Introduction

Light olefins are important raw materials for producing various chemicals, especially propene is one of the most significant building blocks for polypropylene, propylene oxide, and acrylonitrile, etc. [1]. Propene is mainly produced by steam cracking of naphtha and as a by-product of the fluid catalytic cracking of crude oil. The speed of its production cannot meet its increasing demand, especially with the emergence of many new industries. Hence, a more effective and economical approach of propene production has to be developed. In recent years, propane dehydrogenation (PDH) has been considered as one of the most promising alternatives because of its low-cost and easily obtainable materials [2, 3]. However, PDH is a highly endothermic and equilibrium-limited reaction, which usually requires a high reaction temperature of over 550 °C to obtain high yield [4]. Under such harsh reaction conditions, undesirable side reactions such as thermal cracking, hydrogenolysis, and isomerization can occur, which generates lighter hydrocarbons and serious coke deposits, resulting in rapid catalyst deactivation [5]. Hence, it is highly important to develop catalytic materials that possess excellent anti-coking ability, high propene selectivity, and high propane conversion.

Platinum-based catalysts have been widely utilized in PDH processes because of their high catalytic activity, thermal stability, and eco-friendliness; however, they are not very efficient because of their significant deactivation, which results from coke formation [6, 7] and sintering of active sites [4], and their low propene selectivity, which results from C–C bond cleavage [8].

To obtain optimal performance, some promoters, such as Sn, Ga, In, Zn, Cu, and Na, are added to improve the properties of Pt-based catalysts in PDH reactions [9,10,11]. In commercial PDH processes, Pt–Sn catalysts supported on Al2O3 have been extensively used because of their excellent olefin selectivity and catalytic stability [12, 13]. Although the catalysts have substantial improvements, they still possess limited stability and regenerability. Recently, Zn-doped Pt catalyst has been reported to have an enhanced catalytic performance. For example, a Pt–Zn/Al2O3 catalyst can enhance anti-coking ability as well as propene selectivity [14]. Because of their remarkably high activity, selectivity, and stability, Pt–In and Pt–Ga-supported catalysts have also been widely studied [7, 15].

It has been reported that Cu addition can promote catalytic activities of Pt-based catalysts for light hydrocarbon dehydrogenation by forming Pt–Cu alloy [8, 10]. The promoter Cu can suppress coke deposits formation and hydrocarbon cracking, because the intimate interaction of Pt and Cu can inhibit propene adsorption and increase the energy barrier of C–C bond breakage. Han et al. [10] demonstrated that propene selectivity of Pt/Al2O3 catalyst increased from 77.2 to 90.8% after Cu addition, although catalyst dispersion slightly decreased because a fraction of Pt surface was covered by Cu. Veldurthi et al. [16] studied Cu-modified Pd/Al2O3 and Pt/Al2O3 catalysts for n-butane dehydrogenation and reported a superior n-butane conversion, C4 olefin selectivity, and catalytic stability.

Cerium has been widely applied in several catalytic processes such as steam reforming [17], CO oxidation [18], NOx reduction [19], and water gas shift [20]. Cerium (IV) oxide has a high oxygen storage capacity and can suppress coke formation by oxidizing carbon deposition on the surface of a catalyst utilizing released lattice oxygen [21]. According to Zhang et al. [22], Pt–Sn/ZSM-5 catalysts modified by Ce can depress coke deposition, stabilize Pt nanoparticles, and strengthen the interaction between Sn species and support, which would result in a high reaction activity, stability, and propene selectivity. Furthermore, it has been proposed that the presence of Ce over Pt-based catalysts can suppress Pt sintering by the intimate interaction of Pt and CeOx [4].

In addition, it is generally believed that the support does not only affect the catalytic performance of the dehydrogenation reaction, but also influences the properties of the final metal phase [23, 24]. Previous studies indicate that metal oxide supports, such as K-L zeolite, alkali-doped alumina, spinels, and calcined hydrotalcite (HT), can greatly suppress coke formation because of the free acid sites as well as low alkene adsorption [15]. Studies on calcined HT (Mg(Al)O) or HT-like materials are particularly interesting owing to their suitable basic characteristics and high thermal stability. In addition, the surface Al3+ ions can effectively enhance the dispersion of metal particles [9, 25]. Akporiaye et al. [26] found that PtSn/Mg(Al)O catalysts showed advanced activity and stability in a PDH reaction compared to PtSn/Al2O3 prepared by conventional method. In recent years, it has been reported that Pt/Mg(In)(Al)O and Pt/Mg(Ga)(Al)O catalysts with calcined HT supports exhibit good catalytic activity and alkene selectivity [27, 28].

This study aims to optimize the catalytic activity and anti-sintering of Mg(Al)O-supported Pt-based catalysts in PDH reactions by adding Cu, changing the Ce content, and using a more effective preparation method. The relationship between catalytic performance and the structural features of a series of CeO2-modified Pt–Cu/Mg(Al)O catalysts was systematically investigated by complementary characterization techniques.

Experimental Details

Preparation of Catalysts

Materials

All reagents used in this study are analytical grade. The chemicals H2PtCl6·6H2O, Al(NO3)3·9H2O, Cu(NO3)2·3H2O, NaCO3, and NaOH were purchased from Tianjin Yuanli Chemical Co. Ltd, and Mg(NO3)2·6H2O was purchased from Shanghai Aladdin Biochemical Technology Co. Ltd. Chemical Ce(NO3)3·6H2O was purchased from Tianjin Guangfu Fine Chemical Research Institute.

Preparation of PtCu/CexMgAl-R

A series of catalysts were prepared by one-pot synthesis, where 19.23 g of Mg(NO3)2·6H2O, 9.38 g Al(NO3)3·9H2O, 1.55 g Cu(NO3)2·3H2O, 7.23 mL H2PtCl6 aqueous solution (19.3 mmol L−1), and a certain amount of Ce(NO3)3·6H2O were dissolved in 300 mL of deionized water under magnetic stirring at 90 °C for 10 min. Then, a mixed solution of NaOH and NaCO3, with a CO32−/Al3+ molar ratio of 2 and OH−/(2 Mg2+ + 3Al3+) molar ratio of 1.6, was added dropwise to the above solution under vigorous stirring until a pH value of 9.7 was achieved. The formed solution was kept at 90 °C for 6 h, and the precipitate was filtered, washed with deionized water, and dried at 80 °C for 24 h. The obtained precursors (marked as PtCuCexMgAl-P) were calcined at 550 °C for 4 h to obtain Mg(Pt)(Cu)(Ce)x(Al)O catalysts (marked as PtCuCexMgAl-O). Similarly, MgAl-O, PtMgAl-O, Ce3.0MgAl-O, CuMgAl-O, PtCuMgAl-O, PtCe3.0MgAl-O, and their calcined products were also prepared. For all samples, the loadings of Pt and Cu were fixed at 0.6 and 9.0 wt%, respectively. The content of Ce was changed from 0 to 3.0 wt% by varying the amount of Ce(NO3)3·6H2O during the preparation. Prior to the activity test, the sample was reduced in 5 vol% H2/N2, which is denoted as PtCu/CexMgAl-R.

For comparison, imp-PtCu/MgAl-O was prepared by conventional sequential impregnation method and calcination. First, Cu2+ ion was loaded by incipient wetness impregnation of the as-prepared Mg(Al)O support using Cu(NO3)2·3H2O aqueous solution, which was left to stand for 6 h at ambient temperature. After drying at 50 °C for 2 h and 120 °C for 3 h, the sample was calcined at 550 °C for 4 h in a muffle furnace, and Pt was supported on the calcined product by the same procedure as Cu impregnation using H2PtCl6·6H2O as precursor. Prior to the activity test, the sample was reduced in 5 vol% H2/N2, which is denoted as imp-PtCu/MgAl-R.

Characterization

The X-ray diffraction (XRD) patterns were recorded on a D8-Focus (Bruker) equipped with Cu Kα radiation (λ = 0.15418 nm). The catalysts were scanned from the 2θ value of 5°–90°, with a scanning speed of 8° min−1. The X-ray tube was manipulated at 40 kV and 40 mA.

The textural properties of the samples were measured by N2 adsorption–desorption at − 196 °C. Prior to the experiments, the samples were degassed for 4 h at 300 °C. The specific surface areas of the samples were determined by the Brunaure–Emmett–Teller (BET) method. The Barrett–Joyner–Halenda (BJH) method was used to calculate the cumulative pore volume and average pore diameter in accordance with the desorption branches of the isotherms.

The transmission electron microscopy (TEM) images were taken with a JEM-2100F microscope operated at 200 kV. The as-synthesized catalysts were pre-reduced at 580 °C in a flow of 5 vol% H2/N2 for 2 h. Then, the samples were dispersed, sonicated in ethanol, and dropped on carbon-film coated copper grids.

Temperature-programmed reduction (TPR) experiments were carried out in a programmable temperature system, and prior to the experiment the baseline was stabilized for 30 min. A H2-TPR was implemented in 5 vol% H2/N2 (30 mL min−1) with a heating rate of 10 °C min−1 from room temperature to 850 °C. The H2 consumption profile was recorded with a thermal conductivity detector.

X-ray photoelectron spectroscopy (XPS) was carried out on a Perkin-Elmer PHI 5000C ESCA using Al Kα radiation. Prior to the analysis, the sample was reduced under 5 vol% H2/N2 (30 mL min−1) at 580 °C for 2 h, and the binding energies were calibrated using the C 1 s peak at 284.8 eV as an internal standard.

Thermogravimetric (TG) analysis was carried out with a DTG-50/50H thermal analyzer to determine the quantity of coke on the used catalyst and the TG curves were recorded from room temperature to 900 °C at an increasing rate of 10 °C min−1.

The elemental composition of the samples was analyzed by inductively coupled plasma-optical emission spectroscopy (ICP-OES) using VISTA-MPX.

Dispersion of metals was determined by CO chemisorption. For a fresh catalyst, 100 mg of sample was pre-reduced with H2 at 580 °C for 1 h and then flushed with Ar flow at 580 °C for 0.5 h. CO chemisorption was carried out at 30 °C by injecting pulses of CO (50 μL) every 3 min until the peak area became stable. The adsorption capacity of CO is calculated by the following equation:

$${\text{CO uptake (}}\upmu{\text{mol g}}_{\text{cat}}^{ - 1} )= \frac{n \times A - B}{A} \times \frac{50 \, }{22.4} \times 10 ,$$
(1)

where n stands for the number of chemical adsorption peaks, A represents the average value of the peak area when the peak area remains constant, and B stands for the total area of the chemical adsorption peak.

Activity Test for Propane Dehydrogenation

Activity tests were carried out in a fix-bed reactor with an 8 mm inner diameter. Prior to reaction, 0.4 g sample was reduced in 5 vol% H2/N2 (30 mL min−1) at 580 °C for 2 h. Then, a mixture of C3H8, H2 and N2 with a molar ratio of 8:7:35 was introduced into the reactor and the total flow was 55 mL min−1. The weight hourly space velocity (WHSV) of propane was about 3 h−1. The PDH reaction was performed at atmospheric pressure and 600 °C. The reaction products were analyzed by an online gas chromatograph, equipped with an FID detector and Al2O3 column. The catalytic performance of the catalyst was evaluated by C3H8 conversion, C3H6 selectivity, and C3H6 yield, which are calculated by the following equations:

$${\text{Conversion }}\left( \% \right) = \frac{{\sum {{}_{i}n_{i} } }}{{\sum {{}_{i}n_{i} } + \left( {n_{{{\text{C}}_{3} {\text{H}}_{8} }} } \right)_{\text{out}} }} \times 100$$
(2)
$${\text{Selectivity }}\left( \% \right) = \frac{{n_{i} }}{{\sum {{}_{i}n_{i} } }} \times 100$$
(3)
$${\text{Yield }}\left( \% \right) = {\text{Conversion}} \times {\text{Selectivity}} \times 100 ,$$
(4)

where i stands for the hydrocarbon products in the effluent gas stream, ni represents the number of carbon atoms of ingredient i, and \(\left( {n_{{{\text{C}}_{3} {\text{H}}_{8} }} } \right)_{\text{out}}\) stands for the number of carbon atoms of propane in the effluent gas stream, i.e., unreacted propane during the PDH reaction.

Results and Discussion

Structure of the Precursors

The XRD patterns of the as-synthesized samples with different compositions are shown in Fig. 1. All the precursors display a similar pattern to the typical HT material reported in previous papers [21, 29]. For MgAl-P, PtMgAl-P, CuMgAl-P, PtCuMgAl-P, and PtCuCexMgAl-P (x = 0.1, 0.3) samples, only an HT phase (JCPDS file No. 14-0191) can be observed. However, other non-HT phases, such as Ce(OH)3 phase [21], can be indexed at 28.2° in Ce3.0MgAl-P, PtCe3.0MgAl-P, and PtCuCexMgAl-P samples with a relatively high Ce content (x = 0.6, 1.8, 3.0). This is attributed to the larger ionic radius of Ce3+ (0.101 nm) [30] in comparison to that of Al3+ (0.054 nm) [30] in the octahedral coordination [21, 29, 31], which can influence the stability of brucite-like layers and even result in the segregation of partial Ce3+ ions into non-HT phases for higher Ce-containing samples [21]. The diffraction peaks of the HT phases broaden and weaken with the successive incorporation of Pt, Cu, and Ce ions, reflecting the decrease in crystallinity of the HT phase and shrinkage in crystallite size.

Fig. 1
figure 1

XRD patterns of the precursors. (black circle) hydrotalcite phase; (black Club) Ce(OH)3. (inset: a. PtCuCe3.0MgAl-P; b. PtCuCe1.8MgAl-P; c. PtCuCe0.6MgAl-P; d. PtCe3.0MgAl-P; e. Ce3.0MgAl-P)

Full size image

The calculated lattice parameters of the precursors are listed in Table 1. The a values of CuMgAl-P, Ce3.0MgAl-P, and PtCuCexMgAl-P samples increase compared to that of MgAl-P. Additionally, the presence of small amounts of Ce3+ ions gives rise to an increasing value of lattice parameter a across PtCuCe0.1MgAl-P, PtCuCe0.3MgAl-P, and PtCuCe0.6MgAl-P, followed by a decrease with further increase in the Ce content. For CuMgAl-P, the octahedron becomes larger because of the decrease of positive charge density in the layers incorporated with Cu2+ ions. As for the Ce-containing samples, it is mainly because Ce3+ ions have a bigger radius than Al3+ ions, which can broaden the distance between metal cations in the layers and increase the a value [31]. The corresponding lattice parameter c exhibits a similar trend. Although the reasons are different, it can be confirmed that Cu2+ and Ce3+ ions can be introduced in the HT structure layers as the amounts of Cu2+ and Ce3+ ions are small.

Table 1 Lattice parameters of the as-synthesized precursors
Full size table

Compared with MgAl-P sample, the increase of c value for PtMgAl-P is related to the low charge density of the layer. The increase of c as well as the presence of only well crystallized HT phase mean that Pt4+ ions can be inserted into a layer of the HT structure. Indeed, the Pt4+ ions are most likely to form octahedrally coordinated compounds, unlike Pt2+ ions [32]. Furthermore, the enhanced c value implies that the Al3+ ions may not be fully co-precipitated in the HT phase when Pt4+ is in the mother liquid, which is proved by the ICP-OES results in Table 2. The slightly increased lattice parameter a of PtMgAl-P is related to the bigger radius of Pt4+ ions (0.063 nm) [30] in the octahedral coordination than that of Al3+ ions (0.054 nm), but the introduced amount of Pt4+ ions is so small that the effect on a is weak.

Table 2 Chemical compositions of the calcined samples
Full size table

Certainly, we cannot exclude the presence of the partial metal ions in the amorphous phase on the basis of XRD result.

Structure, Composition, and Texture of the Calcined Samples

The element analysis results provided in Table 2 show that the Pt/Cu/Ce/Mg/Al atomic ratios in the samples are different from those employed in the synthesis mixture; especially, the (M4+ + M2+)/M3+ atomic ratio has a significant increase compared with the corresponding feeding amount, indicating an incomplete precipitation of the Al3+ ions. For the one-pot synthesized samples, the result reflects that the big Pt4+ ions (0.063 nm) are incorporated by partial substitution of the small Al3+ ions (0.054 nm) in octahedral coordination, and the loss of the Al3+ ions in imp-PtCuMgAl-O is mainly ascribed to the acidic leak of Al3+ ions from Mg(Al)O support during the impregnation. Furthermore, it can be found that the incorporation of Cu inhibits the insertion of Pt4+ ions in the layer. The Jahn–Teller distortion of the coordination environment around the Cu2+ ions can explain the difference. Subsequently, the addition of small amounts of Ce3+ ions can weaken the Jahn–Teller effect and enhance the capacity of Pt4+ and Cu2+ ions in the layer. The high Pt loading amount for imp-PtCuMgAl-O is mainly due to the weak physisorption during incipient wetness impregnation, which is independent of the formation of chemical bonds with other metal ions.

Figure 2 shows the XRD patterns of the calcined samples. The absence of the characteristic diffraction peaks of HT phase indicates that the layered structure has been destroyed. Instead, strong and sharp diffraction peaks of MgO (JCPDS file No. 04-0829) at 2θ of around 35.6°, 43.4°, and 62.7° are displayed in all samples [9]. Meanwhile, these diffraction peaks slightly shift toward larger angles compared with those of pure MgO crystallite phases (2θ = 42.9° and 62.3°). This is as a result of the incorporation of Al3+ into the MgO structure to form Mg(Al)O solid solution [33,34,35]. No obvious peaks associated with PtOx can be detected for all Pt-containing samples, which can be attributed to the low content, small particle size, or a partial dissolution in the cubic mixed oxides of Pt species [9]. For those Cu-containing samples, except the imp-PtCuMgAl-O, the diffraction peaks of CuO phase cannot be found, indicating that Cu2+ might have been incorporated into the MgO structure with Al3+ to form Mg(Cu, Al)O solid solution, such as Mg(Ni, Al)O, Mg(Co, Al)O, and Mg(Ni, Fe, Al)O, by the calcination of the corresponding HTs [36,37,38]. This is associated with the well-distributed Cu2+ in the brucite-like layers. On the contrary, the sample imp-PtCuMgAl-O has visible diffraction peaks of CuO at 2θ of 35.6° and 38.9°, indicating the formation of a large CuO crystallite.

Fig. 2
figure 2

XRD patterns of the calcined samples. (black diamond) Mg(Al)O periclase; (black heart) CeO2; (black Spade) CuO

Full size image

For the Ce-containing samples, the CeO2 phase could not be detected until the Ce content was increased to above 0.3 wt%. The samples with Ce content greater than or equal to 0.6 wt% display clear peaks at 2θ of around 28.8°, 47.9°, and 56.8°, which can be attributed to the (111), (220), and (311) diffractions of CeO2 (JCPDS file No. 43-1002) [21, 39]. It can be speculated that the CeO2 phase also exists in the samples with low Ce content, although there are no visible diffraction peaks of CeO2, which is due to the low content or small particle size of CeO2 owing to its interaction with Pt and Cu species.

The textural properties of the samples are summarized in Table 3. The one-pot synthesized samples show higher BET specific surface areas, pore volumes, and pore sizes than the imp-PtCuMgAl-O sample prepared by impregnation method. It can be deduced that the metal particle size of the former samples is smaller than that of the latter, which has large CuO nanoparticles (NPs).

Table 3 Textural data of calcined samples
Full size table

Structure of the Reduced Samples

The XRD patterns of the reduced catalysts are shown in Fig. 3. Here, Pt diffraction peaks were still not detected. Compared with the XRD patterns in Fig. 2, there are no changes in the diffraction peaks of Mg(Al)O periclase and CeO2 phase over the reduced samples. This demonstrates that the reduction of these oxides is extremely limited [21]. For PtCu/MgAl-R, PtCu/CexMgAl-R (x = 0.1, 0.3, and 0.6, respectively), the weak and broad diffraction peak at 2θ of 50.2° can be indexed to (200) reflection of cubic Cu (JCPDS file No. 04-0836), indicating that the Cu NPs have small particle sizes and are highly dispersed [40]. However, no obvious Cu diffraction peaks are detected for the samples with Ce content of 1.8 and 3.0 wt%, which can be ascribed to a high content of CeO2 that promotes Cu dispersion. In addition, the imp-PtCu/MgAl-R presents strong and sharp Cu diffractions at 2θ of 43.3°, 50.4°, and 74.1°, which indicates a large crystal size formation and strong Cu crystallization.

Fig. 3
figure 3

XRD patterns of the reduced samples. (black diamond) Mg(Al)O periclase; (black heart) CeO2; (black star) Cu. (inset: a. imp-PtCu/MgAl-R; b. PtCu/MgAl-R; c. Cu/MgAl-R; d. Pt/MgAl-R)

Full size image

From the inset of Fig. 3, the peak at 2θ of 43.1° for PtCu/MgAl-R shifts slightly to the lower 2θ value compared with the position at 43.3° of the Cu diffraction for the Cu/MgAl-R sample. This indicates that Pt and Cu interacts on the formation of Pt–Cu alloy. Meanwhile, for the imp-PtCu/MgAl-R sample, the peak at 43.3° is unsymmetrical and broadens at the lower 2θ value, indicating the co-presence of Cu and Pt–Cu alloy [41,42,43].

The morphologies and particle size distributions (PSDs) of the metal species of the reduced samples are shown in Fig. 4. According to the lattice spacing of 0.226 nm in the HR-TEM image of Pt/MgAl-R, the Pt (111) plane can be confirmed. From the HR-TEM images (Fig. 4d, g, and j) of the other three catalysts, a lattice spacing of 0.215 nm of the metal NPs can be assigned to the (111) plane of the Pt–Cu alloy phase [10]. There is no obvious change in the particle size of the Pt–Cu alloy in all samples. In addition, small size Cu NPs exists in the samples PtCu/MgAl-R and PtCu/Ce0.3MgAl-R. However, in the imp-PtCu/MgAl-R sample, isolated Pt NPs (Fig. 4j) and large Cu NPs (Fig. 4k) can be found, which can be ascribed to the inhomogeneous distribution of metal species.

Fig. 4
figure 4

HR-TEM images and particle size distribution of metal species for the catalysts of a, b Pt/MgAl-R; c, d, e PtCu/MgAl-R; f, g, h PtCu/Ce0.3MgAl-R; i, j, k imp-PtCu/MgAl-R

Full size image

The PSDs reflect a well-dispersed small metal NPs feature for the samples synthesized from HT precursor by one-pot method. The introduction of Cu results in a slight increase of average particle size from 2.65 nm of Pt/MgAl-R to 4.34 nm of PtCu/MgAl-R, and the average particle size further declines to 3.24 nm of PtCu/Ce0.3MgAl-R following the incorporation of Ce. The slightly increased size of PtCu/MgAl-R metal NPs is related to the co-existence of Pt–Cu alloy and isolated Cu NPs. For PtCu/Ce0.3MgAl-R, the reduced size indicates that the Ce-doping facilitates the dispersion of the Pt–Cu alloy and Cu NPs. However, the PSD of imp-PtCu/MgAl-R was split into two areas and the average particle size increased to 6.07 nm, mainly because of the large metal NPs formed by the aggregation of Cu NPs. The HT lattice-confined Pt4+, Cu2+, and Ce3+ ions are beneficial to playing the Ce-dispersing role and forming the interactions among metal NPs, involving the Pt–Cu alloy. On the contrary, the isolated Pt and Cu NPs not confined on the reduced catalysts surface are unstable and tend to aggregate into big particles. Thus, both the preparation method of catalysts and addition of Ce significantly affect the crystal properties of the reduced catalysts.

To identify the elemental distribution of the representative PtCu/Ce0.3MgAl-R catalyst, the TEM-EDS mapping and line scan results are shown in Fig. 5. The results reveal a uniform distribution of Pt, Cu, and Ce on the support surface. This is due to the topotactic transformation feature of HT materials, which keeps the uniform mixing of every metal into the corresponding calcined samples and even reduced products [40].

Fig. 5
figure 5

TEM images with EDS element-mapping analysis (a–d) and line scan profile (e and f) of PtCu/Ce0.3MgAl-R

Full size image

In the line scan profile of PtCu/Ce0.3MgAl-R, the peaks of Pt and Cu appear at the same position, indicating an intimate interaction between Pt and Cu, which clearly demonstrates that the well-dispersed Pt and Cu facilitated the Pt–Cu alloy formation in accordance with the above analyses. The Ce also has the same variation trend as Pt and Cu, suggesting that Ce has a close contact with Cu and Pt, which aids the dispersion and stabilization of Cu and Pt–Cu NPs.

From the CO chemisorption data in Table 4, it can be found that although CO can be adsorbed on both Pt and Cu surfaces, the CO uptake capacity can still reflect the overall metal dispersion of the catalysts. It can be concluded that the Pt/MgAl-R and imp-PtCu/MgAl-R samples have a low metal dispersion, while the Pt–Cu/Ce0.3MgAl-R sample has the highest metal dispersion because Ce can further promote Pt and Cu dispersion.

Table 4 CO uptake on reduced samples
Full size table

Reducible Properties

To examine the reducible properties of the calcined samples, H2-TPR characterization was performed and the curves are shown in Fig. 6. For Ce3.0MgAl-O, no obvious H2 consumption can be observed. A faint bulge can be seen in PtMgAl-O, which is because of the reduction of PtOx species. For the PtCe3.0MgAl-O sample, there are two reduction peaks at 280 and 450 °C, whereby the lower temperature peak can be assigned to the reduction of Pt4+ ions and the higher temperature peak can be attributed to the partial reduction of CeO2 [44]. This indicates that CeO2 NPs disperse the Pt NPs and both of them can promote a mutual reduction [45]. In addition, from the XRD pattern (Fig. 3), the CeO2-containing samples show clear CeO2 diffraction peaks after reduction, which is highly consistent with the TPR results, indicating that the reduction of Ce oxide is extremely limited.

Fig. 6
figure 6

H2-TPR curves for the calcined samples

Full size image

For the Cu-containing samples derived from HT precursors, a prominent hydrogen consumption peak is centered at around 150–350 °C, which is easily attributed to the reduction of surface Cu2+ to Cu0 or the co-reduction of Pt and Cu, and the small and broad high-temperature shoulder at about 580 °C may be because of Cu2+ reduction in bulk. In the absence of noble metal, a strong peak at 290 °C can be observed for the CuMgAl-O catalyst. On the addition of noble metal Pt, this reduction peak shifts toward a lower temperature for the PtCuMgAl-O catalyst. The decreased reduction temperature of Cu2+ ions and the missing reduction peak of Pt4+ ions suggest that a strong interaction exists between Pt and Cu, which is the key to the formation of Pt–Cu alloy [46]. This result is consistent with the XRD and TEM analyses of the reduced samples [47]. However, the reduction behavior of the imp-PtCuMgAl-O shows a relatively higher temperature and broader reduction peak at 300 °C because of the PtOx and large CuO NPs reduction.

In comparison with the PtCuMgAl-O catalyst, a small amount of Ce-doping (PtCuCe0.3MgAl-O) can decrease the reduction peak temperature, indicating that the reducing ability is enhanced with the introduction of Ce species. At 1.8 wt% Ce content, a small frontal peak appears. The frontal peak and main peak temperatures decrease with further increase in Ce content. In addition, the frontal peak area shows an increasing tendency accompanied by the decrease of the main peak area. It can be deduced that the frontal peak belongs to the reduction of Pt4+ ions and the main peak should be related to the co-reduction of Pt and Cu species. This indicates that the Pt and Cu interaction becomes weak because of the dilution effect of excessive Ce. Thus, the Ce species significantly affects the dispersion of other metal species. Its content needs to be controlled to ensure a moderate dispersion and a consequent interaction for other metal species.

X-ray photoelectron spectroscopy (XPS) characterization has been carried out to investigate the chemical state of Cu element in the prepared catalysts. The Cu 2p3/2 XPS spectra are shown in Fig. 7 and the corresponding results of the surface Cu state are listed in Table 5. The XPS spectra of Pt 4f in all samples are not given because of the difficulty in distinguishing Pt 4f peak from Cu 3p and Al 2p peaks caused by their overlapping [46, 48] and the extremely small Pt loading (0.47–0.62 wt%) [4].

Fig. 7
figure 7

XPS spectra of the Cu 2p3/2 in the representative samples: Cu/MgAl-R, PtCu/MgAl-R, PtCu/Ce0.3MgAl-R, and imp-PtCu/MgAl-R

Full size image
Table 5 XPS characteristics of Cu 2p3/2 region for the catalysts
Full size table

In Fig. 7, the obvious satellite peaks located at 939.0–947.0 eV indicate the presence of Cu2+ ions. The source of the satellite peaks is either the promotion of 3d electrons to 4p and/or 4s levels or the charge transfer of electrons to the unfilled 3d, so that the transfer cannot occur in Cu+ and Cu0 because of their completely filled 3d shells [46]. For Cu/MgAl-R, the Cu 2p3/2 peak can be fitted into two peaks at 934.0 and 935.8 eV, corresponding to the Cu0 and Cu2+ ions, respectively. Compared with the Cu/MgAl-R, the Cu 2p level of other samples shifts to the lower binding energy, which is caused by Pt–Cu alloy, and this phenomenon occurred in previous studies. Cho [49] and Kleiman [50] systematically studied the charge transfer in Pt–Cu alloy and revealed that the Cu 2p level exhibited a negative shift upon alloying with Pt. For the PtCu/MgAl-R sample, the peak is deconvoluted into three contributions centered at about 935.1, 933.7, and 932.9 eV. The minor contribution at 935.1 eV is because of the unreduced Cu2+ ions in the bulk which strongly interact with the support and are difficult to be reduced, while the peaks at 933.7 and 932.9 eV are ascribed to the isolated metallic Cu (Cu 0A ) and Cu0 in Pt–Cu alloy (Cu 0B ), respectively. The absence of Cu+ species in the spectra indicates that Cu2+ ions were directly reduced to Cu0. A similar situation was encountered in the PtCu/Ce0.3MgAl-R and imp-PtCu/MgAl-R.

A slightly increased binding energy value for Cu2+ ions can be observed in Table 5, which is in accordance with the order of imp-PtCu/MgAl-R, PtCu/MgAl-R, PtCu/Ce0.3MgAl-R, and Cu/MgAl-R. This is related to an increased surface interaction between Cu2+ ions and other near groups. It signifies that the Cu2+ ions in imp-PtCu/MgAl-R are unstable. Furthermore, according to the proportion of the Cu2+ ions, it is clear that the impregnation method favors Cu2+ species more than Cu0 species, which can be due to the difficult reduction of large CuO NPs in imp-PtCu/MgAl-R. It can be predicted that a large amount of unstable Cu2+ ions tends to shorten the operational lifespan of imp-PtCu/MgAl-R. On the contrary, the binging energy of Cu 0B decreases across imp-PtCu/MgAl-R, PtCu/MgAl-R, and PtCu/Ce0.3MgAl-R. This indicates that the interaction strength between Pt and Cu decreased slightly. However, it is suggested that the chemical activity of Pt and Cu species in these catalysts increases in the same order. In addition, the sequence of the Cu 0B proportion is as follows: imp-PtCu/MgAl-R < PtCu/MgAl-R < PtCu/Ce0.3MgAl-R, indicating an increase of the Pt–Cu alloy sites. However, there is no difference for the binding energy value of the isolated Cu 0A species, which reflects that the interaction between the adjacent metal elements and Cu 0A species is the same among the samples. However, the Cu 0A proportion increases across imp-PtCu/MgAl-R, PtCu/Ce0.3MgAl-R, and PtCu/MgAl-R. Based on the same preparation method, the lesser the amount of Cu 0A , the stronger the stability of the catalysts.

Catalytic Performance

To compare results, the PDH of the reduced catalysts is tested. Figure 8 presents the propane conversion, propene selectivity and yield over the reduced catalysts. The propane conversions of Pt/MgAl-R and Cu/MgAl-R catalysts are below 10%, and the propene selectivities and yields are below 75% and below 10%, respectively. Although the initial conversion of Pt/Ce3.0MgAl-R is close to 100%, the initial propene selectivity and yield are nearly zero; this indicates that the Ce element has a good dispersion effect on the Pt NPs, which results in a high Pt NPs activity and deep dehydrogenation and cracking [7, 51]. It can be deduced that the Pt NPs can be dispersed into Pt single atoms by Ce, which accounts for the 100% propane conversion over Pt/Ce3.0MgAl-R at the initial stage. However, the unstable Pt atoms on Pt/Ce3.0MgAl-R can be rapidly aggregated into large Pt NPs and decrease catalytic performance.

Fig. 8
figure 8

Propane conversion, propene selectivity and yield over reduced catalysts. Reaction conditions: 600 °C, atmospheric pressure, C3H8:H2:N2 (molar ratio) = 8:7:35, and WHSV = 3 h−1

Full size image

The co-introduction of Pt and Cu significantly enhanced the PDH performance, as the interaction between Pt and Cu accounted for the high catalytic activity of PDH [10]. The initial propane conversion of PtCu/MgAl-R is higher than that of imp-PtCu/MgAl-R and slightly lower than that of PtCu/Ce0.3MgAl-R. After spending 360 min in the gas stream, the conversion of PtCu/MgAl-R, PtCu/Ce0.3MgAl-R and imp-PtCu/MgAl-R decreased from 60, 62, and 49 to 44, 45, and 31%, respectively. The slightly higher activity of PtCu/Ce0.3MgAl-R than that of PtCu/MgAl-R can be attributed to the high dispersion of Pt–Cu alloy and Cu NPs promoted by Ce, which is supported by the results of CO chemisorption in Table 4. The corresponding low activity of imp-PtCu/MgAl-R is related to the poor Cu dispersion and lack of interaction between Pt and Cu. The basic reason is the different structural features between the PtCu/Ce0.3MgAl-R and PtCu/MgAl-R obtained by one-pot synthesis from HT precursor and imp-PtCu/MgAl-R prepared by impregnation.

From Fig. 8b, the interaction between Pt and Cu favors the selectivity to propene by inhibiting propene adsorption and improving the energy barrier of C–C bond breakage [10]. During the 360 min reaction time, the PtCu/Ce0.3MgAl-R propene selectivity is similar to that of imp-PtCu/MgAl-R and slightly higher than that of PtCu/MgAl-R. Even their propene selectivity plots cannot be distinguished after the 360 min reaction. According to the structural features of these catalysts, the similar selectivity is only correlated with the presence of Pt–Cu alloy, but independent of the content and interaction strength of Pt–Cu alloy.

The propene yield increases in the following order: imp-PtCu/MgAl-R < PtCu/MgAl-R < PtCu/Ce0.3MgAl-R, indicating that the catalytic performance follows the same increasing trend. During the reaction, the PtCu/Ce0.3MgAl-R catalyst affords the highest propene yield of around 43.5% (around 40.2% for PtCu/MgAl-R and around 39.1% for imp-PtCu/MgAl-R). The presence of a large number of highly dispersed Pt–Cu alloy sites enhances the productivity of PtCu/Ce0.3MgAl-R to propene. However, a sharp decrease of propene yield is observed in imp-PtCu/MgAl-R, hinting that imp-PtCu/MgAl-R has a shorter lifespan than PtCu/Ce0.3MgAl-R and PtCu/MgAl-R. This can be attributed to the less Pt–Cu alloy sites and poor dispersion of metal species on the imp-PtCu/MgAl-R surface, which results in the aggregation and sintering of isolated Pt and Cu NPs during the high-temperature reaction.

The effect of Ce content on the catalytic performance of PDH is illustrated in Fig. 9. Comparing PtCu/Ce0.1MgAl-R and PtCu/Ce0.6MgAl-R, the PtCu/Ce0.3MgAl-R catalyst has a much better conversion and yield, but they both have a similar propene selectivity. This suggests that the Ce species on the surface of the catalysts noticeably influences the catalytic performance. The optimal Ce content was confirmed to be 0.3 wt%, which is related to the formation of a great amount of Pt–Cu alloy sites and high dispersion of active Pt–Cu centers and other metal species on the PtCu/Ce0.3MgAl-R surface.

Fig. 9
figure 9

Variation of a propane conversion, b propene selectivity, c propene yield for PDH with time over PtCu/CexMgAl-R with different Ce contents. Reaction conditions: 600 °C, atmospheric pressure, C3H8:H2:N2 (molar ratio) = 8:7:35, and WHSV = 3 h−1

Full size image

Deactivation Analysis

The TG spectra of the used catalysts are shown in Fig. 10. The calculated weight losses of the used Pt/MgAl-R, PtCu/MgAl-R, PtCu/Ce0.3MgAl-R and imp-PtCu/MgAl-R are 17.1, 47.1, 35.9, and 13.9%, respectively. The weight loss is mainly due to the removal of deposited coke. Usually, the higher the dehydrogenation activity of the catalyst, the easier the coke formation. However, the amount of coke formation over PtCu/Ce0.3MgAl-R is less than that of PtCu/MgAl-R. This suggests that the coke resistance can be enhanced by adding Ce, because the high oxygen storage capacity of CeO2 can promote the gasification of coke on the catalyst by storing and delivering active oxygen species.

Fig. 10
figure 10

TG profiles of the used catalysts

Full size image

The used catalysts were analyzed by XRD and the results are shown in Fig. 11. In comparison with the XRD patterns of the reduced catalysts in Fig. 3, the most remarkable change for the used catalyst is the appearance of a graphitic carbon diffraction at 2θ of about 26°, and the intensity of the peak decreases in the order of PtCu/MgAl-R > PtCu/Ce0.3MgAl-R > PtCu/Ce3.0MgAl-R > imp-PtCu/MgAl-R > Pt/MgAl-R. This order is similar to that of the total amount of carbon formation of these catalysts, indicating that the coke deposit is mainly graphitic carbon. Moreover, an Al2O3 phase appears for all samples, which may be because of the prolonged high-temperature reaction that leads to the phase separation of Al2O3 from Mg(Al)O solid solution.

Fig. 11
figure 11

XRD patterns of the used catalysts. (black diamond) Al2O3; (black heart) CeO2; (black Club) graphitic carbon; (black circle) Cu

Full size image

The TEM images of the used catalysts are illustrated in Fig. 12, where the metal NPs are surrounded by carbon materials with thin graphene sheets [52]. In Fig. 12h and l, few carbon nanotubes can be found on the used PtCu/Ce0.3MgAl-R and relatively big carbon nanotubes can be found on the used imp-PtCu/MgAl-R.

Fig. 12
figure 12

TEM images and particle size distribution of metal particles over the used catalysts of a, b Pt/MgAl-R, c, d, e PtCu/MgAl-R, f, g, h PtCu/Ce0.3MgAl-R, i, j, k, l imp-PtCu/MgAl-R

Full size image

Moreover, a visible change in the dispersion and size of metal NPs in all samples can be observed among the catalysts before and after 360 min of PDH reaction. In fact, there is almost no aggregation of metal NPs on PtCu/Ce0.3MgAl-R. This indicates an excellent anti-sintering ability for Pt–Cu alloy NPs in PtCu/Ce0.3MgAl-R. For sample imp-PtCu/MgAl-R, large Cu NPs were observed accompanied by small Pt–Cu alloy NPs after PDH reaction at 600 °C for 360 min, indicating severe sintering of Cu NPs as well as superior anti-sintering of Pt–Cu alloy. The sintering degree is determined by the amount of the isolated metal ions. In other words, the highly dispersed Pt–Cu alloy NPs are more stable than isolated metal NPs, suggesting that the more the amount of Pt–Cu alloy NPs, the better the stability of the catalyst structures.

Conclusions

A series of CeO2-modified PtCu/Mg(Al)O catalysts (PtCu/CexMgAl-R) with different Ce contents were obtained by one-pot synthesis from HT-like precursors. Sites of Pt–Cu alloy were formed and were highly dispersed by CeO2 NPs which interacted with Mg(Al)O support. The Pt–Cu alloy is the main active phase in PtCu/Ce0.3MgAl-R, unlike the contrast sample with isolated large Cu phase; this leads to a high catalytic activity and propene yield in the PDH reaction. The introduction of Cu can effectively stabilize and disperse Pt NPs by forming Pt–Cu alloy, and the subsequent addition of 0.3 wt% Ce can further disperse Pt by increasing the Pt–Cu alloy sites and improve resistance to sintering and coke deposition. The excellent catalytic behavior of CeO2-modified Pt–Cu alloy in PDH can enhance supported Pt–Cu bimetallic catalysts for a better performance in the dehydrogenation of light alkanes.