Template-free synthesis of mesoporous and amorphous transition metal phosphate materials

: We present how


Introduction:
Transition metal phosphates (TMPs) have been gaining interest as materials for catalysis, electrocatalysis and electrochemistry due to their low-cost, stability, and tunability [1][2][3][4] .For instance, it has been reported that the performance of iron, nickel, and cobalt phosphates in the oxygen evolution reaction (OER) in neutral/alkaline solutions is comparable to IrO2/RuO2.Ni and NixCo1-x phosphates demonstrate excellent supercapacitor characteristics 5,6 , zirconium and vanadium phosphates are used in organic catalysis, and silver phosphate in photocatalysis [7][8][9] .The chemical stability of TMPs over a wide range of temperatures makes them ideal candidates e.g. for fuel cell applications (optimal T-range for many fuel cells 150°C ≤ T ≤ 350°C 10 ) and electrolysis [11][12][13][14] .
Typically, TMPs are synthesized by precipitation from solution, hydrothermal methods, molten salt, microwave, or mechanochemical syntheses [15][16][17][18][19][20] Thereby, highly crystalline and/or amorphous bulk TMPs can be obtained.In particular, the amorphous state is often desired, because the absence of long-range ordering in amorphous TMPs enhances their reaction kinetics compared to the crystalline equivalents 21,22 .Even assuming similar chemical composition, crystalline and amorphous compounds may in general exhibit completely different physicochemical properties such as solubility, reactivity, or adsorption.For instance, amorphization of pharmaceutical ingredients is a well-known strategy to improve their solubility and hence biomedical activity 23 .In the case of electrocatalysis, the amorphicity of the material enhances the overall free Gibbs energy of formation which could potentially lead to significantly improved cell potential and higher open circuit voltages in amorphous electrodes compared to crystalline bulk electrodes 24 .However, the as-synthesized TMP solids, crystalline or amorphous, typically exhibit low surface area, which must be increased by engineering the porosity into the material.
In general, for all applications, a highly porous material with high surface area and pore volume significantly increases the performance majorly because of the enhanced reaction kinetics and more active sites.In this context, a stable mesoporous TMP material (pore sizes between 2 -50 nm, IUPAC 25 ) and also amorphous TMP. could combine all advantages for the applicational use due to a high number of active metal sites, significantly enhanced mass transfer through the compound and a more negative Gibbs energy of formation compared to the crystalline analogue.Typically, mesoporosity is achieved via template-assisted synthesis routes.Despite the huge potential, template-based synthesis of TMPs is highly challenging and expensive limiting the currently available metal compositions and the (up)scalability [26][27][28] .A major issue is the incomplete removal of template/surfactant through acid etching or high temperature calcination which could harm the target phosphate mesostructure and purity.These hurdles have stimulated the development of new strategies for the synthesis of TMPs, with a focus on template-free methods 29,30 .
We present a template-free synthesis method for amorphous and mesoporous TMPs through a simple thermal treatment of a crystalline precursor called M-struvite NH4MPO4•6H2O.In Mstruvite M 2+ is usually Mg 2+ , but other divalent cations such as Ni 2+ , or Co 2+ , can easily substitute Mg 2+ in the struvite structure due to similar ionic radii (where other examples considered elsewhere include among others Zn 2+ , Fe 2+ , Mn 2+ ) [31][32][33][34] .In our recent study, we showed how to precipitate and engineer Ni-and Co-struvites crystals from aqueous solutions and we explored their phase diagrams and growth mechanisms 15 .Importantly, it has been also recently shown that common Mg-struvite converts directly to amorphous regular mesostructured frameworks of high surface area (~300 m 2 /g) 30 by thermal heating.Such thermal composition method has not been used so far to prepare mesostructured TMPs containing Ni and Co.In this study we show how the mesoporous metal phosphate compounds systematically differ in their absolute surface, pore size distribution and temporal evolution from each other although they share the same original isostructural precursor struvite.A direct relationship between amorphicity and mesoporosity is demonstrated.Our findings show that the evolution of mesoporosity is not just a simple case of degassing of ammonia and water from the crystal, but in involves complex restructuring of amorphous solids.Through adjusting the reaction conditions, a high degree of control over the formed porous structure is achieved, which is relevant for a customized applicational use e.g. in electrocatalysis.
The complete synthesis procedure can be found elsewhere 15 .The different salts were dissolved in deionized water as stock solutions, which were further diluted for the experiment.For further discussion on the (Ni,Co)-mixtures, where specifically c(Ni 2+ + Co 2+ ) = 0.02 M, c(DAP) = 0.1 M, we define Ni numbers (Ni#) as they are used (see eq. Error!Reference source not found.])from 0.1 (Co-rich) to 0.9 (Ni-rich).

X-ray diffraction
Powder X-ray diffraction (PXRD, XRD) measurements from the precipitated powders were performed on a D8 Bruker Diffractometer equipped with a LYNXEYE XE-T detector.The diffraction data were collected with Cu-Kα-radiation (1.5406 Å, 40 kV and 40 mA) from 5-60° using a step size 0.015° (2Θ) and a scanning time of 0.5 s per step.All samples were prepared on silicon specimen holders, to minimise the instrumental background, which was especially important for the amorphous phases.

SEM
The scanning electron microscopy (SEM) analysis was performed on an FEI XL 30 tungsten cathode scanning electron microscope operating with 20 keV acceleration voltage and using a secondary electron (SE) detector.Prior to the analysis all samples were coated with a 30 nmthick layer of gold.
The EDS analyses and the SEM imaging of the NiCo-mixtures were performed under Zeiss Supra 40 (Zeiss, Oberkochen, Germany) equipped with a Schottky-field emitter.All samples were coated with carbon before analysis.A 5 kV beam voltage was applied for imaging of all samples.An energy-dispersive X-ray (EDS) system of type UltraDry SDD (Silicon Drift Detector) with a nominal crystal area of 100 mm 2 from Thermo Fisher Scientific (Waltham, MA, USA) was used for the qualitative and quantitative analyses of the elemental composition.For this, highly flat sample areas (10 µm x 10 µm) were selected for an EDS quantitative analysis with the software Pathfinder (v. 1. 3, 2018).The corresponding EDS spectra were collected at 15 keV and a standardless quantification was performed.Due to the rough sample surface morphology, the obtained elemental compositions are estimated to exhibit an error of <15% atom.based on multiple measurements.However, quantitative comparisons of the relative results for different samples obtained under the same experimental conditions (including sample geometries) are more accurate than the absolute values.Qualitative EDS maps were acquired over selected individual crystals indicating the spatial distribution of the elements.

TGA/DSC
Thermal gravimetric analysis (TGA) and differential scanning calorimetry (DSC) were performed in a combined mode on dry M-struvite powders using a heat flux TGA/DSC 3+ device from Mettler Toledo.All measurements were carried out under a nitrogen atmosphere (flow of 100 ml min -1 ).As a reference material for the heat flux DSC an α-Al2O3 corundum crucible was used.All samples were heated from room temperature to 850 °C with a heating rate of 10 °C min -1 .The different endo-or exothermic peaks were fitted with Pseudo-Voigt functions to determine peak onset and centroid temperature.

N2 gas sorption measurements
Nitrogen adsorption/desorption isotherms at -196°C were measured with a Micromeritics ASAP 2020 V4.04 surface area and porosity analyzer.A pre-treatment of vacuum exposure for only one hour was performed to avoid any interactions with the mesoporous sample.By applying the BET (Brunauer-Emmett-Teller) method, the specific surface area of the samples was calculated.The pore size distribution was determined by using the BJH (Barrett-Joyner-Halenda) method with the ASAP 2020 V4.04 software accompanied with the analysis equipment.

SAXS/WAXS
In-situ and time-resolved small-and wide-angle X-ray scattering (SAXS/WAXS) measurements to follow the evolution of mesoporosity were performed at the beamline BM26 37 of the European Synchrotron Radiation Facility (ESRF).Experiments were performed using a monochromatic X-ray beam at 12.4 keV (λ = 1.00 Å) aligned with a capillary (borosilicate glass, Glass Müller, Berlin ID = 1.498 mm) of the furnace LINKAM DCS 600.All the samples and backgrounds were measured with this setup.
Scattered intensities were collected at small-angles with a Dectris Pilatus 1M, and at wideangles with Dectris Pilatus 300k.Sample transmission factors were measured by means of a photodiode installed in the beam-stop of the SAXS detector, and an ionization chamber used as a primary beam monitor.The sample-to-detector distances allowed for a q-range of 0.11 < q < 7.36 nm -1 in SAXS, and of 5.76 < q < 52.64 nm -1 in WAXS.The scattering q-range at smallangles was calibrated against silver behenate and the intensity was calibrated to absolute units against water.The q-range in WAXS was calibrated with a silicon powder reference material (NIST SRM640f).The recorded 2D scattering data from the SAXS and WAXS detectors were reduced using the software package BUBBLE v. 4.2 38 .The reduction steps involved data normalizations and instrumental corrections, correction for transmission, subtraction of a background, followed by integration to 1D scattering curves.Since the covered q-ranges of the SAXS and WAXS overlapped between 5.76 and 7.36, the WAXS data can be normalized on the SAXS data and scaled properly.
In a typical experiment, a capillary filled with a crystalline powder was put into a LINKAM DCS 600 furnace and heated at 20 K/min to 363.15 K.The scattering patterns were measured continuously at 2 fps.SAXS scattering curves were used to extract pore size-distributions as a function of time.
Ex-situ SAXS/WAXS measurements were conducted using the MOUSE instrument 39 .In a microfocus X-ray tube attached with parallelizing and monochromatising multilayer optics, Xrays of a distinct wavelength of Cu Ka (λ = 0.15406 nm) were generated.An in-vacuum Dectris Pilatus 1M, was used to measure the scattered signal.The sample-to-detector distances allowed for a q-range of 0.015 < q < 40 nm -1 in a combined SAXS/WAXS approach.The synthesized powders were sealed on a sample holder from both sides with Scotch© tape and hold into the beam.The obtained data was corrected according to the procedure in the literature 40 .

Scattering Model
The measured in-and ex-situ SAXS data were fitted by McSAS v. 1.3 software 41 under the simplified assumption that the pores could be modelled as hard spheres 42,43 .A mutual interference of the developing pores is assumed to occur in agreement with other SAXS studies of phosphate materials 30,44,45 .Considering this and other factors, the hard sphere model output the most meaningful and accurate fitting data.A full evaluation of the fitting model can be found in the SI.WAXS was used to monitor crystalline-to-amorphous transitions.
The hard sphere structure factor can be described by the Percus-Yevick structure factor 43 : where the variable  represents the local volume fraction of hard spheres, RHS is the hardsphere radius and G(2qRHS) is a dependent function on q,  and RHS (ESI: Equation S1).Thus,  and RHS remain as the only independent parameters in this expression.The variable , although named as the local volume fraction, expresses the packing of the hardspheres/particles at a distance of 2RHS.In the fitting procedure,  is kept constant, and was estimated based on the total pore volume from the gas adsorption measurements (see below).The size distributions of hard sphere pore radii/sizes RHS were calculated from the fits to the SAXS data using the Monte Carlo method.

XAS
XAS (X-ray absorption spectroscopy) measurements both at the near edge structure (XANES) and extended fine structure (EXAFS) were performed at the BAMline (BESSY-II, Helmholtz Centre Berlin for Materials and Energy Berlin, Germany) 46 on the solid powders to determine and compare the coordination environment of the different phases.The beam was monochromatized using a double-crystal monochromator (DCM) with a Si crystallographic orientation of [111].The size of the beam was 3 mm (h) x 1 mm (v).The measurements were performed at Co-K edge (7709 eV) and Ni-K edge (8333 eV) in transmission geometry, with two ionization chambers as detectors.The excitation energy was varied from 7606 eV to 8678 eV for Co and 8230 eV to 9302 eV for Ni, with varying energy steps.For the pre-edge region, the energy was varied in 10 eV steps; for the region around the edge, energy was tuned first in 0.5 eV steps, then in 1 eV steps and in the EXAFS region with a constant step in the k-space of 0.04 Å -1 .The associated uncertainties were experimentally determined by measuring the cobalt and nickel metal foils, 10 times each.A value of ±0.3 eV was obtained for both systems.
For the measurement, the samples were mixed with boron nitride, placed in polycarbonate hole plates with a thickness of 1 mm and sealed with a polyimide tape (Kapton) on both sides.Before collecting the sample spectra, a cobalt or nickel foil was used as a reference for the respective K edges.The relative energies of the spectra were calibrated to the first inflection point from the first derivative of the cobalt/nickel metal absorption edge.
The XAS data were processed by using ATHENA (for normalization and background removal) and ARTEMIS (to fit known models to the experimental EXAFS data).These two software programs belong to the main package IFEFFIT (v.1.2.11) 47 .The respective Ni-phosphates (Nistruvite NH4NiPO4•6H2O, Ni-dittmarite NH4NiPO4•H2O) phases were used as model structures, the scattering paths for first and second coordination sphere were theoretically modelled and fitted to the measured spectra.M-struvite contains water and ammonia in its structure, and therefore potential phase transitions should occur at elevated temperatures as a result of a dehydration and degassing processes.Based on this assumption, to evaluate such thermally-induced phase transitions, a combined TGA/DSC analysis was performed on the dry crystalline struvite powders at T ≤ 300°C.

Crystalline phase composition and their amorphization
Based on those characterizations, we see that the thermal decomposition behavior of the different M-struvites vary significantly.While Co-struvite transformed in a course of a quick, phase transition to Co-dittmarite with an Tonset,1 = 72 °C, Ni-and Mg-struvite evaporated their volatile compounds at slightly higher temperatures at Tonset,1 = 83°C (Ni) and Tonset,1 = 88°C (Mg) (ESI: Table S1).Interestingly, Co-and Co-rich NixCo1-x-struvite show multiple phase transitions indicated by multiple events in DSC (1 st , 2 nd , 3 rd thermal event in Figure 2B).The resulting order of the 1 st M-struvite decomposition is the following: TCo <TNi ≈ TMg (Figure 2B, Table S2).
Considering the mass-loss analysis associated with the DSC (ESI: Fig. S1) the first thermal event corresponds in all compounds to a major phase of degassing with a release of five water molecules and minorly ammonia.Therefore, since all struvite compounds show their main degassing event at around 90°C, the phase composition of heated M-struvite samples were measured in a time-series t = 10 min, 20 min, 40 min, 60 min, 90 min, 120 min, 150 min, and 1d at this temperature.Due to the broad T-interval of the 1 st thermal event, higher temperatures T =120°C and 150°C were also evaluated for the different M-struvite systems to observe potential differences (Figure 2).At temperatures below T < 200°C, Mg-and Ni-struvite were transforming to amorphous phases in dry air, beginning majorly after t = 40 min and completed at t = 120 min, while Costruvite recrystallized quickly after 20 min to Co-dittmarite (Figure 2, ESI: Figure S2).In fact, the phase transformations in transition metals would manifest themselves, just through simple color changes (photographs of powder in circles, in Figure 2B, Figure 2C and Figure 2D).When heated, Ni-and Co-phosphates exhibited a color change from green to yellow (Ni) and from pink to purple/violet (Co).In the Ni-system during degassing of volatile components a transitional stability of Ni3(PO4)2•8H2O had been observed which disappeared again upon further heating.At high temperatures, the amorphous Ni-phases showed a high stability up to 500°C while in the Co-system multiple crystalline-crystalline phase transitions were observed (ESI: Figure S2).
In the case of single-element struvites, all three considered compounds exhibited different thermally-induced phase transitions despite sharing the same crystal structure.Importantly Ni-struvite turned out to convert into a stable amorphous phase, and Co-struvite turned into crystalline Co-dittmarite.The question arose how would mixed NixCo1-x-struvites evolve under similar thermal conditions in particular that the mixed transition metal struvites had different thermal profiles (Figure 1D, ESI: Figure S1) from their mono-cationic analogues.In the first step, we confirmed that for a broad range of compositions we indeed synthesized phase-pure NixCo1-x-struvites with x = 0-1 (ESI: Figure S1).Based on EDS mappings, a complete solid solution of NixCo1-x-struvite with homogenous distribution of both metals regardless of the Ni# could be identified at least on the sub-µm-scale (ESI: Figure S3, Table S2).Upon heating, the synthesized NixCo1-x struvite transformed to either amorphous or crystalline phases after the thermal treatment depending on the Ni# (Eq.[1], Figure 2D).Ni-rich NixCo1-x-struvites above Ni# >0.5 decomposed to an amorphous phase while Co-rich compositions below Ni# ≤ 0.5 transformed to M-dittmarite NH4MPO4•H2O which coexisted with an amorphous phase (Figure 2D).Thus, the materials became phase-separated, despite originating from struvite solid solutions.Heated NixCo1-x-struvites with high Co content Ni# ≥ 0.8 showed no amorphous phases.In a hydrothermal synthesis in a saturated H2O environment at 90°C, Ni-struvite transformed to Ni-dittmarite NH4NiPO4•H2O (Figure 2B).Summing the thermal history of the compounds up, Mg-, Ni-and Ni-rich (Ni# > 0.5) NixCo1-x-struvites decomposed to amorphous phases in dry conditions, whereas Co-and Co-rich (Ni# ≤ 0.5 NixCo1-x-struvite transformed to the M-dittmarite component and other crystalline phases.The occurring phase transitions, crystalline-crystalline or crystalline-amorphous are strongly dependent on the humidity, absolute temperature, heating rate and on the choice of the metal cation in the struvite precursor.We show further that the evolution of amorphicity is a key property of the material.As a stable amorphous phase prevents the structural reconfiguration of the compound and the closing of the pores, it is a necessary condition for the development of mesoporous frameworks.Our results show that thermal evolution of M-struvites is driven by the removal of volatile components from the crystal structure (Figure 1).This in turn, might make one to expect that porosity should develop in the materials over time.To test this, we characterized the evolution of porosity in our samples using gas adsorption.For the evaluation of BET surface areas and pore sizes, N2 gas adsorption-desorption measurements were performed on thermallytreated M-struvites (M = Mg, Ni, Co, and NixCo1-x).Based on these measurements, it was clear that the porosity would indeed develop in the mesoscale, but only in M-struvites which would become amorphous simultaneously (Mg, Ni, NixCo1-x).On the other hand, Co-struvite and Corich NixCo1-x-struvite showing thermally induced crystalline transformations (Fig. 2) did not develop any significant porosity.In Figure 3 we present the BET measurement results from the Ni-struvite thermal series.

Characterization of mesoporous structures
Four different stages can be defined in all obtained isotherms (inset in Figure 3A).Firstly, in stage (I) at low partial pressures (p/p 0 < 0.4) a constant adsorption of N2 can be observed, associated to the formation of a mono-to multilayer on the surface.In the intermediate stage (II) at partial pressures between 0.4 < p/p 0 < 0.55 progressive multilayer condensation and filling of pores smaller size with N2 gas occurs probably.The third (III) and fourth (IV) stages show the absence of significant amounts of macropores as a plateau region occurs at high relative pressures (0.55 < p/p 0 < 0.92) followed by a steep increase due to bulk condensation of N2 at the highest partial pressures (p/p 0 > 0.92).Based on the slope changes in the observed isotherms in the stage (II) and the BJH pore size distribution (Figure 3A), heated Ni-struvite demonstrated type IV isotherms/hysteresis loops indicative of significant amount of mesopores.The surface area and the cumulative pore volume increased from 3.61(1) m²/g, t = 10 min heating time to 90.9(5) m²/g, t = 150 min heating time while the average desorption BJH pore size decreased from 12.3 nm to 5.2 nm with higher heating time (Figure 3B).After extended heating times t = 120-150 min the BET surface area remained nearly the same.As a reference and for comparison of the BET data, Mg-struvite was heated at T = 90°C for one day and measured (last data point in Figure 3B).At a BET surface area of 240.5 (9) m²/g, it demonstrated a significantly higher value, accompanied by an average desorption BJH pore size of 4.2 nm compared to thermally treated Ni-struvite (Figure 3B, ESI: Table S3).The adsorption/desorption BJH pore size distribution of the Ni t-series clearly showed a shift to lower average pore sizes for longer heating times.Looking at higher temperatures T = 120°C, 150°C (ESI: Figure S4, S5, Table S3), the Mg and Ni system showed systematically lower BET surface area (Mg: 195.9(2) m²/g for T =120°C, 138.2(2) m²/g for T =150°C; Ni: 26.9(1) for T = 120°C and 54.5(1) m²/g for T = 150°C).
Clearly, a direct correlation between the amorphicity and mesoporosity of the material exists as only the amorphous compounds were also mesoporous.This correlation starkly demonstrated itself in heated NixCo1-x-struvite mixtures, where Ni# ≥ 50% all showed BET surface areas of around 80-90 m²/g and average pore sizes of 4 nm with type IV hysteresis loops similar to Ni-struvite (ESI: Figure S6, Table S4).Below Ni# ≤ 50%, the BET surface area decreased to 20-30 m²/g while the average pore size increased rapidly from 3-4 nm to tens of nanometers.The heated Co-rich NixCo1-x-struvite and pure Co-struvite demonstrated no significant mesoporosity and low surface area (1.1 m²/g), indicated by type II isothermes (ESI: Figure S6, Table S4).All summarized BET data, additional adsorption/desorption pore size distributions and isothermes of M-struvite can be found in the ESI (ESI: Figure S5 and S6, Table S3 and S4).
However, the presented N2 BET data above lack of considering isolated pores, chemisorption or a change in the pore structure during the measurement due to its ex-situ nature.The associated BJH calculations are based on arguable assumptions i.e. the occurrence of only disconnected parallel cylindrical pores.Therefore, to gain a complementary information of the pore structure, in-situ SAXS/WAXS measurements of isothermally heated M-struvite at T = 90°C were performed allowing a view on the temporal evolution of pores during the thermal treatment ( Figure 4).Here, we want to evaluate how the M-struvites differ so much among each other although they share the same precursor crystal structure.Characteristic steps in the in-situ SAXS data of the mesoporous Mg-and Ni-system were selected and fitted to determine the pore size distribution.By fitting the SAXS curves of selected time steps, a quantitative comparison of the calculated pore size distribution and a temporal evolution of the mesoporous framework can be obtained.All details concerning the fitting procedure can be found in Methods: SAXS/WAXS (ESI: Supplementary Note 1, Equation S1, S2).All SAXS fits and pore size distributions of selected time steps of heated Mg-, Ni-and NixCo1-x-struvites can be found in the ESI (ESI: Figure S7-S13, Table S5 and S6).
Looking at the pore size distribution of thermally treated M-struvite, the mesoporous frameworks of heated Mg-and Ni-struvite differ significantly from each other.Considering the WAXS signal, the Mg-and Ni-system evolve pure amorphous phases during thermal treatment while Co-struvite transforms into crystalline Co-dittmarite ( Figure 4).Again, this is the reason for the absence of a stable mesoporosity in thermally treated Co-struvite.Therefore, only the Mg and Ni SAXS patterns were fitted with the model describing the mesoporosity (see Methods, ESI: Supplementary Note 1).
Upon heating, Mg-struvite exhibits initially a multimodal distribution, a mesoporous population at > 38 nm pore diameter and a meso-to microporous population below 20 nm, respectively at t = 400 s (Figure 5A, Figure 5C, ESI: Figure S8).With progressive heating after t = 2000 s, the mesoporous population of large pore size diameter of around 38 nm disappears while the population of mesopores of small diameter below < 20 nm shift significantly from 13±3 nm to 8±1 nm in size.
At the final stage at t = 5500 s, the mesoporous population becomes narrower with time, as is indicated by a decreasing width of size distributions.The final maxima (1 st , 2 nd and 3 rd maximum) point out to an average and uniform pore size (ESI: Figure S7, S8, S11).Although the microporous population evolves quickly already at t = 398 s, its average pore size remains similar at around 1 nm (1.3±0.3 nm at t = 400 s and 1.0±0.3nm at t = 5500 s).In the Ni-system, beginning with a broad multimodal population in the micro-and mesoporous range at t = 2200 s with a mean pore diameter of 22±6 nm, a bimodal population of micro-and mesopores establishes itself after t = 5500 (Figure 5B, 5D, ESI: Figure S9, S10, S12).Here, the final mesoporous population at t = 5500 s of heated Ni-struvite with a mean pore diameter of 14±4 nm is much broader and larger than the narrow mesoporous population of Mg-struvite of 8±1 nm.In addition, Mg-struvite exhibits a higher amount of mesopores in relation to the amount of micropores compared to Ni-struvite (approximately for Mg: φmicro ≈ 83% to φmeso ≈ 17% and for Ni: φmicro ≈ 60% to φmeso ≈ 40%, ESI: Table S5).Based on the relative volume fraction of meso-and micropores from SAXS, the final heated amorphous magnesium phosphate phase exhibits 2 -2.5 higher amount of mesopores than the final heated Ni-PO4 phase (Mg: φmeso= 40% vs. Ni: φmeso= 17 %).Although a pure Co-struvite precursor did notevolve mesopores in the final stage, a detailed view on in-situ data could reveal transitional pores formed in the heating process, which eventually disappeared.In fact, we could observe short-lived micro-to mesopores of an average size of 2-3 nm at around t = 150 s ( Figure 4C).They evolved during a thermal treatment due to the simultaneous decomposition/degassing of Co-struvite.After this time, all released gases, mainly H2O, had evaporated and Co-dittmarite recrystallized immediately, resulting in a progressive decline of the mesopore signal until its disappearance at around t = 700 s.This points out again to a sharp crystalline-crystalline phase transition between Co-struvite and Co-dittmarite.Interestingly, permanent well defined narrow peaks at 5.1 and 5.5 nm -1 developed rapidly already within the first 150 s and remained present in the scattering pattern.Such peaks did not appear either in the Mg-or in the Ni-system ( Figure 4C).These narrow and sharp reflexes are probably associated with the crystal structure changes rather than with an actual evolving microporosity.In the context of stark differences between Ni-and Co-struvite, we also investigated the evolution of the NixCo1-x-struvites of various metal contents.Namely, for comparison against the pure Ni-and Co-endmembers, exsitu combined SAXS/WAXS measurements were performed on three isothermally heated NixCo1-x-struvite solid solutions with Ni# = 0.8, Ni# = 0.5 and Ni# = 0.2 and fitted with the same model (ESI: Figure S13, Table S6).As indicated by the BET surface measurements, the amount of mesopores decreased with lowering Ni content.Based on SAXS/WAXS it was further evident that also the pore structure itself changed with different Ni# in the material (ESI: Figure S13, Table S6).The Ni-rich composition with Ni# = 0.8 exhibited majorly a micro-to mesoporous population with minor macropores like the Ni endmember.In contrast, the intermediate NixCo1-x-mixtures showed a broad pore size distribution with a significant amount of macropores and a lower amount of mesopores.Finally, the Co-rich NixCo1-x-mixture demonstrated no significant mesoporosity (ESI: Figure S13, Table S6) in correlation with the results from other other methods considered in this work.Cobalt-richer compositions of heated NixCo1-x-struvite shift the pore distribution towards larger pore sizes visible in a scattering signal at lower q and in agreement with the BET data (ESI: Figure S6, Table S4).
Reiterating in the context of our previous results, the amorphicity of the material is necessary to evolve any significant mesoporosity.

Coordination environment of amorphous phases and T-effect on crystalline phases
Based on our SAXS/WAXS results we observed a significant change in the pore size during heating (Figure 5), which is counterintuitive considering just the simple evaporation of ammonia and water.This finding suggests a potential condensation or change of the amorphous structure with extended heating times resulting in a decrease of the pore sizes.Therefore, to characterize the amorphous structure, and explain its correlation with the occurrence of the mesoporous framework, X-ray absorption spectroscopy (XAS) measurements were performed.The obtained spectra of the timeseries of heated Ni-struvite are displayed in Figure 6.They correlate the evolution of mesoporosity with the change in coordination environment.In addition, the selected NixCo1-x-struvites and their heated products at T = 90°C were measured on the Ni-and Co-edge to compare their distortion of the M 2+ -octahedron in a solid solution.Especially the intensity of the pre-peak (in the XANES region) originating from the 1s-3d transitions expresses the degree of centrosymmetry or distortion in the vicinity of the considered ions (inset of Figure 6A, ESI: Supplementary Note 2).As only 3d-4p hybridization enable parity allowed transitions the extent of those is strongly correlated with the coordination geometry.Near-ideal centrosymmetric coordinations show very low pre-peak intensities while more distorted geometries exhibit higher intensities.As Xray spectroscopy considers primarily the core shell of the absorber atom, being element specific, an analysis of crystalline and amorphous samples is possible.In general, highly crystalline phases demonstrate narrower and more intense peaks in R-space compared to amorphous phases because more ions have similar bond distances to the metal cation due to a regular periodic structure (towards a delta-function-like distribution).Conversely, as amorphous phases do not display an established long-range order, they show broader, less intense, and less-defined peaks in the R-space mostly in the low R-region due to a non-regular arrangement of the lattice ions.The XANES region (Figure 6A) clearly prove the exclusive presence of bivalent Ni 2+ and/or Co 2+ species in the materials.No indications for oxidation to Ni 3+ or Co 3+ were observed either in the pure endmember nor in the bicationic solid solutions.Quantitative pre-peak integration results App (area of pre-peak) were calculated for comparison of the distortion in the first coordination sphere (ESI: Figure S14, Table S7-S9).Ni exhibits a slightly higher degree of distortion in the amorphous phase than in the precursor Ni-struvite, which is visible through higher pre-peak intensities (Ni-struvite: Area of pre-peak App (t = 0 min) = 0.0250(1), App (t=150 min) = 0.0298(1), ESI: Figure S14, Table S7).
Looking at the pre-peak region of NixCo1-x-struvite mixtures and their heated mixtures, Ni 2+ forms a near-ideal octahedral coordination in the crystalline NixCo1-x-struvite structure and in the heated products regardless of the Ni#, as is indicated by a low pre-peak intensity (ESI: Figure S14, Table S8,S9).In contrast, Co 2+ shows a slightly higher degree of distortion with increasing Ni content in the struvite structure (precursor Ni0.8Co0.2-struviteand App = 0.032(2) vs. Ni0.2Co0.8-struviteand App = 0.025(1)).In the heated products the degree of distortion decreases with lower Ni# (heated Ni0.8Co0.2-struviteApp = 0.046(2) vs. heated Ni0.2Co0.8-struviteApp = 0.034(2)).Based on XRD results (Figure 2), NixCo1-x-struvites decomposes to an amorphous phase, as a single phase in Ni-rich compositions above Ni# > 0.5 and as a secondary phase at values of 0.2 < Ni# ≤ 0.5.Subsequently to that, M-dittmarite recrystallizes at Ni# ≥ 0.5 as a primary phase and it is the only occurring phase below Ni# <0.2 (ESI: Table S8 and S9).Hence, Ni 2+ tends to be incorporated in the amorphous phase, while Co 2+ favors the crystalline M-dittmarite structure.Surprisingly, the distortion degree of Ni 2+ barely changes in all observed structures, in the precursor M-struvite, M-dittmarite or in the amorphous phase.In contrast, Co 2+ forms a more distorted non-centrosymmetric coordination environment than Ni 2+ in the pure compounds and the distortion increases even more in the mixed NixCo1-x-ones with higher Ni content (ESI: Table S8 and S9).
Based on the comparison of the distortion (Figure 6, ESI: Figure S14, Table S7-S9) and on fitting of the spectra of the Ni time series (ESI: Figure S15 and S16, Table S10-S18, Supplementary Note 2), the structure of the amorphous Ni phases formed at 90°C showed similarities to that of Ni-dittmarite, albeit with a less pronounced long-range order.After a 40-min-long heating time, the transformation was completed, majorly visible in the pre-peak area integration results and calculated bond distances (ESI: Figure S16, Table S10-S18).

Phase transitions and correlated amorphicity/crystallinity
Major observations in the heated metal phosphate compounds were a densification/condensation of the material due to amorphization or recrystallization and the development of meso-to microporous frameworks caused by degassing of water and ammonia.Based on our results, we assume that the stability of the amorphous phase is directly linked to the preservation of the mesoporous structures.
The occurring phase transitions in heated M-struvites (i.e.amorphous-amorphous, crystallineamorphous or crystalline-crystalline) are highly dependent on humidity, temperature, heating rate, and on the metal cation in the precursor struvite.From the same isostructural M-struvite precursor, various amorphous and/or transition metal derivates with different pore structures can be obtained.The main phase transition is assumed to be the removal close to five water molecules per formula at around 70-90°C.Mg and Ni-struvite decompose to an amorphous phase at TMg ≈ TNi ≈ 85°C while Co-struvite transforms through a crystalline-crystalline phase transition to Co-dittmarite already at TCo ≈ 72°C.Probably minor amounts of ammonia are also removed in this step.Both amorphous phases of Mg and Ni exhibit a mesoporous network although their evolution and final pore structure differ significantly.This amorphous phase locally resembles that of the Ni-dittmarite structure but is less ordered, at R > 4 Å.Interestingly, similar effects of humidity were observed in iron phosphate minerals 48,49 .In contrast, Co-struvite decomposed to Co-dittmarite already at room temperature after several days when exposed to air (ESI: Figure S1D, more details in literature 15 ).When heated, this process is accelerated.
Looking at the NixCo1-x-struvite precursors, even low amounts of Ni could prevent this decomposition to the M-dittmarite compound at room temperature.The heated products of the NixCo1-x-solid solutions transform to a single amorphous phase, to single crystalline Mdittmarite or a two-phase mixture of both depending on the Ni# (ESI: Figure S1B).High Ni content is favorable for the formation of the amorphous phase while high Co content leads to the recrystallization of M-dittmarite.Polyamorphism, the occurrence of multiple phases with a different amorphous structure, cannot be completely excluded, especially in the heated NixCo1-x-mixtures, based on the XAS signal.But even if several amorphous phases are present in considerable amounts, they appear to have a similar short-range order.In this regard, the Fourier transformed spectra in R-space demonstrate no peak splitting.

Temporal formation of mesoporosity and structural reconfiguration
As the mesoporosity is directly dependent on the change of the amorphous structure, heated Mg-and Ni-struvite differ significantly in their absolute and temporal evolution of mesoporous frameworks.Primarily, isolated pores form in the course of removal of mainly water and, to a minor extent, ammonia.In the initial degassing step, barely any reconfiguration occurs in the compound, which results in a polydisperse population of larger mesopores (> 10 nm for Mg, > 20 nm for Ni) and micropores.Upon further heating, majorly the building units, PO4 and MO6 begin probably to bind to each other causing a potential shrinkage of the pores (disappearance of larger mesopores), as impled by the structures in XAS.Minorly the isolated ammonium unit decomposes partly to ammonia, and such released protons could form in combination with oxygen free water as a condensation product.As the bond distances were just evaluated exsitu with EXAFS, the dynamic reconfiguration process of the compound is likely to be more far more complex.At final t = 5500 s, Mg-and Ni-struvite exhibit bimodal distributions of pores: a single microporous population of around 1 nm, one other mesoporous population at around 8±1 nm for Mg and 14±4 nm for Ni.We assume that the isolated micropores begin to partly coalesce to mesopores 5-15 nm in size, and we speculate that channel/worm-like, elongated geometries (5-15 nm) of pores also evolve.Such a geometry would explain the pore interconnectivity.The shape of pores has to be further confirmed, and this geometry is not explicitly included in our scattering model due to its mathematical constraints.However, in Mg-struvite these channel-like/elongated pores, if actually present, seem to form fast and gradually increase with higher heating time (progressive increasing intensity of main peak at around q = 0.7-0.8nm -1 ).Therefore, the relative volume fraction of mesopores is much higher than the one of micropores (approximately for Mg: φmicropores ≈ 83% to φmesopores ≈ 17%).Uniform and narrow pore size distributions of 8±1 nm and 1.0±0.5 nm develop through time.
In contrast, Ni-struvite decomposes much slower than Mg-struvite to an amorphous phase and preserves a significant fraction of isolated micropores.Therefore, we assume that Nistruvite forms only a limited amount of interconnected pore frameworks which would explain the significant lower BET surface area compared to a heated Mg-struvite (Mg: 240 m²/g vs. Ni: 90 m²/g) and the higher amount of micropores compared to mesopores (approximately for Ni: φmicropores ≈ 60% to φmesopores ≈ 40%).The ratio of the absolute surface areas, pore volumes of heated Mg-(240 m²/g, 0.31 cm 3 g -1 ) and Ni-struvite (90 m²/g, 0.13 cm 3 g -1 ) and the average pore sizes obtained from BET N2 gas sorption are mostly in the same order of magnitude (Figure 3Figure 5).Small discrepancies of few nm in the calculated BJH pore sizes and the SAXS pore sizes could be originated in the BJH method as it always includes desorption artefacts.Within the NixCo1-x-struvite mixtures, a higher Co 2+ content leads to a stronger polydispersity and higher average pore size (ESI: Figure S13, Table S6).Coincidentally, the absolute BET surface area and pore volume decreases significantly at Ni# <0.5.After thermal treatment, only a limited amount of Co 2+ can be probably incorporated into an amorphous phase, since most of Co tends to form its own crystalline phase, namely Co-dittmarite (ESI: Figure S1).This results in a binary phase mixture: amorphous and crystalline.Based on the EXAFS analysis, the structure of the amorphous phases resembles the local structure of crystalline Ni-dittmarite NH4NiPO4•H2O with a less pronounced long-range order.An indicator for this assumption is the linking of the phosphate units with the metal octahedron associated with the characteristic appearance of R-distances of around 2.7 Å (Figure 6B).Interestingly, Co 2+ exhibits a significant dependency in the NixCo1-x-mixtures of its distortion in the octahedral coordination on the Ni# in the solid solution precursors and heated products while Ni 2+ shows barely changes in comparison.A high Ni content in the precursor struvite and heated products leads to a significant distortion of Co 2+ in its octahedral coordination.Ni 2+ shows barely any correlation with increasing Co content.Co 2+ with a d 7 electron configuration demonstrates a strong Jahn-Teller effect in high-spin octahedral coordination (H2O weak field ligand) due to more unpaired electrons compared to Ni 2+ as a d 8 -ion.Based on the pre-peak integration results (ESI: Figure S14, Table S7-S9), a mutual influence of both 3d metals in the shared solid solution on the coordination sphere is assumed, whether in the precursor M-struvite, in the amorphous phase or M-dittmarite structure.The significant correlation of the distortion degree of Co 2+ in NixCo1-x-mixtures is in agreement with electron spin resonance spectroscopy 50 .As stated also in our previous study 15 , we assume a strong dependency between the stabilities of the different transition metal struvites and their electronical d-configuration.
Regardless of their individual differences in pore size evolution and absolute surface area, heated Mg-, Ni-and Ni-rich NixCo1-x-struvites preserve mesoporous frameworks.On the other hand, Co-and Co-rich NixCo1-xstruvite eliminate any potential mesoporous structures completely due their fast reconfiguration to crystalline M-dittmarite NH4MPO4•H2O.The amorphization to a stable non-crystalline phase during degassing seems to be necessary to obtain any mesoporous system and probably prevents the collapse of mesopores due to the absence of a crystalline reconfiguration.The high stability of an occurring amorphous phase could have stopped further diffusion of volatiles out of the compound.Since a particular mesoporous framework configuration seems to be directly linked to an element specific amorphous phase, and cannot be predicted a priori from the precursor material, we suggest that the small-angle scattering signal (and the resulting pore size distribution) itself might constitute a fingerprint of a specific amorphous phase.Porous amorphous phases could be classified in such a way.While this idea would have to be fully verified, an indicator for this could be observed in the major differences in the mesoporosity of heated Mg-and Ni-struvite, which in turn are directly connected to the stability and composition of the amorphous phases.By adjusting the reaction conditions (Ni/Co ratio in precursor M-struvite, temperature, humidity) mesoporous materials with different pore structures can be obtained.In such a way, an engineering of the mesoporous framework can be performed for a distinct applicational use e.g. in electrocatalysis.The as-obtained surface area values of the mesoporous struvite-based phosphate materials are comparable to other mesoporous phosphates and their related phosphonates synthesized through much more complex chemical procedures [51][52][53][54][55] .Transition metal phosphates are frequently used in electrodes or supercapacitors 3,35,36,56 , the same mesoporous analogues would potentially exhibit a higher performance due to enhanced reactivity.A coated or doped amorphous mesoporous transitional metal phosphate might potentially improve properties by exhibiting e.g.high electrocatalytical activity.However, this aspect needs to be further tested.Shown in a previous study 15 , M-struvite could be precipitated out of industrial or mining waste waters.Due to its fast precipitation and low solubility product Ksp, it is also occurring as a biomineral in specialized bacteria species 57,58 .Through a simple thermal treatment, these compounds containing critical raw materials such as phosphate or cobalt could be re-and upcycled for industrial purposes.

Conclusions
Through a simple thermal treatment of precursor M-phosphate compounds such as Mstruvites NH4MPO4•6H2O (M = Mg, Ni, Ni-rich, NixCo1-x), simultaneously mesoporous and amorphous metal phosphate materials were synthesized.The temporal evolution of mesopores responsible for the high surface area was accompanied by a crystalline-amorphous phase transformation during degassing of volatile compounds, NH3 and H2O.Such thermally induced phase transformation was linked to the response in the coordination environment of Ni 2+ and Co 2+ and their structural reconfiguration in the solid compound.Here, Ni 2+ demonstrated a near-ideal centrosymmetric octahedral coordination sphere compared to a distorted one of Co 2+ .Mg-, Ni-and Ni-rich NiXCo1-x-struvites form amorphous phases during thermal treatment and evolve stable mesoporous frameworks.At first, near spherical micropores form and coalesce later to possibly more elongated mesopores.The condensation degree to mesoporous structures associated with surface area, pore size and pore shape is directly dependent on the complex amorphous structure.Consequently, no significant formation of mesopores occurred in Co-and Co-rich NixCo1-x-struvite after thermal treatment as these compounds exhibit several crystalline-crystalline phase transitions.

Figure 1 :
Figure 1: (A) crystal structure of precursor M-struvite, color of elements: hydrogen = white, nitrogen = dark blue, oxygen = red, P = orange, Metal cation = light blue (B) DSC graphs for pure M-struvite (M = Mg, Ni, and Co) and two NixCo1-x-struvite mixtures (x = 0.3 and 0.7) endothermic events due to degassing of volatile components.

Figure 5 :
Figure 5: in-situ SAXS data and fits of isothermally treated (A) Mg-and (B) Ni-struvite at T = 90°C with a fixed local volume fraction  = 40 %; Resulting pore size distribution of heated (A) Mg-and (B) Ni-struvite with relative and cumulative (final time) volume fraction of pores.

Figure 6 :
Figure 6: (A) normalized XAS spectra of crystalline precursor Ni-struvite (NIS), time series of thermally treated Ni-struvite and hydrothermally synthesized crystalline Ni-dittmarite (NID) compared with multiple M 2+ (M = Ni 2+ , Co 2+ ) standards (only the Ni-foil is shown for clarity) with detailed view of the pre-peak region; (B) Fourier-transformed Ni spectra plotted in R-space; (C) (010) crystal plane of Ni-struvite structure spacegroup Pmn21 with black dashed line of potential near spherical micropores inside the structure and amorphous phase.

Figure 7 :
Figure 7: Summary sketch including the four major processes in the samples: (a) the amorphization of the crystalline Mstruvite , (b) the formation of isolated micropores and the connection to elongated worm-like mesoporous frameworks, (c) the densification and of the material majorly through linking the PO4 -units with the MO6 -units due to loss of coordinating water and minorly of ammonia and (d) a crystalline-crystalline phase transition of X-shaped Co-struvite to 2D plate Codittmarite with no significant mesoporosity.In summary, we observed three major effects in M-struvite-derived materials which are visualized in Figure7: (a) The amorphization and densification of Mg-and Ni-struvite majorly through linking of the PO4 -units with the MO6 -units due to loss of coordinating H2O together with minor NH4,and (b) their formation of isolated micropores and the connection to larger eventually elongated/worm-like mesoporous frameworks.In (c) a complete reconfiguration of X-shaped Co-struvite occurs to crystalline 2D sheet-shaped Co-dittmarite, without the evolution of significant porosity.