3.1. Bulk structure
Table 1 presents the elemental composition. Co3Al1-600 and 27Co3Al1-mel-600 had similar compositions of Co and Al. The undetectable C and N contents of 27Co3Al1-mel-600 implied that melamine is ineffective as the nitriding agent. 27Co3Al1-AC-600 and 27Co3Al1-N@C-600 had higher amounts of Co than the designated value (27 wt%). The oxygen content of 27Co3Al1-N@C (5.6 wt%) was lower than other catalysts (12.4-13.1 wt%), suggesting a higher reduction extent of LDH was achieved by using g-C3N4 as the carrier. The molar ratio of Co/Al of tested catalysts was close to the designated value (3.0), ranging from 2.5 to 3.2.
Table 1 also lists the porosity and Figure S1 shows the isotherms of tested catalysts. The surface area and pore volume showed an increasing trend following the order as Co3Al1-600 (74 m2/g and 0.14 cm3/g) < 27Co3Al1-mel-600 (79 m2/g and 0.28 cm3/g) < 27Co3Al1-N@C-600 (115 m2/g and 0.30 cm3/g) < 27Co3Al1-AC-600 (733 m2/g and 0.61 cm3/g). Type IV isotherm with type H3 hysteresis loop of mixed micro- and mesoporous structures was observed for each catalyst.
Figure 1a shows the XRD patterns of pre-activated Co3Al1-LDH, Co3Al1-LDH/mel, Co3Al1-LDH/AC, Co3Al1-LDH/g-C3N4, melamine, and g-C3N4. Each pre-activated sample showed combined diffractions of LDH (JCPDS card #51-0045) and its C-containing carrier. A close inspection of Co3Al1-LDH/g-C3N4 (Figure 1b) can find that its diffraction of (003) plane of LDH was at a higher angle (2θ = 11.8o) than its LDH-containing counterparts (2θ = 11.7o). Moreover, the I(003)/I(006) ratio of the LDH diffraction showed a decreasing trend as Co3Al1-LDH (2.1) > Co3Al1-LDH/mel (1.9) > Co3Al1-LDH/AC (1.8) > Co3Al1-LDH/g-C3N4 (1.4). This trend indicated that by the in-situ growth of LDH on its carrier, the interlamellar spacing of LDH can be reduced, and using g-C3N4 can reduce the interlamellar spacing to the highest extent.
Figure 1c shows the diffraction patterns of LDH-based samples after subjecting to the activation treatment at 600 oC for 2 h. The diffractions of the (111), (200), and (220) planes of metallic Co (JCPDS card #15-0806) were observed. The Scherrer’s formula estimated crystallite size of Co0 by using the (111) plane showed an increasing trend following the order as Co3Al1-600 (13 nm) < 27Co3Al1-AC-600 and Co3Al1-mel-600 (17 nm) < 27Co3Al1-N@C-600 (19 nm). Moreover, the (111) diffraction of Co0 of 27Co3Al1-N@C-600 showed a low-angle shift to 2θ = 44.1o compared to other catalysts (see Figure 1d). A possible explanation is the presence of heteroatoms (i.e., N) in the lattice plane of Co0, resulting in an increase in its d-spacing of Co0 crystals [38].
Figure 2 shows the uncorrected radial distribution function (RDF) profiles of the Fourier transform of k3-weighted EXAFS data at Co K-edge of activated catalysts and Co foil standard. The fitted curves were also presented in Figure 2 and the fitted results were listed in Table 2. Each catalyst showed a strong Co-Co shell (~2.5 Å) signal of the backscattering response of metallic Co with its CN value in the range of 4.8 to 5.3. The Co-O shell (~2.0 Å) signal of the octahedral Co2+ site of LDH [39] was weak. The fitted results indicated that all catalysts had similar Co coordination environments. Note that a small response that is closer to the Co atom than the Co-O response was identified at the ionic radius of ~1.8 Å for 27Co3Al1-N@C, possibly due to the presence of the Co-N bond [40].
To understand whether the presence of Co-N coordination may affect the ionic state of Co, the normalized Co L-edge XANES spectra were collected, see Figure 3. The surface-sensitive total electron yield (TEY) mode of the L3-edge spectra of all samples showed three characteristic responses (A1, A2, and A3) similar to CoO, indicating Co2+ dominated [41]. A close look can find that the A3 peak of 27Co3Al1-N@C-600 (780.2 eV) was higher than the other catalysts (780.0 eV) and CoO (780.0 eV), indicating the presence of Coδ+ (δ > 2) species [42].
3.2. Surface structure
Figure 4 shows the Co 2p XPS spectra. The Co 2p3/2 photoline can be deconvoluted into 2 peaks at approximately 778.9 eV (Co0) and 780.9 eV (Co2+ and Coδ+). Among them, 27Co3Al1-N@C-600 had a higher relative composition of Co2+ and Coδ+ (86.8%) than those of the other catalysts (78.7% of Co3Al1-600, 70.6% 27Co3Al1-mel-600, and 64.1% of 27Co3Al1-AC-600), in line with the aforementioned results of Co L-edge XAS analysis.
Figure 5 shows the N 1s XPS spectra of 27Co3Al1-N@C-600 and g-C3N4. Pyridinic N (398.7 eV, 82.5%), pyrrolic N (400.4 eV, 14.3%), and graphitic N (401.1 eV, 3.2%) species were observed for g-C3N4 [43]. As for the photoline of 27Co3Al1-N@C-600, two new species, i.e., Co-Nx (399.2 eV, 23.8%) and oxidative N (404.1 eV, 7.0%) [44] were observed. Moreover, the binding energy of pyridinic N of 27Co3Al1-N@C-600 was approximately 0.6 eV lower than that of g-C3N4. A possible explanation is that N is receiving electrons from its neighboring Co species, elevating its electron density to reduce the binding energy of the response of pyridinic N [6,5].
The surface Co0 concentration was estimated by using the CO uptake of pulse titration, shown in Figure S2. The CO uptake was quite low for each sample (29-36 μmol/g), implying a limited surface concentration of Co0 site. The surface acidity was analyzed by using Py-IR (see Figure 5). The Py-IR spectra showed signals of BAS-bonded (ca. 1545 cm-1) and LAS-bonded (ca. 1450 cm-1) pyridine, and Table 3 lists the estimated concentrations of weak and strong acids by using the formula proposed by Emeis [45]. The estimated concentrations of BAS were close at 150 and 300 oC for each catalyst. Still, 27Co3Al1-N@C-600 showed an approximately two-fold higher amount (4.5 μmol/g) than those of the other catalysts (1.9-2.2 μmol/g). The estimated weak (7.0 μmol/g) and strong LAS (9.9 μmol/g) of 27Co3Al1-N@C-600 were much higher than those of the other catalysts (3.1-3.5 μmol/g and 1.3-1.8 μmol/g, respectively). These trends indicated that 27Co3Al1-N@C-600 had higher concentrations of BAS and LAS with stronger strengths.
The surface LBS was analyzed by using CO2-TPD (Figure 6) and CO2-IR (Figure S3). The CO2-TPD profiles were complicated, indicating different types of adsorbed CO2. A close inspection can find that a trivial amount of desorbed CO2 was detected at a temperature higher than 300 oC, excluding that of 27Co3Al1-N@C-600. Moreover, 27Co3Al1-N@C-600 had a higher amount of desorbed CO2 (1.8 mmol/g). This difference suggested that 27Co3Al1-N@C-600 had stronger surface basicity and a higher base site concentration. For CO2-IR spectra, the bands of bidentate (νas = 1650-1670 cm-1) and unidentate carbonates (νs =1400-1420 cm-1 and νas = 1520-1530 cm-1) of LBS-bonded CO2 [46] were observed at 50 oC. When increasing the temperature, the amplitudes of both of these two bands decreased. At temperatures higher than 300 oC, only strongly bonded unidentate were observed. Therefore, we could differentiate weak and strong LBS by using 300 oC as the demarcation by resorting to the CO2-TPD results, shown in Table 3.
3.3. Activity evaluation
Table 4 presents the activity evaluation of tested catalysts. 27Co3Al1-N@C-600 (entry 1) showed the highest GVL conversion (76.2%) and yield of 1,4-PDO (74.2%) with a trace amount of pentanoic acid and 4-hydroxypentanal (<1.0%). For the other catalysts (entries 2 to 4), the conversion of GVL and yield of 1,4-PDO were all lower than 50%. Furthermore, the effects of activation temperature, LDH loading, Co-to-Al ratio, and reduction gas were investigated by treating CoAl-LDH/g-C3N4 at 500 and 700 oC (entries 5 and 6, respectively), immobilizing LDH with 20 wt% and 40 wt% loadings (entries 7 and 8, respectively), varying the Co-to-Al ratios of 4 and 2 (entries 9 and 10, respectively), and changing the activation gas to N2 (entry 11). Still, 27Co3Al1-N@C-600 had the highest activity. Accordingly, we could claim that by using 600 oC as the activation temperature, a LDH loading of 27 wt%, a Co/Al ratio of 3, and using H2 as the activation gas is favored to transform CoAl-LDH/g-C3N4 into the hydrogenation catalyst.
To reveal the influence of N decoration, kinetic analysis of 27Co3Al1-N@C-600 and Co3Al1-600 were performed, shown in Figure S4. By using the pseudo-first-order kinetic model, the derived activation energies (Ea) of 27Co3Al1-N@C-600 and Co3Al1-600 were 57 kJ/mol and 83 kJ/mol, respectively. The apparent Ea of 27Co3Al1-N@C-600 was also lower than other Co-based catalysts, e.g., CoMg-based catalyst (Ea = 84 kJ/mol [47]), underlining the benefit of N decoration.
The catalyst recyclability was evaluated, shown in Figure 7. The recycling tests of 27Co3Al1-N@C-600 and Co3Al1-600 showed that the former had a 15% loss of its original activity (51%) after five consecutive trials; the latter, a 43% loss compared to its fresh form (49%). The recycled catalysts were analyzed by XRD (see Figure S5). 27Co3Al1-N@C-600 showed similar diffractions and Co0 crystallite size (21 nm) as its fresh form (19 nm). Moreover, the amount of leached Co in each cycle was less than 1 ppm. Thus, the loss of activity of recycled 27Co3Al1-N@C-600 was mainly attributed to the unrecovered catalyst after each cycle test. By contrast, recycled Co3Al1-600 showed intensified diffractions with its Co0 crystallite size to be almost doubled (22 nm) compared to its fresh form (13 nm). This result underlined the improved stability of 27Co3Al1-N@C-600, possibly due to the decorated NCx species [5].