ZnO Nanoparticles Encapsulated in Nitrogen-Doped Carbon Material and Silicalite-1 Composites for Efficient Propane Dehydrogenation.

Non-oxidative propane dehydrogenation (PDH) is an attractive reaction from both an industrial and a scientific viewpoint because it allows direct large-scale production of propene and fundamental analysis of C-H activation respectively. The main challenges are related to achieving high activity, selectivity, and on-stream stability of environment-friendly and cost-efficient catalysts without non-noble metals. Here, we describe an approach for the preparation of supported ultrasmall ZnO nanoparticles (2-4 nm, ZnO NPs) for high-temperature applications. The approach consists of encapsulation of NPs into a nitrogen-doped carbon (NC) layer in situ grown from zeolitic imidazolate framework-8 on a Silicalite-1 support. The NC layer was established to control the size of ZnO NPs and to hinder their loss to a large extent at high temperatures. The designed catalysts exhibited high activity, selectivity, and on-stream stability in PDH. Propene selectivity of about 90% at 44.4% propane conversion was achieved at 600°C after nearly 6 h on stream.

Introduction

Propene is the second-largest building block of the chemical industry with a steadily increasing demand owing to its plentiful downstream applications. Stream cracking and fluid catalytic cracking (FCC) of various oil fractions are the most common methods for producing propene. These technologies have several drawbacks such as high-energy consumption and low selectivity to propene. In addition, metathesis of ethylene and 2-butenes (Mol and van Leeuwen, 2008) and non-oxidative propane dehydrogenation (PDH) (Caspary et al., 2008) have been developed for on-purpose propene production. The latter process is the most attractive technology because of the fast exploitation of shale gas providing an exciting opportunity for producing light olefins and aromatics (Wang and Li, 2017, Bruijnincx and Weckhuysen, 2013). For this reason, the PDH technology has attracted increasing attention of researchers from industry and academia around the world (Liu et al., 2016a, Sokolov et al., 2012, Kim et al., 2017, Hu et al., 2018, Zhu et al., 2017, Sattler et al., 2014a). From a fundamental viewpoint, PDH is also of great significance because it is a good model reaction for studying the fundamentals of the activation of C-H bond.

As PDH is a strongly endothermic reaction, it requires high temperatures (>500°C) to achieve industrially attractive degrees of propane conversion. Chromium-based (Mentasty et al., 1999, Weckhuysen and Schoonheydt, 1999) and platinum-based (Jiang et al., 2015, Shi et al., 2015, Li et al., 2017, Xiong et al., 2017) materials are the commercial catalysts used for this reaction, which have, however, shortcomings related to toxicity and high cost of the active components, respectively. To overcome them, a lot of efforts have been put into developing new kinds of alternative catalysts such as gallium-based (Kim et al., 2017, Choi et al., 2017, Sattler et al., 2014b), vanadium-based (Liu et al., 2016a, Hu et al., 2018, Sokolov et al., 2012), tin-based (Wang et al., 2016, Wang et al., 2017), and zirconium-based (Otroshchenko et al., 2015, Otroshchenko et al., 2016, Otroshchenko et al., 2017a, Otroshchenko et al., 2017b) materials. Owing to the recent developments in nanoscience and synthetic technologies, ZnO nanoparticles (ZnO NPs) or ZnO quantum dots as promising semiconductor materials were elaborately synthesized and are widely used in photocatalysis (Etacheri et al., 2012, He et al., 2014), photodetectors (Tang et al., 2018, Shao et al., 2013), degradation of organic pollutants (Akkari et al., 2017), and other low-temperature applications. ZnO-based materials are promising catalysts for PDH because of their costs, environment friendliness, and high efficiency to activate C-H bond (Sun et al., 2014, Schweitzer et al., 2014, Camacho-Bunquin et al., 2017, Liu et al., 2016b, Biscardi et al., 1998). However, ZnO is often unstable under reducing reaction conditions because of the formation of metallic Zn, which melts at about 420°C. To improve the stability of ZnO at high temperatures, synthesis of composite oxides (Sun et al., 2014) and modification of ZnO with Pt (Liu et al., 2016b) have been proposed. To the best of our knowledge, ultrasmall ZnO NPs (<10 nm) have still not been studied in high-temperature (>420°C) reactions. The main reason is related to the difficulties in preparing catalysts possessing highly dispersed NPs with high stability against sintering and Zn/ZnO volatilization under reaction conditions (Anthrop and Searcy, 1964, Sirelkhatim et al., 2015).

Owing to their intriguing topologies and diverse functionalities, metal organic frameworks (MOFs) are widely used for gas separation (Hu et al., 2015, Cadiau et al., 2016, Rodenas et al., 2014), sensors (Campbell et al., 2015), and catalysis (Lee et al., 2009, Zhao et al., 2014, Zhao et al., 2016a, Zhao et al., 2016b, Zhang et al., 2017). In the latter case, MOFs and metals@MOFs are applied in photocatalysis (Nasalevich et al., 2015), organic catalysis (Huang et al., 2017), and other reactions (Yang et al., 2016, Liu and Tang, 2013) as they possess highly dispersed metal sites. A recent attractive direction for MOF applications is their usage as precursors to generate highly dispersed metals or metal oxides@carbon materials (even single-site atoms@carbon material) (Chen et al., 2015, Chen et al., 2017, Yin et al., 2016). The formed carbon material or N-doped carbon material could isolate metal or metal oxide species, and M(metal)-N(N atom)-C(carbon material) bonds can be formed in some cases to prevent the catalytically active species from aggregation. In spite of this significant progress, reports on using non-metal heteroatoms in MOFs to synthesize stable metal and/or metal oxide-based catalysts at high temperatures are still rare. With this thought in mind, here we introduce a methodology for the preparation of composite catalysts with thermally stable ZnO NPs encapsulated into a nitrogen-doped carbon layer (NC layer) on the surface of Silicalite-1. The NC layer is formed through carbonizing zeolitic imidazolate framework-8 (ZIF-8, Zn(2-methylimidazole)2) followed by leaching with nitric acid. Scanning transmission electron microscopy and H2-temperature programmed reduction (H2-TPR) revealed that the presence of N species in the NC layer is decisive for stabilizing the NPs at temperatures up to 700°C. The potential of the so-designed catalysts was validated for the PDH reaction at 600°C. Owing to the stabilizing effect of the NC layer on ZnO NPs, the catalysts showed high on-stream stability and activity.

Results and Discussion

For the preparation of ZnO@NC and Silicalite-1 composite catalysts, Silicalite-1 was chosen as the support because it possesses high specific surface area and unique porosity (Figure S1), and it was synthesized according to the reference (Shen et al., 2013). Moreover, silica-based materials are typically used as supports for preparing selective PDH catalysts as they do not possess acidic sites, which are considered to negatively affect selectivity due to coke formation. To form a ZIF-8 layer on this support, Zn(NO3)2·6H2O and 2-methylimidazole were added to a suspension of Silicalite-1 in methanol. After centrifuging and drying, the obtained solid material was carbonized at 700°C in N2 atmosphere and denoted as ZnO@NC/S-1(0.0). The ZnO@NC/S-1(0.0) sample was further treated by nitric acid for partially dissolving ZnO species and thus generating more uniform ZnO NPs with smaller size. This material was abbreviated as ZnO@NC/S-1(x), where x stands for the concentration of the acid. Zn loading determined by inductively coupled plasma spectrum (ICP) in ZnO@NC/S-1(0.0) and ZnO@NC/S-1(1.0) was 2.8 wt % and 2.0 wt %, respectively. The X-ray diffraction patterns of as-synthesized ZIF-8 (Figure 1A) match well with the simulated pattern, confirming the formation of the ZIF-8 phase (Pan et al., 2011, Kuo et al., 2012). A reflex at 2θ of 7.3° characteristic for ZIF-8 is also presented in the diffractogram of Silicalite-1 coated with a ZIF-8 layer. The coated material maintained the typical MFI topological structure with good crystallinity after carbonizing the ZIF-8 layer followed by acid leaching process. No reflexes characteristic for bulk ZnO or NC material were observed. Thus the presence of large crystalline particles of ZnO and NC in the as-synthesized catalysts can be excluded. To derive an insight into the nature of carbon in the NC layer, we applied Raman spectroscopy. The obtained spectra of ZnO@NC/S-1(0.0) and ZnO@NC/S-1(1.0) are characterized by two evident bands at about 1,340 cm−1 and about 1,590 cm−1 (Figure 1B). These bands can be ascribed to the defective structure (D band) and graphitic carbon (G band), respectively (Lim et al., 2012, Zhang et al., 2014). The relatively low ratio of the D band intensity to that of the G band indicates the high degree of graphitization of the formed NC material. X-ray photoelectron spectroscopy (XPS) was applied to determine catalyst surface composition. All XPS signals were adjusted by the position of Si 1s with a binding energy of 103.3 eV. As seen in Figure 1C, two XPS signals at about 1,022 eV and about 1,045 eV are presented in the spectra of all catalysts and are characteristic for Zn 2p1/2 and Zn 2p3/2, respectively (Ma et al., 2011, Aksoy et al., 2012). However, in comparison with ZnO loaded on Silicalite-1 (ZnO/Silicalite-1, 1,022.2 eV), the Zn 2p1/2 binding energy for ZnO@NC/S-1(0.0) and ZnO@NC/S-1(1.0) is slightly shifted to lower values, i.e., 1,022.0 eV and 1,021.6 eV, accordingly. The energy shift may be caused by the electron donor property of the introduced NC layer. After HNO3 leaching, the binding energy value is also shifted from 1,022.0 eV to 1,021.6 eV, which may be caused by decreasing ZnO content. Detailed N 1s spectra of ZnO@NC/S-1(0.0) and ZnO@NC/S-1(1.0) are presented in Figure S2. On their basis, three different kinds of N species were identified: pyridinic-N, pyrrolic-N, and graphitic-N. Their percentage distribution is given in Table 1. The profiles of temperature-programmed desorption of propene (C3H6-TPD) from these two samples are presented in Figures 1D and S3, and the temperature (Tmax) of maximal desorption is given in Table 1. These data suggest that C3H6 adsorbs weaker on the ZnO@NC/S-1(1.0) sample than on the ZnO@NC/S-1(0.0) sample. Owing to electron-rich property of C3H6, it adsorbs more strongly on the ZnO@NC/S-1(0.0) sample with electron-deficient state (higher values of Zn binding energy) as confirmed by XPS results.

Figure 1

Physicochemical properties of as-synthesized catalysts

(A) X-ray diffraction patterns of as-synthesized catalysts.

(B) Raman spectra of ZnO@NC/S-1(x) samples.

(C) Zn 2p spectra of as-synthesized catalysts.

(D) Temperature-programmed desorption profiles of propene from ZnO@NC/S-1(x) samples.

See also Figures S2 and S3.

Table 1

Specific Surface Area (SBET), Zn Loading, Rate of Coke Formation (rcoke), Temperature (Tmax) of Maximal Propene Desorption, and Surface Distribution of Different N Species

Samples	SBET (m2/g)	Zn Loading (wt %)a	rcoke (g⋅g−1cat⋅h−1)b	Tmax (oC)c	N Speciesd
					Pyridinic N	Pyrrolic N	Graphitic N
ZnO/S-1	426	2.5	0.004	–	–	–	–
ZnO@NC/S-1(0.0)	351	2.8	0.10	324	32%	40%	28%
ZnO@NC/S-1(1.0)	370	2.0	0.01	316	48%	25%	27%

See also Figures S1 and S9.

Zn loading was determined by ICP.

rcoke was calculated on the basis of thermogravimetric analysis (TGA) according to Equation 1. (See it in Supplemental Information)

Determined from temperature-programmed profiles of propene desorption.

Obtained through deconvolution of detailed N 1s peak in the XPS.

To determine the size of ZnO NPs, high-angle annular dark-field scanning transmission electron microscopy (STEM) and bright-field STEM were applied, and the corresponding representative images are shown in Figures 2A and 2B. It can be clearly seen that ZnO NPs are encapsulated into the NC layer. ZnO NPs in ZnO@NC/S-1(0.0) have a broad distribution with an average size of 4.0 nm. Treatment with 1 M HNO3 results in a narrower distribution of ZnO NPs with an average size of 2.6 nm in ZnO@NC/S-1(1.0). This may be due to the partial dissolution of ZnO in HNO3 as concluded from the ICP results shown in Table 1. To elucidate the factors determining the stability of ZnO NPs at high temperatures, the ZnO@NC/S-1(1.0) sample was treated in either a flow of pure N2 (ZnO@NC/S-1(1.0)-N2-700°C) or a mixture of 10 vol % H2 in N2 (ZnO@NC/S-1(1.0)-10 vol % H2-700°C) at 700°C for 2 h. ZnO NPs are hardly visible (Figures 2A III, 2B III, and S4) in the sample treated in the H2-containing mixture, and the ZnO loading is 0.4 wt %. The latter result suggests that most of ZnO was lost during 10 vol % H2 treatment at 700°C. In contrast, ZnO NPs did not disappear after N2 treatment (Figures 2A IV and 2B IV) as concluded from the fact that Zn loading decreased only slightly from 2.0 to 1.8 wt %. The size of Zn NPs also decreased from 2.6 to 2.0 nm. To derive an insight into the effect of H2 treatment on the stability of ZnO NPs, H2-TPR tests were additionally carried out using a fresh sample. When carefully analyzing the H2-TPR results (Figure S5), no signal related to water could be identified, thus indicating that ZnO NPs could not be reduced upon reductive catalyst treatment. As proven by in situ diffuse reflection infrared fourier transform spectrum (DRIFTs) analysis of the catalysts treated in different atmospheres and at different temperatures (Figure S6), no surface functional groups characteristic of aromatic C-N or C=N species could be identified after treatment at 700°C in H2/N2. Such species were, however, present after treatment in N2. Thus we can conclude that N species could be removed in the form of NH3 upon H2 treatment at high temperatures (>650°C). Taking the above results into account, it can be confirmed that the N species can provide protective function for ZnO NPs to a large extent at high temperatures. Consequently, the aggregation and loss of ZnO could be prevented by carbon layer and N species, respectively. Such efficient preparation of ultrasmall ZnO NPs by encapsulating in NC material via post-treatment of MOFs thus presents a platform for maintaining potential stability in PDH at 600°C.

Figure 2

Electron microscopic characterization of as-prepared catalysts

(A) HAADF STEM images and (B) BF-STEM images of as-synthesized catalysts. (I) ZnO@NC/S-1(0.0), (II) ZnO@NC/S-1(1.0), (III) ZnO@NC/S-1(1.0)-10 vol % H2-700°C, (IV) ZnO@NC/S-1(1.0)-N2-700°C, (V) spent ZnO@NC/S-1(1.0) (Scale bars, 50 nm in A [I–IV] and 20 nm in A [V] and B [I–V[).

See also Figures S4–S6.

To check if and how the kind (ZnO@NC-based versus impregnated catalyst) of catalyst and the presence of N in NC layer affect catalytic performance in PDH, catalytic tests were performed at 600°C and atmospheric pressure. Propane conversion and propene selectivity as a function of time on stream are summarized in Figures 3 and S7. Poor catalytic performance of bare Silicalite-1 (S-1) as well as NC and S-1 (NC/S-1) composite material (Figure S7) indicates that the support and the NC material are inactive for PDH, whereas Zn-containing materials were active. Thus ZnO NPs should be the active species. However, catalytic activity depends on how this active component was introduced on the surface of Silicalite-1. For comparison, when a similar amount of ZnO (2.5 wt %) was introduced into Silicalite-1 by incipient wetness impregnation (ZnO/S-1), propane conversion was only 13.7% and the selectivity to propene was below 70% (Figure S7). When ZnO/S-1 was treated with 1 M HNO3, the initial conversion of propane decreased from 13.7% to 6.9%, whereas the selectivity to propene slightly increased. The as-prepared ZnO@NC/S-1(x) composite catalysts performed significantly superior. This can be due to the presence of ultrasmall ZnO NPs. STEM analysis proved that the NPs were stable during the PDH. The size of ZnO NPs in spent ZnO@NC/S-1(1.0) did not change obviously (2.2 nm, Figures 2A V and 2B V). In addition, the conversion of propane and the selectivity to propene increased with an increase in the concentration of nitric acid used for catalyst leaching. Acidic treatment is also important for catalyst stability against deactivation. The higher the acid concentration, the higher the on-stream stability was (Figure S7). For example, the conversion of propane over ZnO@NC/S-1(0.0) dropped from 52.2% to 15.3% within 315 min on propane stream, whereas the corresponding values for ZnO@NC/S-1(1.0) were 56.4% and 44.4%. For the ZnO@NC/S-1(1.0) sample, the selectivity to propene maintained at about 90% during stability test. Other undesired gas-phase products were mainly methane, ethane, and ethene (Figure S8). The selectivity to propene over ZnO@NC/S-1(1.0) was also higher than that over ZnO@NC/S-1(0.0). Such phenomenon may be caused by a decrease in particle size of ZnO after acid catalyst treatment. According to previous studies on PDH reaction (Zhang et al., 2018, Zhu et al., 2015), undesired reactions are inhibited upon decreasing the size of catalytically active species. The amount of coke formed during the PDH reaction was quantitatively determined by thermogravimetric (TG) analysis (Figure S9) and used for calculating the average rate of coke deposition (Table 1). The lowest rate was obtained for ZnO/S-1 due to its very low PDH activity and consequently low concentration of propene responsible for coke formation. Importantly, the rate of coke deposition over ZnO@NC/S-1(1.0) is 10 times lower than that over the ZnO@NC/S-1(0.0) catalysts, although the samples operated with a similarly high initial activity. Taking into account the results of C3H6-TPD (Figure 1D) and TG tests, it can be concluded that C3H6 interacts weaker with ZnO@NC/S-1(1.0) than with ZnO@NC/S-1(0.0). As a consequence, appropriate adsorption/desorption properties of C3H6 species would effectively hinder their transformation into coke precursors (Jiang et al., 2015, Shi et al., 2015). On the basis of the results of carbon and ICP analysis of catalysts before and after reaction (Table 1 and Figure S9), reaction-induced carbon deposition should be the main reason for catalyst deactivation, which is slow over ZnO@NC/S-1(1.0). In addition, although the loss of ZnO has been hindered in the present case, it will also result in a drop in the activity to some extent. Unfortunately, when carrying out oxidative catalyst regeneration, not only carbon deposits but also NC layer will be oxidized, thus resulting in the loss of their protection function. The problem may be solved by selective coke removal by controlling combustion parameters. The related work is ongoing in our group.

Figure 3

Catalytic performance of as-prepared catalysts

(A) Propane conversion and (B) propene selectivity. Reaction conditions: 0.2 g catalysts, 600°C, H2: C3H8: N2 = 1: 1: 5, N2 flow rate was 7.5 mL/min.

See also Figures S7 and S8.

In summary, a facile strategy was developed to prepare catalysts with ultrasmall ZnO NPs being stable up to 700°C. An NC layer in situ formed through carbonization of MOFs is used for encapsulation of such NPs and thus protects ZnO NPs from sintering and volatilization at high-temperature. The so-designed catalysts demonstrate high activity, selectivity, and on-stream stability in PDH. Such efficient utilization of MOFs via simple post-treatment for the construction of stable ZnO NPs may provide new insights into the design of highly effective metal oxide NPs and promote their catalytic applications.

Limitations of the Study

Additional characterization of the ZnO nanoparticles would be warranted, and regeneration of the catalyst would be also valuable.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.