Experimental and theoretical assessment of the mechanism of hydrogen transfer in alkane-alkene ...

12 Pages·2017·2.01 MB·English

by  Michele L. Sarazen

JournalofCatalysis354(2017)287–298
ContentslistsavailableatScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Experimental and theoretical assessment of the mechanism of hydrogen
transfer in alkane-alkene coupling on solid acids
⇑
Michele L. Sarazen, Enrique Iglesia
DepartmentofChemicalandBiomolecularEngineering,UniversityofCalifornia,Berkeley,Berkeley,CA94720,UnitedStates
a r t i c l e i n f o a b s t r a c t
Articlehistory: ExperimentalandtheoreticalmethodsareusedtoprobethemechanismandsiterequirementsforC–C
Received14April2017 bondformationandhydridetransfer(HT)reactionsofalkane-alkenemixturesonsolidacidswithdiverse
Revised2August2017 acidstrengthandconfiningvoids.Suchmethodsprovidequantitativedescriptorsofreactivityintermsof
Accepted6August2017
thepropertiesofmoleculesandsolidsforchemistriesthatenablethepracticeofalkylationandoligomer-
Availableonline19September2017
izationcatalysis.Intheseprocesses,chaingrowthiscontrolledbyHTfromalkanestoalkene-derived
boundalkoxidesformedviaoligomerizationorb-scission.Transitionstate(TS)treatmentsoftheelemen-
Keywords:
tarystepsthatmediatethesereactionsshowthatHTratesdependontheenergiesrequiredtodesorb
Alkeneoligomerization
alkoxidesascarbeniumionsandtocleavetheweakestC–Hbondingaseousalkanes.Theseenergiesserve
Hydridetransfer
asaccuratemoleculardescriptorsofhydridetransferreactivityand,takentogetherwiththeacidstrength
Alkylation
Zeolites andvanderWaalsstabilizationpropertiesofcatalyticsolids,providethekineticdetailsrequiredtopre-
Confinement dicttherelativeratesatwhichalkoxidesreactwithalkenes(toformC–Cbondsandlargeralkenes)or
Acidstrength alkanes(toacceptH-atomsanddesorbasalkanes)forchainswithabroadrangeofsizeandskeletal
Heteropolyacids structure.ConfinementeffectsreflectthesizeoftheTSanditsprecursorsforC–CcouplingandHTrela-
tivetothedimensionsoftheconfiningvoids,whichdeterminehowguestspeciesformvanderWaals
contactswiththehostwithoutsignificantdistortions.Smallervoidspreferentiallystabilizethesmaller
C–C bond formation TS over the larger structures that mediate HT. Acid strength, in turn, influences
thestabilityofconjugateanionsattheion-pairTS:strongeracidsleadtohigherturnoverratesforC–C
coupling and HT, but to similar extents, because their TS structures contain fully-formed framework
anionsthatbenefitsimilarlyfromtheirmorestablecharacterinstrongeracids.
PublishedbyElsevierInc.
1.Introduction stable, larger alkanes as products [3,5,6,12–15]. Hydride transfer
can also lead to the formation of dienes and trienes from H-
The practice of catalytic alkylation upgrades small alkanes by donors; such species can act as precursors to cyclic and unsatu-
incorporating them into alkene oligomerization cycles through ratedmoleculesthatultimatelyformunreactiveresidues,thusren-
thetransferofhydrogentogrowingchains,typicallyusingsulfuric deringacidsitesinaccessibleinsolidacidcatalysts.
or hydrofluoric liquid acids [1,2]. Solid Brønsted acids contain TheseoligomerizationandH-transferreactions,aswellasmost
much lower densities of protons and exhibit faster deactivation other reactions catalyzed by protons, involve full ion-pair transi-
than these liquid acids [3–8], but avoid the environmental risks tion states, which vary in size among the different routes. These
of materials corrosion and toxicity upon release, characteristics size differences allow some control of reactivity and selectivity
that are ubiquitous in the case of liquid acids. The conversion of basedonthesizeofconfiningvoids,asaresultofthepreferential
alkenes to larger chains, useful as chemicals and fuels, involves van der Waals stabilization of transition states of a certain size
concurrent oligomerization, isomerization, b-scission, hydride [16–18]. Turnover rates for each given reaction also depend on
transfer and cyclization reactions [9–11]. Hydride transfer from thechemical stabilityof theanion and thecation in the relevant
alkanesto alkoxideintermediatesprovides strategies for regulat- ion-pairtransitionstate.Thestabilityoftheconjugateanionisdic-
ingchaingrowthviacontrolledterminationaslessreactivealka- tatedbythestrengthofthesolidacid,whichisgiven,inturn,by
nes, while consuming typically unreactive alkanes and forming thedeprotonationenergyoftheacid[19,20].Theorganiccationic
moiety at the transition state becomes more stable as its proton
affinityincreasesand,inthecaseofbimolecularhydridetransfer,
⇑
Correspondingauthor.
astheenergyrequiredtoabstractahydridefromthedonorspecies
E-mailaddress:[email protected](E.Iglesia).
http://dx.doi.org/10.1016/j.jcat.2017.08.002
0021-9517/PublishedbyElsevierInc.
288 M.L.Sarazen,E.Iglesia/JournalofCatalysis354(2017)287–298
(dehydridation energy) decreases. The identity, structure, and and from 2,6-di-tert-butylpyridine titration uptakes for meso-
energy of such cationic moieties are specific to the reactants porousacidsthatallowedthediffusionofsuchlargetitrants,which
involved; their properties become the molecular descriptors of selectively interact with protons without binding on Lewis acid
reactivityandselectivity,asthisstudyshowsthroughacombina- sites. Samples (15–50mg) were placed within a tubular reactor
tionoftheoryandexperimentforalkane-alkenereactionsonsolids with plug-flow hydrodynamics (316 stainless steel; 12mm I.D.)
thatdifferinacidstrengthandinthesizeandshapeoftheconfin- heldwithinathree-zoneresistively-heated furnace(AppliedTest
ingvoidstructure. SystemsSeries3210;Watlowcontrollers;96Series).Thetemper-
Alkene oligomerization turnover rates are limited by the ature was measured with a K-type thermocouple held within an
kinetically-relevantformationofaC–Cbondbetweenanalkoxide internalconcentricthermowellplacedinthemiddleofthecatalyst
andanunboundalkene[16].Thisstudyshowsthatthetransferofa bed.KegginPOMclusters(H-form)weretreatedinflowingHe(50
H-atomfromthealkanetoanalkoxidelimitsalkaneincorporation cm3g(cid:1)1s(cid:1)1;99.999%,Praxair)byheatingto503K(at0.083Ks(cid:1)1)
intoalkeneoligomerizationcycles,leadingtoratesproportionalto toremoveadsorbedspeciesbeforereactionmeasurements.Alumi-
thepressureoftheH-donoralkaneandtotheconcentrationofa nosilicatesweretreatedbeforecatalyticratemeasurementsina5%
given alkoxide. The formation rates of C alkanes, through H- O inHestream(83.3cm3g(cid:1)1s(cid:1)1,Praxair)byheatingto818K(at
6 2
transfer from alkanes to bound hexoxides formed via propene 0.025Ks(cid:1)1)andholdingfor3htoconverttheNH+toH+,andthen
4
dimerization,dependonthesizeandshapeoftheconfiningvoids. cooledto503K.n-Butane(99.9%,Praxair),isobutane(99.9%,Prax-
H-transfer transition state structures are significantly larger than air) or 2-methylbutane (99.5% HPLC grade, Sigma-Aldrich) were
thosefordimerization,leadingtoratiosofH-transfertodimeriza- mixedintoastreamofpropene(99.9%,Praxair,Praxair)inHe.Liq-
tionrateconstants(i.e.,selectivity)thatalsodependsensitivelyon uid2-methylbutanewasevaporatedintoaflowingHestream(UHP
thevoidsizeoftheporousaluminosilicates.Bothtransitionstate Praxair)usingahigh-pressuresyringepump(TeledyneIscoSeries
structures are, however, fully formed ion-pairs. Asa result, they D). The effluent was transferred through heated lines held above
benefittothesameextentfromstrongeracids;consequently,acid 373Kintoagaschromatograph(Agilent6890),whereconcentra-
strength does not influence the relative rates of oligomerization tionsweremeasuredusingflameionizationdetectionafter chro-
andH-transferforprotonsresidingeitherwithinvoidslargerthan matographic separation with a methyl silicone capillary column
thetransitionstatestructures(e.g.,Kegginpolyoxometalates dis- (AgilentHP-1column,50m(cid:3)0.32mm(cid:3)1.05lm film).Reactant
persedonSiO andmesoporousaluminosilicates)orwithinvoids pressureswerevariedbydilutionwithHe(99.999%,Praxair)ata
2
ofsimilarsize,whichprovideweakorsimilarvanderWaalscon- systempressureheldconstantbyadome-loadedregulator(Tem-
tacts,respectively. presco).Molecularspeciationwasconfirmedbytheuseofknown
MoleculardescriptorsofH-transferreactivitymustconsiderthe compounds and by mass spectrometry after chromatographic
stabilityofboththehydrideacceptors(alkoxides)andthehydride separation.
donors(alkanes)astheyrearrangethecationicchargeatthetran- Alkene conversion turnover rates (per proton)were measured
sitionstate;descriptorsmustalsoconsidertherelativesizeofthis intheabsenceco-fedalkanestodeterminealkeneoligomerization
transitionstateanditsrelevantprecursors.Thus,measuredkinetic rates[16].Hydridetransferturnoverratesweredeterminedfrom
rates for hydride transfer, which reflect energy differences the rate of formation of alkanes and reported either as the total
between the transition state and its alkoxide precursor, decrease hydridetransferrate(thecombinedformationratesforallisomers
asthealkoxidebecomessmaller.Theskeletalstructuresofbound withagivennumberofC-atoms)orastherateofformationofa
alkoxidesalsoinfluencetheirpropertiesasH-acceptors,asshown givenskeletalisomer,whichreflectstheH-acceptorpropertiesof
by the different rate constants for the formation of each skeletal the bound alkoxide with that specific skeleton. Selectivities are
hexaneisomerfromtheirrespectiveboundandequilibratedhex- defined as the ratio of H-transfer and oligomerization rate con-
oxideisomers.Similarly,theextentofsubstitutionatthehydride stantsandextractedfromkineticdatafordifferentacceptoralkox-
donorincreaseshydridetransferratesforagivenalkoxide,because idesandH-donoralkanes.
they lead to more stable cations at the bimolecular transition
states that mediate such reactions. The results, calculations, and 2.2.Densityfunctionaltheorymethods
concepts gleaned here through the use and benchmarking of
experiment and theory have allowed the first accurate, quantita- Periodic density functional theory, as implemented in the
tiveassessmentoftherelativereactivityoftherelevantmolecules Viennaabinitio simulation package (VASP)[24–27], was used to
inalkane-alkenereactions.Theseassessmentshaveledtomolecu- optimizestructuresandenergiesforstableintermediateandtran-
lar and catalyst descriptors, which taken together with previous sitionstatesinvolvedintheelementarystepsfordimerizationand
studies [9,16], allow their reliable extrapolation to related reac- hydride transfer. A periodic plane-wave basis-set expansion to a
tions and processes catalyzed by solid acids and relevant to the cutoffenergyof396eVwasusedtorepresentthewavefunctions
practical upgrading and conversion of fuels, energy carriers, and forvalenceelectronsandprojector-augmentedwave(PAW)pseu-
chemicalfeedstocks. dopotentials were used to account for electron-core interactions
[28,29]. Exchange and correlation energies were calculated using
the generalized gradient approximation and the revised Perdew–
2.Experimentalmethods Wang (PW91) functional [30]. A 1(cid:3)1(cid:3)1 Monkhorst–Pack k-
pointmeshwasusedtosamplethefirstBrilliounzone[31].
2.1.Materialsandratemeasurements Keggin POM clusters were described by placing full clusters
(1.1nm diameter) at the center of cubic unit cells (3nm edge
The properties and protocols used in the synthesis of the H+- length) in order to prevent electronic interactions among neigh-
form of BEA, MFI, MOR, TON, and FAU zeolites, mesoporous boringcells[19,32].Minimumenergypathswerecalculatedusing
silica-aluminaandKegginPOMclustersdispersedonmesoporous nudgedelasticband(NEB)methods[33]withstructuresconverged
colloidal silica [21] have been reported elsewhere [9,16] and in to1(cid:3)10(cid:1)4eVforenergiesand0.3eVÅ(cid:1)1forforces.NEB-derived
Table1;thesesampleswerepelleted,crushed,andsievedtoretain transition state structures were then refined using the Dimer
l
180–250 maggregatesbeforeuse.Thenumberofprotonsineach method [34] with more stringent convergence for energies and
samplewasdeterminedfromNH evolutionfromNH+-exchanged forces on each atom (10(cid:1)6 eV and 0.05eV Å(cid:1)1). The bridging
3 4
samplesforzeoliticacids(SupplementalInformation(SI);Fig.S1) O-atomsintheHPWPOMclusters(H PW O )wereusedasthe
3 12 40
M.L.Sarazen,E.Iglesia/JournalofCatalysis354(2017)287–298 289
Table1
Source,Si/Alratios,andprotoncountsforsolidacidsused.
Sample Source Si/Alratio1 H+/T*ratio
a BEA Zeolyst 11.8 0.392
BEA Zeolyst 43 1.02
MOR Zeolyst 10 0.862
MFI Zeolyst 43.8 1.02
TON [22,23] 24 0.402
LZ-210 Engelhard 7.5 0.372
Silica-Alumina Sigma-Aldrich 5.5 0.253
b Sample POMcontentonSilica(%wt.) POMsurfacedensity(POMnm(cid:1)2) Protons(H+/POM)3
H3PW12O40/SiO2 5 0.04 0.72
1 Elementalanalysis(ICP-OES).
2 NH3evolutionfromNH4+-exchangedsamples.
3 2,6-di-tert-Butylpyridinetitration.
initialprotonlocationforoptimizationofboundintermediatesand agesinequilibriumwiththecorrespondinggaseousalkeneviaan
transitionstates.Thechargeateachatominoptimizedstructures adsorptionconstantK.Mixturesofisomerswithdistinctpointsof
i
was determined by transforming converged wavefunctions into attachment at the surface give hKi values averaged over all such
i
localizedquasi-atomicorbitals(QUAMBO)[35–38]. bindingconfigurations(j)or,inthecaseofoligomer-derivedalkox-
ides,overallskeletalisomers,becauseoftheirformationasequili-
bratedmixtures[9,16].
3.Resultsanddiscussion
Thealkoxidescandesorbviadeprotonationtoformallskeletal
anddoublebondalkeneisomers.Thesealkoxidescanalsodesorb
3.1.Hydridetransferratesfromalkanestoalkoxidesduring
as the corresponding alkanes after hydride transfer (HT) when
oligomerizationturnovers
alkane co-reactants are present. This sequence of steps is illus-
trated for isobutane-propene reactant mixtures in Scheme 1. HT
Oligomerization turnovers on Brønsted acids involve the
events introduce chain termination routes that can be exploited
kinetically-relevantadditionofgaseousalkenestoboundalkoxides
tocontrolthenumberofoligomerizationeventsthatoccur,which
(Scheme 1) that are present at near saturation coverages during
woulddependotherwiseonlyontheintrinsicdimerizationreactiv-
catalysis,asshownbykinetic,spectroscopic,andtheoreticalstud-
ity and adsorption-desorption thermodynamics of each alkoxide
ies[16,39].Oligomerizationratesforanalkenewithicarbonatoms
structure.TheseHTprocessesalsoallowtheincorporationofunre-
aregivenby:
activealkanesintooligomerizationcyclesatmodesttemperatures
roligoi ¼kolighoK;i½Ci½iC(cid:4)½C(cid:4)ailk(cid:4)¼kolhigKo;iiK½Ci½C(cid:4)i(cid:4)2¼kohliKgo;iiKi½Ci(cid:4) ð1Þ (a4lk0o0x–i5d0e0,fKor).mHeTdfrreopmlactheseaallkkaennee-vdiaeriiovne-dpaailrkotrxaindseistiownitshtaatensotthheart
i i i i i mediatethetransferofaH-atom.Theprimaryandtertiaryisobu-
where k is the rate constant for the elementary reaction of a toxidesformed(inthecaseofisobutane)candesorbasisobutene,
oligo,i
gaseous alkene (Ci) with a bound alkoxide (Cialk) present at cover- but also react with propene to form C7 alkoxides that can either
Scheme1. Isobutane-propenereactionpathwaysonsolidBrønstedacids.Equilibratedskeletalandregioisomersareindicatedby[...].
290 M.L.Sarazen,E.Iglesia/JournalofCatalysis354(2017)287–298
deprotonatetoformgaseousC7alkenesoracceptahydridefrom keff ¼kHT;nijKij¼exp(cid:2)(cid:1)DG =RT(cid:3) ð4Þ
isobutane to form C alkanes with the skeletal structure of the HT;nij hK i nij
7 3
boundalkoxides.
Theturnoverrates(perH+)fortheformationofC6alkanesfrom DG ¼Gz (cid:1)GðgÞ(cid:1)GðgÞ(cid:1)ðhGalki(cid:1)GðgÞÞ ð5Þ
isobutane-propene mixtures are proportional to isobutane pres- nij HT;nij n ij 3 3
sureonallmicroporousandmesoporousacids(Fig.1;0–250kPa TheformationofC alkenesandalkanesandtheincreaseinthe
7
isobutane; 9kPa propene; 503K). Such rates are consistent with C /C productratioswithincreasingisobutanepressure(Fig.S2,SI)
7 6
a kinetically-relevant bimolecular step in which the isobutane indicatethatisobutane-derivedalkoxidescanalsoreactwithpro-
donortransfersaH-atomtothealkoxideacceptors(e.g.,C6alkox- pene in subsequent oligomerization steps. HT from isobutane to
idesformedbypropenedimerization)onsurfacesnearlysaturated the predominant bound propoxide species also occurs, albeit at
withpropene-derived alkoxidesas in the case of propenedimer- rateslowerthanthoseforlargeralkoxides;suchratesarealsopro-
ization reactions in the absence of H-donor alkanes [16]. The portionaltotheprevalentisobutanepressure(Fig.2;BEA(a)and
resultingrateequationforhydrogentransferis: HPW(b)).
r ¼kij½Con(cid:4)½Caijlk(cid:4)¼kijKij½Con(cid:4)½Cij(cid:4)¼kijKij ½Cij(cid:4)½Co(cid:4) ð2Þ bonCd–Cfobrmonadtiocnlea(vvaiageit(sb-mscicisrsoisocno)poicccruervsercsoen)cutorrefonrtmly walkitohxiCd–eCs
HTnij hK i½C (cid:4) hK i½C (cid:4) hK i ½C (cid:4) n
3 3 3 3 3 3 andalkeneswithintermediatecarbonnumbers(i.e.,notmultiples
wherek istherateconstantfortheelementaryHTreactionfrom of i, the number of carbons in the monomer) [9]; these smaller
ij
analkane(Co)withnC-atomstoanalkoxideisomerj(Calk)withi alkoxidescanalsoundergoHTtoformthecorrespondingalkanes.
n ij
C-atoms,occurringonprotonssaturatedwithalkene-derivedalkox- The linear dependence of the rates of formation of these alkanes
ides[16,39]thatareequilibratedwiththegas-phase(C )viaK .The (e.g. pentanes, Fig. 2; BEA (a) and HPW (b)) is also proportional
ij ij
totalrateofHTforapoolofalkoxideswithiC-atoms(r )is: toisobutanepressure,indicativeofacommonHTmechanismfor
X HTni alkoxide acceptors of different size. HT turnover rates, however,
rHTni ¼ rHTnij ð3Þ depend on the identity of both the alkoxide acceptor and the
j alkanedonorandalsoontheacidstrengthandconfinementprop-
ertiesofthesolidacids,asdiscussedinthesectionsthatfollow.
Themeasuredrateconstant(keff )representsacombinationof
HT;nij
the intrinsic HT rate constant and the thermodynamic constants
foralkoxideformationandreflects,inturn,afreeenergydifference 3.2.Effectsofconfinementandacidstrengthonhydridetransferrates
(DGn;ij)thatisgivenbythedifferencebetweenthetransitionstate andrateconstants
energy (Gz ) and those for the gaseous alkane (GðgÞ) and the
HT;n;ij n TheslopeoftheratedatainFig.1reflectsthecombinationof
alkoxideacceptor(GðgÞÞpresentonsurfacespredominantlycovered
ij kineticandthermodynamicconstantsandofthe[Ci]and[C3]con-
withpropene-derivedC3alkoxides: centrations, as they appear in Eq. (2). These slopes vary among
solidacids,evenamongaluminosilicateswithdifferentvoidstruc-
tures,butcontainingacidsitesofsimilarstrength[40].Theselinear
relations indicate that the concentrations of hexenes and their
equilibrated isomeric hexoxides are unaffected by the presence
or pressure of isobutane, which does not perturb the alkene-
alkoxide equilibrium or the dimerization rates. Also, C alkanes
6
represent a small fraction of the total C products formed, even
6
atthehighestisobutanepressure,becausehexoxidedepletionvia
HTismuchslowerthantheformationanddesorptionofhexenes
via dimerization events. Thus, the concentrations of hexenes and
propene remain unaffected by the isobutane pressure and repre-
sentthekineticdrivingforcesintheslopesextractedfromtherate
datain Fig. 1becausetheydetermine,in turn,thesurface cover-
ages of the corresponding alkoxides. The resulting kinetic con-
stants,afteraccountingforthealkenepressures,solelyreflectthe
kineticandthermodynamicconstantspresentinEq.(4).
These effectiveHT rate constantsfor isobutane reactionswith
the lumped hexoxide or propoxide isomer pools are shown in
Fig. 3 for hexoxides ðkeff Þ (Fig.3a) and propoxides ðkeff Þ
HT;4;6 HT;4;3
(Fig.3b) as a function of the void diameter in the microporous
and mesoporous aluminosilicate hosts. Such diameters, as used
in Fig. 3, have been defined using a purely geometric descriptor
consisting of the largest diameter sphere that either (i) can be
inscribed within the void space or (ii) can traverse throughout
the void structure unimpeded by smaller apertures [41,42]. The
HT rate constants for both hexoxide and propoxide acceptors
increasedasthevoidsizedecreasedandthenreachedamaximum
value. The initial increase in reactivity with decreasing void size
reflectsthepreferentialstabilizationofthelargerbimoleculartran-
sition state involved in these reactions over the smaller alkoxide
Fig.1. C6alkaneformationratesduringreactionsofisobutane-propenemixturesas precursorviavanderWaalsinteractionswiththelatticeoxygens
afunctionofisobutanepressureonaluminosilicateswithdifferentvoiddimensions
butsimilaracidstrengths(BEA(N),FAU(r),SiAl(4),MOR(d),MFI(}),TON(j)) inthealuminosilicateframeworks,asgivenbythechemicaliden-
[9kPapropene,503K,<5%alkeneandalkaneconversions]. tityoftherateandequilibriumconstantscontainedinEq.(4).
M.L.Sarazen,E.Iglesia/JournalofCatalysis354(2017)287–298 291
Fig.2. C6,C5,andC3alkaneformationratesfromhexoxides(N),pentoxides(d),andpropoxides(r),respectively,presentasamixtureofequilibratedisomersmeasured
duringreactionsofisobutane-propenemixturesasafunctionofisobutanepressureon(a)BEAand(b)HPW[9kPapropene,503K,<5%alkeneandalkaneconversion].
Fig.3. First-orderrateconstantforreactionofisobutanewith(a)hexoxidepooland(b)propoxidepoolasafunctionofvoiddiameter[503K,<5%alkeneandalkane
conversion].
The void size corresponding to the largest HT rate constants, isobutanetohexoxidescontainstenC-atoms;itsdiameterforthe
however, is slightly larger for hexoxides (BEA; Fig.3a) than for specific case of the tertiary-2-methylpentoxide on HPW clusters
propoxides(MOR;Fig.3b).Thisdifferencereflectsthesizeoftheir (Fig. 4; right) is 0.73nm. In contrast, there are seven C-atoms in
respectivetransitionstatestructuresandtheirconsequentlydiffer- thetransitionstateforHTtosec-propoxides,leadingtoasmaller
entabilitytoformvanderWaalscontactswiththevoidwalls.The mean diameter (0.66nm) for this transition state than for that
size of the DFT-derived structures for these transition states is involved in isobutane HT to hexoxide acceptors. Consequently,
defined here as the diameter of the sphere of equivalent volume thetransitionstateforHTtopropoxidesbecomesmoststableon
determined from a Connolly surface of the organic moiety opti- MORchannels(0.64nm);thesechannelsareslightlysmallerthan
mizedonHPWPOMclusters.Therelativesizesoftransitionstates the BEA channels (0.67nm) that exhibit the largest HT rate con-
donotvarysignificantlywhenoptimizedonsuchKegginclusters stantsforisobutanereactionswithhexoxides.
or in zeolites, as previously shown for propene dimerization on The matching of voids with transition state structures in size
HPWandTON[16].Thetransitionstateforhydridetransferfrom andshapeleadstooptimalvanderWaalscontacts,whileavoiding
292 M.L.Sarazen,E.Iglesia/JournalofCatalysis354(2017)287–298
Fig. 4. DFT-derived transition state structures for oligomerization (C–C bond
formationbetweensecondarypropoxideandpropene)andhydridetransferfrom
isobutanetotertiary2-methylpentoxideonHPW[PW91;PAW].Axesforaandb
areshown,withcintothepage.Thedimensionsforoligomerizationare[0.63,0.26,
0.43] and are [0.78, 0.53, 0.81] for hydride transfer. Charges from QUAMBO
(Section 2.2) and diameters of spheres with equivalent volume hdi are also
included.
stericeffectsthatcancauseanincreaseinenergybyrequiringsig-
nificantgeometricdistortionsoftheorganicguestortheinorganic
host.Theheuristicgeometricdescriptorsbasedon sphericalcon-
structs(andusedinFigs.3and4)leadtomeandiameters<d>that
cannot capture the most effective ‘‘fit” because neither host nor
guest structures are spherical in shape. For instance,alkyl chains Fig.5. Rateconstantratioofhydridetransferbetweenisobutaneandequilibrated
hexoxidepooltooligomerizationofpropeneasafunctionofvoiddiameter[503K,
inalkoxidesbenefitfromcylindricalvoids,butmayrequiresignif-
<5%alkeneandalkaneconversion].
icantdisruptionsfromtheirmoststableconfigurationstobenefit
from van der Waals contacts with cage-like hosts. Here, mean
diametersareusedasapproximatedescriptorsofsizefortransition donotsenseanypropertyofthehost.Asaconsequence,thetrends
states and bound species. They are appropriate and sufficient for evidentinFig.5mustreflecteither(i)anincreaseinGz through
HT;46
thecomparisonsintended,whichinvolvestructuresthatdiffersig- stericdestabilizationoftheHTtransitionstateor(ii)adecreasein
nificantlyinsizeforthechemicaltransformationsofinterest.The Gz asaresultofthemoreeffectivevanderWaalscontactsofthe
refinementofsuchananalysis,whensorequiredforspecificpre- oligo;3
dimerizationtransitionstatewiththesmallervoids.
dictive purposes, can be carried out using the averaging of van
Measured rate constants for hydrogen transfer (and for
derWaalsinteractionenergiesforagivenorganicmoietythrough
oligomerization[16])alsodependonacidstrength,becauseofits
eachspecificvoidstructureusingmethodsdevelopedandapplied
inherent role in determining the stability of the conjugate anion
inpreviousstudiesofalkenecoupling[16],alkoxidestability[39],
ation-pairtransitionstates.Indeed,HTrateconstantsonHPW,a
andacetonecondensation[43]withinconfiningenvironments.
strongacidwithadeprotonationenergy(DPE;[32,40])of1085kJ
ThetransitionstateforisobutaneHTtotert-2-methylpentoxide
mol(cid:1)1,areabout100-foldlargerthanonweakermesoporousSiAl
(0.73nm) is significantly larger than that for the dimerization of
acids (1220kJ mol(cid:1)1 DPE [40]) for isobutane reactions with both
propenewithsec-propoxide(0.62nm)(Fig.4).Theratiooftherate
hexoxides (Fig.3a) and propoxides (Fig.3b). Stronger acids, with
constants for isobutane HT to C alkoxides (Eq. (2)) to those for
6 morestableconjugateanions,leadtoion-pairtransitionstatesof
propenedimerization(Eq.(1))isgivenby:
lowerenergy,becausethechargeintheorganicandinorganicmoi-
c¼kkoHliTg;o4;63KK63¼expð(cid:1)DGc=RTÞ ð6Þ ebtoiuesnda)realmkouxcihdelaprgreecrutrhsaonrsin(Etqh.e(4e)s)s.eTnhtieaslelyDnFeTu-tdrearliv(ceodvaclheanrtgleys-
are ±0.96 for the isobutane-tert-2-methylpentoxide HT transition
a measure of the probability of terminating chains before further
statesand±0.90forisobutane-sec-propoxideHTtransitionstates,
growth. This ratio is not affected by void size for larger voids
but only ±0.30 for the sec-propoxide precursor in the latter
(BEA,FAU,SiAl),butdecreasessharplyforprotonsresidingwithin
reaction.
smaller voids (TON, MFI, MOR) (Fig. 5), as a result of a larger HT
Similarly, the isobutane tert-2-methylpentoxide HT transition
transitionstatethatbecomes lessstablecomparedwithasmaller
stateandthepropene-sec-propoxidedimerizationtransitionstate
dimerization transition state as the size of voids approaches that
are ion-pairs of similar charge (±0.96 and ±0.94, respectively;
oftheHTtransitionstate.TheeffectsofvanderWaalsinteractions
Fig.4).Asaresult,mesoporousHPW(onSiO )andSiAl,twosolid
c 2
onthisratioofrateconstants( ),usingthevoiddiameterasaproxy
acidsofverydifferentacidstrengthandcontainingvoidsmuchlar-
ofsizeandvanderWaalscontacts,dependssolelyonthedifference
gerthaneitherofthesetransitionstates,givesimilarratiosofHT
infreeenergybetweenthetworelevanttransitionstates: c
anddimerizationrateconstants( ;Eq.(5)),becausethestability
DGc¼GzHT;46(cid:1)Gzoligo;3(cid:1)G4ðgÞ(cid:1)Gð6gÞþ2Gð3gÞ ð7Þ of the conjugate anions influence the GzHT;ni and Gzoligo;3 terms in
Eq.(7)tosimilarextents.WeconcludethatHTandoligomerization
where GzHT;46 and Gzoligo;3 are the respective free energies of the HT rateconstantsdependsimilarlyonacidstrengthwhenprotonsare
and C–C bond formation transition states. The other terms in Eq. located either within large mesoporous voids or within voids of
(7)representpropertiesofthegaseousalkanesandalkenes,which similardimensionandshape,thusrenderingacidstrengthincon-
donotdependonstabilizationbybindingorconfinementandthus sequential in the control of chain length via systematic changes
M.L.Sarazen,E.Iglesia/JournalofCatalysis354(2017)287–298 293
intherelativeratesofHTandoligomerizationinisobutane-alkene hydride transfer from isobutane to each isomer j ðkeff Þ are
HT;4ij
alkylationprocesses.
groupedinFig.7accordingtothenumberofcarbonatomsinthe
bound alkoxide acceptor, which acts here as the initial surrogate
3.3.Effectsofchainlengthandsubstitutioninalkoxideacceptorson for their relative size. Such kinetic constants combine the effects
hydridetransferrates ofchainlengthandextentofsubstitutiononalkoxidereactivity.
Therateconstantsforeachofthealkoxideacceptorsarelarger
In the kinetic treatments and discussion above, HT rates are onHPWthanonSiAl(Fig.7),consistentwithHTtransitionstates
reportedforalumpedpoolofboundalkoxides(hexoxides,propox- thatbenefitfromamorestableconjugateaniontoagreaterextent
ides)consistingofisomersdifferentinskeletonandsurfaceattach- than the bound alkoxide precursors. Isobutane HT rate constants
mentpoint.Suchtreatmentsexploittheequilibratednatureofall foreachalkoxidearelargeronBEAthanonSiAl(Fig.7),indicative
regioisomers and skeletal isomers of alkoxides and alkenes of a ofthemoreeffectivevanderWaalscontactsbetweentheHTtran-
givenchainlength[9].Thislumpingofboundalkoxideacceptors sitionstateandthesmallerchannelsinBEA,inspiteofthesimilar
based on chain length is undone here as a strategy to determine acidstrengthpreviouslydemonstratedforaluminosilicatesofdif-
thereactivityofeachskeletalisomerwithinthisequilibratedpool ferentframeworkstructureandcrystallinity[16,39].Thediversity
with isobutane (and with n-butane and 2-methylbutane in Sec- ofactivityand selectivityofzeolitecatalysts haspreviouslybeen
tion 3.5). Such speciation is made possible because the specific inaccuratelyattributedtodifferencesin acidstrength amongdif-
skeletal alkane isomers formed from each alkoxide skeleton via ferentframeworks,aconsequenceoftheinappropriateuseofthe
HTcanbedetectedinthegaseousproducts. adsorption energy or evolution temperature of bound titrants.
Approachtoequilibriumvaluesforallhexeneisomersformed Suchpropertiesarenotdirectdescriptorsofacidstrengthbecause
from propene dimerization with respect to a given isomer (2- theyalsoreflectthestabilizationofboundspeciesbyelectrostatic
methyl-2-pentene; 2M2P) on TON, MFI, BEA, SiAl, and HPW are andvanderWaalsinteractionsthatareunrelatedtoacidstrength.
near unity [9] even at very low propene conversions (<0.1%; Alkoxideswithdifferentbackbonestructuresdifferinreactivity
Table2;approachtoequilibriumforC5inFig.S3).Theseisomers becauseoftheeffectsofsubstitutioninthestabilityoftheproto-
are grouped in Table 2 according to their skeletal backbone: 2- natedorganicmoietythatformsattheHTtransitionstate.These
methylpentenes(2-MP),3-methylpentenes(3-MP),linearhexenes reactivity trends with alkoxide structure remain mostly
(n-H),and2,3-dimethylbutenes(23-DMB).Eachskeletallumpcon- unchanged, however, with changes in acid strength or confining
tainsalldoublebondandstereoisomers.Only23-DMBisomersin voids (Fig. S4), apparently because the differences in charge and
TON are slightly below equilibrium values, a consequence of the size between the HT transition state and its alkoxide precursor
slow diffusion of the bulkiest C6 skeleton through the smallest dependonlyslightlyontheskeletalstructureoftheboundalkox-
channels,butapproachunitywithincreasingconversion.Theequi- ide acceptors, a concept that we examinein more detailthrough
libration of 23-DMB isomers within zeolite crystals was demon- DFTtreatmentsbelow.
strated by the fast intramolecular scrambling of all products
formedfromsingly-labeledpropene[9].The2,2-dimethylbutenes
(22-DMB) were not detected either because of the difficulty in 3.4.Densityfunctionaltheorytreatmentsofhydridetransfertransition
formingquaternaryCatomsviaskeletalisomerizationorbecause statesonHPWPOMclusters
of its diffusion-enhanced secondary interconversion via subse-
quentisomerizationorb-scission. The elementary steps of hydride transfer between isobutane
Bound alkoxides with different skeletal structures undergo and various alkoxides were investigated on HPW POM clusters
hydride transfer from isobutane (Scheme 2) to form 2- (describedinSection2.2)asanillustrativeexampleofanuncon-
methylpentane (2MP), 3-methylpentane (3MP), 2,3- strainedsolidacid.Fig.8showstheHTreactioncoordinate,defined
dimethylbutane(23DMB),andn-hexane(nH)atratesproportional intermsoftheC–HbondlengthinthedonorthattransferstheH-
totheprevalentisobutanepressure(Fig.6;BEA)butwithratesand atom to the bound alkoxide, for isobutane reactions with sec-
rate constants that differ among these backbones. Equilibrated ondary n-hexoxides or tertiary 2-methyl-pentoxides. The opti-
pentoxides, formed via b-scission of larger oligomers, lead to the mizedstructuresofthetransitionstatesindicatethatthehighest
formation of 2-methylbutane (2MB) and n-pentane (nP) at rates energysaddlepointoccursatdifferentC–Hbondlengthsforthese
also proportionalto the H-donorpressure. The rate constants for two alkoxide acceptors (Fig. 8). In forming 2MP from 2-methyl
Table2
Approachtoequilibriumfortheformationofhexeneisomerswithrespectto2-methyl-2-penteneonthesamplestestedwiththeirfractionalconversionsgiven:TON(0.005),MFI
(0.009),BEA(0.004),SiAl(0.005),andHPW(0.003)usingcalculatedequilibriumconstantsforSiAlatafractionalconversionof0.14.[503K,60kPa].Isomersareseparated
accordingtotheirbackbonestructures:2-methylpentene(2-MP),3-methylpentene(3-MP),linearhexenes(n-H),and2,3-dimethylbutene(2,3-DMB).
Backbone Approachtoequilibrium(w.r.t.2M2P) TON MFI BEA SiAl HPW
2-MP 2M2P 1 1 1 1 1
4M1P 0.96 0.93 0.93 1.09 0.82
c-4M2P 1.00 1.21 1.33 1.00 1.05
t-4M2P 0.96 0.96 0.97 1.05 0.87
2M1P/1-H 0.93 0.91 0.91 0.96 0.90
3-MP 3M1P 0.79 0.76 0.76 1.09 0.67
2E1B 0.87 0.75 0.65 0.92 0.63
c-3M2P 0.92 0.82 0.67 0.97 0.67
t-3M2P 0.93 0.84 0.68 1.00 0.68
n-H c/t-3-H 1.06 0.87 0.87 0.97 1.00
c-2-H 1.03 0.86 0.86 0.98 1.00
t-2-H 1.04 0.85 0.85 0.99 1.01
2,3-DMB 2,3-DM1B 0.35 1.40 1.14 1.00 0.92
2,3-DM2B 0.35 1.13 1.13 1.01 0.92
294 M.L.Sarazen,E.Iglesia/JournalofCatalysis354(2017)287–298
Scheme2. Alkaneincorporationintoequilibratedalkoxideisomerpools.Illustratedhereisthehexoxidepoolthatforms2-methylpentane(2MP),3-methylpentane(3MP),n-
hexane(nH),or2,3-dimethylbutane(2,3DMB)afterhydridetransferfromisobutane.
Fig.6. Formationratesof2-methylpentane(2MP;j),3-methylpentane(3MP;N),
2,3-dimethylbutane(23DMB;r),andn-hexane(nH;d)measuredduringreactions Fig.7. First-orderrateconstantforalkaneformationviahydridetransferbetween
ofisobutane-propenemixturesasafunctionofisobutanepressureonBEA[9kPa isobutaneandpropoxide,pentoxide,orhexoxideisomerpoolsonBEA(N),HPW
propene,503K,<5%alkeneandalkaneconversion]. (j),andSiAl(r)[9kPapropene,503K,<5%alkeneandalkaneconversion].
pentoxides,theH-atomhasalreadyfullytransferredatthetransi- viaBorn-Haberthermochemicalcycles(Scheme3).Insuchcycles,
tion state, while the transition state structure is formed before alkoxidesfirstdesorbastheirrespectiveprotonatedgaseousana-
hydride transfer for n-hexoxide acceptors. The energy barrier for logs at an energy cost given by the carbenium-ion detachment
HTtothetertiary2-methyl-pentoxideis16kJmol(cid:1)1smallerthan energy(CDE).Agaseousisobutane(thedonorn) thentransfersa
for HT to the secondary n-hexoxide, consistent with the more hydridetothesespeciestoformatert-butylgaseouscationanda
stablenatureoftertiarycarbeniumionsrelativetothosesecondary gaseous alkene, with a reaction energy given by the gaseous
carbeniumionsandwiththelocationofthetransitionstatealong hydridetransferenergy(HTE).Thegaseoustert-butylcationisthen
thehydridetransferreactioncoordinate. broughtclosertotheconjugateanioninordertorecoveraninter-
TheactivationenergyðDEHT;nij)forHTtransferfromdonornto action energy (IE) with electrostatic and charge reorganization
acceptori,jcanbedissectedintoindividualcomponentsforhypo- components [44]. The transition state structure for nH formation
theticalconnectingpathsbetweenprecursorsandtransitionstates fromsec-hexoxide(Fig.8)showsthattheH-atomtobetransferred
M.L.Sarazen,E.Iglesia/JournalofCatalysis354(2017)287–298 295
Fig.8. DFT-derivedelectronicenergiesforhydridetransferfromisobutanetosecondaryhexoxide(--)ortertiary2-methylpentoxide(—)relativetoalkoxideandgasphase
isobutaneasafunctionofC–Hbondlengthinthedonor(isobutane)[PW91,PAW].
statesrelativetotheirdifferentrespectiveprecursors.Thesetran-
sition states differ in their size and numbers of C-atoms (Sec-
tion 3.2). Their van der Waals volumes, surface areas, and
equivalent sphere diameters (Section 3.2) are shown in Table 3
forHTreactionsonHPWusingisobutaneastheH-donorandeach
oftheC ,C ,andC skeletally-distinctalkoxideswiththeattach-
6 5 3
ment locations listed (optimized structures in Fig. S5). The mea-
sured HT rate constants on HPW, BEA, and SiAl increased as the
DFT-derived size of the respective transition state structures (on
HPW)increased(Fig.9),eventhoughTSstructureswiththesame
number of C-atoms differ only slightly in size. On mesoporous
acids (HPW and SiAl), such an increase predominately reflects
the greater stability of the cationic moiety as the electron-
donatingalkylchainsbecomelonger,sincevanderWaalsinterac-
tionschangelesssensitivelywithtransitionstatesizeoverflatand
concave unconfined surfaces than over the smaller concave
domainsprevalentwithinzeotypes.
Transition states of similar size by the metrics used here can
exhibit different rate constants when the degree of substitution
at the center of charge in the organic moiety varies (e.g. nP vs.
Scheme3. Thermochemicalcycleforhydridetransferbetweenanalkanedonor(i)
andanalkoxideacceptor(ij)(here,isobutaneandtertiary2-methylpentoxide),in 2MB formation; hdi=0.712 and 0.714nm and keHfTf;45j=33.9 and
termsofcarbenium-iondetachmentenergy(CDE),hydridetransferenergy(HTE), 61.3lmol(H+-s-kPa)(cid:1)1,respectively),becauseofthedifferentsta-
andinteractionenergy(IE).
bility of their respective alkoxide-derived carbenium ions at the
transitionstate.Foreachtypeofbackbone(e.g.,linearalkoxides;
Fig.9,opensymbols),HTrateconstantsfortheformationofpro-
remains with the isobutane donor, indicating that the barrier to
pane,n-pentaneandn-hexaneincreasedmonotonicallywithchain
form this transition state predominantly reflects that required to
lengthonbothHPWandSiAl.Incontrast,suchchainlengtheffects
form the secondary carbenium-ion from the sec-hexoxide (CDE).
became weaker for larger chains on BEA, as steric effects require
Thehydridehasbeentransferredtotheacceptorinforming2MP
distortions in theframeworkand theorganicmoietyand impose
from 2-methyl pentoxide bound through its tertiary C-atom
energetic penalties that ultimately offset the more effective van
(Fig.8).Theseresultsindicatethatsubstitutionathydrideaccep-
derWaalscontactsbetweenvoids andlargertransitionstates,as
torsaffectsthetransitionstateenergy,leadingtotheobserveddif-
discussedearlier(Section3.2).
ferences in measured rates for HT to alkoxides with different
skeletonsfromisobutane(Fig.7).Theextenttowhichtheforma-
tion of the carbenium ion (CDE) from the alkoxide acceptor or 3.5.Effectsofhydridedonorchainlengthandsubstitutiononhydride
thedetachmentandreattachmentoftheH-atombeingtransferred transferrates
(HTE)contributetothistransitionstateenergyreflectstheirrela-
tivemagnitudes,withtheformationofthecarbeniumionbecom- Hydridetransferratesalsodependonthesizeandsubstitution
ing the sole determinant of transition state stability as the inthedonoralkanes.TherateofformationofC alkanesfromall
6
carbeniumionsderivedfromtheboundalkoxidesbecomeincreas- combined isomers of bound hexoxides on BEA are proportional
inglyunstable. to the pressure of 2-methylbutane, isobutane, and n-butane
Measuredrateconstantsforhydridetransferfromisobutaneto (Fig. 10). The first-order rate constants for the formation of each
differentalkoxideacceptorsreflecttheenergiesoftheirtransition C ,C ,C ,andC alkaneisomersfromHTwithn-butane,isobutane,
3 4 5 6
296 M.L.Sarazen,E.Iglesia/JournalofCatalysis354(2017)287–298
Table3
DFT-derivedvanderWaalsvolumes(VvdW),surfaceareas(SAvdW),anddiameters(hdi)ofhydridetransfertransitionstatestructuresforthegivendonorandacceptoronHPW.The
diametersarecalculatedassumingasphereofequivalentvolume.
H-donor(n) H-acceptorC-atoms(i) H-acceptorisomer(j) attachment VvdW[nm3] SAvdW[nm2] hdi[nm]
n-Butane 6 2MP tertiary 0.211 2.18 0.739
Isobutane 6 2MP tertiary 0.203 2.06 0.730
23DMB tertiary 0.202 1.95 0.727
3MP tertiary 0.202 1.95 0.728
nH secondary 0.208 2.18 0.736
5 2MB tertiary 0.191 1.85 0.714
nP secondary 0.189 1.97 0.712
3 prop secondary 0.151 1.68 0.661
2-Methylbutane 6 2MP tertiary 0.222 2.13 0.751
3 prop secondary 0.169 1.85 0.686
and2-methylbutaneonBEAareshowninFig.11.Measuredrates tertiary 2-methylpentoxides on HPW (Fig. 12) indicate that the
andrateconstantsallincreasedasthelengthandsubstitutionof C–Hbondlengthissimilarforthetwodonorsandthatthehydride
the H-donors increased for all alkoxide acceptors. These trends has been nearly fully transferred at the transition state. Their
l
are consistent with the different dehydridation energies of these different measured rate constants on BEA (2.6 versus 6.9 mol
alkanes (Eq. (8); Fig. S6), which is calculated here as the energy (H+-s-kPa)(cid:1)1forn-butaneandisobutane,respectively)thusreflect
requiredtoremovethesecondaryH-atomfromn-butaneandthe the different stabilities of the donor alkane as a cationic moiety,
tertiaryH-atomfromisobutaneand2-methylbutaneasagaseous given that the alkoxide acceptor and the extent to which the
hydrideion: H-atomshavebeentransferredarethesameinHTreactionswith
thesetwoH-donoralkanes.
DehydridationEnergy¼En(cid:1)ðEn(cid:1)HþþEH(cid:1)Þ ð8Þ Thepresenceofanadditionalmethylgroupwhenisobutaneis
Here,En istheenergyofthegaseousalkane,En(cid:1)Hþ istheenergyof replaced with 2-methylbutane as the H-donor leads to an addi-
thecationformeduponhydrideremoval,andEH(cid:1) istheenergyof tionalincreaseinhydridetransferrateconstantsonBEA(6.9and
thehydrideanioninitsgaseousform.Eq.(8)illustratestheeffects 62lmol (H+-s-kPa)(cid:1)1 for isobutane and 2-methylbutane, respec-
of the stability of the alkane-derived cation on hydride transfer tively; Fig. 11). Such an increase with chain length is larger than
energies (HTE) and on the energy of the HT transition state. Such thatobserveduponchangingtheskeletonofC4alkanes(n-butane
effects become stronger for more stable alkoxide acceptors, for vs. isobutane). These differences reflect the combined effects of
which the hydride has already been transferred at the transition cation stability and of the larger size of the 2-methylpentane HT
state,asshownintheprecedingsection.Optimizedtransitionstate transition state, which can more effectively interact with BEA
structures for hydride transfer from n-butane or isobutane to channels. Indeed, the transition state for 2-methylbutane HT to
Fig. 9. First-order rate constant for alkane formation via hydride transfer from
isobutaneandtopropoxide,pentoxideorhexoxideisomerpoolsonBEA(N,4), Fig.10. C6alkaneformationratesfromhydridetransferbetweenhexoxidepool
HPW(j,h),andSiAl(r,})asafunctionofestimatedtransitionstatediameter and2-methylbutane(N),isobutane(r),orn-butane(d)measuredwhenfedwith
(<d>)from Table3.Open symbolsindicate hydride transfer tolinear alkoxides, propeneasafunctionofalkanepressureonBEA[9kPapropene,503K,<5%alkene
whileclosedsymbolsindicatehydridetransfertobranchedalkoxides. andalkaneconversion].