Sodium Manganese Ferrite Water Splitting Cycle: Unravelling the Effect of Solid–Liquid Interfaces in Molten Alkali Carbonates

In this work, the Na2CO3 of the sodium manganese ferrite thermochemical cycle was substituted by different eutectic or eutectoid alkali carbonate mixtures. Substituting Na2CO3 with the eutectoid (Li0.07Na0.93)2CO3 mixture resulted in faster hydrogen production after the first cycle, shifting the hydrogen production maximum toward shorter reaction times. Thermodynamic calculations and in situ optical microscopy attributed this fact to the partial melting of the eutectoid carbonate, which helps the diffusion of the ions. Unfortunately, all the mixtures exhibit a significant loss of reversibility in terms of hydrogen production upon cycling. Among them, the nonsubstituted Na mixture exhibits the highest reversibility in terms of hydrogen production followed by the 7%Li-Na mixture, while the 50%Li-Na and Li-K-Na mixtures do not produce any hydrogen after the first cycle. The loss of reversibility is attributed to both the formation of undesired phases and sintering, the latter being more pronounced in the eutectic and eutectoid alkali carbonate mixtures, where the melting of the carbonate is predicted by thermodynamics.


S4.1. Reversibility under non-oxidative conditions
The cycled 50%Li-Na showed a significant amount of LiFeO 2 (COD 1541312), which explains the lower CO 2 capacity of the mixture.In fact, we recently found that the decarbonation of the MnFe 2 O 4 -Li 2 CO 3 mixture leads to the irreversible intercalation of Li to form LiFeO 2 1 .
Moreover, the cubic spinel phase regenerated during the final carbonation step is characterized by a lattice constant that is significantly smaller than the one observed for the Na and 7%Li The XRD of the cycled Li-K-Na mixture indicates MnFe 2 O 4 as the main phase, followed by traces of LiFeO 2 , Na 2 CO 3 and HK 2 Na(CO 3 ) 2 •H 2 O.This result is somehow in contrast with the low CO 2 capacity observed for this mixture.In fact, for the other mixtures, the reversibility loss was always reflected by the presence of significant amounts of NaFeO 2 and LiFeO 2, i.e. to the incomplete regeneration of the starting reactants.As Li-K-Na shows the lowest reversibility among the four mixtures, higher amounts of these phases would be expected.Moreover, it was previously observed that the MnFe 2 O 4 -K 2 CO 3 mixture loses CO 2 capacity upon cycling due to the formation of potassium beta ferrite (K 2 Fe 10 O 16 ) 1 .However, this phase was not detected in the cycled Li-K-Na.Rather, the presence of HK 2 Na(CO 3 ) 2 •H 2 O suggests that K tends to form a mixed Na-K carbonate that subsequently absorbed moisture before the XRD analysis was performed.Another interesting point is that the cubic MnFe 2 O 4 phase in the cycled Li-K-Na presented the same lattice shrinkage observed in the 50%Li-Na, with an average lattice parameter of 8.46 Å.This suggests that part of Li present in the mixture was incorporated in the spinel phase and explains the low amount of LiFeO 2 .

S4.2. Hydrogen production cycles under oxidative conditions
During the H 2 production experiments, the 50%Li-Na and Li-K-Na exhibited no reversibility in terms of hydrogen production, as they only produced hydrogen in the first cycle.The XRD of both mixtures excludes the formation of appreciable amounts of Na x Mn 3 O 7 (Figure 6).
Rather, the loss of reversibility of 50% Li-Na was clearly due to the formation of LiFeO 2 , which is in line with the 50% drop in the CO 2 capacity observed after the first cycle.Also, K-Li-Na showed the presence of LiFeO 2 , which explains the loss of CO 2 capacity observed by thermogravimetry.These results are in line with the thermodynamic calculations performed under CO 2 (Figure S9, Supportin Information).Indeed, at 750 ºC the high amounts of the LiFeO 2 phase formed during the WS of the 50 %Li-Na and the Li-K-Na mixtures are predicted to be stable even under CO 2 .Table S3.H 2 production for the replicates of the first and second cycle carried out for the 7%Li-Na mixture in order to assess the reproducibility of the hydrogen production measurements.

Figure S1 .Figure S3 .
Figure S1.Binary phase diagram of the Na 2 CO 3 -Li 2 CO 3 system.The compositions and transition temperatures of the eutectic (Na 0.48 Li 0.52 CO 3 ) and eutectoid (Na 0.93 Li 0.07 CO 3 ) mixtures are indicated in blue and red, respectively.

Figure S4 .
Figure S4.Equilibrium calculations of the molar fractions of the different species and phases found at different temperatures under inert atmosphere (Ar) for the Na (a), 7%Li-Na (b) and 50%Li-Na (c) and Li-K-Na (d) mixtures.The calculations were carried out using the FactSage program.

Figure S5 .
Figure S5.Thermograms (red line) obtained for the four mixtures during five H 2 production cycles in isothermal conditions at 750 ºC.The temperature measured during the experiments is also reported (black line).

Figure S6 .
Figure S6.The evolution of the H 2 hydrogen concentration in the exhaust gases for the replicates of the first cycle carried out for the 7%Li-Na mixture in order to assess the reproducibility of the hydrogen production measurements.

Figure S7 .
Figure S7.XRD analysis of the %7Li-Na mixture after the first and the second WS reaction step.

Figure S8 .
Figure S8.SEM analysis of the %7Li-Na mixture after the first (a-b) and the second (c-d) WS reaction step.Secondary electrons (a, c) and (b, d) backscattered electrons images are reported.

Table S2 .
Decomposition temperature of the MnFe 2 O 4 -alkali carbonate mixtures.The values reported in parentheses are the theoretical values obtained by the equilibrium calculations reported in Figure S4.

-
Na mixtures -i.e., 8.46 vs 8.50 Å.A shrinkage in the crystal lattice of the cubic MnFe 2 O 4 phase was already observed after cycling a MnFe 2 O 4 -Li 2 CO 3 mixture and can be attributed to partial Li atomic substitution or intercalation in the MnFe 2 O 4 lattice.