3.1. Characterization of Support Materials and Pd Catalysts
The crystallographic structure and composition of the MNPs synthesized by co-precipitation method were identified using XRD and Mossbauer spectroscopy. The TiO2/MNPs composites prepared by ultrasound-assisted mixing of the metal oxides were also characterized using the XRD method.
Figure 1 shows the XRD patterns of the MNPs, TiO
2, and TiO
2/MNPs composites with different metal oxides’ ratio. The six characteristic peaks at 35.0°, 41.4°, 50.5°, 63.2°, 67.4°, and 74.4° corresponding to the (220), (311), (400), (422), (511), and (440) crystal planes, respectively, of magnetic iron oxides, Fe
3O
4 (JCPDS Card No. 88-0315) or γ-Fe
2O
3 (JCPDS Card No. 39-1346), with an inverse cubic spinel structure were observed in XRD patterns of the MNPs and TiO
2/MNPs composites (
Figure 1a–d) [
24]. At the same time, the XRD patterns of TiO
2/MNPs composites (
Figure 1b–d) exhibited additional characteristic peaks at 29.3°, 44.0°, 56.3°, 63.4°, 64.9°, 74.4°, 81.8 °, 83.8°, and 90.1°, corresponding to the (101), (112), (200), (105), (211), (204), (116), (220), and (215) crystal planes, respectively, of the starting TiO
2 (
Figure 1e) in the anatase phase (JCPDS card No. 21-1272) [
25]. Calculation with Scherrer’s formula for the strongest peaks reveals that the average crystallite sizes of the samples are 10.0 and 15.6 nm for the starting MNPs and TiO
2, respectively [
26]. The calculated sizes of MNPs and TiO
2 particles in composites were near the same as in the starting materials and ranged from 8.7 to 11.6 nm and from 12.4 to 14.5 nm, respectively.
The isostructurality of Fe
3O
4 and γ-Fe
2O
3 and close values of crystal lattice parameters make it difficult to unambiguously identify diffractograms that correspond to magnetic powder particles [
27]. For more accurate identification of the phase composition, the MNPs were analyzed using Mossbauer spectroscopy at 293 K.
The spectrum of starting MNPs is asymmetric and represents a superposition of four sextets (
Figure 2), indicating the magnetically ordered state of the iron ions caused by agglomeration of the iron oxide nanoparticles [
28].
The sextet parameters from the Mossbauer spectrum are presented in
Table 1. The first sextet, having values of δ = 0.30 mm/s, ε = −0.01 mm/s, and H
n = 482 kOe, can be attributed to Fe
3+ in the A-positions (tetrahedral sublattice) of both magnetite (Fe
3O
4) and maghemite (γ-Fe
2O
3) [
29,
30]. The second component (δ = 0.68 mm/s, ε = −0.01 mm/s and H
n = 465 kOe) is attributed to Fe
2.5+ located on B-positions (octahedral sublattice) of the magnetite spinel structure (Fe
3O
4) [
29]. The third sextet has values of δ = 0.43 mm/s, ε = −0.01 mm/s and H
n = 438 kOe, which are characteristic values for Fe
3+ in B-positions (octahedral sublattice) of maghemite (γ-Fe
2O
3) [
30]. The quadrupole splitting of −0.01 mm/s of the first three sextets is indicative of cubic symmetry [
27]. The fourth component with values of δ = 0.44 mm/s, ε = −0.04 mm/s, and H
n = 383 kOe is probably attributed to iron ions located in the surface regions of γ-Fe
2O
3 nanoparticles, which are “depleted” in exchange bonds [
30]. Thus, based on Mossbauer study, it can be concluded that the MNPs are composed of both magnetite and maghemite. The magnetite most likely is located inside and maghemite on the outside of the MNPs.
The room-temperature magnetic hysteresis measurements of the MNPs and TiO
2/MNPs composites were carried out at 300 K in the applied magnetic field, sweeping from −70 to 70 kOe. As shown in
Figure 3 (curve 1), the specific magnetization at 20 kOe, remanent magnetization (M
r), and coercivity (H
C) of the starting sample of MNPs were found to be 63.6 emu/g, 1.83 emu/g, and 16 Oe, respectively. This suggests that the MNPs exhibit weak ferromagnetic and soft magnetic behaviors [
31], which is consistent with Mossbauer spectroscopy data and indicates some agglomeration of MNPs [
28]. The magnetization of the MNPs is lower than their bulk counterparts (86 and 76 emu/g for magnetite and maghemite at room temperature, respectively) and is not far from the value of 63–64 emu/g observed for the iron oxide nanoparticles of 8.3–9.4 nm in size [
32]. The saturation magnetization (M
S) of the TiO
2/MNPs composites was decreased with decreasing the content of MNPs (
Figure 3, curves 2–5).
The more detailed analysis of the hysteresis loops (
Table 2) showed that the M
S and M
r values decreased proportionally to the expected content of MNPs in the composites. For example, a specific magnetization of 50TiO
2/MNPs composite at 20 kOe was 34 emu/g or 53% from the value for starting MNPs. This suggests that metal oxides homogeneously distributed to each other and MNPs (or TiO
2) content in the composites is close to calculated values of 10, 30, 50, and 70%wt.
At the same time, the coercivity of the composites has not significantly changed to compare with changes in M
S and M
r values for corresponding composites. Lee J.S. et al. [
33] demonstrated that the coercivity of magnetic multi-granule nanoclusters composed of nearly the same size of granules (9.0–11.6 nm) can be decreased up to 97% (from 19.1 to 0.57 Oe) when the nanocluster diameter decreased from 53 to 32 nm. This suggests that the sizes of magnetic agglomerates were practically not changed after ultrasound-assisted mixing of MNPs with TiO
2.
The results of nitrogen adsorption–desorption measurements were used to determine the specific surface area (S
BET) of the MNPs, TiO
2, and TiO
2/MNPs composites. Specific surface area (S
XRD) of the samples was also calculated using XRD data by the modified equation, proposed in [
34]:
where a and b—the content (from 0 to 1) of TiO
2 and MNPs in a support material, respectively; d—average crystallite sizes of TiO
2 and MNPs, calculated from XRD data using Scherrer’s formula; 3.9 and 4.9—density of TiO
2 and MNPs.
A comparison of S
XRD and S
BET data from samples is presented in
Table 3.
Values of a specific surface area calculated from XRD data were shown to be higher than those determined with the BET method. The positive values for the interface area among the particles, calculated using the formula (SXRD – SBET)/2, indicate surface blocking of TiO2 and MNPs nanoparticles due to their agglomeration. The (SXRD – SBET)/2 value for MNPs and TiO2/MNPs composites was higher compared with that for TiO2. This suggests that MNP-containing materials possess a higher agglomeration degree, and the surface of TiO2 particles can be blocked with MNPs after their mixing.
Relatively developed specific surface area of the resulting materials makes them good candidates for the preparation of supported Pd catalysts. It should be noted that all samples demonstrated nearly the same BET surface area, which allows to overlook the effect of surface area of a support material when comparing the behavior of the catalysts based on them in the hydrogenation process. In our prior work [
35], it was found that TiO
2 possesses a low affinity to [PdCl
4]
2− ions. Therefore, in this study, Pd catalysts were prepared by deposition of [PdCl
4]
2− ions on the support materials using sodium hydroxide.
Table 4 shows the results of assessing the degree of palladium ions’ deposition on MNPs, TiO
2, and TiO
2/MNPs composites.
According to photoelectric colorimetric (PEC) analysis, the addition of NaOH to a suspension containing [PdCl
4]
2− and a support material promotes almost complete (95–99%) deposition of palladium ions onto the metal oxides (MNPs, TiO
2, and TiO
2/MNPs). The palladium content in the catalysts, calculated based on the PEC data, was about 1%, which is confirmed by elemental EDX analysis (
Table 5). In addition, EDX elemental analysis data were used to calculate the content of TiO
2 in the catalysts. For example, according to calculations, Pd@TiO
2 was composed of 93% TiO
2, 1% Pd, 6% moisture, and other (Al, Si, Cl, etc.) impurities. In the case of Pd@TiO
2/MNPs catalysts, the TiO
2 content was found to be close to calculated (expected) values of 10, 30, 50, and 70%wt. (
Table 5), which is consistent with data obtained from the SQUID magnetometer (
Table 2). It should be noted that Na was not detected in almost all catalysts. That is, the catalysts do not contain NaOH, which is able to affect the catalytic properties of the supported Pd nanoparticles [
18].
In order to assess the state of palladium during the hydrogenation process, the catalysts were treated with hydrogen in the reactor at 40 °C and then studied using XPS. The deconvoluted Pd 3d signals (
Figure 4) clearly illustrate the different oxidation states of Pd existing on the surface of the catalysts. In the Pd@TiO
2 and Pd@70TiO
2/MNPs, Pd 3d
5/2 peaks with binding energies at around 335 eV and 337 eV can be attributed to Pd in 0 and +2 oxidation states, respectively (
Figure 4a,b) [
36,
37], where the metallic Pd was the dominant species (>50%). On contrary, in samples Pd@30TiO
2/MNPs and Pd@MNPs (
Figure 4c,d), palladium was mostly in oxidized form. Moreover, the binding energy of Pd
0 and Pd
2+ have a positive shift (ca. 0.4–0.7 eV) [
37], probably due to electron transfer from Pd species to iron oxide [
38] (
Table 6).
Figure 5 shows Ti 2p and Fe 2p regions of the XPS spectra of the Pd@TiO
2, Pd@70TiO
2/MNPs, Pd@30TiO
2/MNPs, and Pd@MNPs catalysts. The two strong peaks from the Pd@TiO
2 at around 464.1 eV and 458.4 eV with symmetry can be attributed to Ti 2p
1/2 and Ti 2p
3/2, respectively (
Figure 5a). The peak positions and 5.7 eV peak separation of the Ti 2p doublet agree well with the energy reported for TiO
2 nanoparticles [
39]. The Ti 2p regions of the XPS spectra of the Pd@70TiO
2/MNPs and Pd@30TiO
2/MNPs (
Figure 5c,e) were nearly the same as for Pd@TiO
2, except that small shoulders at around 460 eV were observed, indicating the presence of the Ti
3+ in both samples, probably due to an interaction of TiO
2 with MNPs [
40] (
Table 6).
The peaks from the Pd@70TiO
2/MNPs, Pd@30TiO
2/MNPs, and Pd@MNPs (
Figure 5b,d,f) at around 710 eV and 712 eV can be fitted with two configurations of Fe 2p
3/2, which can be ascribed to Fe
3O
4 and γ-Fe
2O
3, respectively. The weak peak at 719 eV is the Fe
3+ shake satellite peak of the γ-Fe
2O
3 [
41]. Thus, the Fe 2p XPS spectra confirm that MNPs are composed of both Fe
3O
4 and γ-Fe
2O
3. The presence of Fe
2+ of Fe
3O
4 on the catalysts’ surface can also be caused by electron transfer from Pd species to MNPs [
38]. In addition, the binding energy of the Fe 2p
3/2 peak at around 710 eV for Pd@TiO
2/MNPs catalysts was shifted towards smaller energies compared with that for Pd@MNPs (ca. 0.1–0.4 eV), confirming that the interaction between TiO
2 and MNPs took place.
Table 6.
Results of XPS analysis of the catalysts.
Table 6.
Results of XPS analysis of the catalysts.
Region | Catalyst | Experimental BE Values, eV | BE Values in Literature, eV | Proposed Metal State | Ref. |
---|
Pd3d | Pd@TiO2 | 335.2 | 335.0 | Pd0 | [36] |
337.1 | 337.3 | Pd2+ | [37] |
Pd@70TiO2/MNPs | 335.0 | 335.0 | Pd0 | [36] |
337.5 | 337.3 | Pd2+ | [37] |
Pd@30TiO2/MNPs | 335.7 | 335.0 | Pd0 | [36] |
337.7 | 337.3 | Pd2+ | [37] |
Pd@MNPs | 335.7 | 335.0 | Pd0 | [36] |
337.8 | 337.3 | Pd2+ | [37] |
Ti2p | Pd@TiO2 | 458.4 | 458.6 | Ti4+ | [40] |
Pd@70TiO2/MNPs | 458.2 | 458.6 | Ti4+ | [40] |
459.3 | 460.2 | Ti3+ |
Pd@30TiO2/MNPs | 458.8 | 458.6 | Ti4+ | [40] |
460.6 | 460.2 | Ti3+ |
Fe2p | Pd@70TiO2/MNPs | 709.9 | 708.5 | Fe3O4 | [41] |
712.3 | 711.0 | γ-Fe2O3 |
Pd@30TiO2/MNPs | 710.2 | 708.5 | Fe3O4 | [41] |
712.3 | 711.0 | γ-Fe2O3 |
Pd@MNPs | 710.3 | 708.5 | Fe3O4 | [41] |
712.4 | 711.0 | γ-Fe2O3 |
Figure 6 shows an HAADF-STEM microphotograph and EDX elemental mapping images of Ti, Fe, O, and Pd from the Pd@70TiO
2/MNPs treated with hydrogen, according to which the catalyst was composed of three phases. The largest particles (15–50 nm), forming an aggregate with clearly traced boundaries, correspond to titanium dioxide. Some sites of this TiO
2 aggregate are covered with more compact aggregates composed of smaller MNPs (8–12 nm). The smallest particles observed on the surface of both TiO
2 and MNPs aggregates belong to Pd
0 (the brightest spots in the HAADF-STEM image) and PdO species. Of note, HAADF-STEM data are consistent with the results of XRD and BET studies, confirming the size and agglomeration of MNPs and TiO
2 particles.
A comparison of HAADF-STEM microphotographs obtained at lower magnification from the Pd@TiO
2, Pd@70TiO
2/MNPs, and Pd@30TiO
2/MNPs catalysts (
Figure 7a,c,e) shows that, in all cases, small spherical Pd nanoparticles (5–6 nm) are uniformly distributed on the surface of the supports. It should be noted that the surface of TiO
2 aggregates in Pd@30TiO
2/MNPs was more significantly blocked with MNPs compared with that of Pd@70TiO
2/MNPs. Consequently, in the case of Pd@30TiO
2/MNPs, the Pd particles mainly located on the surface of MNPs, while a predomination of palladium on the surface of TiO
2 particles was observed for Pd@70TiO
2/MNPs, which is consistent with results of XPS studies. In addition, Pd particles in Pd@TiO
2 (5.3 nm) and Pd@70TiO
2/MNPs (5.2 nm) were slightly smaller than those in the Pd@30TiO
2/MNPs (6.1 nm) catalyst (
Figure 7b,d,f).
3.3. Phenylacetylene Hydrogenation
The catalytic performance of Pd catalysts supported on MNPs, TiO
2, and TiO
2/MNPs was evaluated in the selective hydrogenation of phenylacetylene. The hydrogenation experiments were carried out in ethanol at 0.1 MPa H
2 and 40 °C.
Figure 9a shows the variation in H
2 uptake versus time during the hydrogenation process. The Pd@TiO
2 and Pd@70TiO
2/MNPs demonstrated the higher catalytic activity, reaching the semi-hydrogenation point (50 mL) after 4 and 6 min, respectively (
Figure 9a, curves 1 and 2). The catalytic activity of the rest of the catalysts was lower and reached the semi-hydrogenation point after 8, 9, 10, and 12 min for Pd@50TiO
2/MNPs, Pd@30TiO
2/MNPs, Pd@10TiO
2/MNPs, and Pd@MNPs, respectively (
Figure 9a, curves 3, 4, 5, and 6). It should be noted that in the initial period, the hydrogen uptake for Pd@50TiO
2/MNPs, Pd@30TiO
2/MNPs, and Pd@10TiO
2/MNPs was nearly the same. However, after reaching a semi-hydrogenation point, the rate of hydrogen uptake was different (
Figure 9a, curves 3, 4, and 5). The phenylacetylene hydrogenation rate, calculated from the hydrogen uptake data, is presented in
Figure 9b. In all cases, the reaction rate increased in the first two minutes and remained constant until a semi-hydrogenation point. Then, the reaction rate increased and, after passing a maximum, sharply decreased. The increasing of TiO
2 content in the catalysts was accompanied by increasing of the maximum rate, observed after reaching the semi-hydrogenation point (
Figure 9b).
According to the chromatographic analysis, styrene is accumulated on Pd@MNPs, Pd@TiO
2, and Pd@50TiO
2/MNPs in the initial period and then is reduced to ethylbenzene (
Figure 10a–c). Accumulation of styrene was accompanied by the formation of a small amount of ethylbenzene, and its yield at the semi-hydrogenation point for Pd@MNPs (17.5%) was higher than that for Pd@TiO
2 (3%) and Pd@50TiO
2/MNPs (12%) catalysts. At the same time, the maximum yield of styrene was close to 80% for all catalysts. The composition of the reaction mixture was changed similarly during phenylacetylene hydrogenation on the rest of the Pd@TiO
2/MNPs catalysts. Dependence of selectivity on conversion for all catalysts tested shows that the titania-containing catalysts possessed a better selectivity to styrene to compare with Pd@MNPs (
Figure 10d).
The hydrogenation rate and selectivity to styrene were calculated from the hydrogen uptake and chromatographic analysis data, respectively. A comparison of the catalytic properties of the catalysts during the hydrogenation of phenylacetylene is presented in
Table 7.
Accumulation of styrene on Pd@MNPs occurred selectively (85%) at a rate of 3.5 × 10–6 mol/s, and then the rate increased to 5.3 × 10–6 mol/s, corresponding to hydrogenation of double C–C bond. The WC≡C to WC=C rates ratio was 1:1.5. The Pd@TiO2/MNPs magnetic catalysts showed improved catalytic properties, and their activity increased with increasing the titania content. Hydrogenation rates of triple and double C–C bonds on Pd@70TiO2/MNPs reached 7.0 × 10–6 and 14.3 × 10–6 mol/s, respectively, which are close to WC≡C and WC=C values observed for Pd@TiO2. The WC≡C to WC=C rates ratio was also changed to 1:2. However, the selectivity to styrene on the Pd@TiO2/MNPs and Pd@TiO2 catalysts was higher than 90%.
To determine optimal reaction conditions, a series of experiments on investigation of the kinetics of phenylacetylene hydrogenation over Pd@70TiO
2/MNPs were performed. The reaction parameters such as catalyst dosage (25–100 mg), phenylacetylene amount (0.25–1.00 mL), and temperature (30–50 °C) were varied (
Figure 11).
Figure 11a shows that the reaction rates (W
C≡C and W
C=C) are proportional to the amount of the catalyst in the range of 25–75 mg. The rates of reaction increased linearly with increasing the catalyst amount, which accompanied with change in the W
C≡C to W
C=C rates ratio. A further increase in catalyst amount (100 mg) did not affect the rates of reaction. This result also suggests that measurements under the experimental conditions (50 mg of the catalyst) studied are within the kinetic regime. A variation in the phenylacetylene amount did not affect the rate (
Figure 11b), and the reaction seemed to be of zero order to phenylacetylene under the reaction conditions studied. Increasing the reaction temperature from 30 to 40 °C led to an increase in the hydrogenation rates, while further increase in temperature did not significantly affect the efficiency of the process (
Figure 11c). Thus, based on the results obtained, further catalytic studies were carried out at the following conditions: 50 mg of a catalyst, 0.25 mL of phenylacetylene at 40 °C.
In our prior study [
18], it was shown that the modification of the surface of Pd magnetic catalysts with NaOH led to significant increase in their activity and lifetime. Such behavior can be explained by the effect of pH-dependent surface charging of metal oxides [
44]. Therefore, in this study, the catalysts obtained were also tested in alkali medium. For this purpose, the solvent (ethanol) was adjusted with NaOH to pH = 10 and then used in the hydrogenation process. The results obtained (
Table 7) show the increased activity of the catalysts in alkali medium. Hydrogenation rates of double C–C bonds increased up to 9.8 × 10
–6 and 8.9 × 10
–6 mol/s, while W
C≡C did not increase on Pd@MNPs and Pd@10TiO
2/MNPs catalysts, respectively. It is worth noting that both catalysts demonstrated nearly the same catalytic properties (activity and selectivity). This is consistent with photocatalytic studies data, according to which it was proposed that the surface of 10TiO
2/MNPs composite does not contain available TiO
2 sites. In the case of the rest of the catalysts, increasing both the W
C≡C and W
C=C rates was observed. The W
C≡C to W
C=C rates ratio was also changed to 1:3 for the catalysts with higher MNP content (more than 50%wt.), while this value for Pd@70TiO
2/MNPs and Pd@TiO
2 almost remained unchanged. The selectivity to styrene achieved 95–96% for the catalysts containing more than 30%wt. of titania. The Pd@70TiO
2/MNPs was found to be the most optimal catalyst due to the combination of excellent catalytic properties of Pd@TiO
2 and magnetic properties of Pd@MNPs. The catalyst was recovered with an external magnet and then reused during 12 runs without loss in its activity (
Figure 12, curve 2). The Pd@TiO
2 demonstrated near-same activity during reuse (
Figure 12, curve 1), while the W
C≡C hydrogenation rate on Pd@MNPs was significantly lower and gradually decreased after 8 runs (
Figure 12, curve 3).
Thus, the Pd@70TiO
2/MNPs catalyst demonstrated improved catalytic properties in alkali medium. The values of selectivity to styrene (96%), the reaction rate in terms of TOF (2.0 s
−1), and stability in terms of TON (19,400) were comparable with those indicated for other known Pd catalysts supported on magnetic core–shell composites (
Table 8). Moreover, after the 1st run, the activity of the catalyst increased, and the reaction rate in terms of TOF was found to be up to 3.4 s
−1 for the next 11 runs. Further, the reaction rate decreased, and the TOF value for the 20th run was 1.7 s
−1, which was close to that for the 1st run (
Figure 12).