2.1. Characterization of the Catalysts
All catalysts were prepared by impregnation-reduction procedure. The detailed composition for these catalysts and their surface area (SSA) are given in
Table 1. The content of Rh or other metal cations in catalysts was studied by inductively coupled plasma optical emission spectrometer analysis, demonstrating that the presence of alkali or alkaline earth cations did not affect the practical loading of bulk Rh in the catalysts since the nominal Rh loading kept constant or even a bit higher when alkali or alkaline earth cations exist. This result indicates that the addition of alkali or alkaline earth cations little occupied the active sites of TNTs and may be even in favor of Rh loading. Moreover, the Rh content of Rh0.5/Li-TNTs-B1 and Rh0.5/Li-TNTs-B5 is much lower than that of other catalysts with the same nominal Rh loading. It suggests that the reduce method affected on the loading of Rh. The SSA of catalysts was calculated using Brunauer–Emmett–Teller (BET) model. The SSA of Rh0.5/Mg-TNTs-P5 was close to that of Rh0.5/TNTs-P, indicating that the existence of Mg cations does not affect the structure of support. Such observation was further confirmed by X-ray diffraction characterization.
The XRD patterns of pristine TNTs or cations decorated TNTs supported Rh catalysts are shown in
Figure 1. All the catalysts display the same diffraction peaks compared to the reported pure TNTs [
9]. The assignment of the structure of titanate nanotubular was the subject of intense controversy [
24]. In our previous report, the main diffraction peaks of TNTs were assigned as anatase phase of TiO
2 [
9,
25,
26]. However, the samples indeed present four broad diffraction peaks located at 9.58°, 24.36°, 28.36°, and 48.44° which can be attributed to the (020), (110), (130), and (200) crystal planes of orthorhombic lepidocrocite-type titanate on the basis of literature [
27,
28,
29]. Except the main peaks for the TNTs support, there is no any diffraction peaks related to Rh nanoparticles or other salts in all samples. This indicates that Rh nanoparticles or other salts in the as-prepared catalysts might be well dispersed on the support and the content of Rh nanoparticles or other salts might be too low to be detected by XRD technique [
26].
TEM analysis was subsequently adapted to observe the morphology of catalysts and identify the existence of Rh nanoparticles (
Figure 2). All catalysts display perfect tubular multiwall structure, which further manifests that the introduction of alkali or alkaline earth cations has no apparent effect on the morphology of TNTs (
Figure 2A–F). The interlayer spacing of prepared TNTs is about 0.7 nm and the diameter of TNTs or cation-modified TNTs is about 10 nm [
22] (
Figure 2A–F). As for the catalyst Rh0.5/Li-TNTs-B5 prepared by borohydride-reduction (
Figure 2B), we can see obvious Rh nanoparticles anchored to the surface or the interspace of the supporter of Li-TNTs. The size distribution of Rh nanoparticles was statistically calculated and exhibited in
Supplementary Materials Figure S1. The average size of Rh nanoparticles is about 1.27 nm. However, only some black spots can be seen on the Rh0.5/Li-TNTs-P5 catalyst (
Figure 2A) or other alkaline earth cation-contained catalysts (
Figure 2C–F). This illustrates that the procedure of reduction affects the size of Rh particles in the catalyst, and the much stronger reducibility of borohydride may result in the formation of larger Rh particles, which has been elucidated intensively in our previous report [
26]. It is worth noting that though the size of Rh nanoparticles in catalyst Rh0.5/Li-TNTs-B5 are much bigger than that in the other catalysts, they also don’t not show any diffraction peaks in the XRD spectra probably due to the low concentration and high dispersion of Rh nanoparticles [
10]. Apart from the observation above, it was found that there are no significant difference in morphology for the fresh and used Rh0.5/Sr-TNTs-P5 catalyst (
Figure 2E,F), demonstrating that M-TNTs are quite stable keeping perfect in the process of hydrformylation.
In order to learn about the chemical state of Rh nanoparticles and to explore the interaction among Rh nanoparticles, TNTs supporter and cations, we selected pure Rh catalyst and Li- or Mg-containing catalyst as examples and characterized them with X-ray photoelectron spectroscopy. As shown in
Figure 3, Rh nanoparticles in all samples exist as two chemical states Rh
0 and Rh
2+, in which Rh 3d 5/2 and Rh
2+ 3d 5/2 peaks located at about 306.7 eV and 308.0 eV, respectively [
10,
30]. The oxide species Rh
2+ might be caused by the samples being exposed to air during collection and waiting for characterization. The cations, such as Mg species, they exist as Mg
2+ ions in Rh0.5/Mg-TNTs-P5 catalyst according to the XPS spectra shown in
Supplementary Materials Figure S2 [
31] and they may be adsorbed on the TNTs. As literature reported, the multi-layered nanotube TNTs were constructed by rolling up one [1 0 1] layer of the anatase structure along the [−1 0 1] direction and there are a lot of unsaturated oxygen atom which can be the active site for Rh
2+ ions and promoter Mg
2+ ions [
32,
33].
Compared with pure TNTs supported Rh catalyst, we can find that the binding energy of Rh 3d in Li or Mg-contained catalyst keeps the same value as pure TNTs catalyst. It suggests that the chemical environment surrounding the Rh nanoparticles have no significant change with modifying TNTs support. However, the binding energy of Rh
2+ 3d in Li or Mg-contained catalyst shifts to a higher value, especially Mg-containing catalyst, indicating that the electronic density of Rh
2+ species decreased [
34,
35]. Besides, from the area of the fitted Gauss peaks, it is easy to find that Rh
0 is the major Rh species for all catalysts. However, the proportion of Rh
0 in Mg-contained catalyst is much higher than that in Rh0.5/TNTs-P and Rh0.5/Li-TNTs-P5, indicating that Mg
2+ cations might suppress oxidation of Rh nanoparticles.
It is well known that CO is an important reactant in the hydroformylation. Herein, we use FT-IR to analyze the CO species adsorbed on catalysts. As we can see from
Figure 4, all the samples present the peaks located at 1449 cm
−1, 1549 cm
−1, and 1630 cm
−1, respectively, which can be attributed to the bending vibration of O-H bonds that came from the H
2O molecules adsorbed on surface of the TNTs support [
36]. Except the peaks of water molecules, there are no other peak in the spectra of pure TNTs (
Figure 4A,B). However, pure TNTs supported Rh catalysts and cationic-contained catalysts display CO absorption peak located around 2068 cm
−1, which can be attributed to typical terminal CO adsorbed on the metals (M-CO) [
37]. Additionally, the band intensity of carbonyl vibration shows significantly difference among these catalysts. Compared with pure TNTs supported Rh catalysts (
Figure 4C,D), the intensity and position of CO peak in Li-contained catalyst increases a little (
Figure 4E), while Mg-containing catalyst enhances remarkably (
Figure 4F), so as those of Ca and Sr-containing catalysts (
Figure 4G,H). This result implies that the introduction of alkali or alkaline earth cations, especially alkaline earth cations, can enhance CO adsorption ability. It can be explained by the interaction between these cations and Rh nanoparticles. CO, a π-acid ligand, tends to be absorbed on Rh nanoparticles which can provide feedback electrons and the electrons may transfer from the d-orbital of Rh to the anti-bonding CO molecular orbital. During this procedure, the charge transferring from alkali or alkaline earth promoter to the surface of Rh metal makes the formation of electrons donated more easily, especially when they carry with high charge, as the alkaline earth promoters. Moreover, the high proportion of Rh
0 in Mg-containing catalyst is also a possible reason based on XPS results.
2.2. Catalytic Performances of the Catalysts
To evaluate the catalytic performance of the catalysts, we carried out experiment on hydroformylation of vinyl acetate. In general, the hydroformylation of vinyl acetate can produce two functional products (
Scheme 1). In our experiments, the hydroformylation of vinyl acetate over all catalysts has high regioselectivity for 2-acetoxy propanal (2-acetoxy propanal:3-acetoxy propanal = 100:0) and generates at least other three by-products, ethylene, propanal, and acetic acid. The gas chromatogram of products formed in the hydroformylation of vinyl acetate are shown in
Supplementary Materials Figure S3.
In order to explore the effect of alkali or alkaline earth cations on the reaction, we evaluated the catalytic activity of Rh/TNTs with different Rh loadings firstly. Taking into account the effect of actual Rh loading on catalyst activity on the reaction, we calculated an apparent turnover frequency (TOF) of vinyl acetate consumption over all the catalysts. As shown in
Table 2, the catalytic activity of Rh/TNTs varies with Rh loading (Entry 1–3). It is obvious that TOF increases with the decrease of Rh loading, and so do the selectivity for aldehyde. This means that there should be an optimal Rh loading; when the Rh loading is larger than the optimal one, the superfluous Rh may not increase the amount of catalytic active site efficiently. As for the reason that the selectivity for aldehyde increases as Rh loading decreases, it could be deduced by the good distribution of Rh nanoparticles on the surface of TNTs for the low Rh loading catalyst which benefits the interaction among Rh nanoparticles, TNTs and substrate. Compared to Rh/TNTs, all the alkali or alkaline earth cations modified TNTs supported Rh catalyst show better catalytic activity, both in the conversion of vinyl acetate and the selectivity for aldehyde. Furthermore, the impact level of alkali and alkaline earth cations on the catalytic performance has to do with the preparation method of catalysts, Rh loading and the species of cations. For instance, the catalytic activity of Li-contained catalyst prepared by photoreduction is much higher than that by borohydride reduction (Entry 4–7). The reasonable explanation might be that the actual Rh loading is higher and the size of Rh nanoparticles is smaller in the catalyst prepared by photoreduction (
Table 1,
Figure 2A,B). Comparing Entry 4 with Entry 5, Rh0.5/Li-TNTs-P5 has higher selectivity for aldehyde, and the effect of Li
+ ions become more remarkable with the increase of the molar ratio of Li to Rh. The optimal molar ratio of Li to Rh should be 5:1.
For Rh/Li-TNTs catalysts, the influence of Li
+ varies on the catalytic properties with Rh loading. When Rh nominal loading is 0.5%, the selectivity for aldehyde in hydroformylation of vinyl acetate over Rh0.5/Li-TNTs-P5 (Entry 5) is much higher than that over Rh0.5/TNTs-P (Entry 1); when Rh nominal loading is 0.25%, both the conversion of vinyl acetate and the selectivity for aldehyde over Rh0.25/Li-TNTs-P5 (Entry 8) are much higher than those over Rh0.25/TNTs-P (Entry 2); when Rh nominal loading is 0.1%, the conversion of vinyl acetate over Rh0.1/Li-TNTs-P5 (Entry 9) is higher than that over Rh0.1/TNTs-P (Entry 3), but the selectivity for aldehyde is a bit lower. Overall, Rh in Rh0.25/Li-TNTs-P5 shows highest efficient, and the TOF over Rh0.25/Li-TNTs-P5 (Entry 8, the TOF is 3581 h
−1) is much higher than that over Rh0.25/TNTs-P (Entry 2, the TOF is 2501 h
−1). For Na
+ and K
+ containing catalysts, as Rh nominal loading is 0.5%, the selectivity for aldehyde is improved by alkali metal cations and the effect of cation promotor is in the following order: Li
+ (Entry 5) > Na
+ (Entry 10) > K
+ (Entry 11). This result is in accord with literature reports and can be explained by the polarizing power of alkali cations [
20]. The higher the polarizing effect of the cation, the stronger interaction among alkali cations, TNTs and Rh nanoparticles. As a result of this interplay, the CO adsorption enhances, the rate of CO insertion improves and then the selectivity for aldehyde increases.
As for catalysts containing alkaline earth cations Mg
2+, Ca
2+ and Sr
2+, the catalytic performance is much better than that pure TNTs supported Rh catalysts, but the effect of alkaline earth cations is not as regularly as alkali cations. This might have to do with the actual Rh loading (
Table 1). It is worth noting that when Rh nominal loading is 0.5%, the conversion of vinyl acetate over all of catalysts can reach up to 100%, while the selectivity for aldehyde over Rh0.5/M-TNTs-P5 is much higher than that over Rh0.5/TNTs-P except Rh0.5/Ca-TNTs-P5. Compared with our previous work on Au-Rh/TNTs catalysis system, the conversion of vinyl acetate over cations-containing catalysts improves noticeably though Au could provide free coordination sites for substrates and work synergistically with Rh nanoparticles to enhance the catalytic performance of Rh catalysts [
10]. When Rh nominal loading is 0.1%, the conversion of vinyl acetate over Rh0.1/M-TNTs-P5 (M = Li, Mg, Ca, Sr) is higher than that over Rh0.1/TNTs-P while the selectivity for aldehyde is a bit lower. Even so, the selectivity for aldehyde over Rh0.1/M-TNTs-P5 is much higher than that over Rh-Ru/TNTs catalysis system [
9]. This might be because that alkali or alkaline earth cations are not a catalyst for hydroformylation or hydrogenation while Ru is a catalyst for hydrogenation [
38], as a result, Rh-Ru catalysis system enhance the conversion of vinyl acetate at the meantime strengthen the hydrogenation activity.
On the basis of characterization and experiment results, we propose the possible pathways of the reaction which is shown in
Scheme 2. At first, Rh nanoparticles anchored on M-TNTs turn into Rh(CO)x Hy/M-TNTs after bumping the syngas CO and H
2. Then the substrate vinyl acetate coordinate with the catalyst, and the intermediate are formed. In this step, the introduction of cations plays an important role. As we mentioned in the introduction, the acidity of the support has an influence on the activity of Rh catalysts for hydroformylation of olefins. In Rh/M-TNTs catalysis system, cations can be as an electron-deficient center (Lewis acid site) and the carboxyl in vinyl acetate with π bonds tends to be an electron donor (Lewis base) [
39,
40]. Consequently, vinyl acetate is easily to be adsorbed on the Lewis acid sites with the formation of electron-deficient intermediates and this will help to hydroformylate by CO and H
2. As we proposed in
Scheme 2, there can be two kinds of intermediates which are supposed to form the final product 2-acetoxy propanal and 3-acetoxy propanal. However, there is no 3-acetoxy propanal formed. This is because the intermediate with a five-membered ring is more stable [
41]. During the CO insertion step, the CO adsorption capacity of the catalyst plays a role. The presence of cations (in particular, alkali earth cations) enhances CO adsorption. This might lead to the higher selectivity for aldehyde. During the process of hydroformylation, the interplay accompanied by the existence of cations works synergistically and facilitates the reaction effectively.