What does B3LYP do well? What does it do badly?

The main issue of B3LYP in such systems, is that it is hybrid functional, which are extremely expensive to use in periodic systems. You will ... MatterModelingStackExchangeisaquestionandanswersiteformaterialsmodelersanddatascientists.Itonlytakesaminutetosignup. Signuptojointhiscommunity Anybodycanaskaquestion Anybodycananswer Thebestanswersarevotedupandrisetothetop Home Public Questions Tags Users Unanswered Teams StackOverflowforTeams –Collaborateandshareknowledgewithaprivategroup. CreateafreeTeam WhatisTeams? Teams CreatefreeTeam Teams Q&Aforwork Connectandshareknowledgewithinasinglelocationthatisstructuredandeasytosearch. Learnmore WhatdoesB3LYPdowell?Whatdoesitdobadly? AskQuestion Asked 11monthsago Modified 10monthsago Viewed 2ktimes 25 6 $\begingroup$ IhaveseenageneralcriticismofB3LYPincomputationalchemistry(especiallywiththe6-31G*basisset).Iwouldliketoknowsomeofthemore/lessobviousproblemsthatoccurwhenusingitandwhatalternativesexistforthatleveloftheory.IamawarethatitdoesnotincludevDWinteractions,soIwouldliketoavoiddiscussionofthatunlesscomparingtoanotherfunctionalwhichincludesit. Iamlookingmainlyforanswersintheorganic/inorganicspaceasthisiswhereitisusedforbetterorworse,butwouldbehappytoalsohearaboutsurfacecatalysisforthisfunctional. density-functional-theorycomputational-chemistrysurfacesfunctionalcatalysis Share Cite Improvethisquestion Follow editedMay13,2021at1:00 NikeDattani 1 askedMay1,2021at15:19 TristanMaxsonTristanMaxson 9,45111goldbadge1414silverbadges6767bronzebadges $\endgroup$ 2 3 $\begingroup$ PerhapsIwillhavetimelatertochimeinwithmoredetails,butitisnotoriousforoverstabilizingthehighspinstateinopen-shell3dtransitionmetalcomplexessuchthatthespinsplittingenergy(andoccasionallythegroundstatespinstatealtogether)isoftenincorrect.Decreasingthe%HFexchangeto15%(sometimescalledB3LYP*)canresolvethisquiteabit. $\endgroup$ – AndrewRosen May1,2021at19:13 $\begingroup$ OneofthepracticalproblemswithB3LYPistheconfusionaboutitsdefinitionindifferentprograms,asdiscussedonchem.SE. $\endgroup$ – TAR86 May2,2021at7:17 Addacomment  |  3Answers 3 Sortedby: Resettodefault Highestscore(default) Datemodified(newestfirst) Datecreated(oldestfirst) 17 +200 $\begingroup$ Ithinkthisreview¹byHead-GordonisausefulsupplementtoNike'sanswer.Itscombinesareviewoffunctionaldevelopment,abenchmarkingofvariousfunctionals,andanexplanationofthedesignprocessforthe$\omega$B97functionals.Itsalsoopenaccess,soitsagreatresourceifyouareinterestedinDFTfunctionalsingeneral. Theybenchmarked200DFTfunctionalsonavarietyofdifferenttestsfocusedonisomerizationenergies(I),barrierheights(BH),thermochemistry(TC),nonconvalentdimers/clusters(NCD/NCC),andequilibriumbondenergies/lengths(EBE/EBL).Thedatasetswerealsoclassifiedwithasuffixindicatingwhethertheyareeasy(E)ordifficult(D)examples. Notethecoloringofeachcellgivesaroughsenseofhowthefunctionalcomparesamongall200functionalstested,ratherthanjustthoseinthetable.Thefulllistoffunctionals,alongwiththeranking/RMSDofeachfunctionaloneachdatasetisgiveninthepaper.Evenamongthewholeset,wheremanyoftheotherfunctionalsalsodon'texplicitlyaccountfordispersion,B3LYPtendstobeinthebottomquartileformostofthedatasets. WhatyoucanseeisthatB3LYPdoesfairlypoorly,notablyforisomerizationenergiesandequilibriumbondlength/energy.SoB3LYPonitsowngenerallydoesn'tdogreatacrossawidevarietyofproblems.However,ifoneisabletoincludedispersioncorrections,inthiscaseD3(BJ),B3LYP'sperformanceismuchimproved,becomingbasicallyonparwith$\omega$B97X-D.Theonlynotableexceptionisisomerizationenergieswhereitstillseemstostruggle. Ishouldalsonotethatallthecalculationsweredonewiththedef2-QZVPPDbasissetanda(99,590)DFTintegrationgrid(99radialshellswith590pointspershell).Theuseofalargebasisandaccurategridshouldmakethecomparisonasmuchaspossibleaboutrelativecapabilitiesofthefunctionals,ratherthanbasissetcompleteness.Ontheflipside,ashasbeenmentionedinotheranswers,ifyouareworkingwithcertainsmallerbasissets,youmayobtainmorefavorableresultsthanexpectedduetoerrorcancellation. N.Mardirossian&M.Head-Gordon.MOLECULARPHYSICS,2017 VOL.115,NO.19,2315-2372 Share Cite Improvethisanswer Follow editedMay3,2021at20:47 answeredMay2,2021at3:28 Tyberius♦Tyberius 13.7k33goldbadges2323silverbadges107107bronzebadges $\endgroup$ 3 $\begingroup$ +1butitseemslikeinthistable,thefunctionalslistedafterB3LYPseemlargelytobeimprovedbecauseofdispersioncorrections,whichtheOPalreadynotedissomethingheknowsisashortcomingofvanillaB3LYP.Itstillisaveryvaluablereferencethough! $\endgroup$ – NikeDattani May2,2021at4:49 1 $\begingroup$ @NikeDattaniThetablehighlightstheseparticularfunctionals,buttheoverallcomparisonisdoneacrossthewholesetof200functionals.Evenamongthewholeset,wheremanyoftheotherfunctionalsalsodon'texplicitlyaccountfordispersion,B3LYPtendstobeinthebottomquartileformostofthedatasets.WhilethiswasthemostcompactpresentationoftheB3LYPdatainthepaper,togetabettersenseofhowitcomparedwithaparticularfunctionalofinterestwouldrequiredelvingintosomeofthelargerfiguresinthepaper. $\endgroup$ – Tyberius ♦ May3,2021at20:36 $\begingroup$ Thanksforclarifyingthatinyourlastedit.It'sinterestingthatinthepaperreferencedinmyanswer,B3LYPwasinthe"middleofthepack"exceptforinthecaseofcalculatingreactionenergies(thoughIwasonlycomparingittootherfunctionalsatthesame"leveloftheory",i.e.single-hybridswithoutdispersioncorrections). $\endgroup$ – NikeDattani May3,2021at21:26 Addacomment  |  19 $\begingroup$ B3LYPisstilladecentfunctionalatitsleveloftheory(single-hybridfunctional),butyou'rerightthatthere'sageneralcriticismofit,whichIlargelyhearintheformofpeoplesayingthingslike"alltheydidwasB3LYP/6-31G*"tocriticizenon-expertsthatblindlyusethiscombinationwhichbecamethe"default"inchemistryfornon-expertgroupsformanyyears. SomehistoryaboutB3LYP/6-31G* AsSusiLehtolasaidhere,BLYP/6-31G*performedbetterthanwavefunction-basedmethodsforequilibriumgeometries,dipolemoments,harmonicvibrationalfrequenciesandatomizationenergiesinthis1993paper.Aco-authorofthatpaper(JohnPople)wastheoriginalauthorofGaussian70andwouldsoongettheNobelPrize,soitwasawidelyseenpaper,andthenayearlaterMikeFrisch(currentlyPresidentofGaussian)didacomparisonofBLYPtoB3LYPandLSDAagainwith6-31G*(andonemorebasisset),whichledtotheB3LYP/6-31G*combinationbecomingwidelyadopted. Butwavefunction-basedmethodshaveimproved Intheaboveparagraph"wavefunction-basedmethods"referredonlytoMP2andQCISD(itwas1993afterall)andiftheycouldcomparetoCCSD(T)whichbecameinthe2000sknownasthe"goldstandardofquantumchemistry"foraboutadecade,theywouldn'thavefoundfoundBLYP/6-31G*orB3LYP/6-31G*toperformbetter.MuchofwhatpeopleuseB3LYP/6-31G*forcanbedoneveryfastwithCCSD(T)/cc-pVTZin2021ifpeoplejustknewthat,somanypeoplefinditcringeworthywhenB3LYP/6-31Gisstillused.* Oneshouldconsidermorethanonefunctional TAR86pointedouthere,thatit'simportanttoevaluatemorethanonefunctionalforthesetofmoleculesyou'redealingwith,ortolookatbenchmarkstudieswhichhavealreadydonethis,andmostofthetimewhenyouseepeopleusingB3LYP/6-31*theyareusingitwithnojustificationapartfromitbeingalltheyknow.Thismeansthatpeoplecanagainbegettingworseresultsthanwhattheycouldhavegotwiththesamecost,simplybecausetheydidn'tdoabitmoreresearch. Functionalshavealsoimproved(answeringthecallfor"alternatives") AstheOP(TristanMaxson)saidhere,B3LYP/6-31G*isoldandthere'snowbetteroptions.Asmentionedbymehere,acomparisonof47functionals(2LDA,14GGA,3meta-GGA,23Hybrid,5DoubleHybrid)inanenormousdatasetwasdonebyGoerigkandGrimme.YoucanseeinthefigurebelowthatB3LYPisstillaverydecentfunctional,standinginthe"middle"ofthepackofotherfunctionalsatthesameleveloftheory(single-hybrid),havinganoverallaverageerroragainstCCSD(T)/experimentofslightlylessthan4kcal/molwhileothersingle-hybridscanhaveaverageerrorsofabitmorethan2kcal/molinthebestcaseandalmost5kcal/molintheworsecase. ButyoucanindeedseeseveralotherfunctionalsatthesameleveloftheoryofB3LYP: Soevenifwewanttoblindlyusejustonefunctionalasablack-box(withoutcomparingagainstotherfunctionalsorlookingforabenchmarkstudyontheclassofmoleculeswe'redealingwith),thingslikeM06andPW6B95seemtoworkbetterbetteronaveragethanB3LYPoverallonacomprehensivedatasetliketheoneusedinthestudyabove.Theydidn'tuse6-31G*becausesomeoftheelementsincludedinthedatasetdon'thave6-31G*setsavailable,buttheydidsaythat(aug-)def2-SV(P),whichtheydiduse,isusuallycomparableto6-31G*. SpecificproblemswithB3LYP Thelastsectionaboveansweredyourcallfor"alternativesatthesameleveloftheory"sonowI'dliketoansweryourcallfor: "someofthemore/lessobviousproblemsthatoccurwhenusingit" Thefigurebelow,alsofromthesamepaperastheabovefigure,showsthatB3LYPis: Quitegoodcomparedtoothersingle-hybridsforbasicproperties(inthispaperexamplesincludedatomizationenergies, electronaffinities,ionizationpotentials,protonaffinities,SIE relatedproblems,barrierheights) Abitworsebutstill"inthemiddleofthepack"fornon-covalentinteractions(inthispaper:waterclusters,conformationalenergies,andinter-andintramolecularLondon-dispersioninteractions) Oneoftheworstoverallforreactionenergies(inthispaperexamplesincludedisomerizations,Diels–Alderreactions,ozonolyses,reactions involvingalkalinemetals) SoreactionenergieswouldbeoneoftheplacestobecarefulaboutusingB3LYP,inadditiontoRESP(RestrainedElectrostaticPotential)calculationsasnotedinB.Kelly'swarningthatitwillcausepartialchargestobeaddedimplicitly(there'salotmoredetailsintheoriginalanswerthough),andforotherthingssuchasinAndrewRosen'scommentwhichIwouldbethrilledtoseeturnedintoanansweriftimepermits. Share Cite Improvethisanswer Follow editedMay2,2021at7:20 TAR86 1,55322silverbadges1515bronzebadges answeredMay2,2021at2:01 NikeDattaniNikeDattani 1 $\endgroup$ Addacomment  |  16 $\begingroup$ Idon'thavetoomuchtoaddtotheanswersofNikeDattaniandTyberius,butIthinkthecruxisthatitscapabilitieshavebeenhistoricallyoverestimated. OneparticularfailingofB3LYPisthatittendstounderestimatebondenergies.However,sincethesmall(andfast)6-31G*basissetwillleadtooverbinding,thefamouscombinationB3LYP/6-31G*endedupbeinggood"byaccident."Atleastforsmallmolecules,becauseitturnsoutthattheperformanceofespeciallytheLYPcorrelationpartworsensinlargersystems.Maybenottoosurprising,sinceLYPisdesignedtoreproducecorrelationenergyoftheHeatom. Thegeneralissue,then,isthatevenifB3LYP/6-31G*mayworkbecauseoferrorcancellation,thisisnotareliableassumptioningeneral.Afterall,youwanttoberightforrightreasons! Youalsoaskaboutsurfacecatalysis.ThemainissueofB3LYPinsuchsystems,isthatitishybridfunctional,whichareextremelyexpensivetouseinperiodicsystems.YouwillalmostalwaysseepeopleuseGGAormeta-GGAfunctionalsinsuchsystems,someofwhichareactuallyverygoodforsuchproblems(butthisdependsstronglyonthesystem!). Share Cite Improvethisanswer Follow answeredMay3,2021at16:37 KristofBalKristofBal 56611silverbadge66bronzebadges $\endgroup$ 2 3 $\begingroup$ +1Welcometoournewcommunityandthankyouforyourcontributionhere!Wehopetoseemuchmoreofyouinthefuture!YouranswerremindsmeofsomethingAxelBeckesaidataconferenceinTorontosomeyearsago:"you'dbeafoolnottotakeadvantageofthiserrorcancellation".Hewastalkingaboutasystematicerrorincorrelationtreatmentbeingcancelledoutbyasystematicerror(oftheoppositesign!)frombasissetincompleteness.Iwasgoingtomentionthis,butmyanswerwasquitelonganyway,soI'mgladyoumentionedit:) $\endgroup$ – NikeDattani May3,2021at17:11 3 $\begingroup$ thanks!I'vebeenlurkingabitinthepastandthoughttogiveansweringatry.Whodoesn'tloveerrorcancellation?AnotherniceoneisvdWinteractioninDFT(orHF):youmightgetreasonablenon-bondedinteractionenergiesifyouuseabasissetwithsufficientlylargeBSSE. $\endgroup$ – KristofBal May3,2021at17:42 Addacomment  |  YourAnswer ThanksforcontributingananswertoMatterModelingStackExchange!Pleasebesuretoanswerthequestion.Providedetailsandshareyourresearch!Butavoid…Askingforhelp,clarification,orrespondingtootheranswers.Makingstatementsbasedonopinion;backthemupwithreferencesorpersonalexperience.UseMathJaxtoformatequations.MathJaxreference.Tolearnmore,seeourtipsonwritinggreatanswers. Draftsaved Draftdiscarded Signuporlogin SignupusingGoogle SignupusingFacebook SignupusingEmailandPassword Submit Postasaguest Name Email Required,butnevershown PostYourAnswer Discard Byclicking“PostYourAnswer”,youagreetoourtermsofservice,privacypolicyandcookiepolicy Nottheansweryou'relookingfor?Browseotherquestionstaggeddensity-functional-theorycomputational-chemistrysurfacesfunctionalcatalysisoraskyourownquestion. Linked 3 WhyisthereadeviationbetweentheexperimentallycalculatedvaluesandtheoreticaldonebyDFTcalculation? 41 Whataresomerecentdevelopmentsindensityfunctionaltheory? 21 WhatarethepitfallsfornewusersofDFT? 11 Whatdoes"leveloftheory"mean? 15 Whataretheparametersthatneedtobecalculatedtodesignthestructureofamolecule? 10 WhichDFTfunctionalshouldIusetosimulatespectraofpesticides? 11 DFTfunctionalsfromMP2methods Related 48 Whatistheclosestthingwehaveto"the"universaldensityfunctional? 25 Whatarethebestfunctionalsfortransitionmetalcompounds? 17 WhatisthedifferencebetweenBDFTandCDFT(magnetic-fieldDFTandcurrentDFT)? 19 Whatsoftwareisneededinchemoinformaticsormaterialsmodeling,thatasoloprogrammercouldworkon? 14 Whatistheroleofachemistincomputationalchemistry? 16 Howwellcanwemodelchemicalsynthesis? 12 Whatdoesadensityofstates(DOS)plotshow? 7 Whatdoes"Testjobnotarchived"mean? 8 Quantummechanicalantireflectioncoating;whatquantitywouldbeanalogoustoindexofrefraction? HotNetworkQuestions Whenshowinginterestinpeople,howisthegenderinferred? "Systemisreadytointoshippingmode"afterchangingsettingsinBIOS? AmIallowedtotellATC"unable"ifaskedtoclimbtolessthan12,500withoutsupplementaloxygenaboard? ShouldIapplyformydreamjobevenifIexpecttobeonmaternityleave? WhyisAvogadroconstantusedtocalculatethenumberdensity? WhydidtheRussianFederationneverstresseconomicgrowthasChinadid? Wouldallowingbarbarianstomaintainrageasanactionbeunbalancedorpermitshenanigans? Yogurt+tahini=paste? WhendidtheCatholicChurchdecide'dailybread"isitsEucharist? Understandingunderpaymentpenalty:implicationsandhowtoavoiditinthefuture Mapatwoargumentfunctiontoeveryelementonlist HowshouldIwritealettertoasktheauthorofaresearchpapersomequestions? VehicleACworksinitially,butafterdrivingforawhile(hour+),theairflowdropstothepointbeingineffective Whatkindofeventwouldwipeout90%ofthehumanpopulation? What'stheoppositeofcrowded,whenspeakingofroads? Howtoprovenovelconductionpathinweakerbatterycellsintwo4s-1pcarbonzincbatteriesdischargedat1mAfrom6Vto3V? Blendernotsavingneworientationafterapplyingalltransformswhensavingas.OBJfile Funnyteleporterinventionstory howtoget"curvedgraph"ingeometrynodes? ShouldItelltheeditorI'mdecliningareviewinvitationbecausetheauthorsarehostileaboutfeedback? HowtoaccessspecialkeysymbolsonMagicKeyboardUKlayoutusedwithWindows10 Howwouldanandroid'sabilitytothinkfreelybeaffectedifitcannotquestionitsprimarygoal? Simulatetherippleeffect Canweplayanysonginanykeyonthepianoandbywhatprincipleisthat? morehotquestions Questionfeed SubscribetoRSS Questionfeed TosubscribetothisRSSfeed,copyandpastethisURLintoyourRSSreader. default Yourprivacy Byclicking“Acceptallcookies”,youagreeStackExchangecanstorecookiesonyourdeviceanddiscloseinformationinaccordancewithourCookiePolicy. Acceptallcookies Customizesettings