The correlation between nitrogen species in coke and NO x formation during regeneration☆

Teng Li,Chaohe Yang,Xiaobo Chen*,Libo Yao,Wei Liang,Xuemei Ding

State Key Laboratory of Heavy Oil Processing,China University of Petroleum,Qingdao 266580,China

1.Introduction

The nitrogen-containing compounds should not be neglected due to the increasing level of impurities in the feedstock for fluid catalytic cracking(FCC).As an important source of FCC feedstock,a typical vacuum gasoil(VGO)only contains approximately 25%-30%of the nitrogen existent in crude oil.However,in the last few years,more fluid catalytic crackers began to process vacuum residua(VR)containing 70%-75%of the nitrogen present in crude oil[1].Moreover,coker gas oil(CGO),which is characterized by high nitrogen compounds and polynuclear compounds,was blended into the FCC feedstock as well[2-4].Table 1 compares the nitrogen content of VGO,VR and CGO from three typical Chinese refineries and indicates that,using FCC technology to process vacuum residua or coker gasoil has to deal with the increasing nitrogen content in heavy feedstock.

Studies on the deactivating effect of nitrogen compounds during FCC process have been carried out for several decades,and earlier work on this subject concluded that,nitrogen bases could deactivate FCC catalysts by interacting with the acid sites responsible for cracking reaction[5-12].Zhao et al.studied the nitrogen balance during the cracking process and found the percentage of nitrogen being converted to coke approximates the percentage of basic nitrogen in the FCC feedstock[13].This result illustrates basic nitrogen compounds are comparatively easier to be part of coke.

Apart from the deactivation of catalysts caused by nitrogen compounds,nitrogen oxides(NOx)emission during the regeneration of coked catalysts is an environmental issue.Atypical FCC section contributes about 50%of NOxemission in a refinery[13,14].In order to comply with environmental legislation,researchers try to find ways to control NOxemission from adopting novel additives or regenerator design.

Regarding the formation of NOx,the reaction route of fluidized bed coal combustion was introduced for reference[15,16].HCN and NH3are considered the intermediates for the formation of NOx[17,18].It is also well-known that the reduction of NO by CO is inhibited when CO combustion promoter is used[19-21].Therefore,a new generation of combustion promoter focuses on two aspects,i.e.promoting CO combustion and maintaining NO emission level[22,23].Moreover,NOxreduction can be achieved by changes in the regenerator design[24,25].However,considering the huge cost,the majority of refineries have to rely on operation changes or catalytic approaches.

As a model successfully applied to the kinetic calculation of coke formation,the multi-layer structure of coke molecule has been proposed for several decades[26].But from a regeneration point of view,littleliteracy ever introduced this model to control NO emission in practice.Based on the multi-layer structure model,the aim of this study is to identify the correlation between nitrogen species in coke and NOxformation during regeneration,hoping to find a new strategy to control NOxemission.

Table 1 Nitrogen content of various feed stocks(wt%)

2.Experimental Section

The catalysts used in the experiments were commercial USY zeolite based equilibrium FCC catalysts supplied by Changqing re finery of China National Petroleum Corporation(CNPC)and had been pretreated under air flow at 700°C for 2 h before usage.

First,three coked catalysts were obtained in a fixed bed by using model compounds(o-xylene and quinoline)as reactants.The tests were performed at 500°C and atmospheric pressure.Prior to reaction,the catalyst of(5±0.005)g was pretreated for 0.5 h in a 40 ml·min-1nitrogen flow at 500°C to eliminate water adsorbed.After reaction,nitrogen flow was used to sweep the products located on the catalyst surface.For coked Catalyst A,1.0 g mixture of o-xylene and quinoline(0.8 and 0.2 g respectively)was injected with the weight hourly space velocity(WHSV)of 0.1 min-1by maintaining the flow constant(0.5 g·min-1).For coked Catalyst B,0.2 g quinoline was firstly injected followed by 0.8 g o-xylene.Coked Catalyst C was obtained by reversing feeding order with 0.8 g o-xylene in advance.Elemental analysis(C,N)of coke on the catalyst was carried out on an Elemental Analyzer Vario ELIIICNHS/O.

Surface acidity of the catalysts was confirmed by IR spectra of pyridine adsorption with a Mercury Cadmium Telluride(MCT)detector using a Nexsus?FT-IR spectrometer.The spectra were recorded with 4 cm-1and 64 scans.The concentrations of the Br?nsted and Lewis sites were determined by the absorbance of the bands respectively at 1545 cm-1and 1450 cm-1[27,28].

X-ray Photoelectron Spectroscopy(XPS)characterization of coke was performed on a PHI-5000 Versaprobe.C 1s and N 1s envelopes were subjected to a de-convolution procedure by using a minimal number of peaks,changing FWHM,position and intensities of peaks to generate the best NLLS fit(χ2＜0.99).A mix of Gaussian(80%)and Lorentzian(20%)functions was adopted.

The regeneration process was achieved through Thermogravimetry coupled with Mass spectroscopy(TG-MS)and the gas composition during the oxidation could be monitored continuously.During the temperature programmed process,0.1 g of sample was loaded into the apparatus at first,in contact with a mixture of 1 vol%O2/Ar,and then heated to 850°C at a fixed rate of 20°C·min-1.The effluent gas was analyzed by an Aeolos32 Quadruple MS.

3.Results and Discussion

3.1.Coke formation and characterization

Table 2 shows the result of elemental analysis for coked catalysts.The contents of carbon element in coked catalysts are arranged in the following order:Catalyst C＞Catalyst A＞Catalyst B.Obviously,the feeding order of the same feedstock has an effect on the coke content.Under present reaction conditions,quinoline molecules hardly crack on catalysts,which has been proved by few gas products detected by GC in our previous experiments.This can be attributed to its nature of aromaticity and chemical stability.Therefore,it is rational to assume thatquinoline deposited on catalysts still preserves its aromatic structure,that is,one nitrogen atom is still corresponding with nine carbon atoms.As a consequence,the carbon atoms in the coked catalysts can be divided into two categories according to different sources and calculated based on the above assumption,as shown in Table 3:the carbon atoms from quinoline and those from o-xylene.The coked Catalyst B presents the least carbon content from o-xylene,since the injection of quinoline before o-xylene caused the loss of more acid sites,preventing the subsequent reaction of o-xylene to form coke.On the contrast,the coked Catalyst A has more carbon atoms originated from o-xylene,which might imply that the multi-layer structure of coke on coked Catalyst A is more obvious.

Table 2 Elemental analysis of coked catalysts(‰)

Table 3 The carbon content of coked catalysts(‰)

Coke formed on the catalyst was analyzed by two methods,i.e.FT-IR and XPS characterizations.Fig.1 represents the spectra of infrared framework transmission of equilibrium catalyst(e-cat)and coked catalysts.Band at1639 cm-1is typical of aromatic cycles and the increase in the band intensity with the carbon content is clearly visible.On all catalysts,bands at 3400 and 1080 cm-1are observed,which are respectively related with the intrinsic-OH structure and the matrix of catalyst.Compared with equilibrium catalyst,new bands at 2923 and 2854 cm-1appear for coked catalysts,and these bands are due to the existence of-CH3or-CH2.It is obvious that coke formed on Catalyst B owns the maximum proportion of alkyl side-chain among three coked catalysts,which suggests a relatively low degree of polymerization.

Fig.1.FT-IR framework transmission spectra of catalysts.

Fig.2 shows the decrease in the intensity of bands corresponding to-OH(3400-3800 cm-1)and acid sites(Br?nsted and Lewis acid sites).Clearly,a very strong consumption of active-OH groups occurs when coke is produced on the catalysts.It is noticeable that the band at 3743 cm-1,characteristic of Si-OH group in defects and on the external surface[27],decreases greatly for all coked catalysts.The decreased amount of silanol group is normally due to the formation of bulkier coke molecules that grow out to the external surface.For this reason,it suggests that the coke molecules formed on Catalyst A have larger molecular weight since the loss of Si-OH group is more serious.

Fig.2.FT-IR diffuse reflection spectra of catalysts.

The amounts of Br?nsted or Lewis acid sites are compared by integral method of the corresponding bands.Both of the two types of acid sites decrease in density.For coked Catalyst A,it suffers a more severe loss of Lewis acid sites.However,the consumption of Br?nsted acid sites on Catalyst A is the least among the three coked catalysts.Actually,it is still not completely clear which acid site plays a role in the coke formation,the Br?nsted acid site,the Lewis acid site or both[29-32].And the explanation about the change of acid site density is beyond our scope.

In order to differentiate and quantify carbon or nitrogen species in the coke,XPS spectra in the region of C1s or N1s BE were recorded for coked catalysts.As is shown in Fig.3,on Catalyst A,two types of carbon species can be differentiated by the de-convolution method described in the Experimental Section.The first peak at about 284.8 eV is representative of carbon atoms bonded to other carbon or hydrogen atoms(1st C in Fig.4).The second peak that represents specific carbon atoms bonded to nitrogen atoms(2nd C in Fig.4),appears at 286.9 eV.Both Catalysts B and C show two types of carbon species,but the ratio of each peak is different from Catalyst A.

In reality,the ratio of different carbon atoms can be determined through integration of each profile and the result is shown in Table 4.From the view of the molecular structure for quinoline,one nitrogen atom is bonded to two carbon atoms.On that assumption,the ratio of different carbon atoms due to their respective positions can be calculated through elemental analysis as well.For all coked catalysts,the XPS result is almost consistent with the calculating result from elemental analysis.This agreement illustrates that the XPS technology is a good tool to differentiate and quantify specific atoms.Moreover,the result also indicates that the N-heterocyclic structure of quinoline is not destroyed during coke formation.

Fig.3.Carbon types in FCC catalyst coke by XPS.

Fig.4.Carbon and nitrogen atoms in the coke molecule.

Table 4 The ratio of 1st carbon to 2nd carbon atoms

Fig.4 shows that nitrogen species in the coke also contain two types.The first peak centered at399.2 eVis representative of nitrogen atoms in a normal hydrocarbon environment(1st N in Fig.4).The second peak(2nd N in Fig.4)has a higher binding energy at about 401.7 eV.And the variation of binding energy can be ascribed to the interaction with the Br?nsted or Lewis acid site.The shift of electron density away from the nitrogen atom to the acid site leads to an increase in N1s binding energy.Under the present experimental system,the ratios of two types of nitrogen species do not differ much at all.However,this result is due to a large amount of quinoline added to o-xylene in the feed mixture because the XPS characterization has limited detection sensitivity.Therefore,the coked catalysts obtained from fixed bed tests have some discrepancies from industrial coked catalysts.Qian studying coke formation in the FCC process concluded that most nitrogen containing coke is formed in the earlier stages of cracking while hydrocarbons are the primary contributors to the coke yield in the later stages[33].This can be interpreted by the stronger adsorption ability of basic nitrogen compounds on acid sites in comparison with hydrocarbons.From an industrial point of view,the second type of nitrogen species interacting with acid sites should take a bigger percentage.In another word,the distribution of carbon and nitrogen atoms in the multi-layer structural coke is unbalanced.As to the three coked catalysts prepared in the experiment,coke molecule on Catalyst A is more in line with an industrial multi-layer structural coke model since its second nitrogen atom processes a bigger share(Fig.5).

Fig.5.Nitrogen types in FCC catalyst coke by XPS.

3.2.Coke regeneration

Partial oxidation of coke was completed in 1%O2/Ar atmosphere.Complete coke removal was observed within the temperature interval 350-825°C for all coked catalysts.Evolution of gaseous products except H2O for coked Catalyst A is shown in Fig.6.The main products were CO,CO2,N2,and NO and small amounts of NO2were also observed.Similar products were observed as well for Catalyst B and Catalyst C.

Fig.6.Evolution of gaseous products during TPO for Catalyst A.

As is shown in Fig.6,all gas products'ion currents appear at the temperature of350 to 450°C following the order:CO≈CO2＜N2＜NO≈NO2.The maximum in CO release is observed around 525°C.After that,N2and CO2reach the peak at about680 and 700°C respectively.NO is the last one to reach the peak,with the temperature being about 740°C.It is noticeable that in company with the decrease in CO,an increase in NO release is observed.One possible reaction route during regeneration is like Fig.7.When CO decreases,NO cannot be reduced to N2.Thus the amount of NO release increases.

Fig.7.Reduction of NO to N2.

Through the integration of ion current versus time,the amount of each gas product is obtained.The relative proportion of N2and NO+NO2is presented in Table 5.What is more,the distribution of nitrogen atoms in the multi-layer structural coke by XPS characterization is also presented.NOxtakes a minority of nitrogen-containing gas products since the oxygen concentration is relatively low.However,an interesting phenomenon is observed by comparing the distribution ofnitrogen atoms with the variation trend of nitrogen-containing gas products.From Catalysts A to C,with the decline in the proportion of second nitrogen species,the percentage of NO+NO2also shows a downward trend.This indicates that the second nitrogen species in the multi-layer structural coke has a closer relationship with NOxformation.

Table 5 Distribution of nitrogen atoms and nitrogen containing gas products

In order to identify the correlations between nitrogen species and different nitrogen-containing products,an overall analysis should be taken.In the analyzing process,two factors play important roles,i.e.unbalanced distribution of atoms and oxidation sequence.Compared with the outer layer structure of coke,the inner part has more nitrogen atoms and less carbon atoms.On the contrast,the outer layer is rich in carbon atoms(Fig.8(a)).Usually,oxidation of the multilayer structural coke starts from the outer edge.Thus,at the early stage of regeneration,CO concentration is relatively high and NOxcould be reduced to N2immediately.As the oxidation process goes on,the inner part of coke begins to be exposed to O2(Fig.8(b)).At this stage,with the decrease in CO concentration,NO retains for lack of reducing agent.In a word,the unbalanced distribution of carbon and nitrogen atoms in the multilayer structural coke leads the nitrogen atoms at the inner part to contribute more to NOxformation.

Based on the above analysis,the multi-layer structure of coke molecule could be applied into control NOxemission.At first,a new generation of additives which selectively catalyze the reaction between CO and NO should be developed.Then suitable reaction conditions including the temperature and oxygen pressure,which do not only fit for this specific reaction but also satisfy the regeneration process,should be studied.Apart from catalytic approach,regenerator design and operation can be used as well.Since the 1970s,two-stage regeneration technology has been explored and put into practice in many Chinese refineries.And the mature staked two-stage regeneration technology,which has the advantages of less air consumption,simple flue structure,and easy for pressure controlling,could be used as part of the strategy to control NOxemission[34,35].According to different oxidation sequences of the multilayer structural coke,one can divide the regeneration process into two parts and modify reaction conditions for a specific situation.Especially,in the second regeneration stage which retains more potential nitrogen species producing NO,proper conditions for the reaction between CO and NO,could be adapted to fulfill the function of catalytic additives.By combining catalytic approach and regeneration operation,we hope to control NOxemission effectively.At present,relevant work is being done to prove the practicability of this strategy.

4.Conclusions

By using model compounds and changing the feeding order of reactants,it is possible to obtain different multilayer structural coke with higher nitrogen content,which is beneficial to XPS characterization.All coked catalysts present two types of carbon species.This is due to different chemical environments where carbon atoms are located.Two types of nitrogen species were also observed and the type with a higher binding energy is ascribed to the inner part nitrogen atoms interacting with acid sites.Due to the stronger adsorption ability of basic nitrogen compounds on acid sites in comparison with hydrocarbons,the multilayer structural coke has unbalanced distribution of carbon and nitrogen atoms between the inner part and the outer edge,which means that the inner part has more nitrogen atoms and the outer edge is rich in carbon atoms.During the regeneration process,the oxidation of coke can be divided into two parts according to different oxidation sequences.At the later stage,the inner part of coke is exposed to O2.At this period,the formation of CO decreases due to lack of carbon atoms,which does not facilitate the reduction of NO to N2.Therefore,nitrogen species in the inner part of multilayer structural coke contributes more to NOxformation.

At last,based on the multilayer structure model of coke molecule,a possible strategy to control NOxemission by combining catalytic approach and regeneration operation is discussed.However,further work is necessary to prove it.

Fig.8.Regeneration process of the multi-layer structural coke:(a)early stage(b)later stage.

[1]G.Caeiro,A.F.Costa,H.S.Cerqueira,Nitrogen poisoning effect on the catalytic cracking of gas oil,Appl.Catal.A Gen.320(2007)8-15.

[2]G.Wang,Y.Liu,X.Wang,C.Xu,J.Gao,Studies on the catalytic cracking performance of coker gas oil,Energy Fuel 23(2009)1942-1949.

[3]Q.Shi,C.Xu,S.Zhao,K.H.Chung,Y.Zhang,W.Gao,Characterization of basic nitrogen species in coker gas oils by positive-ion electrospray ionization Fourier transform ion cyclotron resonance mass spectrometry,Energy Fuel 24(2009)563-569.

[4]G.Wang,Z.K.Li,H.Huang,X.Lan,C.M.Xu,J.S.Gao,Synergistic process for coker gas oil and heavy cycle oil conversion for maximum light production,Ind.Eng.Chem.Res.49(2010)11260-11268.

[5]C.M.Fu,A.M.Schaffer,Effect of nitrogen compounds on cracking catalysts,Ind.Eng.Chem.Prod.Res.Dev.24(1985)68-75.

[6]J.Scherzer,D.P.McArthur,Catalytic cracking of high-nitrogen petroleum feed stocks:effect of catalyst composition and properties,Ind.Eng.Chem.Res.27(1988)1571-1576.

[7]A.Corma,V.Fornes,J.B.Monton,A.V.Orchilles,Catalytic cracking of alkanes on large pore,high SiO2/Al2O3zeolites in the presence of basic nitrogen compounds.Influence of catalyst structure and composition in the activity and selectivity,Ind.Eng.Chem.Res.26(1987)882-886.

[8]T.C.Ho,A.R.Katritzky,S.J.Cato,Effect of nitrogen compounds on cracking catalysts,Ind.Eng.Chem.Res.31(1992)1589-1597.

[9]R.Hughes,G.Hutchings,C.Koon,B.McGhee,C.Snape,A fundamental study of the deactivation of FCC catalysts:a comparison of quinoline and phenanthrene as catalysts poisons,Stud.Surf.Sci.Catal.88(1994)377-384.

[10]Z.k.Li,G.Wang,Q.Shi,C.m.Xu,J.s.Gao,Retardation effect of basic nitrogen compounds on hydrocarbons catalytic cracking in coker gas oil and their structural identification,Ind.Eng.Chem.Res.50(2011)4123-4132.

[11]J.O.Barth,A.Jentys,J.Lercher,On the nature of nitrogen-containing carbonaceous deposits on coked fluid catalytic cracking catalysts,Ind.Eng.Chem.Res.43(2004)2368-2375.

[12]Z.K.Li,J.S.Gao,G.Wang,Q.Shi,C.M.Xu,Influence of nonbasic nitrogen compounds and condensed aromatics on coker gas oil catalytic cracking and their characterization,Ind.Eng.Chem.Res.50(2011)9415-9424.

[13]X.Zhao,A.Peters,G.Weatherbee,Nitrogen chemistry and NOXcontrol in a fluid catalytic cracking regenerator,Ind.Eng.Chem.Res.36(1997)4535-4542.

[14]J.O.Barth,A.Jentys,J.Lercher,Elementary reactions and intermediate species formed during the oxidative regeneration of spent fluid catalytic cracking catalysts,Ind.Eng.Chem.Res.43(12)(2004)3097-3104.

[15]R.Bassilakis,Y.Zhao,P.R.Soloman,M.A.Serio,Sulfur and nitrogen evolution in the argonne coals:experiments and modeling,Energy Fuel 7(1993)710-720.

[16]M.A.Wojtowicz,J.R.Pels,J.A.Moulijn,Combustion of coal as a source of N2O emission,Fuel Process.Technol.34(1993)1-71.

[17]E.Furimsky,A.Siukola,A.Turenne,Effect of temperature and O2concentration on N-containing emissions during oxidative regeneration of hydroprocessing catalysts,Ind.Eng.Chem.Res.35(1996)4406-4411.

[18]J.P.Hamalainen,M.J.Aho,Effect of fuel composition on the conversion of volatile solid fuel-N to N2O and NO,Fuel 74(12)(1994)1922-1924.

[19]T.P.Kobylinski,B.W.Taylor,The catalytic chemistry of nitric oxide.I.The effect of water on the reduction of nitric oxide over supported chromium and iron oxides,J.Catal.31(450-458)(1973).

[20]T.P.Kobylinski,B.W.Taylor,The catalytic chemistry of nitric oxide.II.Reduction of nitric oxide over noble metal catalysts,J.Catal.33(1974)376-384.

[21]X.Chu,L.D.Schmidt,Intrinsic rates of NO x-carbon reactions,Ind.Eng.Chem.Res.32(1993)1359-1366.

[22]A.W.Peters,G.D.Weatherbee,X.Zhao,Origin of NO x in the FCCU regenerator,Fuel Reformulation 5(3)(1995)45-50.

[23]A.W.Peters,M.Koranne,C.Pereira,G.D.Weatherbee,X.Zhao,J.Haley,S.Davey,B.Lakhanpal,Control and origin of NOx in the FCCU regenerator,ACS Annual Meeting,Orlando,FL,1996.

[24]R.Mann,Fluid catalytic cracking:some recent developments in catalyst particle design and unit hardware,Catal.Today 18(1993)509-528.

[25]R.B.Miller,T.E.Johnson,C.R.Santner,A.A.Avidan,J.H.Beech,Comparison between single and two-stage FCC regenerators,Proceedings of the 1996 Annual NPRA Meeting,San Antonio,Texas,1996.

[26]S.Lin,K.Wang,G.Jia,J.Shi,J.Zhang,G.Yang,The kinetics of coke deposition and regeneration of catalysts,Acta Pet.Sin.S1(1982)93-102.

[27]G.Caeiro,J.M.Lopesa,P.Magnoux,A FT-IR study of deactivation phenomena during methylcyclohexane transformation on H-USY zeolites:nitrogen poisoning,coke formation,and acidity-activity correlations,J.Catal.249(2007)234-243.

[28]H.S.Cerqueira,P.Ayrault,J.Datka,m-Xylene transformation over a USHY zeolite at 523 and 723 K:influence of coke deposits on activity,acidity,and porosity,J.Catal.196(1)(2000)149-157.

[29]J.Abbot,Role of Br?nsted and Lewis acid sites during cracking reactions of alkane,Appl.Catal.47(1)(1989)33-44.

[30]J.Abbot,F.Guerzoni,Roles of Br?nsted and Lewis sites during cracking of n-octane on H-mordenite,Appl.Catal.A Gen.85(2)(1992)173-188.

[31]S.M.Babitz,M.A.Kuehne,H.H.Kung,J.T.Miller,Role of Lewis acidity in the deactivation of USY zeolites during 2-methylpentane cracking,Ind.Eng.Chem.Res.36(8)(1997)3027-3031.

[32]A.Vimont,F.Thibault-Starzyk,J.Lavalley,Structural Lewis sites in zeolite beta-role on coking of the catalyst,Stud.Surf.Sci.Catal.130(2000)2963-2968.

[33]K.Qian,D.C.Tomczak,E.F.Rakiewicz,Coke formation in the fluid catalytic cracking process by combined analytical techniques,Energy Fuel 11(1997)596-601.

[34]J.Huang,Application of FCC 2-stage regeneration in China,Pet.Process.Petrochem.35(4)(2004)34-40.

[35]S.Yan,Staked two-stage regeneration technology in heavy oil FCCU,Pet.Process.Petrochem.33(8)(2002)7-9.