinesilver(I) permanganate compounds found during the syntheses of FC. 8

Syntheses
Compound 1 is prepared from the pyridine solution of AgMnO 4 by dilution with water until reaching 10 % pyridine concentration following the method described in the literature. 8 The blackish purple block-shaped crystals are stable for a month at room temperature in the dark, but after 1-2 weeks a shiny silvery colour appears on the surface of the crystals, which is not accompanied by bulk compositional change (ESI1).

Vibrational modes of the coordinated and non-coordinated permanganate anions
Correlation diagrams for MnO 4anions at S4 (non-coordinated permanganate ion) and C1 (permanganate coordinated to [Ag(py) 2 ] + cation) sites of compound 1 lattice are given in ESI2, the observed IR and Raman frequencies and their assignations are collected in Table 2. There are 9 internal modes of the MnO 4anions at S 4 sites. The B and E modes of the factor group (f-g) are IR active, whereas all f-g modes (of A, B and E symmetry) are Raman active. E modes are doubly and F modes are triply degenerated. There are 9 internal vibrational modes of the MnO 4anions at C 1 sites (permanganates are coordinated to [Ag(py) 2 ] + cations). Each mode from the local (site) group splits into 3 components in the factor group. Thus, there are 9 A modes, 9 B modes and 9 E modes (the latter are doubly degenerated) giving rise to 36 vibrational degrees of freedom due to the 4 MnO 4anions on general positions (i.e. positions with C 1 symmetry). There are 6 external vibrations of the MnO 4anions at S 4 sites; 3 and 4 bands are expected in the far IR region, i.e. the low-frequency part of the Raman spectra, respectively. The external vibrations of the MnO 4anions at C 1 sites result in a total of 24 vibrational degrees of freedom, which is in accordance with the 4 tetrahedral anions on general symmetry positions. The external vibration bands are also expected in the far IR and the low-frequency region of the Raman spectra. No predictions of the band intensities can a priory be given. Table 2. IR and Raman wavenumbers of permanganate anions in compound 1 at room temperature vs-very strong, m-medium, w-weak, vw-very weak; According to the factor group analysis (unit-cell group analysis), the two types of permanganate ions result in 4 vibrational modes (Table 2). However, the distinctions of permanganate bands belonging to S 4 and C 1 sites are challenging. One would expect a strong doublet due to the symmetric stretch (the breathing mode) in Raman, however, the observed spectrum does not show this expected feature (ESI 3). Three bands appear as a result of splitting of the  3 (MnO 4 ) modes instead of the expected twice three components due to the pronounced distortion of the permanganate moieties at C 1 symmetry. 17a 17c The  s (AgN) mode appearance in the IR spectrum may result from the deviation of the ideal (linear) N-Ag-N angle of [Ag(py) 2 ] + ion. Although the N-Ag-N unit slightly deviates from linear (173.9(2)° instead of 180°), this deviation may be sufficient to activate the  s (AgN) mode in the IR spectrum. 13 Since the Ag-N distances in compounds 1 (ESI9) and 1-ClO 4 12 are practically the same, the difference between the  s (Ag-N) band positions in 1 and 1-ClO 4 suggests that the  s (AgN) mode of compound 1 is coupled with (AgO) modes

UV-spectral studies
The diffuse reflection UV-Vis spectrum of compound 1 contains a band system consisting of pyridine n-*, Ag +pyridine MLCT and permanganate t 1 -4t 2 , 3t 2 -2e transitions. 18 The UV-silent counterion containing [Ag(py) 2 ]NO 3 and KMnO 4 spectra confirm the assignations and multi-complex nature of the UV bands given for compound 1 in Table 4 and ESI4. The band maxima and their assignations are shown in Table 4. The number of bands belonging to each transition of compound 1 listed in Table 4 can be multiplied due to the presence of two kinds of pyridine containing cations ([Ag(py) 2 ] + and [Ag(py) 4 ] + ) and the coordinated/non-coordinated type of permanganate anions. The 219.9 nm band may be assigned as frequencies for combined bands consisting of Ag-py (CT) and MnO 4 -( 1 A 1 -1 T 2 (t 1 -4t 2 )), whereas the 258.4 nm band contains the MnO 4 -( 1 A 1 -1 T 2 (3t 2 -2e)) transitions and the components of the pyridine ( 1 A 1 -1 B 2 (n→*)) transitions. 219.9 and 258.4 nm bands are close to those found for [Ag(py) 2 ]NO 3 (219.4 and 252.2 nm, respectively) and KMnO 4 (227.3 and 259 nm, respectively) as well. The main text of the article should appear here with headings as appropriate.
The bands found at 521.9 and 710.1 nm belong to the components of permanganate ( 1 A 1 -1 T 2 )(t 1 -2e) and 1 A 1 -1 T 1 (t 1 -2e) transitions, respectively. The hypsochromic band shift Please do not adjust margins Please do not adjust margins belonging to compound 1 toward 1 A 1 -1 T 1 (t 1 -2e) transition at 521.9 nm of KMnO4 might arise from the coordinated nature of permanganates in the lattice of compound 1.

Thermal decomposition
To initiate the redox reaction between the redox active cationic and anionic parts of compound 1, we heated the material in inert and O 2 -containing atmospheres. The thermal decomposition process of compound 1 in an inert atmosphere is proved to be a strongly exothermic explosion reaction; a part of the decomposition products practically burns out from the crucible. Therefore, the sample has to be diluted with an inert heat absorbing material (70 % alumina). The reaction proceeds with 48.0 % (wt.) mass loss at 85 °C peak (TG and DTG) temperature (ESI5 and Table 5.). The same exothermic characteristic is observed in the experiment performed in air flow (Table 5 and ESI5), thus the oxygen of the air oxygen is not essential for initiating the decomposition process. The second decomposition step is a slower process in comparison to the first one and occurs between 410 and 500 °C and results in 8.0 % (wt.) mass loss (DTG peak temperature corresponds to 428 °C).

The autocatalytic character of the decomposition intermediates
The total mass loss in the decomposition reaction of compound taking into consideration the complete lack of oxygen evolution (Fig. 7), the I-300 phases could unambiguously be formed only from the reduction reactions of permanganate ions. The MnO 2 IR band disappears on further heating (it is the strongest oxidant among the Mn-oxides present), and the intensity of IR bands belonging to carbon-rich residues in I-300 (aromatic C=C and C=N bonds at 1500-1600 and ~2300 cm -1 , respectively) decreases/disappears with the appearance of the oxidative cleavage products of aromatic C=C bonds (C-O-C at ~1083 cm-1) (ESI6).  26 or AgMnO 4 (T dec =135). 27 The decomposition of compound 1 is consistent with a solid phase quasi-intramolecular redox reaction and not with a ligand loss/permanganate decomposition followed by consecutive oxidation of the liberated pyridine. The solid phase quasi-intramolecular redox behaviour can be explained by the presence of the weak hydrogen bond interactions between the -CH hydrogen of pyridine ring in the [Ag(py) 2 ] + units and the permanganate ion (Fig. 2)  The pyridine permanganate = 2.4 molar ratio in compound 1 (CH:MnO 4 =12:1 → CH/O=3) suggests that only a part of the pyridine can be oxidized in inert atmosphere by the oxygen content of compound 1, whereas in air, the oxygen content of the atmosphere plays a role in the pyridine oxidation process confirmed by the DSC experiments performed under pure N 2 and O 2 ( Fig. 8 and ESI7). The decomposition heats are -756.94 and -895.02 kJ/mol under N 2 and O 2 , respectively. However, sample aging fundamentally influences the decomposition process. The decomposition starting temperature of the fresh material under N 2 is 107 °C, whereas it is 101 °C for the one-month-old, Koerbl-catalysts containing sample.

Evolved gas composition and the mechanism of the redox reactions
The decomposition heat of one-month-old sample under N 2 slightly increases (-850.35 kJ/mol), which may be attributed to either different reaction products or different distribution of the same products as for the fresh sample. A secondary process may also be found at 129 °C which we explain as the reaction of residual organic content. Under O 2 , the catalyzing effect of decomposition products becomes obvious, the shape of DSC curve and the decomposition heat are strongly altered (Fig. 8). The unusual sigmoid character of the DSC curve is the consequence of the asymmetric heating of the crucible resulted from the non-symmetrical ignition profile of the sample. The decomposition (including the ignition) heat increases significantly, it is -1651.23 kJ/mol respect to -756.94 kJ/mol of fresh sample. This increase unambiguously confirms both the role of outer oxygen in the decomposition process and the catalytic effect of the intermediates formed under storage. The appearance of NO and C 4 H 4 fragments in the TG-MS of compound 1 also suggests that redox reactions of the pyridine rings should proceed similarly to that of compound 2. 8 Pyridine reduces the permanganate into manganese oxides, which is not accompanied by O 2 evolution; the permanganate disappears even before its expected decomposition temperature. 7a The appearance of NO and C 4 H 4 fragments in the TG-MS of compound 1 also suggests that redox reactions of the pyridine rings should proceed similarly to that of compound 2. 8 Pyridine reduces the permanganate into manganese oxides, which is not accompanied by O 2 evolution; the permanganate disappears even before its expected decomposition temperature. 7a

Enigmatic carbon dioxide in the Ag/manganese oxide matrix
Not only the lack of O 2 evolution but also the appearance of CO 2 evolution at 400-430 °C during the decomposition of compound 1 deserves recognition. This temperature range coincides with the thermal decomposition temperature of MnCO 3 into MnO and CO 2 , however, there is no indication of MnCO 3 by PXRD in I-300 and there is no observed reduction of CO 2 by MnO into CO as it should occur in an inert gas stream: 28 MnCO 3 = MnO + CO 2 3MnO + CO 2 = Mn 3 O 4 + CO The basic silver carbonate would similarly decompose with CO 2 evolution in this temperature range ,29 however, neither PXRD nor IR studies show any silver carbonate 30 compounds in I-300. Therefore, the in-situ formed carbonates as sources of CO 2 evolution could completely be excluded.

Figure 8. DSC curves of fresh ((red line) and one-month-old (black line) samples of compound 1 under O 2 atmospheres
The oxidation of carbonaceous materials with MnO 2 (or partly with Mn 2 O 3 ) starts above 300 °C. However, the IR spectrum of I-300 unambiguously shows the presence of intercalated gaseous CO 2 (2350 cm-1 ) 31 and its amount decreases significantly during heating up to 700 °C (ESI6). The formation of this gas can be attributed to the solid-phase redox reaction, during which the formed manganese oxide matrix encloses the locally evolved gas. The high-temperature CO 2 formation would suggest a strong bond between CO 2 and the porous manganese oxide sorbents, 32a which could occur via the Mn=O + CO 2 = Mn(CO 3 ) chemisorption reaction. However, the BET surface area does not support strong interaction between the CO 2 and the Ag/Mn-oxide matrix as the N 2 and CO 2 adsorption measurement shows only 6 and 3 m 2 /g, respectively. The CO 2 sorption-desorption isotherms of I-300 (ESI8) also suggest that the sample contains a simple gas inclusion, which is consistent with the gas-impermeable character of the formed mixture.
In the O 2 atmosphere, three small peaks of CO 2 formation could be detected during decomposition of compound 1. In addition to the peak of CO 2 formed at 430 °C in an inert atmosphere, two additional peaks appear at 200 °C and 750 °C due to the formation of carbon dioxide in the aerial oxidation of organic residues. In the O 2 atmosphere, three small peaks of CO 2 formation could be detected during decomposition of compound 1. In addition to the peak of CO 2 formed at 430 °C in an inert atmosphere, two additional peaks appear at 200 °C and 750 °C due to the formation of carbon dioxide in the aerial oxidation of organic residues.

The role of the oxygen gas flow on the thermal decomposition process of compound 1
DSC study of compound 1 at 10 °C/min heating rate under Ar shows that compound 1 has no polymorphic phase transition between -150 °C and the decomposition temperature. The decomposition peak temperatures under Ar or O 2 atmosphere are found to be almost identical, 101 and 108 °C, respectively (Fig. 8., ESI7). Under O2 gas almost twice more heat is evolved than under Ar, and two very small exothermic peaks (93 and 98 °C) also appear as a result of aerial oxygen during decomposition.

Organic oxidation reactions
Firouzabadi synthesis results in a mixture of compounds 1 and 2, and the recrystallization of the mixed raw product from benzene gives rise to the formation of pure [Ag(py) 2 ]MnO 4 ·0.5py (compound 4). 8 In order to clarify which compound or which component of the reaction mixture prepared and reported by Firouizabadi 1a is responsible for the mild and useful oxidation effect, the oxidation abilities of compounds 1, 2 and 4 have to be studied on well-selected test materials systematically. As a part of this systematic study, the oxidation ability of compound 1 has been tested in the oxidation reaction of benzyl alcohols (BzOH, 2-and 4-nitro and 2-methoxy-substituted benzyl alcohol) in various organic solvents, at room and reflux temperatures (CHCl3-61 °C, CCl4-76 °C and benzene-80 °C). The results of oxidation reactions with selected solvents are given in Tables 7 and 8.

Oxidation of benzyl alcohol by 1
Under analogous conditions used by Firouzabadi, 1a benzaldehyde (PhCHO) and benzoic acid (PhCOOH) are formed together in a PhCHO/PhCOOH = 0.5 ratio. Firouzabadi reported 96 % isolated PhCHO yield without by-products, 1a in contrast, our experiment unambiguously showed that compound 1 could oxidize the PhCHO formed from BzOH further to PhCOOH. A further difference is the appearance of a small amount of diphenyl (PhPh). Increasing reaction time (from 30 to 120 min in benzene as solvent) under reflux slightly influences the distribution of the oxidation products (Table 6). In contrast, the reaction temperature greatly influences the distribution of the products. At room temperature the PhCHO/PhCOOH acid ratio is ca. 6 and the BzOH conversion is ca. 20 %. The appearance of PhPh is observed only at reflux temperature (80 °C).

Scheme 1. Oxidation reactions of 1 towards benzyl alcohol
Since carbon tetrachloride does not result in PhPh formation even at reflux temperature (76 °C) in either 30 or 120 min reaction time, thus the influence of solvents also has to be tested on the formation of PhCOOH and PhPh (Table 6). *From GC-MS ion chromatograms The relative error of measurements was below +/-0.4%. **Using freshly prepared compound 1 without silvery colour; ***In the presence of an artificial silver mirror prepared from diamminesilver(I) chloride and glucose, 33 Compound 1 is purple when freshly prepared but gets a shiny silvery colour in due time, which is attributed to the formation of a thin surface film of metallic silver, similarly to the AgMnO 4 -sourced silver particles formation. 7a Silver nucleating centers result in silver mirror formation on the wall of the vessel (both in CCl 4 and benzene). Since the PhPh formation is observed only in benzene solvent, the decarboxylation of PhCOOH and dimerization of phenyl radicals are improbable, thus benzene is the key factor in the PhPh formation. The silver catalyzed reaction of benzene with PhCOOH in the presence of a strong oxidant (e.g., persulphate) has already been investigated, 34 thus the PhPh formation is supposed as a result of the reaction of benzene and PhCOOH with permanganate as oxidant and silver as a catalyst. To confirm this hypothesis, fresh compound 1 (without silvery lustre) was prepared and used it immediately. During 240 min reflux in benzene, it quantitatively yields PhCHO, without diphenyl or benzoic acid formation. Preparing a silver mirror from diamminesilver(I) chloride and glucose as reducing agent, 35 and using the silver-coated vessel to perform the reaction of BzOH and freshly prepared compound 1 (without silvery lustre), under analogous conditions, a complete BzOH conversion with PhCHO/PhCOOH=~2:1 ratio and several percent of PhPh formation could be observed. This experiment unambiguously shows the catalytic effect of silver in the PhPh and PhCOOH formation reaction. Testing of PhCOOH decarboxylation in benzene under reflux for 240 min without any oxidizing agent (e.g., compound 1) fails to form PhPh. Thus, not only the benzene as solvent and metallic silver as a catalyst play key role in the formation of PhPh but also the presence of oxidant is essential to form PhPh.
As it was revealed that the polarity of the solvent and the reaction temperature play an essential role in the conversion of benzyl alcohol and the distribution of the oxidation products, a low-boiling and oxidation-resistant but polarsolvent (CHCl3) was also tested in the reaction between compound 1 and various benzyl alcohols (Table 7). Roughly 90 % conversion was found in 30 min at reflux temperature, notwithstanding at room temperature, 180 min reaction time gave an almost complete conversion of BzOH into PhCHO without PhCOOH formation.

Substituted benzyl alcohols oxidation with compound 1
Since the chloroform was found to be the best choice among the tested solvents, with the aim of preparation of the appropriately substituted benzaldehydes, a couple of oxidation reactions were examined. In particular, the 2-and 4nitrobenzyl alcohols (electron-withdrawing substituents) as well as 2-methoxybenzyl alcohol (electron-donating substituent) substrates were tested. The oxidation reactions were performed in chloroform at room and reflux temperatures ( Table 7). The phenyl ring substituents in benzyl alcohol increased the reactivity towards oxidation with compound 1. Independently from the nature and position of the substituents, practically complete conversion of the benzyl alcohols into the appropriate benzaldehyde occurred even at room temperature in 30 min. To examine the effect of temperature, the oxidation reaction of 4-nitrobenzyl alcohol was tested at reflux temperature as well, but the oxidation of the formed 4-nitrobenzaldehyde did not occur at all. These preliminary results confirmed the oxidation ability of compound 1. However, similar structural and reactivity studies of compound 2 are necessary in order to use FC for preparing commercial pharmaceuticals. (The relative error of measurements was below +/-0.4%).

Experimental
In the synthesis and analytical experiments analytical grade pyridine, silver(I) nitrate, 40 % aq. NaMnO 4 and solid KMnO 4 , twice distilled water, hydrochloric acid (37 %) and nitric acid (67 %) (Deuton-X Ltd) were used. All experiments with pyridine ligand containing silver permanganate and perchlorate complexes have to be performed very carefully due to the existing a possible hazard of explosion. The procedure to prepare the compound 1 can be safely, without risk of explosion, performed in the following way: Freshly prepared wet silver(I) permanganate (2.27 g, precipitated in the reaction of saturated aq. AgNO 3 and 40 % aq. NaMnO 4 solutions, at 1:1 Ag:MnO 4 ratio, with washing the precipitate with copious amounts of cold water) was dissolved in pure pyridine, then the saturated purple solution formed was immediately diluted with water to reach a pyridine content of 10%. A precipitate was immediately formed which proved to be pure 1 8 . Using wet AgMnO 4 is essential, because old-samples of AgMnO 4 can ignite the pure pyridine due to the catalytic effect of the silver manganese oxides formed on the surface of the old and dry samples.
The digestion of samples for ICP measurements was done in the 1:1 mixture of 67 % nitric acid and 37 % hydrochloric acid. The organic (benzyl alcohol, o-and p-nitrobenzyl alcohol and o-methoxybenzyl alcohol) were reagent grade chemicals (Deuton-X Ltd). [Ag(py) 2 ]NO 3 was prepared according to the method given by Jörgenson 35 .
The organic oxidation reactions have been performed with 0.01 mol of benzyl alcohol derivative dissolved in the appropriate organic solvent (CHCl 3 , CCl 4 or benzene) and 1.5fold molar excess of compound 1. The mixture was stirred at room temperature for 30 or 120 min, and another portion of the reaction mixture was refluxed (CHCl 3 -61 °C, CCl 4 -76 °C and C 6 H 6 -80 °C) for 30 or 120 min. The conversion was followed by GC-MS ion intensities until reaching the complete conversion. The partial conversion data was calculated from ionchromatogram intensities which define a rough estimation of molar conversions. The calibration was performed using 2,4dinitrophenylhydrazones. Manganese(III) (Mn 2 O 3 and Mn 3 O 4 ) content of the thermal decomposition products were determined by reacting the mixtures containing these materials with oxalic acid in the presence of 20 % sulfuric acid, with 2 h boiling then the oxalic acid excess was titrated back with 0.002 M potassium permanganate solution.
The Ag and Mn content of compound 1 were determined by atomic emission spectroscopy using a Spectro Genesis ICP-OES (SPECTRO Analytical Instruments GmbH, Kleve, Germany) simultaneous spectrometer with axial plasma observation. The multielement standard solution for ICP (Merck Chemicals GmbH, Darmstadt, Germany) was used for calibration. The carbon, hydrogen and nitrogen content were measured by elemental analysis (Fisons model CHN 1018S). The phase purity of compound 1 was checked by powder X-ray diffractometry. X-ray powder diffraction measurements were performed using a Philips PW-1050 Bragg-Brentano parafocusing goniometer equipped with a Cu tube operated at 40 kV and 35 mA power. A secondary beam graphite monochromator and a proportional counter were also equipped. Scans were recorded in step mode. Full profile fitting techniques were used for the evaluation of the diffraction patterns. The FT-IR spectrum of compound 1 was recorded in the attenuated total reflection (ATR) mode on a Bruker Tensor 27 Platinum ATR FT-IR spectrometer (2 cm -1 resolution) between 4000 and 400 cm −1 . The far-IR measurement was performed on a BioRad-Digilab FTS-30-FIR spectrometer for the 400-40 cm -1 range in polyethylene pellet. The Raman measurement was performed using Horiba Jobin-Yvon LabRAM-type microspectrometer with external 532 nm Nd-YAG laser source (~40mW) and Olympus BX-40 optical microscope. The laser beam was focused by an objective (10X) and a D1 intensity filter decreased the laser power to 10 % to avoid thermal degradation. The confocal hole of 1000 µm and 1800 groove mm -1 grating monochromator were also used in a confocal system and for light dispersion, respectively. The spectral range of 100-4000 cm -1 was detected with 3 cm -1 resolution. Each spectrum was collected at 240 s per point. Diffuse reflectance spectrum in the UV-Vis region (200-800 nm) was detected at ambient conditions by a Jasco V-670 UV-Vis spectrophotometer equipped with NV-470 type integrating sphere using the official BaSO 4 standard as a reference. Thermal data were collected using a TA Instruments SDT Q600 thermal analyzer coupled to a Hiden Analytical HPR-20/QIC mass spectrometer. The decomposition was followed between room temperature and 800 o C at 2 o C min -1 heating rate in He and air as carrier gas (flow rate = 50 cm 3 min -1 ). The crystal contains two Ag complex cations with pyridine molecules as ligands and two permanganate anions. The ratio of the two complexes is 1:4 in the double salt. The lattice has high I-4 symmetry. It results in low data to parameter ratio. In case of one tetrahedral cation and one tetrahedral anion there is only one-fourth of the molecule in the asymmetric unit. Numerical absorption correction was applied to the data (the minimum and maximum transmission factors were 0.8567 and 0.9965). The structure was solved by direct methods. 36 Anisotropic fullmatrix least-squares refinement 36  The maximum and minimum residual electron density in the final difference map was 0.386 and -0.274e.Å -3 . The weighting scheme applied was w = 1/[ 2 (F o 2 ) + (0.02940.0000P) 2 + 0.0000P], where P = (F o 2 +2F c 2 )/3. Hydrogen atomic positions were calculated from assumed geometries. Hydrogen atoms were included in structure factor calculations, but they were not refined. The isotropic displacement parameters of the hydrogen atoms were approximated from the U(eq) value of the atom they were bonded to. The liquid products were analyzed using a GC-MS (Shimadzu QP2010, He as the carrier gas) equipped with an RXi-5SIL MS capillary column (30 m x 0.2 mm x 0.25 μm). The column temperature was raised from 50 to 340 °C with a heating rate of 10 °C/min. CCDC-1879263 (1) contains the supplementary crystallographic data for this paper (ESI9). These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.

Conclusions
The The enigmatic CO 2 is an intercalated gas in the manganese oxide matrix instead of the expected chemisorbed carbonate form. Compound 1 is a mild and efficient oxidant toward benzylic alcohols (unsubstituted, 2-MeO, 2-NO 2 and 4-NO 2 ). A solvent-and temperature-dependent oxidation takes place in all reactions with an almost quantitative yield of benzaldehydes. Chloroform is found to be the best solvent. The reaction of benzyl alcohol in CCl 4 and benzene results in benzaldehyde and benzoic acid, whereas in benzene diphenyl formation occurs due to the oxidative coupling of benzoic acid and benzene in the presence of metallic silver catalyst.

Conflicts of interest
"There are no conflicts to declare".