Metal organic framework derived NaCoxOy for room temperature hydrogen sulfide removal

Novel NaCoxOy adsorbents were fabricated by air calcination of (Na,Co)-organic frameworks at 700 °C. The NaCoxOy crystallized as hexagonal microsheets of 100–200 nm thickness with the presence of some polyhedral nanocrystals. The surface area was in the range of 1.15–1.90 m2 g−1. X-ray photoelectron spectroscopy (XPS) analysis confirmed Co2+ and Co3+ sites in MOFs, which were preserved in NaCoxOy. The synthesized adsorbents were studied for room-temperature H2S removal in both dry and moist conditions. NaCoxOy adsorbents were found ~ 80 times better than the MOF precursors. The maximum adsorption capacity of 168.2 mg g−1 was recorded for a 500 ppm H2S concentration flowing at a rate of 0.1 L min−1. The adsorption capacity decreased in the moist condition due to the competitive nature of water molecules for the H2S-binding sites. The PXRD analysis predicted Co3S4, CoSO4, Co3O4, and Co(OH)2 in the H2S-exposed sample. The XPS analysis confirmed the formation of sulfide, sulfur, and sulfate as the products of H2S oxidation at room temperature. The work reported here is the first study on the use of NaCoxOy type materials for H2S remediation.

. While the calcination of monometallic MOFs yields single-phase oxides, the presence of a second metal in the MOF has unpredictable outcomes. Huang 21 . Thus, the formation of metal oxides as composites, single-phase, or doped oxides is highly unpredictable.
In the literature, the role of sodium in MOFs is largely unexplored, with some reports on sodium metal-organic frameworks 23 . In the present study, the unexpected presence of sodium in Co-based MOFs has played a decisive role in the formation of MOF-derived oxides. The presence of Na in Co-MOF resulted in the formation of single-phase NaCo x O y as opposed to Co 3 O 4 /Na 2 O as the air calcination product. Apart from unique microsheets like morphology, the material showed exceptionally high H 2 S removal capacity in ambient conditions. The parameters associated with the column studies were optimized. Moreover, the H 2 S removal mechanism was studied in detail using different microscopic and spectroscopic analyses.

Materials and methods
Chemicals. Benzene-1,4-dicarboxylic acid (H 2 BDC), benzene-1,3,5-tricarboxylic acid (H 3 BTC), and cobalt(II) nitrate hexahydrate (Co(NO 3 ) 2 ·6H 2 O) were purchased from Sigma Aldrich. Ethanol, methanol, N,Ndimethyl formamide (DMF), and sodium hydroxide (NaOH) pellets were acquired from Samchun Pure Chemicals, Korea. H 2 S gas (500 ppm balanced with N 2 gas) was procured from Union gas, Korea. Material characterization. The surface morphology of materials was probed by field emission scanning electron microscopy (FE-SEM, Hitachi S-4300, Japan). A gold-platinum alloy was coated on dried samples using an E-1048 Hitachi ion sputter. The transmission electron microscopy (TEM) was conducted on a field emission TEM (FE-TEM, JEM-2010F, JEOL, Japan). Elemental mapping was done by energy-dispersive X-ray spectroscopy (EDAX, X-Maxn 80 T, Oxford, UK). N 2 adsorption-desorption isotherms were recorded at − 196 °C over a Gemini series Micromeritics 2360 instrument and analyzed by the Brunauer-Emmett-Teller (BET) equation. Samples were pre-heated at 200 °C for 8 h for degassing purpose. XRD patterns were recorded on an X-ray diffractometer (Ultima IV Rigajku, Japan) with Cu Kα and a Ni filter. Fourier-transform infrared (FTIR) spectra were recorded on a Cary670 FTIR spectrometer after pelletization with KBr. For XPS analyses, a K-alpha XPS instrument (Thermo Scientific Inc., UK) with a monochromatic Al Kα X-ray source and 4.8 × 10 −9 mbar of pressure was used. Spectra were charge-corrected to the main line of the carbon 1s spectrum (aromatic carbon) set to 284.7 eV. Spectra were analyzed using CasaXPS software (version 2.3.14).

Synthesis of adsorbents.
Breakthrough studies. H 2 S breakthrough studies were carried out in a fixed bed micro-reactor at 25 °C. A known mass of an adsorbent packed between glass wool was supported on silica beads in a pyrex tube (height: 50 cm, diameter: 1 cm). The adsorbents were tested in dry and moist (by passing moist air through the adsorbent bed for 0.5 h with 0.3 L min −1 of flowrate) conditions. The concentration of the outgoing gas was measured by a multi-gas analyzer (GSR-310, Sensoronic, Korea) every 15 s until the effluent concentration reached 10 ppm (2% was the breakthrough condition). The adsorption capacity of an adsorbent (q, mg g −1 ) was calculated by integration of the area above the breakthrough curve.

Results and discussion
Characterization of adsorbents. The SEM and TEM micrographs of Co-MOFs and derived oxides are shown in Fig. 1. CoBDC has a cluttered sheet-like morphology (Fig. 1a,e). CoBTC has distorted hexagonal microcrystals surrounded by nanothreads (Fig. 1b,f). The oxides formed by the calcination of MOFs have similar morphology. The NCO-D has smooth intercalated hexagonal sheets with some deposition of nanoparticles (Fig. 1c,g). The hexagonal microsheets and nanothreads in CoBTC were transformed to irregular hexagonal microsheets and polyhedral nanoparticles, respectively, in NCO-T (Fig. 1d,h). The TEM-EDS analyses of Co-MOFs and derived oxides are shown in Figs. S1 and S2. The EDS analysis confirmed the presence of Co, C, and O with an additional Na peak at ~ 1.0 eV for Co-MOFs (Fig. S1). For CoBTC, a metal-to-ligand ratio of 1:1 www.nature.com/scientificreports/ (opposed to the conventional 3:2) was adopted in the present study. The deficient Co ions for metal-ligand interactions in CoBTC was balanced by Na-ligand interactions 24 . The probable consumption of some of the cobalt hydroxide to form cobalt oxide was compensated by Na ions, which interacted with the carboxylate groups in a strong alkali medium 25 . The oxides derived from the calcination of Co-MOFs (NCO-D and NCO-T) ( Fig. S2) have all peaks except for carbon. Based on EDS analysis, a compositional formula of NaCo 0.7 O 2.4 and NaCo 1.1 O 3.3 was assigned to NCO-D and NCO-T, respectively (Table S1). The excess oxygen could be from the hydroxyl density, adsorbed molecular oxygen, and mixed-valence states of cobalt ions. The surface and pore properties of MOFs and oxides were evaluated by N 2 adsorption-desorption isotherms (Fig. 2a). The MOFs exhibited adsorption-desorption isotherms for mesoporous materials. The surface area of these MOFs (6.9-18.3 m 2 g −1 ) was in agreement with the reported values 25 . The low surface area of MOFs was due to the formation of microparticles 26 . The MOF-derived binary metal oxides exhibited curves for macroporous or non-porous materials 27 . Metal oxides have a lower surface area than MOFs due to complete loss of mesoporosity after high-temperature calcination. The measured surface area of NCO-D and NCO-T was 1.15 and 1.90 m 2 g −1 , respectively (Table S2).
The PXRD pattern of MOFs and derived oxides are shown in Fig. 2b. The PXRD pattern of CoBDC matched with the one reported by Ma et al. with a slight variation in the peak intensity at 28.5°2 8 . Moreover, the peak at 9.0° has a minor split due to a slight distortion in the symmetry of MOF 29 . The PXRD pattern of CoBTC largely matched with the reported MOF by Nowacka et al. with an additional presence of diffraction peaks in the 5°-10° range 30 . Since MOFs were synthesized in a strong alkali medium, Na + ions in the MOFs as the nodes had a strong www.nature.com/scientificreports/ impact on their PXRD pattern 31 . The calcination of these MOFs results in the formation of conventional oxides like Co 3 O 4 32 . On the other hand, mixed metal ions in a MOF yields binary metal oxides 33 . In the present study, (Na,Co)-based MOFs oxidized to yield single-phase NaCo x O y materials and not Co 3 O 4 /Na 2 O composite. The possible reason for single-phase NaCo x O y formation was a high calcination temperature of 700 °C and a long heating time of 24 h. In the literature, numerous reports are available on the fabrication of sodium cobalt oxides by the solid-state synthesis method with Na 2 O 2 and Co 3 O 4 as precursors 34 . During the calcination of (Na,Co)-MOFs, the oxide formation occurs through the initial growth of oxide nanoparticles on the MOF surface. These nanoparticles served as seeds for the development of microsheets 35 . In the case of the delocalized distribution of Na and Co in the MOF structure, both the metal cations took part in the formation of oxide to yield NaCo x O y type materials. On the contrary, localized distribution of Na and Co in the MOF may have formed Na 2 O/Na 2 O 2 and Co 3 O 4 nanoparticles, which served as the precursors for NaCo x O y type materials at a high temperature of 700 °C for 24 h. Thus, in both cases, NaCo x O y formation was possible. For this reason, the PXRD pattern of NaCo x O y matched with the pattern of NaCo 2 O 4 36 with the absence of Na 2 O or Co 3 O 4 (Fig. S3). The FTIR spectra of MOFs and oxides are shown in Fig. 2c. The band at 3432 cm −1 was assigned to the stretching vibrations of O-H stretching vibrations of adsorbed water molecules. The high-intensity bands at 1579, 1387, and 1357 cm −1 were due to the asymmetric and symmetric O-C-O stretching of organic linkers. The band at 1502 cm −1 was for C=C stretchings of the aromatic skeleton. The mid-intensity bands at 825, 808, and 753 cm −1 were attributed to the C-H bending modes 37,38 . The band at 457 and 510 cm −1 were due to the Co-O stretching 39 . For CoBTC, additional peak at 1668 and 1707 cm −1 were possibly due to the Na-bound carboxylate groups. In the FTIR spectra of NaCo x O y , the band at 1639 cm −1 was due to the bending mode of the adsorbed water molecules. The bands at 881 and 1442 cm −1 were ascribed to the asymmetric stretching Co-OH and Na-O vibrations, respectively. The band at 561 cm −1 was due to the Co-O stretching vibrations 36,40 . The full scan XPS survey of MOFs and derived oxides confirmed the presence of Na in the materials along with C, O, and Co (Fig. 2d). The C peak in NaCo x O y materials was due to the adventitious carbon. Na 1s and Na KLL peak intensity increased in the NaCo x O y compared to the MOFs due to the loss of carbon after calcination.
The HRXPS spectra of NCO-D and NCO-T are shown in Fig. 3, and the curve-fitting parameters are listed in Tables S6 and S7. The HRXPS Na 1s spectra of NCO-D (Fig. 3a) and NCO-T (Fig. 3d) has a peak at 1070.6 and 1070.8 eV, respectively, for Na + ions 43 . The HRXPS Co 2p spectrum of NCO-D has peaks at 779.8 and

H 2 S breakthrough studies. The breakthrough curves for MOFs and oxides in dry and moist conditions
are shown in Fig. 4. CoBDC and CoBTC showed a low adsorption capacity of 1.6 and 5.7 mg g −1 , respectively, in dry condition (Fig. 4a). The studies dealing with Co-based MOFs for H 2 S removal are absent in the literature. The closest study is the role of Co ions in UiO-67(bipy) for H 2 S removal. The post-synthetic inclusion of Co in the MOF could not substantially improve its H 2 S removal capacity. The Co-sites in a highly porous MOF with a surface area of ~ 2500 m 2 g −1 failed to interact with H 2 S gas 46 . The low adsorption capacity of Co-MOFs in the present study was probably due to the low Co-density and poor accessibility of Co-sites in MOFs for H 2 S interaction. The NCO-D and NCO-T had an adsorption capacity of 133.9 and 134.6 mg g −1 , respectively, which was a significant improvement compared to MOF precursors. In the moist condition, the adsorption capacity of MOFs and derived oxides decreased. In general, the presence of moisture plays a positive role in the H 2 S adsorption process by dissociating H 2 S molecules in the water film 37,47 . The adsorption capacity could decrease due to the competitive behaviour of water molecules for the adsorption sites 48 . Moreover, the formation of sulfuric acid could lower the structural integrity of the adsorbent and decrease its adsorption capacity. In the present case, moisture alone can destroy the adsorbent structure due to the hygroscopic nature of sodiated transition metal oxides 49 . Nevertheless, the H 2 S adsorption capacity was satisfactorily preserved even in the presence of moisture. The higher adsorption capacity of NCO-T was due to its comparatively higher surface area than NCO-D. The surface area of nonporous adsorbent is highly relevant in the adsorption of gases. Zheng et al. reported an increased CO 2 adsorption capacity (1.02-2.83 cm 3 g −1 ) in KNbWO 6 ·H 2 O pyrochlore with an increase in the surface area (1.82-2.90 m 2 g −1 ) after Sn 2+ substitution 50 . Thus, the surface area plays a decisive role in the gas adsorption capacity of nonporous materials. The N 2 adsorption-desorption isotherms of spent NCO-D and NCO-T are available in Fig. S5. The surface area of spent NCO-D and NCO-T was 0.94 and 1.16 m 2 g −1 , respectively. The deposition of elemental sulfur and sulfate species was responsible for the decreased surface area. The decreased surface area after H 2 S exposure further supported the fact that a low surface area restricts the gas diffusion process and limits the adsorption capacity. The effect of dosage and flowrate on the adsorption capacity was studied for NCO-D and NCO-T (Fig. 5). The adsorption capacity was negatively impacted by the increasing adsorbent mass (Fig. 5a,b). The adsorption capacity of 154.6 and 168.2 mg g −1 with a 0.2 g dosage reached 117.6 and 118.9 mg g −1 with 0.4 g dosage for NCO-D and NCO-T, respectively. The decreased adsorption capacity was due to the formation of dead zones in the bed with the increasing bed loading, which remained unused during the initial phase of the adsorption process 51 . The adsorption capacity decreased with the increasing H 2 S flow rate (Fig. 5c,d). The adsorption capacity of 154.6 and 168.2 mg g −1 (0.1 L min −1 ) for NCO-D and NCO-T dropped to 134.8 and 133.9 mg g −1 (0.3 L min −1 ), respectively. The decrease in the adsorption capacity with the increasing flow rate was due to the insufficient contact time for adsorbate-adsorbent interactions at a higher flow rate. The impact was stronger due to the low surface area and porosity of the metal oxide adsorbents 52 .
The adsorption capacity of synthesized oxides was compared with the Co-based adsorbents reported in the literature ( Table 1). The H 2 S uptake capacity of NCO-D and NCO-T were superior to many of the reported adsorbents. The higher adsorption capacity of the Co 3 O 4 -SiO 2 composite was due to its high porosity. Only Zn-Co hydroxide had a higher capacity than NCO adsorbents. Nevertheless, the MOF-derived NaCo x O y  www.nature.com/scientificreports/ adsorbents reported in these studies are unique and highly effective in removing H 2 S from effluent gases at room temperature.
Adsorption mechanism. The distribution of sulfur in the spent NCO-D was probed through TEM-EDS analysis (Fig. 6a). The EDS map has peaks for Co, O, and Na with a new high-intensity peak for S at ~ 2.1 keV. The 2D elemental mapping confirmed the wide distribution of Na, Co, and O in the NCO-D adsorbent. Apart from the constituent elements, uniform distribution of S was observed in the spent adsorbent. The PXRD patterns of fresh and spent NCO-D are shown in Fig. 6b 14 . The PXRD pattern of spent NCO-D has sharp peaks in the entire 15°-70° range. These peaks were assigned to Co 3 S 4 , CoSO 4 , Co 3 O 4 , and Co(OH) 2 phases. The XPS surveys of fresh and spent NCO-D are shown in Fig. 6c. The XPS survey of spent NCO-D has peaks for S 2p and S 2s at ~ 165 and ~ 234 eV, respectively. The S present in the spent sample accounted for 14.2% of the total atomic composition. www.nature.com/scientificreports/ The HRXPS Na 1s peak at 1070.6 eV for fresh NCO-D shifted to 1071.1 eV for spent NCO-D (Fig. 7a). The shift in the binding energy of Na 1s peak by 0.3 eV was probably due to the redistribution of electron density after the formation of different sulfide and oxide phases of cobalt. In the HRXPS Co 2p spectrum of spent NCO-D, the peaks at 780.3 and 796.7 eV were assigned to the Co 2p 3/2 and Co 2p 1/2 , respectively (Fig. 7b). The HRXPS Co 2p 3/2 peak has two contributions at 779.7 and 781.2 eV for Co 3+ (39.9%) and Co 2+ (60.7%) sites, respectively. In the fresh NCO-D, the Co 3+ and Co 2+ contribution was 36.4 and 63.6%, respectively. The variation in the contributions of Co 3+ and Co 2+ sites in the NCO-D was due to the involvement of the Co 3+ /Co 2+ catalytic cycle in the H 2 S oxidation process. The HRXPS O 1s spectrum of spent NCO-D has two contributions at 530.0 and 531.0 eV for O-Na/O-Co and O-S/O-H, respectively (Fig. 7c). The increase in the hydroxyl density over the adsorbent was most likely due to the interaction of H 2 S molecules with the surface lattice O 2− to yield HSand -OH groups 55 . Moreover, the absence of surface H 2 O in the spent sample hinted towards its full utilization to form sulfates and Co(OH) 2 . The HRXPS S 2p spectra of spent NCO-D is shown in Fig. 7d. The spectrum was fitted into six peaks for different sulfur species. The peaks at 162.0 and 163.2 eV were assigned to the S 2p 3/2 and S 2p 1/2 of sulfide species, respectively, which were bound to Co ions in Co 3 S 4 15,56 . The sulfide species was primarily formed due to the reactive interaction of adsorbed H 2 S molecules with the lattice oxygen. The peaks at 164.1 and 165.3 eV were attributed to the S 2p 3/2 and S 2p 1/2 of elemental sulfur, respectively 18 . The peaks at 168.1 and 169.2 eV were assigned to the S 2p 3/2 and S 2p 1/2 of sulfate ions, respectively. These sulfate ions were considered as CoSO 4 , observed in the PXRD pattern as well 18,37 . Thus, all three major sulfur species were conclusively detected in the XPS analysis. Moreover, the sulfide, sulfur, and sulfate contribution in the total sulfur content were 39.1, 11.1, and 49.8%, respectively. Based on the above discussion, the following reactions have been proposed for the H 2 S removal over NCO-D.  www.nature.com/scientificreports/

Conclusion
We have reported a novel approach for the fabrication of NaCo x O y adsorbents by air calcination of (Na,Co)organic frameworks. NaCo x O y were formed irrespective of the type of organic linkers used in the MOF precursor. Moreover, the oxides crystallized as microsheets of 100-200 nm thickness with the presence of some polyhedral nanocrystals. These macroporous oxides have a surface area in the range of 1.15-1.90 m 2 g −1 . X-ray photoelectron spectroscopy (XPS) analysis confirmed the near equal presence of Co 2+ and Co 3+ sites in MOFs, which were largely preserved in the NaCo x O y . The maximum adsorption capacity of 168.2 mg g −1 was recorded for NCO-T in dry conditions. The competitive nature of water molecules led to the decrease in adsorption capacity in moist condition. The adsorption capacity decreased with the increasing flow rate and bed loading due to the insufficient contact time for adsorbate-adsorbent interactions and the formation of dead zones, respectively. TEM-EDAX analysis confirmed abundant and uniform distribution of sulfur in the adsorbent. PXRD analysis of the spent sample suggested the formation of Co 3 S 4 , CoSO 4 , Co 3 O 4 , and Co(OH) 2 after the H 2 S exposure. The products of the H 2 S adsorption-oxidation process were further confirmed by XPS analysis. Thus, we have reported highly efficient adsorbents for the adsorptive-oxidative removal of H 2 S gas in ambient conditions.