Deoxygenation of Stearic Acid over Cobalt-Based NaX Zeolite Catalysts

For the production of sustainable biofuels from lipid biomass it is essential to develop non-noble metal catalysts with high conversion and selectivity under inert gas atmospheres. Herein, we report a novel cobalt-based catalyst supported on zeolite NaX via ion-exchange synthesis. The resultant bifunctional cobalt-based NaX zeolite catalyst displayed high conversion of stearic acid to liquid fuels. In addition, the effect of reaction temperature and catalyst loading was studied to evaluate the order of reaction and activation energy. Decarboxylation and decarbonylation were the dominant deoxygenation pathways. Stearic acid was successfully deoxygenated in N2 atmospheres using Co/NaX catalysts with a conversion as high as 83.7% and a yield to heptadecane up to ~28%. Furthermore, we demonstrate that higher reaction temperatures resulted in competing pathways of decarboxylation and decarbonylation. Finally, the fresh and recycled catalysts were characterized showing modest recyclability with a ~12.5% loss in catalytic activity.


Introduction
Fatty acids are readily extracted from non-edible plants [1,2], algae oil [3], and tallow [4].These abundant compounds make ideal candidates for the production of biofuels and chemicals.Homogeneous base catalysis of fatty acids yields first-generation fatty acid methyl esters (FAME) by transesterification.Unfortunately, FAMEs have a lower energy density, higher cloud and smoke point, and poor cold flow properties compared to conventional fuels limiting their potential as a "drop-in" fuel solution [5].As a result, many researchers have turned to heterogeneous catalysts for the selective deoxygenation of fatty acids with the aim of producing diesel range alkanes.Two predominant pathways for the production of alkanes are hydrodeoxygenation resulting in the retention of the parent molecule's carbon number under hydrogen replete atmospheres and decarboxylation which removes one carbon atom as a carboxylate group under hydrogen deplete atmospheres [5].
Aspects of interest in the valorization of fatty acids include (i) one pot continuous conversion of both saturated and unsaturated fatty acids; (ii) employing non-noble metals for cheaper catalysts; and (iii) decreasing the cost of reaction conditions by lowering temperature/pressure and eliminating hydrogen [6].With respect to (i), it has been shown that a higher degree of unsaturation in the feedstock results in lower conversion and selectivity [7][8][9].Regarding (ii), lower selectivities to alkanes have been shown when removing Pt or Pd from the catalyst [10,11].Finally, (iii) has proven challenging as hydrogen is required for the saturation of unsaturated fatty acids, leading to attempts to produce in-situ hydrogen from solvents or from sacrificial reactions [11,12].Recent work has attempted to tackle these concerns by introducing bifunctional catalysts which provide both hydrogenation activity and solid acid activity [13][14][15][16][17][18].Zeolites offer highly desirable properties such as high surface areas, good chemical and thermal stability, and strong acidity as a support material for metals [19].
In addition, zeolites have shown industrial maturity and scalability in petroleum processes giving further practicality to their use [20].
Zeolites can be tuned to have specific acidity controlled by the ratio of silica to alumina (Si/Al).More framework aluminum results in a higher concentration of negative charges and higher acidity.Cobalt can be introduced to zeolite supports by ion-exchange synthesis, by which the pore filling agent, in our case, sodium exchanges with the cationic divalent metal in solution.This synthesis method has been applied successfully with cobalt and nickel to a number of zeolites [21,22] and has shown significant benefits when compared to incipient wetness impregnation for the deoxygenation of fatty acids [23].
The zeolite support NaX provides a large surface area, playing an important role in the dispersion of cobalt.Additionally, zeolite NaX is well established as an ideal candidate for ion-exchange due to its large unit cell capacity of approximately 86 cations [24].Brønsted and Lewis acidity in ion-exchanged NaX zeolites have been well studied [25][26][27].In a recent study focusing on monovalent and divalent ion-exchange in NaX it was shown that the intrinsic Brønsted acidity of the zeolite (160 µmol/g) decreased as the degree of divalent exchange increased.The NaX reached and maintained a minimum Brønsted acidity of approximately 25-40 µmol/g after the divalent ion had replaced over 60% of the Na + [27].This residual Brønsted acidity coupled with presence of metallic acid sites is hypothesized to provide bifunctionality in our catalyst system.Our group [28,29] and independent groups [30,31] have observed that acidic supports enhance metal dispersion resulting in improved catalytic activity for decarboxylation and hydrodeoxygenation reactions.The catalytic ability of Co based catalysts for hydrodeoxygenation and decarboxylation reactions has been documented [32][33][34][35][36].For example, Centeno et al. reported the use of Co-noble metal based catalysts for the hydrodeoxygenation reaction of different model molecules containing carbonyl, carboxyl, hydroxyl and methoxy groups [33].The authors found that the presence of Co favored the decarboxylation activity.A synergistic effect between Co and a noble metal was responsible for this catalytic behavior.Wang et al. attributed also a positive synergistic effect when Co was incorporated into Mo based catalysts leading to an enhanced catalytic activity for hydrodeoxygenation of bio-oils [32].Furthermore, it has been reported that cobalt-based catalysts can enhance the C(sp 2 )-O cleavage via direct deoxygenation pathway [34].Zhang et al. found that Co promoted by MoS 2 favored hydrodecarbonylation yielding more alkene products compared to linear alkanes [36].Herein, we report a bifunctional cobalt based zeolite NaX (denoted as Co/NaX) supported catalyst implemented under mild reaction conditions and an inert reaction atmosphere yielding high conversion and modest selectivity to heptadecane.

Nitrogen Physisorption Isotherms
Nitrogen isotherms were collected to estimate surface area, pore volumes, and pore diameters of the parent zeolite NaX, the fresh Co/NaX, and recycled Co/NaX catalysts after use in the deoxygenation reaction (Figure 1).The BET [31] surface area of the NaX zeolite was 788 m 2 /g in close agreement with previous literature [37].The BET surface area of Co/NaX and recycled Co/NaX were 447 and 391 m 2 /g, respectively (Table 1).A clear trend is the reduction in the low pressure uptake of nitrogen, correlated to the surface area, due to the incorporation of cobalt in the zeolites under the alkaline synthesis conditions.Other researchers have seen a similar trend, with a loss in microporous surface area due to metal incorporation in basic synthesis solutions [38,39].The Co/NaX suffers a 13% loss in surface area after one reaction with stearic acid, which could be tentatively attributed to surface carbon coking the active sites [40].The Co/NaX also exhibited a ~13% loss in total pore volume after one reaction.Another important trend is the appearance of mesoporosity after the alkaline synthesis of Co/NaX.Under alkaline synthesis conditions (pH = 8) in low Si/Al zeolites, desilication and mesopore formation has been shown to occur [41].Mesopore formation may also be attributed to the incorporation of the divalent Co 2+ cation in exchange for monovalent Na + , which could result in framework alteration [27,42,43].By introducing mesoporosity in the catalyst, diffusion pathways of large fatty acid molecules are likely less inhibited, giving an argument for the compromise of high surface area. in framework alteration [27,42,43].By introducing mesoporosity in the catalyst, diffusion pathways of large fatty acid molecules are likely less inhibited, giving an argument for the compromise of high surface area.Pore diameter was evaluated using BJH method on the desorption branch of the isotherm (Figure 2).A high nitrogen uptake at ~15 Å is highly pronounced in the parent NaX zeolite in good agreement with previous literature [37].This uptake reduces in the Co/NaX and recycled samples, indicating fewer micropores.A secondary peak at 45 Å shows increasing intensity in the Co/NaX and recycled samples, correlating well with the increase in mesoporosity.  External surface area and micropore volume were calculated by the t-plot method,2 total volume was calculated from the quantity adsorbed at P/P 0 = 0.975,3 mesopore volume = total volume − micropore volume, 4 mesopore volume was calculated by BJH method.
Pore diameter was evaluated using BJH method on the desorption branch of the isotherm (Figure 2).A high nitrogen uptake at ~15 Å is highly pronounced in the parent NaX zeolite in good agreement with previous literature [37].This uptake reduces in the Co/NaX and recycled samples, indicating fewer micropores.A secondary peak at 45 Å shows increasing intensity in the Co/NaX and recycled samples, correlating well with the increase in mesoporosity.

Powder X-Ray Diffraction
The synthesized NaX zeolite exhibits high crystallinity matching well with the simulated X-ray diffraction (XRD) pattern (Figure 3).After ion-exchange synthesis with Co(NO3)2, calcination, and reduction, the presence of strong Co3O4 peaks were observed at 36.49° and 43.20° corresponding to (331) and (400) planes, respectively (JCPDS # 76-1802).The cobalt oxide peaks gave a calculated average nanoparticle size of 17.9 and 19.6 nm for the fresh and recycled Co/NaX catalysts respectively from the Scherrer equation (davg = Κλ/ωcosϴ, where davg is the estimated average particle size, Κ = 0.9 is the shape factor, λ = 1.5406Å is the Cu Kα radiation, ω is the full width of the half maximum intensity, and ϴ is the Bragg angle).The increase in nanoparticle size may be due to sintering of metallic species during the regeneration steps.The strong cobalt oxide peak is characteristic of high cobalt loadings in the pore of the zeolite.It should be noted that complete reduction of the sample was not achieved at the conditions employed.Use of higher reduction temperatures would promote metallic species formation, but may also result in larger particle agglomeration [44].

Powder X-Ray Diffraction
The synthesized NaX zeolite exhibits high crystallinity matching well with the simulated X-ray diffraction (XRD) pattern (Figure 3).After ion-exchange synthesis with Co(NO 3 ) 2 , calcination, and reduction, the presence of strong Co 3 O 4 peaks were observed at 36.49• and 43.20 • corresponding to (331) and (400) planes, respectively (JCPDS # 76-1802).The cobalt oxide peaks gave a calculated average nanoparticle size of 17.9 and 19.6 nm for the fresh and recycled Co/NaX catalysts respectively from the Scherrer equation (d avg = Kλ/ωcosθ, where d avg is the estimated average particle size, K = 0.9 is the shape factor, λ = 1.5406Å is the Cu Kα radiation, ω is the full width of the half maximum intensity, and θ is the Bragg angle).The increase in nanoparticle size may be due to sintering of metallic species during the regeneration steps.The strong cobalt oxide peak is characteristic of high cobalt loadings in the pore of the zeolite.It should be noted that complete reduction of the sample was not achieved at the conditions employed.Use of higher reduction temperatures would promote metallic species formation, but may also result in larger particle agglomeration [44].

Thermogravimetric Analysis
Catalysts were evaluated for thermal stability by thermogravimetric analysis (TGA) (Figure 4).Weight loss up to 250 °C was attributed to physisorbed water, and loss between 250-700 °C was attributed to carbonaceous coke on the spent sample [45].Fresh and recycled samples exhibit high thermal stability up to 700 °C.The spent catalyst, after a typical reaction (T = 280 °C, Loading = 250 mg), showed a weight loss of 36% attributed to coke deposited on the surface.High coke contents have been reported in zeolite materials for biomass upgrading and can lead to deactivation of the catalyst [45,46].

Thermogravimetric Analysis
Catalysts were evaluated for thermal stability by thermogravimetric analysis (TGA) (Figure 4).Weight loss up to 250 • C was attributed to physisorbed water, and loss between 250-700 • C was attributed to carbonaceous coke on the spent sample [45].Fresh and recycled samples exhibit high thermal stability up to 700 • C. The spent catalyst, after a typical reaction (T = 280 • C, Loading = 250 mg), showed a weight loss of 36% attributed to coke deposited on the surface.High coke contents have been reported in zeolite materials for biomass upgrading and can lead to deactivation of the catalyst [45,46].

Thermogravimetric Analysis
Catalysts were evaluated for thermal stability by thermogravimetric analysis (TGA) (Figure 4).Weight loss up to 250 °C was attributed to physisorbed water, and loss between 250-700 °C was attributed to carbonaceous coke on the spent sample [45].Fresh and recycled samples exhibit high thermal stability up to 700 °C.The spent catalyst, after a typical reaction (T = 280 °C, Loading = 250 mg), showed a weight loss of 36% attributed to coke deposited on the surface.High coke contents have been reported in zeolite materials for biomass upgrading and can lead to deactivation of the catalyst [45,46].

Attenuated Total Reflectance Fourier Transform Infrared Spectroscopy
For the evaluation of the surface composition of the synthesized catalysts, attenuated total reflectance Fourier transform infrared spectroscopy was employed (ATR-FTIR).ATR-FTIR of ion-exchanged zeolite materials have been modeled and studied by Król et al. [47].They suggest that the band located at 560 cm −1 is sensitive to the ion-exchange synthesis and acts as an indicator of the sorption process.Shown in Figure 5A, both Co/NaX and recycled catalysts show major increases at the stretch at 560 cm −1 , indicating a successful ion-exchange of Na + for Co 2+ .

Attenuated Total Reflectance Fourier Transform Infrared Spectroscopy
For the evaluation of the surface composition of the synthesized catalysts, attenuated total reflectance Fourier transform infrared spectroscopy was employed (ATR-FTIR).ATR-FTIR of ionexchanged zeolite materials have been modeled and studied by Król et al. [47].They suggest that the band located at 560 cm −1 is sensitive to the ion-exchange synthesis and acts as an indicator of the sorption process.Shown in Figure 5A, both Co/NaX and recycled catalysts show major increases at the stretch at 560 cm −1 , indicating a successful ion-exchange of Na + for Co 2+ .Coking on zeolitic materials can be qualitatively observed by ATR-FTIR.Generally, bridging hydroxyl groups are observed in the region between 3200-3800 cm −1 in zeolites, shifting with a dependence on cavity size, Si/Al ratio, and cation occupancy [48].The υ(O-H) region around 3430 cm −1 has been shown to adopt this position in highly acidic zeolites [49].Shifting of this peak in the cobalt exchanged catalyst to 3340 cm −1 is hypothesized to be caused by interactions of the hydroxyls with the new metal cation (Figure 5B).The recycled catalysts loss of intensity could indicate graphitic coke coverage on the bridging hydroxyl groups [50].Coke composition on the catalyst is discussed in section 3.2.

Field-Emission Scanning Electron Microscopy
Field-Emission scanning electron micrographs of the parent NaX, the fresh Co/NaX, and the recycled Co/NaX catalyst are shown in Figure 6.A clear morphology shift from smooth octahedral NaX zeolite to a roughened surface containing cobalt is the result of the alkaline synthesis.This is in good agreement with the increase in external surface area calculated from the N2-isotherm.The location of the cobalt species and oxygen species were elucidated by energy-dispersive X-ray spectroscopy (EDX) shown in Figure S1.Clearly, the cobalt was well dispersed across the surface of the zeolite with no major agglomeration.Additionally, the cobalt appeared to be dispersed similarly to the oxygen, silicon, and aluminum, suggesting that the cobalt was located similarly throughout the cavities of the zeolite.
Using EDX quantification, the fresh and recycled Co/NaX cobalt loading was calculated to be similar at ~23% and ~26%, respectively (Figure S2).The slightly higher loading in the recycled catalyst was within the experimental error range of EDX.Nevertheless these results indicate that Co leaching was negligible due to the reaction.The Si/Al ratio of the parent NaX, fresh Co/NaX, and recycled Co/NaX was found to be 1.3, 1.6, and 1.5, respectively in close agreement with previously reported ion-exchanged NaX materials [27].The Na/Al ratio was used to observe the extent of ion-exchange Coking on zeolitic materials can be qualitatively observed by ATR-FTIR.Generally, bridging hydroxyl groups are observed in the region between 3200-3800 cm −1 in zeolites, shifting with a dependence on cavity size, Si/Al ratio, and cation occupancy [48].The υ(O-H) region around 3430 cm −1 has been shown to adopt this position in highly acidic zeolites [49].Shifting of this peak in the cobalt exchanged catalyst to 3340 cm −1 is hypothesized to be caused by interactions of the hydroxyls with the new metal cation (Figure 5B).The recycled catalysts loss of intensity could indicate graphitic coke coverage on the bridging hydroxyl groups [50].Coke composition on the catalyst is discussed in Section 3.2.

Field-Emission Scanning Electron Microscopy
Field-Emission scanning electron micrographs of the parent NaX, the fresh Co/NaX, and the recycled Co/NaX catalyst are shown in Figure 6.A clear morphology shift from smooth octahedral NaX zeolite to a roughened surface containing cobalt is the result of the alkaline synthesis.This is in good agreement with the increase in external surface area calculated from the N 2 -isotherm.The location of the cobalt species and oxygen species were elucidated by energy-dispersive X-ray spectroscopy (EDX) shown in Figure S1.Clearly, the cobalt was well dispersed across the surface of the zeolite with no major agglomeration.Additionally, the cobalt appeared to be dispersed similarly to the oxygen, silicon, and aluminum, suggesting that the cobalt was located similarly throughout the cavities of the zeolite.
Using EDX quantification, the fresh and recycled Co/NaX cobalt loading was calculated to be similar at ~23% and ~26%, respectively (Figure S2).The slightly higher loading in the recycled catalyst was within the experimental error range of EDX.Nevertheless these results indicate that Co leaching was negligible due to the reaction.The Si/Al ratio of the parent NaX, fresh Co/NaX, and recycled Co/NaX was found to be 1.3, 1.6, and 1.5, respectively in close agreement with previously reported ion-exchanged NaX materials [27].The Na/Al ratio was used to observe the extent of ion-exchange trending downward from the parent zeolite NaX to the fresh and recycled Co/NaX giving 0.72, 0.23, 0.26, respectively.This trend indicates that the cobalt replaced the sodium in the zeolite cages.

Deoxygenation of Stearic Acid over Co/NaX in N2
Deoxygenation of stearic acid was conducted in the presence of nitrogen over Co/NaX catalysts.The reaction temperature was varied between 260-300 C.Table 2 shows the catalytic performance of the parent zeolite NaX and Co/NaX at various reaction temperatures and catalysts loadings.The conversion correlated to the reaction temperature, with the lowest conversion at 260 °C and the highest conversion at 300 °C.Selectivity to heptadecane (C17) was highest at 280 °C.Yield to C17 was highest at 280 °C followed closely by 300 °C giving 27.9 and 25.7%, respectively.The reaction conducted at 280 °C was repeated to ensure reproducibility.Two major reaction pathways for the Co/NaX catalyst included decarboxylation and decarbonylation evidenced by the dominant formation of C17 and C17-ene products.Similar results were found by Zhang et al. showing selectivity losses to alkene products [36].Additionally, some octadecene was observed, a result of the hydrodecarbonylation pathway [51].The formation of octadecene requires the presence of Brønsted acid sites [52], which have been shown to be present in ion-exchanged zeolite NaX [27].We hypothesize that both the Brønsted acidity of the support and residual surface hydrogen on the reduced Co/NaX catalyst provided protonic H + [53] which could have participated in the hydrodecarbonylation reaction.This would also explain the observed formation of the intermediate hydrodecarbonylation species, octadecanol (observed in the "other" column in Table 2).The proposed pathway is outlined in Scheme 1. Overall, the non-noble metal Co/NaX catalyst provided good conversion and modest selectivity to heptadecane.The catalyst loading had an effect on the selectivity.At low loadings, low conversion was observed, along with very high selectivity to C17 products.This trend may be due to the relatively high loading of cobalt on the catalyst, which promotes decarbonylation in competition with decarboxylation. .Additionally, some octadecene was observed, a result of the hydrodecarbonylation pathway [51].The formation of octadecene requires the presence of Brønsted acid sites [52], which have been shown to be present in ion-exchanged zeolite NaX [27].We hypothesize that both the Brønsted acidity of the support and residual surface hydrogen on the reduced Co/NaX catalyst provided protonic H + [53] which could have participated in the hydrodecarbonylation reaction.This would also explain the observed formation of the intermediate hydrodecarbonylation species, octadecanol (observed in the "other" column in Table 2).The proposed pathway is outlined in Scheme 1. Overall, the non-noble metal Co/NaX catalyst provided good conversion and modest selectivity to heptadecane.The catalyst loading had an effect on the selectivity.At low loadings, low conversion was observed, along with very high selectivity to C 17 products.This trend may be due to the relatively high loading of cobalt on the catalyst, which promotes decarbonylation in competition with decarboxylation.

Effect of Catalyst Loading and Temperature
To better understand the behavior of the Co/NaX catalyst under nitrogen atmosphere, first the loading of the catalyst was varied.When comparing the catalyst loading (15, 150, 250 mg) to a first order rate expression calculated from the stearic acid conversion, a linear relationship was observed (R 2 = 0.999), indicating a pseudo-first-order reaction (Figure 7A).Therefore, based on the first-order nature of the disappearance of stearic acid, we calculated the apparent activation energy of the reaction.An Arrhenius plot (Figure 7B) was prepared from the explored reaction temperatures (533, 553, 573 K) with all other reaction conditions held constant.From a linear regression (R 2 = 0.9948) the apparent activation energy was calculated to be 93.5±0.5 kJ/mol.This value for the decarboxylation of stearic acid (the most dominant pathway) is in good agreement with values found by other researchers [54][55][56].

Effect of Catalyst Loading and Temperature
To better understand the behavior of the Co/NaX catalyst under nitrogen atmosphere, first the loading of the catalyst was varied.When comparing the catalyst loading (15, 150, 250 mg) to a first order rate expression calculated from the stearic acid conversion, a linear relationship was observed (R 2 = 0.999), indicating a pseudo-first-order reaction (Figure 7A).Therefore, based on the first-order nature of the disappearance of stearic acid, we calculated the apparent activation energy of the reaction.An Arrhenius plot (Figure 7B) was prepared from the explored reaction temperatures (533, 553, 573 K) with all other reaction conditions held constant.From a linear regression (R 2 = 0.9948) the apparent activation energy was calculated to be 93.5±0.5 kJ/mol.This value for the decarboxylation of stearic acid (the most dominant pathway) is in good agreement with values found by other researchers [54][55][56].

Catalyst Recyclability
Physically, the catalyst has many desirable properties including thermal stability and maintenance of crystallinity evidenced by the XRD and TGA results.In addition, the catalyst did not exhibit significant metal leaching according to the loading from EDX.However, after calcination and

Catalyst Recyclability
Physically, the catalyst has many desirable properties including thermal stability and maintenance of crystallinity evidenced by the XRD and TGA results.In addition, the catalyst did not exhibit significant metal leaching according to the loading from EDX.However, after calcination and reduction, the catalyst exhibited a loss of catalytic activity.From the TGA analysis, it was apparent that a large amount of coke (36 wt.%) was present on the catalyst.Further analysis of the coke as described by Migliori et al. provided some insight into the species present at the surface [57,58].The carbonaceous coke was mostly composed of large aromatic species and oxygenates (Table 3).The graphitic compounds likely remained on the active sites even after regeneration [59] contributing to the ~12.5% loss in activity when compared to the fresh catalyst.The ATR-FTIR spectra indicated that carbonaceous coke was likely present on the recycled catalyst surface evidenced by the disappearance of the broad stretch in the 3200-3500 cm −1 region.Such coke deposits have been widely shown to cause deactivation in heterogeneous catalysts.This is supported by the lower surface area displayed by the recycled catalyst.4 compares the state-of-the-art Co-based catalysts for deoxygenation of fatty acids to liquid hydrocarbons.Our catalysts show comparable conversions to previous reports in which hydrogen was employed as the gas atmosphere, and the best performance under inert conditions (nitrogen).Although higher yields have been observed under hydrogen environment, the use of hydrogen in the conversion of fatty acid involves processing safety concerns on scale-up along with potentially negative environmental impacts due to the energy intensity involved in obtaining the gas [60,61].For these reasons, we believe that the Co/NaX catalysts may contribute in a meaningful way to the advance of biofuel production under inert environments.Nevertheless, more research is needed to improve both the yield to deoxygenation products, and to improve the catalytic stability of Co/NaX catalysts.

Zeolite NaX Synthesis
Zeolite NaX was synthesized in a molar ratio of 3.5Na 2 O:Al 2 O 3 :4SiO 2 as described elsewhere [68].Briefly, a solution of sodium aluminate and an aqueous solution of sodium hydroxide was mixed homogeneously.Then, Ludox AS-40 was added and stirred for 40 min at room temperature.Next, 40 mL of solution was transferred into a 100 mL PTFE lined Parr autoclave and heated to 90 • C for 15 h at 1 • C/min heating and cooling rates.After, the zeolite was washed by centrifugation until the solution was below a pH = 8.The zeolite was then dried overnight at 100 • C and used for the synthesis of Co/NaX.

Co/NaX Ion-Exchange Synthesis
An aqueous solution of distilled water and Cobalt (II) nitrate hexahydrate (0.28 M, 50 mL) was stirred until homogeneous.Next, 5.025 g of zeolite NaX prepared as described above was added to the cobalt nitrate solution and stirred homogenously.Dropwise addition of 3% ammonia solution was conducted until pH = 8.The solution was covered and stirred at room temperature for 180 min.Finally, the catalyst was collected by vacuum filtration and washing three times with distilled water.The resulting solid was dried overnight at 110 • C, calcined at 450 • C with a 1 • C/min ramp rate, and reduced under 150 mL/min H 2 at 460 • C with a 1 • C/min ramp rate.After reduction, the obtained Co/NaX was employed as catalyst for the deoxygenation of stearic acid.

Catalyst Characterization
Fresh and recycled catalysts were characterized by the following analyses: Textural properties were evaluated by collecting nitrogen adsorption-desorption isotherms.Prior to these experiments, samples were degassed at 300 • C for 5 h (ASAP 2020 porosimeter, Micromeritics, Norcross, GA, USA).The crystalline phase of the catalysts was determined by XRD (Siemens, Munich, Germany, Kristalloflex800, 25 mA, 30 kV, CuKα radiation).FE-SEM images were obtained using a field emission gun and accelerating voltage of 10 kV (JEOL ISM-7000F, Peabody, MA, USA).ATR-FTIR was conducted using 10 scans, 4 cm −1 resolution, DTGS detector (Thermo Scientific Nicolet iS50 FT-IR, Madison, WI, USA).TGA was conducted with at least 10 mg of sample, ramped from room temperature to 800 • C, 10 • C/min, 50 mL/min N 2 , 50 mL/min air (TA Instruments, New Castle, DE, USA).For coke compositional analysis, methods were adapted from Migliori et al [57].Briefly, 15 mg of Co/NaX was completely dissolved in 1 mL of 48% hydrofluoric acid after 15 min.Then, 5 mL of methylene chloride was added and mixed for 2 min.Finally, the solution was allowed to separate into the two phases for 1 h.After, 2 µL of the organic phase was sampled for GC-MS, following the procedure outlined in the liquid product analysis.

Catalytic Reaction: Deoxygenation Experiments
In a typical reaction, 250 mg of Co/NaX catalyst was loaded with 1000 mg of stearic acid in a 100 mL Parr batch reactor.10 mL of hexane was added and the reactor was sealed.The reactor was then pressurized with nitrogen to 10 bar and flushed three times to remove stagnant air.The reactor was finally pressurized to 10 bar N 2 and sealed.Stirring at 600 rpm was initiated and the reactor was heated rapidly to 280 • C and held for 2 h.The reaction conditions were shown to be in the kinetic regime (η = 0.99, Table S1) [69][70][71].The pressure after the heating phase was noted to be ~35 bar on reaching 280 • C and 38 bar after 2 h at 280 • C.After the reaction was complete, the reactor was cooled with forced air until at room temperature.Finally, the chamber was depressurized and the resultant liquid products were sampled.The catalyst was recovered from the liquid products by suspension in 20 mL of hexane and sonicated for 5 min.The sonicated solution was then centrifuged, decanted, and the resultant solid (spent catalyst) was calcined at 400 • C for 4 h, 5 • C/min ramp, and reduced at the same conditions described in Section 4.3.The obtained recycled Co/NaX was then characterized and evaluated for an additional reaction.

Liquid Product Analysis
Products were weighed and added along with 10 µL of BTSFA, 5 µL of n-Hexadecane, and 1000 µL of hexane into a glass 2 mL vial.The contents were sealed and placed in an oven at 60 • C for 60 min to allow the silylation of the fatty acids.A gas chromatography (Agilent GC, 6980N, Santa Clara, CA, USA) mass spectrometer (Agilent MSD, 5973N) equipped with a HP-5 molecular sieve column (30 m × 250 µm × 0.25 µm) in split mode with carrier gas He (10:1) was used for product detection.The conversion of stearic acid was calculated by the removal of the carboxylic acid groups by the acid value test (ASTM D5558).The acid value was calculated by Equation (1): where N is the normality of the sodium hydroxide solution, w is the mass of the sample, and V f and V o are the initial and final volume of sodium hydroxide.The acid value of stearic acid was found to be 194 in good agreement with literature values.The conversion, X TOT , of stearic acid was calculated by Equation (2): X TOT = (AV SA − AV Prod )/AV SA * 100 where AV SA and AV Prod were the acid value of stearic acid and the products, respectively.

Conclusions
In summary, we demonstrate that cobalt-based catalysts supported on zeolite NaX displayed catalytic activity for the deoxygenation of stearic acid to heptadecane in the presence of inert environment.The high conversion of the Co/NaX catalyst was attributed to the mesoporosity introduced in the alkaline synthesis shown in the N 2 -isotherm, the bifunctionality of the acidic zeolite support, and the high loading of active Co metal.Heptadecane yields as high as 27.9% were observed at an optimal 280 • C under nitrogen atmospheres through the decarboxylation pathway.The highest conversion, 95%, was observed at the highest explored temperature, 300 • C, with some heptadecane selectivity losses to alkenes.Some hydrodecarbonylation products were observed and were attributed in part to the hypothesized Brønsted acidity of the support and residual H + protons on the catalyst after partial reduction.The catalyst was regenerated and tested for recyclability resulting in activity loss of ~12.5%, which was attributed to coke deposition.

Figure 3 .
Figure 3. Powder XRD of the catalysts.Cobalt oxide peaks are denoted by ♠️.

Figure 4 .
Figure 4. TGA of the spent catalyst shows a significant accumulation of coke.

Figure 3 .
Figure 3. Powder XRD of the catalysts.Cobalt oxide peaks are denoted by ♠.

Figure 4 .
Figure 4. TGA of the spent catalyst shows a significant accumulation of coke.Figure 4. TGA of the spent catalyst shows a significant accumulation of coke.

Figure 4 .
Figure 4. TGA of the spent catalyst shows a significant accumulation of coke.Figure 4. TGA of the spent catalyst shows a significant accumulation of coke.

Figure 5 .
Figure 5. ATR-FTIR of the catalysts where (A) a large shift and increasing intensity of the band at 560 cm −1 indicated ion-exchange of Na with Cobalt and (B) loss of the 3430 cm −1 band (10x magnification) could indicate coke coverage of the hydroxyl groups.

Figure 5 .
Figure 5. ATR-FTIR of the catalysts where (A) a large shift and increasing intensity of the band at 560 cm −1 indicated ion-exchange of Na with Cobalt and (B) loss of the 3430 cm −1 band (10x magnification) could indicate coke coverage of the hydroxyl groups.

Catalysts 2018, 8 ,
x FOR PEER REVIEW 7 of 16 trending downward from the parent zeolite NaX to the fresh and recycled Co/NaX giving 0.72, 0.23, 0.26, respectively.This trend indicates that the cobalt replaced the sodium in the zeolite cages.

Figure 6 .
Figure 6.Representative SEM micrographs of (A) synthesized NaX zeolite, (B) fresh Co/NaX, (C) fresh at a larger scale, (D) recycled Co/NaX.2.6.Deoxygenation of Stearic Acid over Co/NaX in N 2 Deoxygenation of stearic acid was conducted in the presence of nitrogen over Co/NaX catalysts.The reaction temperature was varied between 260-300 • C. Table 2 shows the catalytic performance of the parent zeolite NaX and Co/NaX at various reaction temperatures and catalysts loadings.The conversion correlated to the reaction temperature, with the lowest conversion at 260 • C and the highest conversion at 300 • C. Selectivity to heptadecane (C 17 ) was highest at 280 • C. Yield to C 17 was highest at 280 • C followed closely by 300 • C giving 27.9 and 25.7%, respectively.The reaction conducted at 280 • C was repeated to ensure reproducibility.Two major reaction pathways for the Co/NaX catalyst included decarboxylation and decarbonylation evidenced by the dominant formation of C 17 and C 17 -ene products.Similar results were found by Zhang et al. showing selectivity losses to alkene products [36].Additionally, some octadecene was observed, a result of the hydrodecarbonylation pathway[51].The formation of octadecene requires the presence of Brønsted acid sites[52], which have been shown to be present in ion-exchanged zeolite NaX[27].We hypothesize that both the Brønsted acidity of the support and residual surface hydrogen on the reduced Co/NaX catalyst provided protonic H +[53] which could have participated in the hydrodecarbonylation reaction.This would also explain the observed formation of the intermediate hydrodecarbonylation species, octadecanol (observed in the "other" column in Table2).The proposed pathway is outlined in Scheme 1. Overall, the non-noble metal Co/NaX catalyst provided good conversion and modest selectivity to heptadecane.The catalyst loading had an effect on the selectivity.At low loadings, low conversion was observed, along with very high selectivity to C 17 products.This trend may be due to the relatively high loading of cobalt on the catalyst, which promotes decarbonylation in competition with decarboxylation.

Scheme 1 .
Scheme 1. Conversion of stearic acid over Co/NaX lead to the formation of hydrodecarbonylation, decarbonylation, and decarboxylation products.

16 Figure 7 .
Figure 7. Variation of the (A) catalyst loading (T = 280 °C) showing first-order behavior for the disappearance of stearic acid compared to the catalyst loading and the (B) temperature (loading = 250 mg) showing Arrhenius behavior for increasing temperatures.Error bars are the range about the average of the two data points collected at 250 mg catalyst and T = 280 °C.

Figure 7 .
Figure 7. Variation of the (A) catalyst loading (T = 280 • C) showing first-order behavior for the disappearance of stearic acid compared to the catalyst loading and the (B) temperature (loading = 250 mg) showing Arrhenius behavior for increasing temperatures.Error bars are the range about the average of the two data points collected at 250 mg catalyst and T = 280 • C.

Table 1 .
Physical properties of the parent zeolite and the fresh and recycled Co/NaX catalyst.

Table 1 .
Physical properties of the parent zeolite and the fresh and recycled Co/NaX catalyst.

Table 2 .
Product distribution for the deoxygenation of stearic acid over Co/NaX catalyst in N2.

Table 3 .
Composition of soluble coke by GC-MS analysis on Co/NaX after a typical reaction.

Table 4 .
State-of-the-art comparison of Co-based catalysts in the deoxygenation of vegetable oils.