Introduction to Relativistic Quantum Chemistry This page intentionally left blank Introduction to RELATIVISTIC QUANTUM CHEMISTRY Kenneth G. Dyall Knut Fægri, Jr. 2007 OxfordUniversityPress,Inc.,publishesworksthatfurther OxfordUniversity’sobjectiveofexcellence inresearch,scholarship,andeducation. Oxford NewYork Auckland CapeTown DaresSalaam HongKong Karachi KualaLumpur Madrid Melbourne MexicoCity Nairobi NewDelhi Shanghai Taipei Toronto Withofficesin Argentina Austria Brazil Chile CzechRepublic France Greece Guatemala Hungary Italy Japan Poland Portugal Singapore SouthKorea Switzerland Thailand Turkey Ukraine Vietnam Copyright©2007byOxfordUniversityPress,Inc. PublishedbyOxfordUniversityPress,Inc. 198MadisonAvenue,NewYork,NewYork10016 www.oup.com OxfordisaregisteredtrademarkofOxfordUniversityPress Allrightsreserved.Nopartofthispublicationmaybereproduced, storedinaretrievalsystem,ortransmitted,inanyformorbyanymeans, electronic,mechanical,photocopying,recording,orotherwise, withoutthepriorpermissionofOxfordUniversityPress. LibraryofCongressCataloging-in-PublicationData Dyall,KennethG.,1955– Introductiontorelativisticquantumchemistry/KennethG.Dyall, KnutFaegri,Jr. p.cm. Includesbibliographicalreferences. ISBN:978-0-19-514086-6 1.Quantumchemistry.2.Quantumfieldtheory.I.Faegri,Knut. II.Title. QD462.D932006 541'.28–dc22 2006014117 9 8 7 6 5 4 3 2 1 PrintedintheUnitedStatesofAmerica onacid-freepaper Preface Theemergenceofrelativisticquantumchemistryhasbeenoneofthemoreremarkable developments within computational chemistry over the past decades. Since the early workofDirac,relativityhasalwaysbeenapartoftheoverallquantumchemicalpicture, butithasmostlybeenneglectedonthegroundsthattheeffectswereconsideredsmall andthemethodstotreatthemwerepoorlydevelopedandexpensivetouse.However, as nonrelativistic quantum chemistry became more powerful and accurate, the lower rowsoftheperiodicsystemcamewithinreachofcomputationalstudies,anditbecame clear that relativistic effects had a significant influence on a number of physical and chemical properties. The start of the “modern” era of relativistic quantum chemistry may be traced back to a review article by Pyykkö (1978) and to articles by Pitzer (1979)andbyPyykköandDesclaux(1979). The developments in the field have been well documented through articles in sci- entific journals, conferences and symposia, and review volumes. Unfortunately, this specialist literature is not easily accessible to newcomers to the field. For many years the book by Moss, Advanced Molecular Quantum Mechanics (1973), served as an introduction to the field, but today this book suffers from two major drawbacks—it is out of print and it does not cover the developments of the past three decades. We thereforedecidedthattherewasaneedforabooktofillthegapbetweenthestandard texts on quantum mechanics, which have little if anything on relativity, the advanced texts,whichtreatrelativityindetailbuthavelittleconnectionwithquantumchemistry, andtheliterature,wherethereisalargeamountofboththeoryandapplications. Ourambitionistoprovideamodernintroductiontothefieldofrelativisticquantum chemistry, aimed at the advanced student and the practicing nonspecialist researcher. The material has been divided into five parts. Parts I and II provide the necessary background from classical physics, relativistic quantum mechanics, and group theory. Part III covers the application of these principles to fully relativistic methods for quantum chemistry within a four-component framework. Part IV deals with the main vi PREFACE approximatemethodsthathavebeendeveloped,andPartVtreatstheconsequencesof relativityforchemicalbonding. Thebookisintroductoryinthesensethatitintroducesmanyoftheconceptsneeded for a firm background in relativistic theory and quantum chemistry but without going intoallthedetails.Referencesaregiventofullertreatmentsoftheintroductorymaterial. Ourintentionisnottogiveallthedetailsofproofsthatcanbefoundinotherstandard works on relativistic quantum mechanics, but rather to present the relevant parts for the purpose of constructing a relativistic quantum chemistry. It is also introductory in the sense that the details of many of the methods that are found in the literature are reproduced or elaborated in this book, so that the educated quantum chemist does not have to search through the literature for them. Finally, it is introductory in the sense thatitcontainsdescriptivematerialtodowithrelativisticeffectsonbonding,structure, and energetics of molecules. We have no ambitions of providing an extensive review ofwhathasbecomealargeandquiteheterogeneousfield,norofprovidingahistorical overviewofthedevelopmentofrelativisticmethods.Thespecialistwillprobablyhave nodifficultyidentifyingoneormore“petsubjects”or“keyreferences”thataremissing. Wedoprovideaguidetothereviewliteratureinthefield,andalsointhisrespectthe workisintroductory. The book started out as a set of lecture notes by KGD for a 1995 graduate course givenat(then)OdenseUniversity.Thesenotesdevelopedthroughfurtherpresentations, and the process of turning them into a book was begun in 1996 during a research visit by KGD to the University of Oslo. Since then the writing has been a gradual process,hamperedbytheacademicworkuncertaintiesofoneofusandbytheextensive administrative load carried by the other. The work would not have been possible had it not been for the support of our employers through this period, for KGD: Eloret and Schrödinger; for KF: the University of Oslo. Support by the Research Council of Norway,NASA,andDOEisgratefullyacknowledged,asisthehospitalitybothofus enjoyedduringresearchvisitstotheIRSAMCofUniversitéPaulSabatierinToulouse. Mostofall,thisbookcouldneverhavebeenwrittenwithoutsupportandcriticism fromandfrequentdiscussionswithourfriendsinthefield.Theseincludeinparticular Harry M. Quiney, Luuk Visscher, Trond Saue, Hans Jørgen Aa. Jensen, and Trygve Helgaker, as well as most of the people involved in the DIRAC and MOLFDIR col- laborations. Peter Schwerdfeger did a wonderful job of commenting on a late version of the manuscript. Trond Saue and Luuk Visscher also provided useful comments on substantialpartsofafinaldraft.Otherswhohaveverykindlyreadandcommentedon parts of the manuscript in various stages of completion are Joost v. Stralen, Raimo v. Handel, LeifVeseth, Werner Kutzelnigg, Timo Fleig, and Hans JørgenAa. Jensen. In addition to the direct contributions to the book, there are many who have in one way or another influenced our thinking or contributed to the work on which this book was based.AmongthosenotalreadymentionedwewouldliketoacknowledgePeterTaylor, Jeppe Olsen,Wim Nieuwpoort, and GustavoAucar. Finally, we would like to express our gratitude to Pekka Pyykkö and Ian Grant, both pioneers and leaders in this field, andtoBerndA.Hess,JaapSnijders,JanAlmlöf,andOddGropen,whoarenolonger withus,butwhoallfourinvariouswayshavehelpedusinoureffortsinthisfield. Inourfamiliesthebookhastakenonalmostmythicalstature.Itiswithsomerelief thatwearenowabletopresentafinalproduct.Wedothisingratefulacknowledgement oftheiralmostinfinitetoleranceandsupport. Notation Conventions Wehaveadoptedanumberofconventionsinthisbookinordertomaintainaconsistent, clear,andidentifiablenotation.Asfaraspossiblewehavekepttocommonconventions for symbols and quantities in quantum chemistry. We have also tried to avoid the duplication of symbols where possible. These goals conflict to some extent, so some quantitiesaregivenunconventionalsymbols.Thefollowinglistidentifiessymbolsand typographyusedthroughoutthebook. r,ϑ,ϕ forsphericalcoordinates Ψ forageneralone-particleormultideterminantmany-particlewavefunction Φ foraone-determinantorCSFmany-particlewavefunction Ψfortime-dependent4-spinors θ(t)forthetimepartofaspinororageneraltime-dependentfunction ψfortime-independent4-spinors,thatisΨ=ψθ ΨX fortime-dependent2-spinors,withX =L, S ψX fortime-independent2-spinors,withX =L, S ξ(ϑ,ϕ,τ)for2-spinorangular-momentumfunctions(includingspin) φ,φX for2-spinorsinamodifiedrepresentation χX for2-spinorbasisfunctions µ ψ,ψXτ forscalarfunctions η,α,βforspinfunctions τ,α,β forspinlabels;α andβ alsodenotethespinfunctions,andβ oneofthe Diracmatrices; φ,φXτ forscalarspacefunctions;thesuperscriptsX andτ areforcomponent andspin. χX,χX¯ forscalarbasisfunction µ µ A,Ω boldfaceformatrices a,σ boldfaceforvectors viii NOTATIONCONVENTIONS a Romantypeforfour-vectors ˆ ˆ F,Ω hatsforoperators h forone-electronHamiltonianmatrixelements pq H forN-electronHamiltonianmatrixelements PQ In addition to these font conventions, we have adopted some conventions for the indices of functions. In general, Roman letters are used for Fock space functions (i.e. orbitalsorspinors),whileGreeklettersareusedforbasisfunctions.Specificrangesof lettersareusedasfollows: p,q,r,s,...generalorbitalorspinorindices i,j,k,l,...occupiedorinactiveorbitalorspinorindices a,b,c,d,...virtualorbitalorspinorindices t,u,v,w,...activeorbitalorspinorindices κ,λ,µ,ν,...basisfunctionindices Inthisbookwehaveusedtwosystemsofunits.ThefirstistheSIsystem,whichwe use in the early chapters of the book and in particular for electromagnetic quantities. Factors of c therefore always represent the speed of light and never a conversion factor for magnetic units. The second is the Hartree atomic units system, defined by (cid:1) = e = m = 1. In these units, c ∼ 137. Many physics texts use the system e (cid:1)=e =c =1, since they are dealing with particles of different masses. Our concern isprincipallywiththeelectronandchemistry,andthesizeofrelativisticeffects,which aremeasuredbyc,soHartreeatomicunitsaremoreappropriate.However,tokeepthe connection with SI units and to track quantities that involve the charge, the mass, or spin,thesymbols(cid:1),e,andm≡m areretainedinmuchofthedevelopment,whereas e 1/4π(cid:17) isusuallyomittedforclarity. 0 Therearesomesituationswhere,intheinterestsofclarity,wehaveallowedsome inconsistencyorsacrificedsomerigorofexpression.Wedonotusuallymultiplyscalars by the unit matrix in expressions where the context would demand it—such as where an operator is a combination of scalar and spin-dependent operators or in a matrix expression—and we do not always indicate the rank of the unit matrix or the zero matrixbyasubscript.InsomeplacesthenotationwouldbeoverloadedifI wereused 2 insteadof1,butthematrixnotationistobeinferredfromthecontext. We have also used both Hˆ and hˆ for the one-electron Hamiltonian operator. The latter is used for the free-particle Dirac Hamiltonian where a distinction between it and the full one-electron Hamiltonian is necessary, and is also used in a sum over one-electron Hamiltonians for a single electron. The former is usually used in formal developments, and to represent the total Hamiltonian. In many of the formal devel- opments, the total Hamiltonian is simply the one-electron Hamiltonian, so Hˆ is used. However,fortheone-electronHamiltonianmatrixelements,lowercaseisalwaysused, andfortheN-electronHamiltonianmatrixelements,uppercaseisalwaysused. Contents Notation Conventions, vii I: Foundations 1 Introduction, 3 2 Basic Special Relativity, 6 2.1 Inertial Frames and Newtonian Mechanics, 6 2.2 Relativistic Coordinate Transformations, 7 2.3 Transformation of Lengths and Relativistic Invariants, 9 2.4 Transformation of Velocities, 11 2.5 Transformation of Mass, 13 2.6 Relativistic Energy, 14 2.7 Relativistic Momentum, 15 3 Relativistic Electromagnetic Interactions, 17 3.1 The Maxwell Equations, 18 3.2 Potentials and Gauge Transformations, 19 3.3 The Relativistic Potential from a Moving Charge, 22 3.4 The Potential Experienced by a Moving Charge, 24 3.5 The Interaction of Two Charged Particles, 26