Shape-selective alkylation of benzene with ethylene over a core-shell ZSM-5@MCM-41 composite material

Zhenhao Shen, Chongwei Ma, Darui Wang, Junlin He, Hongmin Sun, Zhirong Zhu,Weimin Yang,*

1 Shanghai Key Lab of Chemical Assessment and Sustainability, School of Chemical Science and Engineering, Tongji University, Shanghai 200092, China

2 Shanghai Research Institute of Petrochemical Technology, Sinopec, Shanghai 201208, China

3 School of Chemical Engineering, East China University of Science and Technology, Shanghai 200237, China

Keywords:Core-shell ZSM-5@MCM-41 Coating Ethylbenzene Para-diethylbenzene Shape selectivity Deactivation

ABSTRACT A series of ZSM-5@MCM-41 core-shell composite materials prepared via a multi-cycle-sol-gel coating strategy is investigated as the catalyst for benzene alkylation with ethylene,in which both ethylbenzene and para-diethylbenzene(p-DEB)are aimed as the target products.With multi-cycle-sol-gel coating,the external acid sites on the samples are gradually passivated by the inert MCM-41 shell.As a result, the shape selectivity to p-DEB is greatly enhanced.Nevertheless, the coating of mesoporous MCM-41 shell on ZSM-5 accelerates deactivation of the catalyst only due to the dilution effect of ZSM-5 content in the catalyst at the same space velocity,which is a reason that core-shell ZSM-5@MCM-41 will potentially be a practical catalyst in shape selective alkylation of benzene.In order to enhance the yield of p-DEB on ZSM-5@MCM-41, the reaction conditions at the fixed bed reactor including temperature, the molar rate of benzene to ethylene and GHSV, are also optimized.

1.Introduction

Para-diethylbenzene (p-DEB) is a high value-added chemical intermediate and often used as a superb desorbent in the adsorptive separation process of xylene isomers.Moreover, it is also a raw material for the production of copolymers, such as ionexchange resins and cross-linking agents [1].Recently, the preparation of p-DEB has been reported mainly based on shapeselective alkylation or disproportionation of ethylbenzene (EB).Between the two reactions, it is considered that alkylation has higher molecular efficiency than disproportionation [2,3].As the raw material of p-DEB preparation, EB is an important bulk chemical,which is mainly consumed to produce styrene monomer.Practically, EB is usually produced via alkylation of benzene with ethylene in vapor phase over MFI-type Zeolite(ZSM-5)or in liquid phase over *BEA-type or MWW-type zeolites [4].

For a qualified industrial catalyst of benzene alkylation with ethylene, it is of great importance to ensure a high ethylene conversion, preferably higher than 99%, so as to avoid raw material waste.DEB isomers are inevitably formed by consecutive alkylation of EB with ethylene and can be recycled to recover EB by trans-alkylation with benzene in another reactor, but the yield of DEB should be controlled as little as possible in the purpose of energy saving.The alkylation product stream contains three DEB isomers, i.e.m-DEB (d ≈0.72 nm), p-DEB (d ≈0.56 nm) and o-DEB (d ≈0.78 nm), among which p-DEB is the most valuable isomer.If the shape selectivity to p-DEB in DEB isomers is high enough, typically higher than 90%, it will be a more economical solution that p-DEB can be acquired as a co-product in EB producing process, in which trans-alkylation procedure will be no more required.Because of the smallest size among the three DEB isomers,the diffusion rate constant of p-DEB on 10-MR zeolites,such as ZSM-5(0.53 nm×0.56 nm,0.51 nm×0.55 nm),is about 1,000 times higher than that of m-DEB or o-DEB [5].It is a reason that shape selective alkylation or disproportionation can be realized to get p-DEB on ZSM-5 referring to the principle of restricted transition state selectivity.However, non-shape-selective isomerization takes place on the external acid sites, where p-DEB is turned to other DEB isomers and a composition of quasi-thermodynamic equilibrium (m-:p-:o-DEB ≈6:3:1) is acquired.Modification, such as chemical vapor deposition (CVD) of silica [6], chemical liquid deposition (CLD) [7]of silica, oxide impregnation [7,8],pre-coking[9],etc.,can be used to passivate the external acid sites or shrink the zeolitic channels in order to avoid isomerization.It is reported by Zhu that the p-DEB shape selectivity can reach 98% in ethylbenzene disproportionation on a ZSM-5 modified by SiO2CLD and MgO[7].However,these methods of modification will usually form a dense inert shell with small orifices and narrow down the zeolitic pore opening [2].Once coking takes place, the diffusion pathway will easily be blocked leading to a rapid deactivation of the catalyst.

Core-shell composite material, especially core-shell zeolitic material, has been an interesting topic for more than 20 years.Either the‘core’phase or the‘shell’phase in a core-shell composite material can play a respective role and make a synergistic effect with each other, which may be useful in catalysis [10-21].It has also been reported that some types of core-shell zeolitic composite material are good shape selective catalysts in alkylation or disproportionation of aromatics.This kind of materials is usually made of a zeolitic core and an inactive shell, e.g.HZSM-5/Silicalite-1[17,18], Al-MWW/B-MWW [19], HZSM-5/mesoporous SiO2[20,21],etc.MCM-41,firstly discovered by the researchers of Mobil in 1992 [22,23], has a hexagonal array of uniform mesopores and large surface areas above 700 m2·g-1.Because of the siliceous mesoporous wall,MCM-41 shows very weak acidity,which is considered to be inactive in some acid catalyzed reactions, such as alkylation and isomerization of aromatics.If MCM-41 is introduced as the outer shell of an acidic ZSM-5 core, the external acid sites could be eliminated to avoid non-shape-selective isomerization,meanwhile the diffusion of the reactant molecules of product molecules will probably not be affected due to the large mesopores of MCM-41.

In the paper, a series of ZSM-5@MCM-41 core-shell composite material was prepared via a sol-gel coating strategy,and the morphology and other physicochemical properties were investigated.The materials were applied to catalyze alkylation of benzene with ethylene to produce EB,as well as p-DEB as a co-product.The catalytic performance of the core-shell structured catalysts,including the shape selectivity,the stability and the optimized reaction conditions, was studied.

2.Experimental

2.1.The preparation of ZSM-5@MCM-41

Na-form ZSM-5 powder(with a Si/Al molar ratio of 60)supplied by Sinopec Catalyst Company(SCC)was transformed to H-form by three times of consecutive ion exchange in 0.1 mol·L-1NH4NO3aqueous solution at 363 K, subsequent drying at 423 K for 24 h and calcination at 823 K for 6 h, successively.The ZSM-5@MCM-41 core-shell composite material was prepared via a sol-gel coating strategy using the H-form ZSM-5 as the core.Firstly, 20.0 g aqueous ammonia (AR, 28%) and 6.0 g cetyltrimethylammonium bromide (CTAB, CP) were dispersed dropwise in a solution of 640 g ethanol and 1000 g deionized water, respectively.H-form ZSM-5 powder (typically 6.0 g) was then added in under continuous vigorous stirring.After 2 h of stirring, 8.0 g of tetraethylorthosilicate (TEOS, AR) was pumped in at a flow rate of 0.1 ml·min-1and was hydrolyzed slowly.The mixture was then aged at 298 K for 12 h under stirring.Finally, the precursor of ZSM-5@MCM-41 was filtered, washed, dried and calcined at 823 K for 5 h.A cycle of coating procedure was finished and the sample named Z5@M1 was prepared.The above mentioned coating procedure was repeated on the coated samples to get Z5@Mx.The letter‘x’is the number of cycles.For catalytic testing,the samples were tableted,crushed and screened,successively,to φ0.9-1.7 mm granules.

2.2.Characterization

XRD (X-ray diffraction) patterns of the samples were gathered by a Bruker D8Advance diffractometer using Cu Kα radiation at 40 kV and 40 mA with the scanning rates of 0.5 (°)·min-1(2θ = 1°-5°)and 4(°)·min-1(2θ=5°-50°).SEM(Scanning electron microscope) and TEM (transmission electron microscopy) photos were collected by an FEI Nova NanoSEM 450 and a JEOL-2010F operated at 200 kV, respectively.N2adsorption/desorption isotherms were got by a Micromeritics ASAP 2010 porosimeter.The samples were degassed at 623 K for 3 h in vacuum.The specific surface area was calculated by BET method.BJH method is used to determine pore size distribution patterns, and t-plot method (p/p0= 0.05-0.35) is used to calculate the micropore volume.The Si/Al molar ratios were determined using chemical elemental analysis assisted by inductively coupled plasma(ICP)method.In order to determine the acidic properties of the samples, NH3-TPD (temperature programmed desorption) and probe molecules cracking test.NH3-TPD was carried out on a customized chemisorption apparatus to investigate the acidity of the ion-exchanged samples, which were pre-treated at 873 K for 1 h, and then cooled to 373 K for NH3adsorption.Finally, desorption was carried out at 373-873 K with a heating rate of 10 K·min-1, and the desorbed NH3was detected by a TCD (thermal conductivity detector).Cracking of isopropylbenzene (IPB) and 1,3,5-tri-iso-propylbenzene (TIPB) were carried out on a fixed bed micro-reactor equipped with an on-line GC.Reaction conditions:catalyst loading=100 mg,injection quantity of IPB or TIPB=10 μl,flow rate of argon carrier=20 ml·min-1.

2.3.Catalytic performance test

Shape-selective benzene alkylation with ethylene was carried out in a stainless fixed bed reactor (i.d.= 12 mm).Samples to be tested were pretreated at 723 K for 2 h under N2flow,before benzene was introduced.When a stable vapor flow and pressure is maintained, ethylene was added to start the alkylation.The reaction conditions included:T=593-593 K,P=1.0 MPa,Bz/C2=molar ratio = 1-9 and total GHSV = 1600-9000 h-1.The products were analyzed by an on-line GC (gas chromatograph, Agilent 7890B)equipped with a FID (flame ionization detector).Ethylene conversion,ethylation selectivity,product yield and DEB shape selectivity were calculated according to the mass content (X) of each component and the formulas below.

3.Results and Discussion

3.1.Physicochemical properties of core–shell ZSM-5@MCM-41

Fig.1.XRD patterns of the parent ZSM-5 and the ZSM-5@MCM-41 samples.

XRD patterns of the samples in Fig.1 show characteristic peaks of MFI topology at 2θ=8.0°,8.9°,23.1°,24.0°,etc.The intensities of the peaks decrease simultaneously when the cycles of coating increases.On the other hand, for the coated samples (Z5@M1,Z5@M2, Z5@M3 and Z5@M4) a characteristic peak at 2θ = 2.5°(d ≈ 3.5 nm) appears referring to the hexagonal mesopores,(1 0 0) face,of MCM-41.The intensity of this peak increases when the cycles of coating increases.Meanwhile, a broad peak at 2θ =4°-5° appears referring to the (2 0 0) and (1 1 0) face of MCM-41 crystals [23].It reveals the formation and the growth of mesoporous MCM-41 phase around the ZSM-5 crystals.

Photos captured by SEM and TEM are shown in Figs.2 and 3,respectively.According to the SEM photos, the crystals of the parent ZSM-5 sample show hexagonal morphology with a smooth surface,whose size is around 200-300 nm.The particle size grows up and the surface becomes rough by one cycle of MCM-41 coating.The particle grows bigger by more than two cycles of MCM-41 coating,while its morphology becomes sphere with a rough surface.TEM results review the core-shell structure of the ZSM-5@MCM-41 composite materials.In the TEM photo of Z5@M1, Z5@M2 or Z5@M3, there is an obvious boundary between the ZSM-5 ‘core’and MCM-41 ‘shell’.The thickness of the MCM-41 shell can be increased by about 50 nm by each cycle of coating.

The parent ZSM-5 is considered to be a conventional microporous material which shows a type-I isotherm in Fig.4(a).But the isotherms of the core-shell composite materials (Z5@Mx)belong to the combination of type I and IV.The inflection point of relative pressure(p/p0)appears at about 0.3,which is obviously higher than that for the parent ZSM-5.These phenomena indicate the co-existence of microporous and mesoporous textural properties in the core-shell composite materials.The maximum N2adsorption amount raises as the cycle of coating increases resulting a dramatic increase of total pore volume.Due to the formation of micro-/mesoporous combination, the specific BET surface area increases from 356 m2·g-1(parent ZSM-5) to 787 m2·g-1(Z5@M4).Because of the increasing content of mesoporous silica,the microporous surface area decreases from 257 m2·g-1(parent ZSM-5) to 111 m2·g-1(Z5@M4), and the microporous volume decreases from 0.124 cm3·g-1(parent ZSM-5) to 0.035 cm3·g-1(Z5@M4).As shown in Fig.4(b), the mesoporous distribution of the parent ZSM-5 is mainly located at 2 nm referring to the mesopores created by crystal heaping.For the remaining samples, an additional peak appears at about 3 nm reflecting the mesopore of MCM-41, which is in accord with the results of XRD (Fig.1).

Fig.2.SEM photos of the samples.(a) Parent ZSM-5; (b) Z5@M1; (c) Z5@M2; (d) Z5@M3.

Fig.3.TEM photos of the samples.(a) Parent ZSM-5; (b) Z5@M1; (c) Z5@M2; (d) Z5@M3.

Fig.4.(a) N2 ad-/desorption isotherms of the samples, (b) Pore distribution patterns of the samples.

The chemical composition of the samples are determined by ICP-AES, and the results are shown in Table 1.According to the results, Si/Al molar ratio of the parent ZSM-5 is 59.8.When the cycles of coating increases, the content of silica is accordingly raised, e.g.the Si/Al molar ratio of Z5@M2 is 92.6 indicating that an additional part of silica (55%) is loaded as the shell-phase by two cycles of coating.

Table 1 Physicochemical properties of the samples tested by chemical analysis and nitrogen ad-/desorption analysis

Fig.5.NH3-TPD curves of the parent ZSM-5 and the ZSM-5@MCM-41 samples.

Table 2 Results of the probe reaction tests on the samples: iso-propylbenzene (IPB) cracking and 1,3,5-tri-iso-propylbenzene (TIPB) cracking

3.2.Acidity measurement

The NH3-TPD results of the parent ZSM-5, Z5@M1, Z5@M2 and Z5@M3 are summarized in Fig.5.The peak of weak acid sites is located around 523 K, while the peak of medium-strong acid sites is located between 723 K and 773 K.Because of the dilution effect from the mesoporous silica (MCM-41), the areas of both peaks decrease when adding the coating cycles.But the peak of medium-strong acid sites (723-773 K) is just slightly moving to the lower temperature referring to the slightly reduction of acid strength by coating.Medium-strong acid sites play a decisive role in many acid catalyzed reactions, including alkylation of benzene with ethylene [16].

Cracking of iso-propylbenzene (IPB, d ≈0.68 nm) and cracking of 1,3,5-tri-iso-propylbenzene (TIPB, d ≈0.94 nm) are often used to characterize bulk acidity and external acidity, respectively, of zeolites [24,25].Cracking ability of IPB and TIPB on various catalysts are tested, and the results are listed in Table 2.When the coating cycles are added,both IPB conversion and TIPB conversion drop.An important reason is the dilution effect in coated samples.Besides, it can be seen that TIPB conversion is dramatically reduced, while the descend range of IPB conversion is relatively smaller, referring to the stepped-up losing of external acid sites caused by the growth of silica shell.Although the mesoporous channel of MCM-41 is larger than TIPB molecule, according to the characterization results,there is a fraction of amorphous silica depositing on the core-shell interface, which can passivate the external surface of acidic ZSM-5.

3.3.Catalytic performance

The catalytic performance of the samples in shape-selective benzene alkylation with ethylene was investigated,and the performance data of the initial reaction stage,including product distribution and calculated results, are summarized in Table 3.Under the reaction conditions mentioned in Table 3,the ethylene conversion for the parent ZSM-5 sample is almost complete.When the cyclesof MCM-41 coating are added, ethylene conversion drops obviously, which is mainly contributed by the decrease of the zeolite content.Besides, ethylation selectivity rises slightly with the increase of coating cycle, and more yield of EB will be produced due to the lower selectivity to DEB.It is remarkable that paraselectivity (p-DEB/DEB) is obviously raised from 30.3% to 79.9%for Z5@M2 and 95.5% for Z5@M4, respectively.The yield of p-DEB is accordingly elevated from 4.24% to 9.45% for Z5@M2 and 8.86% for Z5@M4, while the yield of m-DEB and o-DEB decreases dramatically.In order to further identify the effectivity of MCM-41 coating on the improvement of para-selectivity, the catalytic performance of Z5@M2 is once more tested at a lower GHSV(2060 h-1) so as to ensure the same loading of zeolitic phase with the parent ZSM-5.It is found that p-DEB/DEB is 60.2% on Z5@M2(2060 h-1),which is obviously higher than that(30.3%)on the parent ZSM-5.Ethylene conversion,ethylation selectivity and EB yield on Z5@M2 (2060 h-1) only changes slightly compared with those on the parent ZSM-5.According to the performance test results,it is proved that MCM-41 coating, namely the formation of ZSM-5@MCM-41 core-shell structure, can improve the para-selectivity of catalyst in benzene alkylation with ethylene, which is contributed by the passivation of acidity on the ZSM-5@MCM-41 interface.

Table 3 Results of the performance test on the samples.The data was gathered at TOS = 4 h.Reaction conditions: T = 673 K, P = 1.0 MPa, Bz/C＝2 molar ratio 4:1

Fig.6.Deactivation rate comparison of the samples:yield of EB+DEB as a function of time on stream in rapid aging experiment at the reaction conditions: T = 653 K,P = 1.2 MPa, total GHSV = 9000 h-1, Bz/C 2＝ molar ratio 1:1.

Post modification on ZSM-5, such as CVD, often cause the narrowing of zeolitic pore opening by the SiO2debris decomposed by its precursor.Because of that, the catalysts will easily be deactivated by coking in alkylation or disproportionation of aromatics[2].In order to evaluate the negative effects of MCM-41 coating on the stability of the catalysts in benzene alkylation with ethylene,rapid aging experiments are carried out under a reaction condition with lower B/E ratio and higher GHSV,which can accelerate the coking on the catalysts.The (EB + DEB) yield versus time on stream curves of the parent ZSM-5 (9000 h-1) and Z5@M2(9000 h-1and 5800 h-1) are summarized in Fig.6.It is found that the activity losing rate of Z5@M2 (9000 h-1) is obviously faster than that of the parent ZSM-5.But when the parent ZSM-5(9000 h-1) and Z5@M2 (5800 h-1) are compared, i.e.the same amount of ZSM-5 component is loaded in the reactor, the activity losing rates of them are almost equally.According to the results,the weakening of stability on the MCM-41 coated samples is caused only by decreasing ZSM-5 amount of the catalyst loaded during the performance test.And deposition of mesoporous silica debris at the ZSM-5@MCM-41 interface does a very little effect on the catalytic stability.On the other hand,this indicates that coking over the acidic sites in ZSM-5 channels causes the deactivation.

Fig.7.The effect of reaction temperature on the catalytic performance of Z5@M2(ZSM-5@MCM-41), in the form of yield of EB, m-&o-DEB, p-DEB and other by-products.Other reaction conditions: P = 1.0 MPa, total GHSV = 3200 h-1,Bz/C＝2 molar ratio 4:1.

Fig.8.The effect of benzene to ethylene (B/E) molar ratio on the catalytic performance of Z5@M2(ZSM-5@MCM-41),in the form of yield of EB,m-&o-DEB,p-DEB and other by-products.Other reaction conditions:T=673 K,P=1.0 MPa,total GHSV = 3200 h-1.

It is mentioned in the first chapter of the article that it is of great importance to ensure a high ethylene conversion for a qualified industrial catalyst of benzene alkylation with ethylene so as to avoid raw material waste and that the yield of DEB should be controlled as little as possible in a conventional EB producing process.However, the situation will become more complicated, if p-DEB is seen as a co-product with EB.In the purpose of getting more p-DEB in benzene alkylation with ethylene, reaction condition study was carried out on the ZSM-5@MCM-41 (Z5@M2).The effects of reaction temperature, benzene to ethylene (B/E) molar ratio and gas hourly space velocity (GHSV) were summarized in Figs.7-9,respectively.

Fig.9.The effect of gas hourly space velocity (GHSV) of the reactants on the catalytic performance of Z5@M2 (ZSM-5@MCM-41), in the form of yield of EB,m-&o-DEB, p-DEB and other by-products.Other reaction conditions: T = 673 K,P = 1.0 MPa, Bz/C＝2 molar ratio 4:1.

As shown in Fig.7,high ethylene conversion(＞99%)can be realized at the temperature above 673 K resulting a high yield of(EB+DEB).Ethylene conversion drops as the reaction temperature goes down, while the p-DEB shape selectivity is raised due to the deceleration of non-shape-selective DEB isomerization.Although the highest p-DEB yield is got at 633 K, the ethylene conversion is only about 90% at that temperature, and the EB yield is negatively affected.Hence, the optimized reaction temperature is considered to be higher than 633 K, typically around 673 K.

It was reported that the optimized B/E molar ratio is around 7 to avoid side reactions,e.g.ethylene oligomerization and consecutive alkylation,and to ensure a high selectivity to EB[4,16].In the present study, the effect of B/E ratio on the catalytic performance of Z5@M2 is shown in Fig.8.As the B/E ratio goes up, the content of by-products decreases, which is coherent with the reference reports.Meanwhile, the yield of EB increases, but the yield of DEB decreases,which is caused by that less consecutive alkylation takes place at a lower ethylene content.However,the shape selectivity to p-DEB (p-DEB/DEB) is not affect by B/E ratio.As a result,more p-DEB will be got at a lower B/E molar ratio, preferably at 3-5,which is obviously lower than the optimized one in a conventional EB producing process.

The effect of GHSV is also investigated, and the results are shown in Fig.9.As we know, increasing GHSV means shortening the contact time of the reactants over catalyst bed.It is a reason that both ethylene conversion and yield of DEB decrease, and p-DEB shape-selectivity increases,when the GHSV goes up.As GHSV is higher than 4800 h-1,ethylene conversion on Z5@M2 drops dramatically,resulting the decrease of p-DEB yield.In order to get the highest yield of p-DEB, the optimized GHSV is around 3200 h-1.

The 1000-hour-lifetime test of Z5@M2 is carried out and is compared with that of the parent ZSM-5.According to the results in Fig.10 (left), ethylene conversion on Z5@M2 drops slightly in the first 1000 hours, but has a dramatic decrease beyond that time(This indicates the porous channels of Z5@M2 are seriously blocked by increasing coke), while deactivation on the parent ZSM-5 is not obvious.The stability loss of Z5@M2 is mainly contributed by the dilution effect of MCM-41 component on ZSM-5 content in the catalyst,which can be instructed by the rapid aging experiments (Fig.6).As the time on stream goes up, p-DEB yield rises gradually in 1000 hours due to the increasing p-DEB shapeselectivity caused by coking, and it drops at 1200 hours because of the decrease of ethylene conversion (See Fig.10 (right)) As far as catalytic performance is concerned, the ZSM-5@MCM-41 sample (Z5@M2) is qualified for the industrial use.

4.Conclusions

A series of ZSM-5@MCM-41 core-shell composite materials is prepared via a sol-gel coating strategy with multiple coating cycles, and is studied as the catalyst for benzene alkylation with ethylene.In the processing for EB production, p-DEB is also aimed as a target co-product.The composite material has a typical coreshell structured morphology.The shell phase shows MCM-41 characterized mesopores,which are found to donate abundant external surface area and pore volume.After multiple cycles of coating,the external acid sites are gradually passivated by the inert MCM-41 shell.As a result, the non-shape-selective isomerization of DEB on ZSM-5 surface is obviously restricted, and the shape selectivity to p-DEB is greatly enhanced.The shape selectivity reaches about 80% over a twice coated sample, while ethylene conversion is higher than 99%.It is found that the coating of mesoporous silica on ZSM-5 accelerates deactivation of the catalyst only due to the dilution effect of ZSM-5 content at the same space velocity indicating that coking over the acidic sites in ZSM-5 channels causes the deactivation, and passivation of acid sites on the interface of core-shell ZSM-5@MCM-41 plays little role in stability loss.In order to enhance the yield of p-DEB on ZSM-5@MCM-41, the effect of reaction conditions is investigated on catalytic performance, and the optimized temperature, B/E molar ratio and GHSV are suggested.

Fig.10.Results of the 1000-hours-lifetime test.Left: ethylene conversion as a function of time on stream, a comparison between the parent ZSM-5 and Z5@M2 (ZSM-5@MCM-41).Right:yield of EB,m-&o-DEB,p-DEB and other by-products on Z5@M2 versus time on stream.Reaction conditions:T=673 K,P=1.0 MPa,Bz/C＝2 molar ratio 4:1,total GHSV = 3200 h-1.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

The authors wish to acknowledge the financial support by the National Natural Science Foundation of China (91534115) and the National Key Research and Development Project(2016YFC1102300).