Math & Engineering

The structure-activity relationship of Fe nanoparticles in CO adsorption and dissociation by reactive molecular dynamics simulations

Description
The structure-activity relationship of Fe nanoparticles in CO adsorption and dissociation by reactive molecular dynamics simulations
Published
of 11
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  The structure–activity relationship of Fe nanoparticles in CO adsorptionand dissociation by reactive molecular dynamics simulations Kuan Lu a,b,c , Chun-Fang Huo b, ⇑ , Yurong He a,b,c , Wen-Ping Guo b , Qing Peng d, ⇑ , Yong Yang a,b ,Yong-Wang Li a,b , Xiao-Dong Wen a,b, ⇑ a State Key Laboratory of Coal Conversion, Institute of Coal Chemistry, Chinese Academy of Sciences, Taiyuan, Shanxi 030001, People’s Republic of China b National Energy Center for Coal to Clean Fuels, Synfuels China Co., Huairou District, Beijing 101400, People’s Republic of China c University of Chinese Academy of Sciences, No. 19A Yuquan Road, Beijing 100049, People’s Republic of China d Nuclear Engineering and Radiological Sciences, University of Michigan, Ann Arbor, MI 48109, United States a r t i c l e i n f o  Article history: Received 14 January 2019Revised 8 April 2019Accepted 8 April 2019Available online 9 May 2019 Keywords: Structure-activity relationshipNanoparticleFe, CO dissociationReaxFF a b s t r a c t Thestructure–activityrelationshipiscrucialincatalyticperformanceandmaterialdesignbutstill largelyobscure due to the complexity of heterogeneous catalytic systems. CO activation occurs widely inFischer–Tropsch reactions and pyrometallurgy, and it is a key to understanding carburization. Here, weinvestigate the structure–activity relationship in Fe nanoparticles by reactive molecular dynamics simu-lations. We focus on two activities, the adsorption and dissociation of CO, and four structural character-istics, morphologies, sizes, defects, and heteroatoms. The results show that CO adsorption anddissociation varies with the change of nanoparticles. Line dislocation and vacancies can strikingly boostCOdissociation, suggesting aneffective way totune the COdissociation rate. Further analysis shows thatthe Eley–Rideal mechanism possibly works in the early periods, followed by the Langmuir–Hinshelwoodmechanism in the later periods for CO 2  formation. Our results shed light on the mechanism and possibleoptimization of the carburization of iron.   2019 Published by Elsevier Inc. 1. Introduction Despite great progress in experimental technologies and meth-ods including transmission electron microscopy (TEM), scanningelectron microscopy (SEM), atomic force microscopy (AFM), andatom-probe tomography (APT), the understanding of the struc-ture–activity relationship in heterogeneous catalytic systems isstill an extraordinary challenge. Generally, catalysts experiencechanges in morphology and sizes, as well as the formation of numerous dislocations in reaction processes. These changes signif-icantly affect their catalytic performance by altering the facets,exposedarea, andconfigurationsof the surface, whichare stronglyrelated to the active sites and their number. However, thesechangesoccur at averysmall scale(atomisticscale) andextremelyfast, in picoseconds or even femtoseconds, posing huge difficultiesin experimental characterization, as well as computationalinvestigation.Morphology changes such as relaxation and reconstruction areinevitable, which leads to instantly varying, complex, and hard-to-control processes. The morphology changes of catalysts havethree main causes: chemical potential, chemical reaction paths,and phase transformation. Yoshida et al. [1] reported the surfacerelaxationofAunanoparticlesonaCeO 2  substrateinaCO/airenvi-ronment, compared with its surface under vacuum during CO oxi-dation. Hansen et al. [2] obtained the dynamic shape responses of Cu nanoparticles on a ZnO substrate to H 2 , CO, and H 2 O. Whenexposed to a mixture of H 2 O and H 2 , Cu nanoparticles tend totransform into a round shape, opposed to a more disclike shapein a mixed atmosphere of CO and H 2 . An in situ study [3] showedthat CO dissociation at high temperatures can induce the recon-struction of Co (0001). The strong metal–support interaction alsohas animportant effect. Zhanget al. [4] reporteda shapechangeof PdnanoparticlesinducedbytheformationofaTiO  x  overlayerfromthe reductionof TiO 2 . Duanet al. [5] reportedthat water vapor canpull supported Cu nanoparticles up onto the ZnO (0001  ) surface,which may lead to new active sites at the perimeter interfacebetween nanoparticles and support. These consequent changes inthe morphology of nanoparticles may play a critical role in cat-alytic performance. Vendelbo et al. [6] reported that spherical Pt https://doi.org/10.1016/j.jcat.2019.04.0100021-9517/   2019 Published by Elsevier Inc. ⇑ Corresponding authors at: State Key Laboratory of Coal Conversion, Institute of Coal Chemistry, Chinese Academy of Sciences, Taiyuan, Shanxi 030001, People’sRepublic of China. E-mail addresses:  huochunfang@synfuelschina.com.cn (C.-F. Huo), qpeng@umich.edu (Q. Peng), wxd@sxicc.ac.cn (X.-D. Wen).  Journal of Catalysis 374 (2019) 150–160 Contents lists available at ScienceDirect  Journal of Catalysis journal homepage: www.elsevier.com/locate/jcat  nanoparticles transform into a faceted shape when CO conversionincreases in Pt-catalyzed CO oxidation.Besides the effect of morphology, the size of nanoparticles alsohas a significant effect on catalytic performance. Nanoparticleshave a large variety of sizes, which keep changing during instanta-neous reactions. For example, Ostwald ripening drives particles tobe larger by a sintering process to lower the surface energy.McDonald et al. [7] studied the influence of the iron particle sizeinanFe/MgOcatalyst.Theirresultsshowedapreferenceofsmalleriron particles to form the  e 0 -Fe 2.2 C phase over the  v -Fe 5 C 2  phasedue to the greater extent of carburization and lower hydrogenchemisorption potential. Guczi et al. [8] observedthat smaller ironparticles are less susceptible to deactivation.It is worth mentioning that experimental samples contain var-ious defects, from line defects including edge dislocations [9,10]and screw dislocations [9–11] to two-dimensional defects includ-ing grain boundaries between polycrystals [9,12]. These defectshaveacriticaleffectonthephysical,mechanical,andchemicalper-formance [13–15] of materials, which, however, have always beenignored because of the limitations of experimental approaches[15,16].Thevacanciesgenerallycantraptheatoms,suchascarbon,due to their strong binding energy [17,18]. Surface vacancies pro-foundly influence the carburization of iron [19].Although in situ methods can observe the changes of catalystsand characterize their performance in real time, the results areplausible, as the great pressure gap between in situ and industrialenvironments causes a substantial gap in chemical potential in thevicinityofthecatalyst,resultinginacorrespondingchangeofmor-phology and catalytic performance. Moreover, the space–time res-olution of the in situ characterization method is at thenanometer/nanosecond level. Nevertheless, the bond breaking orformation of the reaction is at the femtosecond level. The reactiveforcefield(ReaxFF)maybethebestchoicetobridgethegaps,sinceit can simulate the chemical reaction by forming and breaking thebondsinclosetorealconditions. Inaddition,itallowsdirectobser-vation of the structure–activity relationship of the catalyst andinference of the reaction mechanism. Despite extensive attention[20–22], the structure–activity relationship of catalysts is still notwell established.Carbon monoxide is an industrial gas that has extensive appli-cations in bulk chemicals manufacture. Owing to its strong reduc-tion, it has been used in pyrometallurgy since ancient times toreduce metals from ore. Syngas, a mixture of CO and H 2 , is con-verted catalytically into a wide spectrum of hydrocarbons in Fis-cher–Tropsch synthesis (FTS) [7,23,24]. In a related reaction, thehydrogenation of carbon monoxide is coupled with C–C bond for-mation. CO activation mechanisms on surfaces of solid catalystsare critical for understanding the above processes.Carburizationof ironanddepositionof carbonaretwoessentialprocesses in both the Fischer–Tropsch process and blast furnacesteelmaking,whereironcarbides[25,26]areregardedastheactivephases and as useful phases [27,28] to enhance the strength of steel, respectively. Ding et al. [29] investigated CO adsorption ondifferent iron phases under Fischer–Tropsch conditions, showingthat Fe 0 (neutral) species canbe carburized completelyto ironcar-bides quickly after CO adsorption. The ‘‘competition model” [24]proposes that most of the carbon atoms are consumed by carbonpenetration into the bulk in the early stage of the synthesis, withthe assumption that the CO dissociation is relatively slow. Mat-sumoto and Bennett [23] prepared a catalyst consisting of a cleaniron surface with an underlying bulk carbide structure. The resultsshow that the surface iron can be carburized quickly, which isagreement with both the carbide model [30] and the competitionmodel. At the same time, the surface of the active catalyst coveredmostlyby a carbonintermediatehas muchhigher Fischer–Tropschactivity than that of carbon added to the catalyst by CO carburiza-tion. Apart from its own complexity, the sintering [31] and oxida-tion [32] of active phases and so on may also contribute to thedifficulty of research, which results in ambiguous understandingof carburization. Therefore, we choose CO activation and oxidationas a model reaction to investigate the structure–activity relation-ship in Fe nanoparticles.As part of the effort, we have recently reparametrized the Fe–C[33] and Fe/C/O [34] interatomic reactive force fields (ReaxFF) and demonstrated their applications in our previous work. Using thenewly developed potential, here we examine CO adsorption anddissociation on the different nanoparticles, and then analyze thecorrespondingCOconversion,aswellasthebehaviorofadsorptionand dissociation. Then the structure–activity relationship is exam-ined by mapping analysis. Last, we present the mechanism of CO 2 formation in the simulation process. 2. Methods  2.1. The ReaxFF method The ReaxFF force field was srcinally developed by van Duin[35] and to help bridge the gap between quantum-mechanical(QM) methods and experiment. Similarly to the traditional poten-tials, ReaxFFis anempirical potential, but abond-order-dependentforcefield withinstantaneousconnectivityfor the chemical bonds,depending on the local atomic environment. ReaxFF employs adistance-corrected Morse potential for the van der Waals energyto properly describe the short-range interactions. The use of theelectronegativity equilibration method (EEM) with shieldingmakes it possible to calculate the charge distribution updatedevery iteration during the ReaxFF molecular dynamics (MD) simu-lation that determines the geometry of the material. Thus, ReaxFFis reliable for examining the reaction process and describing theenergetics of various reaction intermediates. For the details of the energy formulas, the reader is referred to Refs. [35,36].  2.2. ReaxFF MD simulations ReaxFF MD simulations were performed to investigate thestructure–activity relationship of CO adsorption and dissociationon the  a -Fe nanoparticles. All MD simulations were performedby LAMMPS [37] and the structures were identified by the OVITO[38]. The ReaxFF parameters srcinate from Ref. [34]. The three- dimensional periodic boundary condition was used in10.7   10.7   10.7nm box, which consists of 500 CO molecules Fig. 1.  The reaction model consists of 500 CO molecules and an  a -Fe nanoparticle. K. Lu et al./Journal of Catalysis 374 (2019) 150–160  151  and an  a -Fe nanoparticle (Fig. 1). The crystal parameter of   a -Fe is2.84Å. All the atomic structures were initially optimized throughthe conjugate gradient method to a thermal structures, followedby canonical ensemble (NVT) at  T   =300K for 100ps for thermody-namics equilibrium. These well-equilibrated configurations arethen subjected to further study at 800K. The simulation timewas 300ps with a 0.25fs time step for every reaction. The simula-tions employed the velocity Verlet integrator. 3. Results  3.1. CO conversion rate To systematically present the effect of nanoparticle structuresonCOadsorptionanddissociation,weclassifyourmodelsintofourgroups to include the effects of morphologies, sizes, defects, andheteroatoms. For morphology(Fig. 2a), weconstructfour differentconfigurations, including the quasi sphere consisting of (100),(110), and (111) surfaces, the cubic structure consisting of pure(100) surface (11745 Fe atoms), the rhombus consisting of pure(110) surface (7095 Fe atoms), and the octahedron (octa) consist-ing of pure (111) surface (11431 Fe atoms), which have a similarspecific surface area of about 0.12Å  1 (the ratio of surface area tovolume). The surface energy is consistent with the reference value[39]. For size (Fig. 2b), we include three quasi spheres with differ- ent diameters, which are about 2nm (1641 Fe atoms), 5nm (6520Fe atoms), and 7nm (19409 Fe atoms), respectively. For defects(Fig. 2c), wepreparedfourtypesof defects:surfacevacancies, edgedislocations, screw dislocations, and grain boundaries. Similarly, italso has the same specific surface area, around 0.12Å  1 . The con-centration of surface vacancies, namely the quantitative ratio of vacant sites to total surface atoms, is about 0.09. The other defectsofthespherecontainoneedge,onescrew,andthreegrains,respec-tively. The details of the construction of these defects werereported in our previous work [15]. For hetero atoms (Fig. 2d), about 0.5 monolayer (ML) of surface carbon or oxygen atoms cov-ered the surface, where one monolayer stands for the number of total surface atoms.When the reaction reaches 300ps, we count the CO conversionofdifferentFenanoparticles,asshowninFig.3.The7-nmnanopar-ticle has the highest CO conversion (99%), indicating that 300ps isadequate for all CO reactions with Fe nanoparticles. With increas-ingsizeandexposedsurface,theCOconversionincreases.Thecon-version of CO varies with the change of morphologies. The 5-nmquasi sphere has the lowest CO conversion (85%). The cubic andoctahedral nanoparticles have CO conversion up to 98% and 97%,respectively. The CO conversion of the rhombus is 88%. The occur-rence of defects on the catalyst promotes CO conversion in com-parison with the 5-nm quasi sphere, except that grain boundary.As expected, the adsorption of carbon or oxygen reduces CO con-version by blocking the active sites.  3.2. CO adsorption and dissociation 3.2.1. Site of adsorption The interaction of CO on iron surfaces is of great importance inunderstandingthe initial steps iniron-basedFTS, since the adsorp-tion and dissociation of CO are believed to be essential for CH x  for-mation [40]. CO has various adsorption configurations, includingsingle adsorption, twin adsorption (Twin), and multiple adsorp-tion. As shown in Fig. 4, one CO adsorbs on fourfold sites (4-Fold)and twofold sites (Bridge-100) of Fe (100) surfaces, which is Fig. 2.  The setup of four groups of models: (a) morphologies with four configurations, (b) sizes with two configurations, (c) defects with four configurations, and (d) heteroatom with two configurations. Fig. 3.  The CO conversion of different Fe nanoparticles at 300ps. The values in thecolumnsstandfortheconversionfactorofareasbasedon5-nmquasispheresforallnanoparticles.152  K. Lu et al./Journal of Catalysis 374 (2019) 150–160  consistent with experimental observations [41,42] and theoreticalcalculated configurations [43,44], respectively. CO simultaneouslyadsorbs and dissociates on the surface (3-Fold and Dissociation),in agreement with previous experiments [45]. Multiple CO coad-sorption on top sites (Top) and two-CO adsorption on bridge sites(Bridge-110) of Fe (110) are also observed in the reaction process,agreeing well with experiment [46] and DFT investigations [47]. Owing to the lateral interactions, there is a certain separationbetween adsorbed CO molecules. It is worth noting that twinadsorption on undercoordinated atoms may be the precursor of  Fig. 4.  Snapshots of different adsorption sites observed in the reaction process. Color scheme: iron atoms (blue), oxygen atoms (red), and carbon atoms (gray). Fig. 5.  The adsorption isothermof CO for four groups of Fe nanoparticles. The inset graph in b is the enlarged figure of corresponding data. The ‘‘Ads.” stands for ‘‘adsorbed.” K. Lu et al./Journal of Catalysis 374 (2019) 150–160  153  carbonyl iron [48], which may take away the iron atom and causethe loss of iron catalyst.  3.2.2. Different adsorption behavior of CO Fig. 5 shows the evolutionof CO adsorption on the nanoparticlesurfaceinthereactionprocess. Toaccountthedifferenceinsurfaceareas of different nanoparticles, we normalize the number of adsorbedstateCOofaconfiguration,referringtothenumberbasedonthequasi-spheresurfaceareaof5nmNP. WeonlycounttheCOin the form of molecular adsorption. The normalization or conver-sion factor of area is listed in Fig. 3. The same conversion had alsobeen used in analysis of molecular number. All curves showa gen-eral trend that increases at the beginning followed by gradualincrements (Fig. 5). This behavior can be attributed to the reduc-tion of gas pressure. The numbers of adsorbed state CO on bothrhombus and quasi sphere nanoparticles increase with the lapseof time; the former is larger, as shown in Fig. 5a. Fast adsorptiononthecubicandoctahedralnanoparticleswasobservedinthefirst50ps and then it reached equilibrium.The surfaceenergy of nanoparticles decreases withthe increaseof nanoparticle sizes with decreasing surface–volume ratio. As aconsequence, the average coordination number (CN) may increasein correspondence to structure change. As shown in Fig. 5b, thenumber of adsorbed state CO decreases with their increasing size.Thedefectshaveasmall effect onCOadsorptionexceptat grainboundaries (Fig. 5c). The number of adsorbed state CO on thenanoparticles with grain boundaries is the minimum. Surprisingly,thenanoparticles withoxygenatomsadsorbedonthesurfacehavethe largest number of adsorbed state CO, followed by the surfaceswith carbon atoms adsorbed. The reason might be that theadsorbed atoms block the surface sites, making it more difficultto dissociate CO on the surface and having more molecular COadsorption.  3.2.3. Different dissociation behavior of CO We analyzed the CO dissociation behavior in the reaction pro-cess. Though all morphologies of nanoparticles show the sameapproximate CO dissociated rate (Fig. 6a), a small difference hasbeen observed. The rhombus and octahedron have sharperdecreasing trends than the 5-nm quasi sphere, which leads to thenumber of CO dissociations having a relatively larger differenceat final state, as shown in Fig. 7a. Similarly to the CO adsorption,the quasi sphere and the rhombus have a relatively larger dissoci-ation number than the cubic and octahedral forms due to the dif-ference in surface-active sites. The quasi sphere has the largestCO dissociation number, followed by the rhombus. This can beattributed to a highly active Fe (100) surface for CO activationand a high carbon penetration rate of Fe (110) surface than othersurfaces [34]. Furthermore, referring to the rhombus and the octa-hedron, the cubic has the smallest proportion (11.5% vs. 17.5% and18.9%) of highly active edge and corner sites in the initial period,which explains why the cubic has the lowest CO dissociation rate.The proportion here was defined the ratio of the number of edgeandcorneratomstothenumberofsurfaceatoms.Thoughtheocta-hedronpossessesthelargestproportionofhighlyactivesites,ithasthe most dramatic surface reconstruction simultaneously, due to Fig. 6.  The dissociation rate curves of CO for four different groups of Fe nanoparticles. The inset graph in a and c is the dissociation rate curves of CO in 300ps.154  K. Lu et al./Journal of Catalysis 374 (2019) 150–160
Search
Similar documents
View more...
Tags
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks
SAVE OUR EARTH

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!

x