Thermal Decomposition of [AH][M(HCOO)3] Perovskite-Like Formates

A systematic study of the thermal decomposition of hybrid perovskites of formula [AH][M(HCOO)3] under inert atmosphere was performed by means of thermogravimetry and simultaneous infrared spectroscopy of the evolved gases. The influence of: (i) the metal ion of the [M(HCOO)3]- framework and (ii) the guest [AH]+ cation, in the composition of the final residue was evaluated. In this work, it has been demonstrated that these materials can be used as precursors of metal or metal-oxide compounds—obtained free of carbon—, and that the composition of the final residue is determined by the standard reduction potential of the metal cation of the framework.


Introduction
Coordination polymers are hybrid inorganic/organic structures formed by metal cation centers that are linked by organic ligands, in the form of one-, two-, or threedimensional crystalline structures. Moreover, guest molecules weakly bonded to the framework can be located inside the structure (see Figure 1, as an example).
Solids2021, 1, FOR PEER REVIEW 2 In this context, the study of the thermal decomposition of coordination polymers with perovskite-type structures as potential self-template precursors of metal or metal oxide materials is of interest. Hybrid perovskites are some of the most studied covalent frameworks due to their interesting functional properties reported along the last two decades. Among them, compounds of the formula [AH][M(HCOO)3] (Figure 1) have attracted great interest due to their dielectric [18][19][20], magnetic [21,22], and multiferroic [23][24][25][26] properties. Here, a systematic study of the degradation of the [AH][M(HCOO)3] crystalline materials containing different M 2+ cations, connected through formate (HCOO -) ligands, and hosting different protonated amines (AH + ) is presented. The aim is to understand the decomposition mechanisms and the factors which determine their final residue.

Results
The thermal degradation of two groups of perovskite-like formates [AH][M(HCOO)3] was studied following the next strategy: On one side, in order to evaluate the role of the metal cation on the thermal stability and degradation process, we compare a group of [CH3NH3][M(HCOO)3] perovskites The thermal stability of a coordination polymer is a key parameter which determines its optimal range for practical applications [1] including gas separation [2], ion exchange [3], water desalination [4], and moderate-temperature heat storage [5,6].
Usually, during the thermolysis of coordination polymers the guest molecules (if they exist) are the first ones being eliminated since they are weakly bonded to the framework. Afterwards, the framework collapses and the organic linkers are released as CO 2 or converted into a carbon matrix. The research in this field has been focused on the use of porous coordination polymers (also known as metal-organic frameworks, MOFs) thermally treated under inert conditions for the preparation of core-shell nanocomposites which consist in metal or metal oxide nanoparticles embedded in carbon porous nanostructures, respectively named M@C or MO@C [16,17]. However, for some applications (e.g., in magnetic or in piezoelectric devices), the carbon matrix can be more of a handicap than an asset. For these applications, it is worth looking for different coordination polymers formed by ligands of variable length to evaluate the possibility of obtaining metal (or metal oxide) products free of carbon.
In this context, the study of the thermal decomposition of coordination polymers with perovskite-type structures as potential self-template precursors of metal or metal oxide materials is of interest. Hybrid perovskites are some of the most studied covalent frameworks due to their interesting functional properties reported along the last two decades. Among them, compounds of the formula [AH][M(HCOO) 3 ] ( Figure 1) have attracted great interest due to their dielectric [18][19][20], magnetic [21,22], and multiferroic [23][24][25][26] properties.
Here, a systematic study of the degradation of the [AH][M(HCOO) 3 ] crystalline materials containing different M 2+ cations, connected through formate (HCOO − ) ligands, and hosting different protonated amines (AH + ) is presented. The aim is to understand the decomposition mechanisms and the factors which determine their final residue.

Results
The thermal degradation of two groups of perovskite-like formates [AH][M(HCOO) 3 ] was studied following the next strategy: On one side, in order to evaluate the role of the metal cation on the thermal stability and degradation process, we compare a group of [CH 3 3 ] compounds. TGA results reveal that these materials start to decompose between 150 • C (Cu 2+ ) and 230 • C (Ni 2+ ). The obtained residues after the complete thermal decomposition were analyzed by XRPD at room temperature. According to the XRPD results, the analyzed compounds can be divided in three groups in function of their final decomposition products: (i) M 2+ = Mg 2+ , Mn 2+ , and Zn 2+ formates decompose to the corresponding MO materials (Figures S1-S3 of ESI); (ii) M 2+ = Ni 2+ and Cu 2+ perovskites decompose to the pure M species (Figures S4 and S5); and (iii) Co 2+ compound decomposes to a mixture of 0.29 CoO + 0.71 Co (w/w) ( Figure S6). These results are in agreement with those reported in the literature for anhydrous and dehydrate metal formats [27][28][29][30].  Table 1 recompiles the most relevant information obtained by TGA and XRPD relative to the thermal decomposition of this group of compounds.
To deep further into the mechanism of the observed thermal decomposition for these compounds, the released gases were analyzed by infrared spectroscopy. The IR spectra obtained at different temperatures was compared against the reference spectra from the National Institute of Standards and Technology (NIST) database [31]. The corresponding IR spectra are shown in Figure 2b and Figures S9-S18.  Table 1 recompiles the most relevant information obtained by TGA and XRPD relative to the thermal decomposition of this group of compounds. To deep further into the mechanism of the observed thermal decomposition for these compounds, the released gases were analyzed by infrared spectroscopy. The IR spectra obtained at different temperatures was compared against the reference spectra from the National Institute of Standards and Technology (NIST) database [31]. The corresponding IR spectra are shown in Figures 2b and S9-S18.
[CH 3 NH 3 ][Mg(HCOO) 3 ] is stable up to 200 • C, and above this temperature it shows a two-step degradation process, until it finally decomposes into MgO ( Figure S1). During the first decomposition step (200-280 • C) it degrades to the Mg(HCOO) 2 intermediate ( Figure S7), releasing CH 3 NH 2 and HCOOH as the main gases, together with a minor portion of CO 2 (Figures 2b(i) and S9). In the second step (370-470 • C), the intermediate decomposes eliminating CO 2 together with some CO and HCOOH (Figures 2b(ii) and S10).
[CH 3 NH 3 ][Zn(HCOO) 3 ] is stable up to 160 • C. It degrades in a smooth two-step process to ZnO ( Figure S3). The decomposition starts with the releasing of CO 2 together with CH 3 NH 2 and H 2 O. The amine is eliminated through the whole degradation process. Above 240 • C, the formate linkers are removed or degraded not only as CO 2 but also as HCOOH, CO, and H 2 O (Figures S13 and S14).
[CH 3 NH 3 ][Co(HCOO) 3 ] is stable up to 215 • C. This Co 2+ compound degrades in a smooth two-step process to Co (71.1%) mixed with CoO (28.9%) ( Figure S6). CO 2 is the main gas observed from the beginning and up to the end of the degradation process (steps 1 and 2). During the first degradation step CH 3 NH 2 and H 2 O are observed, while CH 3 NH 2 , H 2 O, and CO are detected during the second step (Figures S15 and S16). [

CH 3 NH 3 ][Ni(HCOO) 3 ] and [CH 3 NH 3 ][Cu(HCOO) 3 ]
show a similar single-step degradation process. The Ni 2+ material is stable up to 230 • C and the Cu 2+ one up to 150 • C. The two compounds decompose to pure Ni ( Figure S4) and Cu ( Figure S5) by releasing CO 2 together with CH 3 NH 2 and H 2 O during the whole degradation processes (Figures 2b(iii), S17 and S18). Table 2 contains a list of the volatile species observed at the different decomposition steps of the [CH 3 NH 3 ][M(HCOO) 3 ] compounds. The gases from the second step are in agreement with gases reported by Shishido and Masuda for the anhydrous formates [32].   Table S2 shows a summary of the thermal decomposition results for these compounds.
Solids2021, 1, FOR PEER REVIEW 5 °C. The two compounds decompose to pure Ni ( Figure S4) and Cu ( Figure S5) by releasing CO2 together with CH3NH2 and H2O during the whole degradation processes (Figures 2b-iii, S17 and S18). Table 2 Table S2 shows a summary of the thermal decomposition results for these compounds. [AH][Cd(HCOO)3] compounds, with AH + = NH4 + , CH3NH3 + , and (CH3)2NH2 + guest cations, degrade in a four-step process. According to the IR spectra of the released gases ( Figures S20-S24), the amine is released during the first decomposition step together with a formate group of the framework as HCOOH and/or CO2.

Discussion
The obtained results have allowed us to determine the decomposition mechanism for [CH 3 3 ] is summarized in Scheme 1. In these compounds, the decomposition starts with the elimination of the CH 3 NH 3 + cations which are linked to the anionic [M(HCOO) 3 ] − framework presumably through hydrogen bonds. The framework is no longer stable and one of the formate HCOOgroups is also released. By H-transfer between the HCOOand CH 3 NH 3 + species, CH 3 NH 2 and HCOOH are obtained as gas products. Formic acid (HCOOH) can decompose following two different possible mechanisms, decarboxylation (CO 2 + H 2 ) and dehydration (CO + H 2 O) [34]. Very interestingly, we have determined that there is a direct correlation/link between the observed route and the standard reduction potential of the metal cation of the framework (Figure 4). Metal cations with a standard reduction potential of −0.27 V or higher (−0.27 V for Co 2+ , −0.23 V for Ni 2+ , and 0.34 V for Cu 2+ ) are reduced to pure metal species, whereas metal cations with a standard reduction potential lower than −0.27 V (−0.40 V for Cd 2+ , −0.76 V for Zn 2+ , −1.18 V for Mn 2+ , and −2.37 V for Mg 2+ ) are not reduced to zero oxidation state, but tend to combine with the oxygen from the ligands to yield the metal oxides.

NH 3 ][M(HCOO) 3 ] and [AH][Cd(HCOO) 3 ] compounds. The proposed degradation for [CH 3 NH 3 ][M(HCOO)
The presence of CoO as the minor phase (28.9%) of decomposition of [CH3NH3][Co(HCOO)3] could be related with the fact that Co 2+ has a standard potential value of −0.27 V, which coincides with the limit potential that determines if the reduction product is favoured.
In addition, it has been observed that lower reduction potentials sequentially stabilize the intermediate M(HCOO)2, so that Cu 2+ -and Ni 2+ -compounds show only one degradation step, Co 2+ -and Zn 2+ -compounds show two degradation steps and finally the M(HCOO)2 species can be isolated for Mg 2+ -and Mn 2+ -materials ( Figure 4).  3 ], we have elucidated that the M 2+ cation of the covalent bonded framework plays a crucial role on the degradation pathway, which can occur thought two different routes: one corresponding to MO oxides and a second one corresponding to pure M metal. In particular, Mg 2+ , Mn 2+ , Zn 2+ , and Cd 2+ compounds degrade to the corresponding MO oxides, while Co 2+ , Ni 2+ , and Cu 2+ compounds are reduced to the respective pure M species.
Very interestingly, we have determined that there is a direct correlation/link between the observed route and the standard reduction potential of the metal cation of the framework (Figure 4). Metal cations with a standard reduction potential of −0.27 V or higher (−0.27 V for Co 2+ , −0.23 V for Ni 2+ , and 0.34 V for Cu 2+ ) are reduced to pure metal species, whereas metal cations with a standard reduction potential lower than −0.27 V (−0.40 V for Cd 2+ , −0.76 V for Zn 2+ , −1.18 V for Mn 2+ , and −2.37 V for Mg 2+ ) are not reduced to zero oxidation state, but tend to combine with the oxygen from the ligands to yield the metal oxides. It is worth noting that these results are in agreement with the ones observed by Das for porous coordination polymers (MOFs) containing elongated organic linkers [35], whose decomposition gives rise to porous M@C nanocomposites, and porous MO@C nanocomposites [12]. In a subsequent review, Gascon and co-workers, rationalized that in those MOFs yielding metal species by thermolysis, the carbon matrix (obtained as by-product) acts as the reducing agent (through the C→CO2 conversion) [12]. In contrast, for the compounds presented here, In an inert atmosphere and above the decomposition temperature, the Gibbs free energy/reduction potential represented on the Ellingham diagram predicts the metallic species/metal oxides which are favoured.
Gascon and co-workers explained the tendency of a metal cation in a MOF to be reduced by using the Ellingham diagram ( Figure 5): if the metal lies below the C/CO2 line, the stable product is the oxide [12]. The presence of CoO as the minor phase (28.9%) of decomposition of [CH 3 NH 3 ] [Co(HCOO) 3 ] could be related with the fact that Co 2+ has a standard potential value of −0.27 V, which coincides with the limit potential that determines if the reduction product is favoured.
In addition, it has been observed that lower reduction potentials sequentially stabilize the intermediate M(HCOO) 2 , so that Cu 2+ -and Ni 2+ -compounds show only one degradation step, Co 2+ -and Zn 2+ -compounds show two degradation steps and finally the M(HCOO) 2 species can be isolated for Mg 2+ -and Mn 2+ -materials ( Figure 4).
It is worth noting that these results are in agreement with the ones observed by Das for porous coordination polymers (MOFs) containing elongated organic linkers [35], whose decomposition gives rise to porous M@C nanocomposites, and porous MO@C nanocomposites [12]. In a subsequent review, Gascon and co-workers, rationalized that in those MOFs yielding metal species by thermolysis, the carbon matrix (obtained as byproduct) acts as the reducing agent (through the C→CO 2 conversion) [12]. In contrast, for the compounds presented here, [AH][M(HCOO) 3 ], no carbon matrix is produced during the decomposition. We propose that for [CH 3  It is worth noting that these results are in agreement with the ones observed by Das for porous coordination polymers (MOFs) containing elongated organic linkers [35], whose decomposition gives rise to porous M@C nanocomposites, and porous MO@C nanocomposites [12]. In a subsequent review, Gascon and co-workers, rationalized that in those MOFs yielding metal species by thermolysis, the carbon matrix (obtained as by-product) acts as the reducing agent (through the C→CO2 conversion) [12]. In contrast, for the compounds presented here, [ In an inert atmosphere and above the decomposition temperature, the Gibbs free energy/reduction potential represented on the Ellingham diagram predicts the metallic species/metal oxides which are favoured.
Gascon and co-workers explained the tendency of a metal cation in a MOF to be reduced by using the Ellingham diagram ( Figure 5): if the metal lies below the C/CO2 line, the stable product is the oxide [12]. In an inert atmosphere and above the decomposition temperature, the Gibbs free energy/reduction potential represented on the Ellingham diagram predicts the metallic species/metal oxides which are favoured.
Gascon and co-workers explained the tendency of a metal cation in a MOF to be reduced by using the Ellingham diagram ( Figure 5): if the metal lies below the C/CO 2 line, the stable product is the oxide [12].
For the [AH][M(HCOO)3] perovskites, a similar approach can be implemented: metals such as Mg, Mn, and Zn are oxidized because their ΔG line lies below the CO/CO2 line, while Ni and Co are stable because their ΔG lines lie above it. This result is in accordance with the explanation gave by Baraldi for the hydrated metal formates [27].   For the [AH][M(HCOO) 3 ] perovskites, a similar approach can be implemented: metals such as Mg, Mn, and Zn are oxidized because their ∆G line lies below the CO/CO 2 line, while Ni and Co are stable because their ∆G lines lie above it. This result is in accordance with the explanation gave by Baraldi for the hydrated metal formates [27].

Synthesis
[AH][Cd(HCOO) 3 ] compounds containing different amine cations (AH + ) inside their pores (AH + = NH 4 + , CH 3 NH 3 + , (CH 3 ) 2 NH 2 + ), degrade all in a four-step process to CdO as the main residue. The decomposition of the Cd(HCOO) 2 compound goes through a carbonate intermediate (Scheme 3) as it was reported in detail by Małecka and coworkers [33]. This pathway was here confirmed by the presence of CdCO 3 traces in the final residue of decomposition of [CH 3 NH 3 ][Cd(HCOO) 3 ] ( Figure S19). Solids2021, 1, FOR PEER REVIEW 8 For the [AH][M(HCOO)3] perovskites, a similar approach can be implemented: metals such as Mg, Mn, and Zn are oxidized because their ΔG line lies below the CO/CO2 line, while Ni and Co are stable because their ΔG lines lie above it. This result is in accordance with the explanation gave by Baraldi for the hydrated metal formates [27].    3 ] perovskite like formates, or at least, that it plays a less important role than the metal (M 2+ ).  3 ] materials, where AH + = NH 4 + , CH 3 NH 3 + , and (CH 3 ) 2 NH 2 + , were synthesized under solvothermal conditions followed by the slow evaporation of the mother liquor [21], or by slow diffusion method [36]. Table S1 of supplementary materials summarizes the list and quality of the starting reagents.

Materials and Methods
Thermal Studies: Thermogravimetric analyses (TGAs) were carried out in a SDT2960 TGA-DTA from TA Instruments (Waters Corporation, Milford, Massachusetts, USA). For these experiments, approximately 27 mg of each sample were heated at a rate of 5 K/min from room temperature to 600 • C using corundum crucibles under a flow of dry nitrogen. The TGA instrument was coupled to a Fourier transform infrared (FT-IR) spectrometer Bruker VECTOR22 working over the wavenumber range from 400 to 4000 cm −1 .
X-Ray Powder Diffraction: (XRPD) was performed in a Siemens D-5000 diffractometer at room temperature using CuKα radiation (λ = 1.5418 Å). XRPD patterns were refined by the Rietveld method using the software X'Pert HighScore Plus. The background was fitted with a Chebyshev function with 6 terms. The peak profiles were modelled using a pseudo-Voight function. The cell parameters, the phase fraction, and the zero shift were also refined.

Conclusions
The exothermic decomposition of [AH][M(HCOO) 3 ] perovskite-like coordination polymers under inert atmosphere takes place in a single-or multi-step process to the corresponding MO or M residue. The composition of the final residue is determined by the standard reduction potential of the metal in the framework. This means that these coordination polymers can be used as self-templates precursors, and that it is possible to predict if the metal or the metal oxide will be obtained.
In contrast with the decomposition of coordination polymers containing long C-rich organic ligands (MOFs), thermolysis of perovskite-like coordination polymers provides a clean and direct method towards the synthesis of metal oxides and metals, avoiding the production of a carbon matrix or other impurities that are not desired for specific applications.  3 ] materials, Figure S1: XRPD pattern obtained after heating the [CH 3 NH 3 ][Mg(HCOO) 3 ] sample to 600 • C. Comparison against the trace reported in the literature for MgO [37], Figure S2: XRPD pattern obtained after heating the [CH 3 NH 3 ][Mn(HCOO) 3 ] sample to 600 • C. Comparison against the trace reported in the literature for MnO [37], Figure S3: XRPD pattern obtained after heating the [CH 3 NH 3 ][Zn(HCOO) 3 ] sample to 600 • C. Comparison against the trace reported in the literature for ZnO [38], Figure S4: XRPD pattern obtained after heating the [CH 3 NH 3 ][Ni(HCOO) 3 ] sample to 600 • C. Comparison against the trace reported in the literature for Ni [39], Figure S5: XRPD pattern obtained after heating the [CH 3 NH 3 ][Cu(HCOO) 3 ] sample to 600 • C. Comparison against the trace reported in the literature for Cu [40], Figure S6: Rietveld refinement of the XRPD pattern obtained after heating the [CH 3 NH 3 ][Co(HCOO) 3 ] sample to 600 • C. Literature source for CoO [37] and Co [41]. Figure 3 ] compounds, Figure S19: Rietveld refinement of the XRPD pattern obtained after heating the [CH 3 NH 3 ][Cd(HCOO) 3 ] sample to 600 • C. Literature source for CdCO 3 [44] and CdO [45], Figure   Data Availability Statement: The datasets generated and analysed during the current study are available from the corresponding author on reasonable request.