ThePLATONcheckCIFandSQUEEZETools

TonSpekUtrechtUniversity,NL

Rigaku/OxfordDiffracIonUsermeeIngWroclaw,june20,2016

QUESTION:IsthisStructureCorrect?TheAug.2015IUCr/checkCIFreportsnoseriousALERTS

NoVoidsNoUnusualContactsNormal?DifferenceDensityRange

R1=0.0111wR2=0.0339S=1.042Rhomin-0.43Rhomax0.47

No,TheStructurewasDeliberatelyINVENTEDThisdeviousstructurewascleverycreatedbyNatalieJohnsonetal.,Newcastle,UK,aimingtobeatcheckCIF.ItwaspresentedasanexcellentPosterduringthe2015ECMCongressinRovinj,CroaIa.But:Everycrimeleavesitstraces…..

UnusualActualdifferencemapDensity

Cleartracesofthe‘Crime’areintheDifferenceDensityMap

DifferencemapintheCH2planeTheCH2HydrogenatomsatcalculatedposiIonsaredefinitelynotinF(obs)

Expectedtypeofdifferencemap

NATALIE

YLID

HowwasStructure‘Natalie’created?

YLID

‘StarIngMaterial’(Bruker)

Chemicalissue:Se+

NoexampleintheCSDforthe6-memberedringSystemMogul:Se–Coutlier

í

ë

NoH-density

NoDensityonBonds

CURRENTPLATON/CHECKCIFVALIDATIONREPORTFOR‘NATALIE’

WhatisPLATON

•  AcollecIonofSHELXcompaIbletoolsbundledinasingleprogram

•  TheToolsetincludes:– CheckCIF (StructureValidaIon)– SQUEEZE (HandlingDisorderedSolvents)– ADDSYM (MissedSymmetry)– TwinRotMat (TwinningDetecIon)– Bijvoet (AbsoluteStructure–Hoom(y))

PLATON/CHECKCIF•  IscurrentlythemajorvalidaIonenginebehindtheIUCrdrivencheckCIFvalidaIonprojectofcrystalstructures.

•  ItChecksandReportson(withALERTS)–  CompletenessofthesuppliedinformaIon–  Problemswiththeexperimentaldata–  ProblemswiththeinterpretaIonofthedata–  Problemswiththereportedstructuremodel–  Unusualstructuralfeatures

StructureValidaIonnowover20Years

•  TheintroducIonoftheCIFstandardfordataarchivalmadeautomaIcstructurecheckingpossible.

•  StructurevalidaIonprovidesauthors,refereesandreaderswithalistofpossiblyinteresIngissueswithastructurereportthatmightneedtobeaddressed.

•  Currentlyabout500testshavebeenimplementedincheckCIFandthatnumberissIllincreasingonthebasisofnewlydetectedissueswithsuppliedCIF’s.

•  ALERTSarenotnecessarilyerrors.TheyomenpointatinteresIngstructuralfeaturestobediscussed.

SomeValidaIonIssues•  ACIFessenIallyarchivestheauthorsinterpretaIonof

theunderlyingexperimentaldiffracIondata.•  ArchivedreflecIondataareneededforameaningful

evaluaIonofunusualresultsandfortestcalculaIons.•  ArchivalofFo/Fcdata(FCF)alreadysolvespartofthis

issue(sideeffectofitsavailability:detecIonandproveofcasesofafewhundredseriouslyfraudedstructuresinActaCryst.E)

•  Recently:embeddingofrefinementinstrucIons(res)andunmergedreflecIondata(hkl)intheCIFurged.

•  Theembeddingofthe.res&.hklintotheCIFwithSHELXL2014currentlyvoidstheneedtosupplyanFCF.

•  IUCr/checkCIFandhkldeposiIonarenowalsopartoftheCSDdeposiIonandarchivalprocedures.

FCF-ValidaIonadds:

•  Analysisofthequalityoftherefinement•  Analysisofthedifferencemap(peaks)•  DetecIonofvoidcontent(SQUEEZE)•  DetecIonofmissingreflecIons•  DetecIonofoutliers•  DetecIonofmissedtwinning•  Checkoftheabsolutestructure

BenefitsoftheSHELXL2014standardembeddingof.res,.hkl(&.fab)

•  NoneedtosupplyanFCFfilealongwiththeCIFaspartofadatadeposiIon.TheFCFcanbecreatedfromtheembeddeddata.

•  Whensupplied,theFCFshouldbe‘LIST4’or‘LIST8’type•  SHELXL2014CIF+FCFaretherecommendedfilestoruntheSQUEEZEtoolsincetheoriginalreflecIondataareretainedintheCIF

STATISTICSPREPAREDBYTHECCDCONSQUEEZEdorOLEX2-MASKdSTRUCTURESintheCSD

TheSQUEEZETool

PLATON/SQUEEZE•  ThecurrentimplementaIonoftheSQUEEZEtooltohandledisorderedsolventsisthethirdgeneraIonofamethodpublishedbyusmorethan25yearsago.

•  InterfacingwithSHELXL2014refinementsolvesmanyearlierissueswithSHELX76&SHELXL97using.res&.hkldata.[e.g.ModificaIonoftheobserveddata]

•  DocumentaIonoftherecommendedprocedure:A.L.Spek(2015)ActaCryst.C71,9-18

•  htp://www.platonsom.nl/PLATON_HOW_TO.pdf

TheDisorderedSolventProblem•  ThecalculatedstructurefactorFccanbespitintotwoparts:Fc=Fc(model)+Fc(solvent)

•  Fc(solvent)canbeparametrizedwithan(elaborate)disordermodelandrefinedalongwiththeothermodelparameters.

•  Fc(solvent)canalsobeapproximatedwiththeSQUEEZEtoolandusedasafixedcontribuIontothestructurefactorsintherefinement.

•  Insimplecases,thefirstapproachispreferred

ThePLATON/SQUEEZETool•  SQUEEZE,asimplementedinPLATON,analysesthecontentofsolventaccessibleVOID(s)inacrystalstructure.(Q:arethevoidsempty?).

•  TheVOIDcontentwillgenerallyinvolve(heavily)disorderedsolvent(s)thatmightbedifficulttoparameterizemeaningfully(e.g.unknownsolvents).

•  ThesolventcontribuIontothecalculatedstructurefactorsisapproximatedbyFouriertransformaIonofthedensityintheVOID(s)aspartoftheleast-squaresrefinementofthemodelparameters.(.fab)

•  SQUEEZEdoesnotrefinetheFc(model)

TheProperuseofSQUEEZE•  ItisimportantthatthefinalCIFarchivesboththedetailsoftheSQUEEZEcalculaIonandtheunmergedreflecIondata.Inthatway,thecalculaIonscanbereconstructedand/oralternaIverefinementmodelsatempted.

•  SHELXL2014offersallwhatisneededforthat.•  SQUEEZEusesthemodelparameterstakenfrom.cifandmergedobservedstructurefactorsfromtheLIST4orLIST8.fcftocalculatesolventF(calc)on.fab.

•  FinalSHELXLrefinementwillbebasedontheCIFembedded.res,.hklfilesalongwiththe.fabfile.

HowtoSQUEEZEwithSHELXL2014

1.  Refineanon-solventmodelwithname.ins&name.hkl(IncludeACTArecord,NOLIST6).

2.  RunPLATON/SQUEEZE,basedonname.cif&name.fcffrom1as‘platon–qname.cif’.

3.  ConInueSHELXLrefinementwiththefilesname_sq.ins,name_sq.hkl&name_sq.fabfrom2as‘shelxlname_sq’

4.  Inspectthe.lis&.lstfilesandValidate

SQUEEZEDisorderedSolvent+Twinning

•  Step1:SHELXLrefinementbasedaname.ins(thatshouldinclude‘ACTA’,‘LIST8’,‘BASF’and‘HKLF5’[or‘TWIN’]records)andaname.hklfile

•  Step2:RunSQUEEZEwiththename.cifandname.fcffilesproducedinStep1(i.e.run:platon–qname.cif)

•  Step3:ConInueSHELXLrefinementwiththefilesname_sq.ins,name_sq.hklandname_sq.fabproducedbyPLATONinstep2èname_sq.cif&name_sq.fcf

•  Note:Thename_sq.fabfilecontainsthesolventcontribuIontotheSFandthedetailsofSQUEEZE.

SQUEEZE-2016Example:CoordinaIonCompound

SpaceGroupP21Z=4,Z’=260:40TwinTwinaxis:(001)150KTWINABShklf5dataAcetonitrilsolvateStep1(SHELXL2014)èR1=0.047,wR2=0.1445Step2(SQUEEZE)è177electronsfoundinunitcellStep3(SHELXL2014)èR1=0.0275,wR2=0.0679,S=1.064

AcetonitrilModel:R=0.0323,wR2=0.0889,rho(max)=1.34e/A-3

EffectofonR(F)beforeandamerSQUEEZEasafuncIonofsin(theta)/lambda

Requirements•  Thereshouldbenoresidualunresolveddensityinthediscretemodelregionofthestructurebecauseofitsimpactonthedifferencemapinthesolventregion.

•  ThedatasetshouldbereasonablycompleteandwithsufficientresoluIon[i.e.sin(theta)/lambda>0.6].

•  Thereshouldbenounresolvedchargebalanceissuesthatmighteffectthechemistryinvolved(e.g.Thevalencyofametalintheorderedpartofthestructure)

LimitaIons•  ThereportedelectroncountinthesolventregionismeaningfulonlywiththesupplyofacompleteandreliablereflecIondataset.•  TheSQUEEZEtechniquecannothandleproperlycasesofcoupleddisordereffecIngboththemodelandthesolventregion.•  Thesolventregionisassumednottocontainsignificantanomalousscaterers(Friedelsaveraged)

Thankyou!

a.l.spek@uu.nlMoreinfo:www.platonsom.nl(includingthispowerpointpresentaIon)

PleasesendsuggesIonsandexamples(withdata)of(annoying)issuesto: