IN PRESS – J. Appl. Cryst. Ce-exchange capacity of zeolite L in different cationic forms: a structural investigation

Synopsis: Zeolite L was Na-and NH 4 -exchanged and the resulting samples were characterized by structural and chemical analyses. From the perspective of the possible use of zeolite L for rare earth element recovery (in particular Ce) and to unveil the differences in the afﬁnity for Ce dictated by the presence of different counter-cations, the three samples with LTL framework type were Ce 3+ -exchanged and fully characterized.


Introduction
The scarcity of rare earth elements (REE) in some regions, such as Europe (Massari & Ruberti, 2013;Charalampides et al., 2015), and the well known balance problem (i.e. the balance between natural abundance and market request) (Binnemans et al., 2018;de Boer & Lammertsma, 2013) are pushing the scientific community to find alternative ways to supply REE.The most promising solution is to recycle them from waste, which in the last few years has become an urgent requirement for society (Jowitt et al., 2018;Rademaker et al., 2013;Balaram, 2019).Beyond the economic benefits, this process is also necessary for the ecological preservation of the natural environment (Balaram, 2019).The mining, extraction and refining of these REE? elements are in fact associated with high environmental costs (Charalampides et al., 2016;Haque et al., 2014).Despite the fact that eco-friendly politics is supportive of and encourages the development of recycling solutions, the current REE recovery percentage is still lower than 1%, and is limited to a few and very specific applications such as recovery from magnets and polishing compounds (Chakhmouradian & Wall, 2012;Royen & Fortkamp, 2016).The goal for the near future is, thus, to increase the REE recycling rate by finding an economical and convenient way to recover these elements from waste.Leached liquors obtained from waste can be a rich secondary source of REE and many methods have been tested to recover them, i.e. solvent extraction (Meshram & Abhilash, 2020;Deshmane et al., 2020;Ni'Am et al., 2020), precipitation (Zhou et al., 2018;Porvali et al., 2018;Diaz et al., 2016), ion exchange (Royen & Fortkamp, 2016).The latter can exploit various materials such as active carbon (Gad & Awwad, 2007), resins (Manos & Kanatzidis, 2016) or microporous materials (Kavun et al., 2021;Royen & Fortkamp, 2016) such as zeolites.
In this paper, we describe a preliminary test of the effectiveness of different cationic forms of zeolite L (LTL framework type) (Baerlocher et al., 2007) for cerium recovery.This framework type has been reported as suitable for REE recovery (Duplouy, 2016), and Ce is one of the most exploited REE in terms of volume (Binnemans et al., 2018;Charalampides et al., 2015), being present in a huge variety of technological applications [i.e. as fluid catalytic cracking (f.c.c.) catalysts, glass polishing, fluorescent lamps, NiMH batteries etc. (Meshram & Abhilash, 2020)].We examine the use of synthetic zeolite L in its as-synthesized K form and in the Naand NH 4 -exchanged ones.The three cationic forms are tested in experiments of Ce exchange from a highly concentrated solution.The obtained results contribute to our understanding of whether zeolite L is suitable for Ce exchange and recovery, and whether Ce shows a different affinity for the diverse exchanged forms.The study was carried out by a combination of different techniques, including structural refinement of high-resolution X-ray powder diffraction experiments.The exploitation of zeolites as cation exchangers cannot disregard the mechanisms involved in the exchanges and the interaction occurring among extra-framework cations, framework and H 2 O molecules.Thus, a structural study unravelling from the atomistic point of view the cation exchange mechanism and the sites involved is fundamental for the choice of a powerful protocol of exchange.Indeed, the determination, from chemical analysis, of the maximum amount of Ce incorporated into the zeolite pores would give only a partial result since it would not shed light on the dynamics of Ce adsorption.Structural analyses reported here are fundamental to determine the interactions between Ce cations and the zeolite framework, and will pave the way to further studies devoted to the investigation of zeolite selectivity towards Ce.This structural interpretation can be exploited for targets other than REE recovery -for example, Ce-exchanged zeolites are very promising catalysts for the selective catalytic reduction of N x O (van Kooten et al., 1998) or for the sterification of glycerol with acetic acid (Gautam et al., 2020).

Zeolite L
Zeolite L [K 6 Na 3 (H 2 O) 21 Al 9 Si 27 O 72 , a = 18.4,c = 7.52 A ˚, = 120 (Baerlocher et al., 2007)] (LTL framework type) (Fig. 1) is built by the linking of columns formed by the stacking of cancrinite cages and double six-membered rings (6MR), leading to the formation of a 1D porous system formed from [12-membered ring] 12MR and [eight-membered ring] 8MR channels running along the c axis.The structure is described using the P6/mmm space group.In zeolite L, three sites, named KB, KC and KD, are occupied by cations (Gigli et al., 2013).KB is placed at the centre of the cancrinite cage, KC at the centre of the elliptical 8MR channel, while KD is close to the wall of the 12MR channel.
In the present work zeolite L, purchased from the Tosoh Corporation (Japan) in its K form (from now on termed K-LTL), was initially characterized and used as starting material for the ion-exchange experiments.

Ion-exchange experiments
2.2.1.Na + and NH 4 + exchange experiments.With the purpose of obtaining different cationic forms, K-LTL was ionexchanged with two solutions, containing Na + and NH 4 + ions.The two solutions were prepared by dissolving NaNO 3 (sodium nitrate 99.5% RPE-ACS, Carlo Erba) and NH 4 Cl (ammonium chloride 99.5% BAKER ANALYZED ACS, J. T. Baker) in double-distilled water, obtaining a 0.8 M concentration of sodium and ammonium.The high concentration of the two solutions was chosen to obtain the maximum exchange of the extra-framework cations of K-LTL.
The pristine sample was sieved so as to keep only particles with a uniform mesh size (!2.5 mm), dried for 24 h at 60 C, and then put in contact with the two solutions under stirring for 5 days; the solution was changed every day to favour the ion-exchange process.The liquid/solid ratio was 16 ml/g, the temperature was set at 25 C and no buffer solution was used to control the pH to avoid interference in the exchange.Exchanged zeolites were rinsed three times with doubledistilled water using an ALC multi-speed centrifuge PK 121 at 5000 r min À1 for 10 min and then dried for 24 h at 60 C. The cationic forms obtained in this way are labelled Na-LTL and NH 4 -LTL.A second experiment was performed for sodium exchange at 80 C to evaluate the influence of temperature on the ion-exchange efficiency.The obtained sample is labelled Na-LTL-80 C.
2.2.2.Ce 3+ exchange and recovery.A Ce solution 0.48 M was prepared by dissolving Ce(NO 3 ) 3 Á6H 2 O [cerium(III) nitrate hexahydrate, REacton, 99.5% (REO), Alfa Aesar] in double-distilled water.This concentration was chosen because it allows the number of Ce ions in the solution to be equal to three times the cation content of the zeolite, hence maximizing the Ce incorporation.The pristine K-LTL and samples Na-LTL and NH 4 -LTL were put in contact with the Ce solution (liquid/solid ratio = 20 ml/g) for 5 days at room temperature with no buffer solution.The solution was changed every day to favour the exchange.The samples were rinsed three times with double-distilled water using an ALC multi-speed centrifuge PK 121 at 5000 r min À1 for 10 min and then dried for 24 h at 60 C. The obtained zeolites are labelled K-Ce-LTL, Na-Ce-LTL and NH 4 -Ce-LTL.For the last sample a further exchange, aimed at Ce recovery from the zeolite pores, was performed.NH 4 -Ce-LTL was put in contact with a solution [NH 4 + ] = 0.8 M prepared by dissolving NH 4 NO 3 (ammonium nitrate, 99%, Carlo Erba) in double-distilled water; it was left for 24 h under stirring, at room temperature.The obtained zeolite, labelled NH 4 -Ce-LTL-rev, was then rinsed as previously reported.

Analytical methods
Thermogravimetric analysis.Thermogravimetric analyses (TGA) were performed in the temperature range 25-1050 C with a ramp of 10 C min À1 under air flow, using a Seiko SSC/ 5200 thermal analyser.TGA were performed for all phases except Na-LTL-80 C, for which the water content was determined by loss on ignition (LOI).
Elemental analysis.Elemental analyses (EA) were performed for samples NH 4 -LTL and NH 4 -Ce-LTL to evaluate nitrogen content.A Flash 2000 CHNS/MAS200R instrument, equipped with a highly sensitive thermal conductivity (TCD) detector, was used.FLASH dynamic combustion (modified Dumas method) allows a temperature of 1800 C to be reached, in order to release elementary gases from the samples.
X-ray fluorescence analysis.X-ray fluorescence (XRF) analyses of samples K-LTL and Na-LTL were performed using a wavelength-dispersive X-ray fluorescence Philips PW1480.Samples were ground and pressed with boric acid to obtain powder pellets.Data were collected and corrected on the basis of standard materials (Leoni & Saitta, 1976;Franzini et al., 1975) and assuming the water content determined by LOI obtained by heating the powders in a furnace at 1000 C or by TGA.
Scanning electron microscopy with energy-dispersive spectroscopy (SEM-EDS).The chemical composition of samples exchanged with Ce (i.e.K-Ce-LTL, Na-Ce-LTL and NH 4 -Ce-LTL) and NH 4 -LTL was checked using a scanning electron microscope Nova NanoSEM 450 equipped with an X-EDS Bruker QUANTAX-200 detector.Powders were compressed into thin self-supporting discs coated with gold.Chemical data were obtained by averaging the results obtained on five areas for each sample.
High-resolution X-ray powder diffraction (XRPD).Diffraction data were collected at ID22, the high-resolution beamline at the European Synchrotron Radiation Facility (ESRF), Grenoble, France, using the new Extremely Brilliant Source (EBS).The samples were ground and then used to fill borosilicate capillaries.Using a channel-cut Si(111) crystal monochromator the wavelength was set to be equal to 0.354496 A ˚; diffraction intensities were recorded by a bank of nine detectors, each preceded by a Si(111) crystal analyser.
Structure refinement.Structure refinements were performed using Rietveld methods through the GSAS package (Larson & von Dreele, 1994) with the EXPGUI interface (Toby, 2001).The structure reported by Hirano et al. (1992) was used as a starting model.The peak profile was fitted using the Thompson pseudo-Voigt function (Thompson et al., 1987), while the background was refined using a Chebyshev polynomial function with 24 coefficients.Si-O distances were refined using 'soft constraints' [ok as edited?],gradually decreasing the weight of the constraint (up to 10 [unit?]) after the initial stages.Extra-framework species, cations and H 2 O molecules were located in the zeolite pores by careful inspection of the Fourier difference map of the electronic density; the species were assigned considering the bond distances, the site-occupancy factors and the mutual exclusion rules for distances that are too short [ok as edited?].

Chemical characterization
Fig. 2 shows the TGA for all the exchanged samples and the obtained water content is reported in Table S4 in the supporting information.
The first peak of all the DTG [please define] curves, positioned below 300 C, represents the water loss, while peaks at higher temperatures, observed only for samples NH 4 -LTL and NH 4 -Ce-LTL, are related to the ammonium content.In particular, the peak near 650 C in the NH 4 -LTL DTG (Fig. 2) curve could derive from the condensation of residual H + ions.S1 and  S2 in the supporting information); (ii) the NH 4 amount determined by elemental analysis (Table S3); (iii) the H 2 O content (Table S4) determined by TGA or LOI analyses.
The chemical compositions reported in Table 1 give a clear idea about the exchange capacity of the samples involved in this study.From the comparison of the two Na-exchanged samples at different temperatures, it is possible to conclude that this variable is not critical in the improvement of the exchange effectiveness, as the two samples present the same final composition.For this reason, all the other experiments were performed at room temperature.Comparing Na-LTL and NH 4 -LTL, we can deduce that the exchange is higher for NH 4 than for Na (2.25 and 6.00 potassium ions are not exchanged from the initial 9.03 in the two samples, respectively).Once the samples are exchanged with Ce 3+ no clear affinity between Ce and one of the three cationic forms is observed.The three Ce 3+exchanged samples show, in fact, a similar amount of Ce (ranging from 1.16 to 1.26 atom p.u.c.[please define]), and also almost the same amount of H 2 O molecules.
In order to recover Ce from zeolite pores, NH 4 -Ce-LTL was further exchanged with a NH 4 + solution.The amount of Ce 3+ , obtained by EA, in the final NH 4 -Ce-LTL-rev (0.09 p.u.c.) was compared with that of the NH 4 -Ce-LTL (1.17 p.u.c.) proving the almost complete release of Ce and the restoration of the initial amount of ammonium as well as the H 2 O molecules.

Structure analysis
Structure analysis was performed to determine the amount and position of the extra-framework species, in particular of Ce, in the zeolite pores, and to understand the guest-guest and host-guest interactions.Cell parameters are reported in Table 2; structural data, as well as refinement details and diffraction Table 1 Chemical formulas of the whole sample set.

Name
Chemical patterns are reported in Tables S5-S17 and in Figs.S1-S6 in the supporting information.For selected samples (K-LTL and K-Ce-LTL), observed and calculated diffraction patterns obtained by Rietveld refinement are reported in Fig. 3.

Na + and NH 4
+ exchange.K-LTL and Na-LTL show very similar unit-cell parameters (Table 2), while we can observe slightly higher values for NH 4 -LTL.Apart from this, the Ce-free cationic forms of zeolite L display similar structures (Tables S6-S11).No new extra-framework sites were located compared with those reported in the literature (see Section 2.1); all the cations are sited in the KB (at the centre of the cancrinite cage), KC (at the centre of the elliptical 8MR channel) and KD (close to the wall of the 12MR channel) sites [see the structure of K-LTL (Fig. 4, left panel)].KB is bonded to the framework oxygen atoms O3 with six distances of about 2.9 A ˚and KC is placed at about 2.9 A ˚from four framework oxygen atoms O5 and at a larger distance from eight O4 atoms.Cations in KD, instead, display a solvation sphere constituted by W5 and W3 H 2 O molecules and bonds with the framework oxygen atoms O4 and O6 (Fig. 4, left panel).
H 2 O molecules are all placed in the main channel; they interact with each other, forming a cluster (Fig. 4, left panel).Two W4, at the centre of the 12MR channel, are at bonding distance and are also bonded to W1, W2, W3.Site W5, almost fully occupied in all samples, is located in the external part of the H 2 O cluster at bonding distance with KD, W3 and W6.The only interaction between the framework oxygen atoms and H 2 O molecules is observed for W2 with O1.In Table 3, the number of H 2 O molecules in the six sites W1-W6 is reported for all the samples.
3.2.2.Ce 3+ exchange.All the zeolites exchanged with Ce show a reduction of cell volume (due to the decrease of the a parameter) if compared with their Ce-free counterparts (Table 2).However, as already observed for the three Ce-free zeolites, NH 4 -Ce-LTL shows a higher volume.When K, Na and NH 4 cations are exchanged with Ce, similar structural modifications occur in the three samples.All the patterns collected for Ce-exchanged samples show changes in the relative peak intensities compared with their Ce-free counterparts, especially in the low-angle region, clearly indicating different cation contents (see as an example NH 4 -LTL and NH 4 -Ce-LTL patterns in Fig. 5).
In the Ce-exchanged forms, in addition to KB, KC and KD cation sites, a new one appears in the middle of the 12MR channel (labelled Ce).In the Ceexchanged samples the occupancy of KD and Ce sites was refined using the Ce scattering factor.The Ce site-occupancy factor ranges from 88 (K-Ce-LTL) to 97% (NH 4 -Ce-LTL), indicating that this latter site is filled almost exclusively with Ce in all the exchanged zeolites (Fig. 4, right panel).This new site coordinates with W2 and W3.The 5 of 9   3 and S12, S14, S16), creating a cluster that interacts slightly with the framework and bonding Ce site.4), we can assume the cation distribution in the three sites reported in Table 5.Since it is not possible to distinguish between different cations occupying the same site from the XRPD data, the comparison between XRPD and chemical data reported in Table 4 is only based on the number of electrons.

Discussion
K-LTL presents a certain amount of Na (0.41 p.u.c.) which remains in part in all the exchanged samples, with the exception of the NH 4 -exchanged one (see Table 1) where NH 4 is able to remove all the Na.Since the KB site is always occupied by non-exchangeable K, these Na ions should be distributed in KC and KD sites.Na-LTL zeolite shows an exchange Na + !K + of about 27%, according to Sato et al. (1990), so we hypothesize a prevalent substitution of K ions in the main channel (KD), while the small amount of Na + in KC is that already present in the pristine K-LTL.In sample NH 4 -LTL, K + ions are found mainly in the KB site and partially in KC, with an exchange NH 4 + !K + of about 73%, in very good agreement with the results reported by Dyer et al. (1993).The higher value of this last exchange seems related to the ionic radii of the species involved (Dyer et al., 1993).Indeed, NH 4 + and K + show a more similar size when compared with the couple Na + -K + ; thus NH 4 + can enter in site KD without affecting the framework stability.The presence of NH 4 + ions both in the 8MR and 12MR channels is also demonstrated by the broad peak (between 290 and 450 C) in DTG data related to the NH 4 + loss (Fig. 2).Indeed, the two sites occupied by this cation present different interactions with the framework and extra-framework species and hence exhibit different temperatures of release.In addition, in NH 4 -LTL, we can notice an enlargement of the cell volume, and that site KD is positioned farther from the framework in comparison with the other samples (i.e.KD-O4 = 3.18 and 3.20 A ˚in K-LTL and Na-LTL, respectively?, and 3.25 in NH 4 -LTL; KD-O6 = 2.98 and 2.99 A ˚in K-LTL and Na-LTL, respectively?, and 3.11 in NH 4 -LTL).This can be a direct consequence of the slightly larger ionic radius of NH 4+ (1.40A ˚) (Sidey, 2016) compared with K + (1.37A ˚) (Shannon, 1976) and Na + (0.99 A ˚) (Shannon, 1976) and/or due to the hydrogen bonds occurring between the framework oxygen atoms and this cation.
Regarding H 2 O molecules, in NH 4 -LTL the shift towards lower temperature of the first peak of the DTG curve (Fig. 2), related to H 2 O loss, can be caused by longer bonds between W2 and the framework [W2-O1 is 2.65 (6) A ˚in K-LTL, 2.61 (7) A ˚in Na-LTL and 2.81 (6) A ˚in NH 4 -LTL].In general, for all the samples, the total amount of cations and H 2 O    molecules obtained by structural refinement is in good agreement with those values resulting from chemical analysis (Table 4).In sample Na-LTL, the number of cations from structure refinement is slightly overestimated, but this could be due to the simultaneous presence of cations and H 2 O molecules on the KD site.This is consistent with the finding that the number of H 2 O molecules obtained is underestimated with respect to the TGA data.The fact that KD can be shared by the different species (also the case in the other samples) is also suggested by the high value of its thermal parameter.

Ce 3+ exchange and recovery
All Ce-exchanged zeolites show a similar amount of Ce p.u.c.(1.1-1.2 atoms p.u.c.) (Table 1) (exchange degree from 38% to 42%) and similar cation distribution.On the basis of the cation distribution reported in Table 5, Ce mainly occupies the site Ce and, to a lesser extent, site KD.From Table 4, structural analysis always shows an overestimation of the cations, accompanied by an underestimation of the number of H 2 O molecules.Thus, we can assume that the cation over-estimation is due to the co-presence of H 2 O molecules in sites Ce and KD; these H 2 O molecules [ok as edited?] have migrated from the original sites W1 and W4, which are empty in these samples.This hypothesis is confirmed by the high thermal parameters of these two sites, indicating a high degree of structural disorder.In comparison with the pristine materials, Ce-exchanged zeolites show a higher amount of H 2 O (Table 1), according to: (i) the bigger solvation sphere of this cation [solvation spheres: Ce 3+ = 9.1 (Lutz et al., 2012), K + = 7.5 (Duignan et al., 2020), NH 4 + = 5.8 (Chang & Dang, 2003), Na + = 5.6 H 2 O molecules (Duignan et al., 2020)]; (ii) the larger space available due to the lower amount of trivalent cations present in the main channel.The H 2 O content is similar for the three samples, due to the same amount of Ce incorporated, as well as the temperature of its release (see the first two peaks of the DTG, Fig. 2).It is interesting to observe that these temperatures appear higher than in pristine materials, due to the strong and short bonds occurring between the Ce site and W2 and W3 (respectively, about 2.3 and 2.6 A ˚). Ce penetration, indeed, leads to a reorganization of the position of the H 2 O molecules and their bonds with this element have a different coordination sphere with respect to the original K + , Na + and NH 4 + cations.This, beyond the intrinsic properties of zeolite L, can play an important role in the exchange capacity.
The complete recovery of cerium from the NH 4 -Ce-LTL was achieved through a further exchange with a NH 4 solution.Chemical results demonstrated complete recovery of the initial amount of ammonium and water molecule in the zeolite with an almost complete release of cerium.As further demonstration, the two XRPD patterns, reported in Fig. 6, of the initial NH 4 -LTL and the NH 4 -Ce-LTL-rev after cerium recovery show the same features, suggesting the same structure and the same type and number of cations.

Conclusions
The aim of this work was the evaluation of the Ce-exchange capacity of three cationic forms of LTL zeolites (the assynthesized K form and Na-and NH 4 -exchanged ones) using a highly concentrated solution of Ce.The effect of the Ce exchange in the different cationic forms was evaluated through chemical analyses and XRPD structural investigations.Indeed, structural studies can give new insight into the sites involved in the exchange, and this is fundamental for the choice of a powerful protocol of exchange for future application such as Ce recovery from diluted solution.The results reported here also have important implications for catalytic applications, since many zeolite catalysts are prepared through Ce exchange.
The obtained results show a Na + !K + substitution of about 27%, while NH 4 + !K + reached 73%, because the cations in the extra-framework site KC -in the 8MR channelare exchanged only if K + is replaced by NH 4 + ions due to their similar ionic size.The cation-exchange degree for Ce ranges from 38 to 42% and the LTL cation-exchange capacity for Ce is not influenced by the nature of the cations in the pristine sample.In contrast to what was observed for Na and NH 4 exchange, in which the cations occupy the pre-existing potassium positions, the Ce cations occupy a new site in the 12MR site and induce a reorganization of the H 2 O molecules.This may be due to the higher ionic potential of Ce atoms, which tend to form a stronger solvation sphere that hinders cation mobility.The Ce-exchange capacity of zeolite L and its affinity with this element can thus be driven not only by the intrinsic property of the LTL framework, but also by the coordination sphere of Ce compared with the original K + , Na + and NH 4 + cations.With the aim of possible REE recovery from zeolite pores, the Ce-exchanged NH 4 -LTL was tested through exchange with a NH 4 + solution, leading to the 100% recovery of this element.The results presented here are promising for future Ce-recovery experiments from very diluted solutions.

Figure 1 (
Figure 1 (a) LTL framework type; (b) columns constituting the framework, formed by the stacking of cancrinite cages and double six-membered rings.

Figure 2 TGA
Figure 2 TGA (solid line) and DTG (dotted line) curves for: K-LTL, Na-LTL and NH 4 -LTL samples in black, red and green, respectively (left panel), and K-Ce-LTL, Na-Ce-LTL and NH 4 -Ce-LTL samples in black, red and green, respectively (right panel).

Figure 3
Figure 3Zoomed-in view of observed (blue symbols) and calculated (red line) diffraction patterns and final difference curve (grey line) from Rietveld refinement of K-LTL and K-Ce-LTL.

Figure 4
Figure 4 Structure of K-LTL (left panel) and K-Ce-LTL (right panel).Cations are drawn as incomplete spheres as a function of their fractional occupancies.For the H 2 O molecules, only the positions actually occupied are shown, rather than all the partially occupied symmetrically equivalent positions.
Na and NH 4 zeolites show similar structures, but the occupancy factors of sites KB, KC and KD are different due to the different types and amount of exchanged extra-framework cations.Combining the information obtained by XRPD and chemical analysis (Table

Table 1
reports the chemical formulas of the whole sample set, obtained considering: (i) the results obtained from XRF or SEM-EDS (Tables

Table 3
H 2 O p.u.c. in each W site for the whole sample set.The low amount of H 2 O found by XRPD in Ce-exchanged zeolites is explained in Section 4. K-LTL Na-LTL NH 4 -LTL K-Ce-LTL Na-Ce-LTL NH 4 -Ce-LTL

Table 5
Cation distribution over the extra-framework sites deduced from structural and chemical results.

Table 4
Number of electrons (e) for cations (cat.) and H 2 O molecules derived from structural (XRPD) (number of atoms Â Z depending on the scattering factor used for the occupancy factor refinement of each cation site) and chemical analyses (number of atoms from the chemical formula Â Z). differences between the total number of electrons for cations and H 2 O molecules obtained by the two techniques are also reported (e cat.XRPD À e cat.chemical analysis; e H 2 O XRPD À e H 2 O chemical analysis).