Adsorption Properties of Ce5(BDC)7.5(DMF)4 MOF

In this article we report on the spectroscopic and adsorptive studies done on Ce(III)-based MOF possessing, upon desolvation, open metal sites, and a discrete surface area. The Ce-based MOF was synthesized from terephthalic acid linker (H2BDC) and Ce3+ cations by the classical solvothermal method. Preliminary powder X-ray diffraction analysis showed that the obtained materials corresponded to the ones reported by other authors. Spectroscopic techniques, such as XAS and in situ FTIR with probe molecules were used. In situ FTIR spectroscopy confirmed the successful removal of DMF molecules within the pore system at temperatures above 250 ◦C. Moreover, the use of CO as a probe molecule evidenced the presence of a Ce3+ open metal sites. Detailed volumetric and calorimetric CO2 adsorption studies are also reported.


Introduction
Cerium is the most abundant lanthanide element present in the earth crust [1,2] and the ores that are mined for the extraction of more rare and precious rare earth elements (REEs) are also rich in Ce; thus, its cost is relatively low. Its oxide, CeO 2 , commonly named ceria, is particularly relevant for redox chemistry, being a catalyst for oxidation and reduction reactions [3], for example, combustion catalysis [4] and photocatalysis [3].
The present work reports on the synthesis, spectroscopic characterization, and adsorption properties of Ce 3+ -based MOFs with terephthalic acid (H 2 BDC) as the linker. This material has been already

Results and Discussion
Ce5(BDC)7.5(DMF)4 MOF [16] optimized synthesis is reported in the Supplementary Materials. The Ce5(BDC)7.5(DMF)4 PXRD pattern was coincident with the one reported by D'Arras et al. [11]. The hypothesized structure was taken from Reference [11] and it is reported in Figure 1, for clarity. The asymmetric unit of crystalline structure shows chains of five independent cerium atoms arranged linearly and surrounded by BDC 2− and DMF molecules (see Figure 1a). The two terminal atoms of the group are coordinated by eight oxygen atoms in a distorted square antiprismatic shape: six oxygen atoms belong to BDC 2− and two oxygen atoms to DMF molecules, while the "central" three cerium atoms are coordinated with nine oxygens, all coming from the ligands, in an uncommon distorted shape. There are mono-dimensional or 1D channels formed in the structure parallel to the Ce chains along the crystallographic 11 0 direction. As depicted in Figure 1b, these micropores are mainly occupied by the coordinated DMF molecules, which protrude into the pores. The surface area and porosity of the material were clarified by the N2 adsorption experiments (vide infra). The powder diffraction pattern of the as synthesized material, as shown in the synthesis development reported in the Supplementary Materials ( Figures S1-S3), was compared to the calculated powder pattern obtained from the crystal structure reported by D'Arras et al. (Figure S3). SEM images of the synthesized powder are available in Figure S4.
A variable temperature powder X-ray diffraction (VTXRD) experiment in N2 flow in the RT-600 °C temperature range (see Figure 2) showed that the material maintained the crystallinity until 475 °C, undergoing some changes in the XRD pattern, especially from 200 to 250 °C (highlighted in blue color), which could be due to the solvent removal, as strongly suggested by the TGA measurements reported in Figure S5.
The solution of the crystal structure of the desolvated material was out of the scope of the present work. At 500 °C the MOF started decomposing and at 525 °C the formation of broad diffraction peaks The powder diffraction pattern of the as synthesized material, as shown in the synthesis development reported in the Supplementary Materials ( Figures S1-S3), was compared to the calculated powder pattern obtained from the crystal structure reported by D'Arras et al. (Figure S3). SEM images of the synthesized powder are available in Figure S4.
A variable temperature powder X-ray diffraction (VTXRD) experiment in N 2 flow in the RT-600 • C temperature range (see Figure 2) showed that the material maintained the crystallinity until 475 • C, undergoing some changes in the XRD pattern, especially from 200 to 250 • C (highlighted in blue color), which could be due to the solvent removal, as strongly suggested by the TGA measurements reported in Figure S5. The solution of the crystal structure of the desolvated material was out of the scope of the present work. At 500 • C the MOF started decomposing and at 525 • C the formation of broad diffraction peaks due to cerium dioxide was visible. The broadness of the peaks testifies that the particles were nanometric in size. Scherrer's equation [3], corrected by instrumental broadening using a Si standard from NIST, suggested a size for the cerium dioxide particles of 5 ± 1 nm.
Inorganics 2020, 8,9 3 of 10 due to cerium dioxide was visible. The broadness of the peaks testifies that the particles were nanometric in size. Scherrer's equation [3], corrected by instrumental broadening using a Si standard from NIST, suggested a size for the cerium dioxide particles of 5 ± 1 nm. The 3+ oxidation state of Ce in the as-synthesized state and activated at 350 °C material was confirmed by means of XAS spectroscopy, in the XANES region comparison with Ce 3+ and Ce 4+ standards. XANES spectra are reported in Figure 3. The 3+ oxidation state was also maintained after activation at 350 °C in the He stream directly in the measurement cell, in agreement with the results reported by the XPS experiments of D'Arras et al. [11]. The 3+ oxidation state of Ce in the as-synthesized state and activated at 350 • C material was confirmed by means of XAS spectroscopy, in the XANES region comparison with Ce 3+ and Ce 4+ standards. XANES spectra are reported in Figure 3. The 3+ oxidation state was also maintained after activation at 350 • C in the He stream directly in the measurement cell, in agreement with the results reported by the XPS experiments of D'Arras et al. [11]. Inorganics 2020, 8,9 4 of 10 N2 adsorption volumetric isotherms were measured on the material in order to point out the specific surface area and porosity of the MOF. Thermal treatments were performed in vacuo for 3 h (a longer time in comparison with those ones performed in case of IR or XRD because of the bigger amount of sample) and were made in a consecutive way. From the isotherms reported in Figure 4 it is clear that upon the solvent loss (occurring in the 200-250 °C range) N2 adsorption grew dramatically, showing microporosity (as the isotherm is a Type I) and higher surface area (more than 200 m 2 /g) due to the accessible pores.
BET and Langmuir adsorption models for the surface area were applied; the results are summarized in Table 1. Generally, the reported value can be quite modest for MOF materials, compared with typical MOF surface areas (thousands of m 2 /g) [17].  N 2 adsorption volumetric isotherms were measured on the material in order to point out the specific surface area and porosity of the MOF. Thermal treatments were performed in vacuo for 3 h (a longer time in comparison with those ones performed in case of IR or XRD because of the bigger amount of sample) and were made in a consecutive way. From the isotherms reported in Figure 4 it is clear that upon the solvent loss (occurring in the 200-250 • C range) N 2 adsorption grew dramatically, showing microporosity (as the isotherm is a Type I) and higher surface area (more than 200 m 2 /g) due to the accessible pores.
BET and Langmuir adsorption models for the surface area were applied; the results are summarized in Table 1. Generally, the reported value can be quite modest for MOF materials, compared with typical MOF surface areas (thousands of m 2 /g) [17]. N2 adsorption volumetric isotherms were measured on the material in order to point out the specific surface area and porosity of the MOF. Thermal treatments were performed in vacuo for 3 h (a longer time in comparison with those ones performed in case of IR or XRD because of the bigger amount of sample) and were made in a consecutive way. From the isotherms reported in Figure 4 it is clear that upon the solvent loss (occurring in the 200-250 °C range) N2 adsorption grew dramatically, showing microporosity (as the isotherm is a Type I) and higher surface area (more than 200 m 2 /g) due to the accessible pores.
BET and Langmuir adsorption models for the surface area were applied; the results are summarized in Table 1. Generally, the reported value can be quite modest for MOF materials, compared with typical MOF surface areas (thousands of m 2 /g) [17].   As synthesized Ce 5 (BDC) 7.5 (DMF) 4 showed the typical mid-IR spectrum for a solvated MOF (see Figure 5). The typical vibrational fingerprints due to DMF solvent molecules mainly inside the pores and to H 2 O molecules adsorbed from the atmosphere can be recognized: a broad band centered at 3400 cm −1 due to the hydrogen-bonded H 2 O molecules and sharp features at frequencies lower than 3000 cm −1 , in the range of the aliphatic C-H stretching mode, and a very intense band centered at 1670 cm −1 , in the range of the carbonyl stretching mode, due to DMF molecules. Upon progressive outgassing, also by increasing the temperatures, vibrational signals due to DMF and H 2 O disappeared and the typical spectrum of a carboxylate-based MOF material was shown: very intense bands in the 1650-1250 cm −1 range due to carboxylate stretching modes (both symmetrical and asymmetrical) and sharp features at frequencies higher than 3000 cm −1 , in the range of aromatic C-H stretching mode due to BDC 2− .
These data support the VTXRD, TGA (see Figure S5), and SSA experiments and therefore we can affirm the material is not destroyed even if activated at 450 • C, maintaining crystallinity and surface area, even if its crystal structure undergoes a phase transition that has not been determined in this study.  As synthesized Ce5(BDC)7.5(DMF)4 showed the typical mid-IR spectrum for a solvated MOF (see Figure 5). The typical vibrational fingerprints due to DMF solvent molecules mainly inside the pores and to H2O molecules adsorbed from the atmosphere can be recognized: a broad band centered at 3400 cm −1 due to the hydrogen-bonded H2O molecules and sharp features at frequencies lower than 3000 cm −1 , in the range of the aliphatic C-H stretching mode, and a very intense band centered at 1670 cm −1 , in the range of the carbonyl stretching mode, due to DMF molecules. Upon progressive outgassing, also by increasing the temperatures, vibrational signals due to DMF and H2O disappeared and the typical spectrum of a carboxylate-based MOF material was shown: very intense bands in the 1650-1250 cm −1 range due to carboxylate stretching modes (both symmetrical and asymmetrical) and sharp features at frequencies higher than 3000 cm −1 , in the range of aromatic C-H stretching mode due to BDC 2− .
These data support the VTXRD, TGA (see Figure S5), and SSA experiments and therefore we can affirm the material is not destroyed even if activated at 450 °C, maintaining crystallinity and surface area, even if its crystal structure undergoes a phase transition that has not been determined in this study. In situ CO adsorption FTIR spectra at −196 °C were recorded on a sample activated at 350 °C for 1 h (see Figure 6). A pressure of 5 mbar of CO was dosed (black spectrum). The lowest frequency peak, at 2131 cm −1 , was assigned to the physi-sorbed CO in the pores, as the first one to be desorbed. The other two bands at higher frequencies (2161 cm −1 and 2152 cm −1 ) required more time for complete desorption, and for this reason they can be assigned to CO adsorbed on Lewis acidic sites.
Upon outgassing, the initial activated MOF spectrum was obtained due to the total reversibility of CO adsorption. Because the CO vibrational mode on the MOF open metal sites is intermediate between metals in oxides and metals grafted in different systems [18,19], the doublet can be assigned In situ CO adsorption FTIR spectra at −196 • C were recorded on a sample activated at 350 • C for 1 h (see Figure 6). A pressure of 5 mbar of CO was dosed (black spectrum). The lowest frequency peak, at 2131 cm −1 , was assigned to the physi-sorbed CO in the pores, as the first one to be desorbed. The other two bands at higher frequencies (2161 cm −1 and 2152 cm −1 ) required more time for complete desorption, and for this reason they can be assigned to CO adsorbed on Lewis acidic sites.
Upon outgassing, the initial activated MOF spectrum was obtained due to the total reversibility of CO adsorption. Because the CO vibrational mode on the MOF open metal sites is intermediate between metals in oxides and metals grafted in different systems [18,19], the doublet can be assigned to CO adsorbed on Ce 3+ sites, since CO interacting with Ce 4+ is expected to give bands at frequencies higher than 2156 cm −1 [20,21], in agreement also with XANES results. With the crystal structure of the desolvated material unknown, we can only hypothesize the presence of at least two different Ce 3+ probed sites [20][21][22]. Only two out of the five different cerium atoms among the crystallographic asymmetric unit underwent the removal of DMF molecules during activation, however for an in-depth understanding of the desolvation process more investigation is needed. It is of relevance also the overall low intensity of the bands associated with CO adsorption could be due to a low accessibility of the uncoordinated sites. In case of FTIR CO adsorption on MOF-76-Ce [5], two quite intense bands at 2155 cm −1 and 2149 cm −1 , due, respectively, to the probe interacting with one Ce site and bridged on two close sites, were reported, testifying the completely different local structure of the metal in MOF-76-Ce and in the present Ce 5 (BDC) 7.5 (DMF) 4 .
Inorganics 2020, 8,9 6 of 10 to CO adsorbed on Ce 3+ sites, since CO interacting with Ce 4+ is expected to give bands at frequencies higher than 2156 cm −1 [20,21], in agreement also with XANES results. With the crystal structure of the desolvated material unknown, we can only hypothesize the presence of at least two different Ce 3+ probed sites [20][21][22]. Only two out of the five different cerium atoms among the crystallographic asymmetric unit underwent the removal of DMF molecules during activation, however for an indepth understanding of the desolvation process more investigation is needed. It is of relevance also the overall low intensity of the bands associated with CO adsorption could be due to a low accessibility of the uncoordinated sites. In case of FTIR CO adsorption on MOF-76-Ce [5], two quite intense bands at 2155 cm −1 and 2149 cm −1 , due, respectively, to the probe interacting with one Ce site and bridged on two close sites, were reported, testifying the completely different local structure of the metal in MOF-76-Ce and in the present Ce5(BDC)7.5(DMF)4. CO2 volumetric isotherms at various activation temperatures are reported in Figure 7. The same powder was heated in vacuo for three hours at the next activation temperature in a consecutive way.
CO2 uptake showed a clear increase with the activation temperature by reaching a plateau at the 350-450 °C temperature range, as reported in Figure 7 and Table 2.   Figure 7. The same powder was heated in vacuo for three hours at the next activation temperature in a consecutive way.
Inorganics 2020, 8, 9 6 of 10 to CO adsorbed on Ce 3+ sites, since CO interacting with Ce 4+ is expected to give bands at frequencies higher than 2156 cm −1 [20,21], in agreement also with XANES results. With the crystal structure of the desolvated material unknown, we can only hypothesize the presence of at least two different Ce 3+ probed sites [20][21][22]. Only two out of the five different cerium atoms among the crystallographic asymmetric unit underwent the removal of DMF molecules during activation, however for an indepth understanding of the desolvation process more investigation is needed. It is of relevance also the overall low intensity of the bands associated with CO adsorption could be due to a low accessibility of the uncoordinated sites. In case of FTIR CO adsorption on MOF-76-Ce [5], two quite intense bands at 2155 cm −1 and 2149 cm −1 , due, respectively, to the probe interacting with one Ce site and bridged on two close sites, were reported, testifying the completely different local structure of the metal in MOF-76-Ce and in the present Ce5(BDC)7.5(DMF)4. CO2 volumetric isotherms at various activation temperatures are reported in Figure 7. The same powder was heated in vacuo for three hours at the next activation temperature in a consecutive way.
CO2 uptake showed a clear increase with the activation temperature by reaching a plateau at the 350-450 °C temperature range, as reported in Figure 7 and Table 2.  Inorganics 2020, 8, 9 7 of 10 CO 2 uptake showed a clear increase with the activation temperature by reaching a plateau at the 350-450 • C temperature range, as reported in Figure 7 and Table 2. Calorimetric data, reported in Figure 8, were recorded for the adsorption of CO 2 at 30 • C in the 0-90 mbar range. The adsorbed quantity and the heat released by the adsorption as a function of the pressure are plotted, respectively, in Figure 8a,b. Both these curves are nearly Henry-type, as confirmed by the first part of our CO 2 volumetric adsorption isotherms collected with a different instrument at 25 • C (Figure 7). The studied pressure range was too low to observe saturation of the adsorbing sites and the temperature difference of 5 • C between these two experiments was due to technical requirements. The differential heat of adsorption at low coverages was about 32-33 kJ/mol and it is quite typical for the interaction of CO 2 with an open-metal site, as in MOF-76-ds [5], HKUST-1 [23], Mg-MOF-74 [17]. It is worth noting that the differential heat curve (Figure 8c) is characterized by an abrupt diminishing with the adsorbed quantity; this can be ascribed to the overall low number of strongly coordinating sites present in the desolvated material, as highlighted by our CO FTIR experiment ( Figure 6). The total reversibility of the adsorption of CO 2 upon outgassing at 30 • C was testified thoroughly by the perfect recovery of the adsorption properties between the primary and secondary adsorption cycles.  Calorimetric data, reported in Figure 8, were recorded for the adsorption of CO2 at 30 °C in the 0-90 mbar range. The adsorbed quantity and the heat released by the adsorption as a function of the pressure are plotted, respectively, in Figure 8a,b. Both these curves are nearly Henry-type, as confirmed by the first part of our CO2 volumetric adsorption isotherms collected with a different instrument at 25 °C (Figure 7). The studied pressure range was too low to observe saturation of the adsorbing sites and the temperature difference of 5 °C between these two experiments was due to technical requirements. The differential heat of adsorption at low coverages was about 32-33 kJ/mol and it is quite typical for the interaction of CO2 with an open-metal site, as in MOF-76-ds [5], HKUST-1 [23], Mg-MOF-74 [17]. It is worth noting that the differential heat curve (Figure 8c) is characterized by an abrupt diminishing with the adsorbed quantity; this can be ascribed to the overall low number of strongly coordinating sites present in the desolvated material, as highlighted by our CO FTIR experiment ( Figure 6). The total reversibility of the adsorption of CO2 upon outgassing at 30 °C was testified thoroughly by the perfect recovery of the adsorption properties between the primary and secondary adsorption cycles.

Materials and Methods
The variable temperature X-ray diffraction patterns (VTXRD) were collected with an X'Pert PRO MPD diffractometer from PANalytical (Almelo, The Netherlands), working in Bragg-Brentano geometry equipped with a Cu Kα source using about 10 mg of sample. Scattered photons were revealed by an X'celerator linear detector (PANalytical, Almelo, The Netherlands) equipped with a Ni filter to attenuate Kβ. A non-ambient chamber XRK900 from Anton Paar (Graz, Austria) with Be windows was used to collect diffractograms as a function of temperature in a flow of dry N2 (20 mL/min). The temperature program was set to measure a pattern every 25 °C, waiting 25 min at the target temperature before collecting the data. The temperature was increased at a rate of 2 °C/min.

Materials and Methods
The variable temperature X-ray diffraction patterns (VTXRD) were collected with an X'Pert PRO MPD diffractometer from PANalytical (Almelo, The Netherlands), working in Bragg-Brentano geometry equipped with a Cu Kα source using about 10 mg of sample. Scattered photons were revealed by an X'celerator linear detector (PANalytical, Almelo, The Netherlands) equipped with a Ni filter to attenuate Kβ. A non-ambient chamber XRK900 from Anton Paar (Graz, Austria) with Be windows was used to collect diffractograms as a function of temperature in a flow of dry N 2 (20 mL/min). The temperature program was set to measure a pattern every 25 • C, waiting 25 min at the target temperature before collecting the data. The temperature was increased at a rate of 2 • C/min. X-Ray absorption spectra at the Ce L 3 edge (5723 eV) were collected at the BM23 beamline of the European Synchrotron Radiation Facility (ESRF). Data were acquired up to the Ce L 2 edge (6164 eV), which limited the EXAFS signal down to k ≈ 10 Å −1 .The acquisition step was set to 0.3 eV in the near-edge region and ∆k = 0.035 Å −1 in the EXAFS part of the spectrum. We used three He/N 2 -filled ionization chambers as I 0 , I 1 , and I 2 detectors, placing chromium foil between I 1 and I 2 for energy calibration. XANES and EXAFS data were analyzed by using the Demeter 0.9.20 package. Sample treatment was carried out in an in situ cell under a flow of He (80 mL min −1 ). During activation, the sample was heated to 350 • C after a ramp heating at 4 • C min −1 , to then be cooled to 30 • C.
Adsorption isotherms were collected on an ASAP 2020 apparatus from Micromeritics (Norcross, GA, USA) using a liquid nitrogen bath at −196 • C, albeit CO 2 isotherms were collected filling the same dewar vessel with water at 25 • C. About 150 mg of sample was heated in dynamic vacuum at 350 • C for 3 h prior to measuring the isotherm. Langmuir fit was made in the 0.05 < p/p 0 < 0.2 range, while the BET analysis was carried out in a very low-pressure region, as prescribed by the so-called Rouquerol rules [24], in order to obtain a positive C value.
FTIR spectra were collected on a Nicolet 6700 from Thermo Scientific (Waltham, MA, USA) equipped with an MCT detector in the 4000-400 cm −1 range with a resolution of 2 cm −1 . The sample was prepared by pressing a thin self-supporting pellet (10 mg of sample) and using a vacuum line and a jacketed IR cell of local construction capable of cooling down the sample with liquid nitrogen and permitting the dosing of probe molecules (i.e., CO). The experiment was run by activating the pellet in dynamic vacuum at 350 • C for 1 h, then dosing about 15 mbar of CO on the sample to record spectra during cooling to the liquid nitrogen temperature, desorption, and then heating back to RT.
Adsorption heats were measured simultaneously with the adsorption isotherms by means of a C80 Tian-Calvet microcalorimeter from Setaram (Caluire-et-Cuire, France) at a temperature of 30 • C, coupled with a glass vacuum line of local construction. The procedure is thoroughly described in references [25,26], and in this case required thermal activation under dynamic vacuum at 350 • C for 3 h, then an overnight outgassing at 30 • C in the calorimeter before measuring the primary and the secondary adsorption runs in order to determine the non-desorbable (irreversible) fraction.

Conclusions
The synthesis of Ce 5 (BDC) 7.5 (DMF) 4 was successful, starting from Ce(NO 3 ) 3 ·6H 2 O and H 2 BDC in solvothermal conditions in DMF at 140 • C. We obtained the same crystal structure reported by D'Arras et al. [11] using a Ce 3+ source directly, conversely to the previous contribution. The thermal stability (up to 475 • C in an inert atmosphere) previously observed by D'Arras et al. [11] was confirmed by our VTXRD and TGA measurements. Differently from D'Arras et al. [11] we found a discrete surface area due to microporosity through N 2 adsorption isotherms at −196 • C (about 220 m 2 /g) after thermal activation in the 250-450 • C range. XAS confirmed the presence of Ce 3+ in the material also after desolvation. FTIR spectroscopy confirmed the successful removal of DMF within the pore system at temperatures above 250 • C, and by means of low-temperature CO adsorption evidenced the presence of a Ce 3+ open metal sites. The interaction of the desolvated material with CO 2 was characterized by volumetric and calorimetric measurements, finding a modest capacity of adsorption (about 3.5 wt % at 1 bar and 25 • C) but a relevant enthalpy (32-33 kJ/mol) for the very first dose, compatible with the presence of open metal sites. This work can open the way to a deep understanding and description of this MOF crystal structure and phase changes upon activation.