Ab Initio Quantum Chemistry - Methodology and A

CURRENT ISSUE (/CONTENT/CURRENT) ARCHIVE (/CONTENT) NEWS & MULTIMEDIA (/MULTIMEDIA) AUTHORS (/SITE/AUTHORS/INDEX.
XHTML)
ABOUT (/SITE/ABOUTPNAS/INDEX.XHTML) COLLECTED ARTICLES (/SITE/MISC/COLLECTEDPAPERS.XHTML) BROWSE BY TOPIC (/SEARCH)
EARLY EDITION (/CONTENT/EARLY/RECENT) FRONT MATTER (HTTP://FRONTMATTER.PNAS.ORG)
ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica This Issue

www.pnas.org
(/) >CurrentIssue(/content/102/19.toc) >vol.102no.19
>RichardA.Friesner,66486653,doi:10.1073/pnas.0408036102
Ab initio quantum chemistry: Methodology and applications

RichardA.Friesner(/search?author1=Richard+A.+Friesner&sortspec=date&submit=Submit)*
AuthorAffiliations
(/content/102/19.toc)
EditedbyBruceJ.Berne,ColumbiaUniversity,NewYork,NY,andapprovedMarch11,2005(receivedforreviewNovember4, May 10, 2005
vol. 102 no. 19
2004)
Masthead (PDF)
(/content/102/19/local/masthead.pdf)
Abstract (/content/102/19/6648.abstract) Full Text Authors & Info Metrics Table of Contents
(/content/102/19.toc)
Related Content (/content/102/19/6648.full/?tab=related) PDF (/content/102/19/6648.full.pdf)
PREVARTICLE NEXTARTICLE
Abstract (/CONTENT/102/19/6647.SHORT)
(/CONTENT/102/19/6654.SHORT)
This Perspective provides an overview of stateoftheart ab initio quantum chemical methodology and
applications.Themethodsthatarediscussedincludecoupledclustertheory,localizedsecondorderMoller Don't Miss
Plesset perturbation theory, multireference perturbation approaches, and density functional theory. The
accuracyofeachapproachforkeychemicalpropertiesissummarized,andthecomputationalperformance
is analyzed, emphasizing significant advances in algorithms and implementation over the past decade.
Incorporation of a condensedphase environment by means of mixed quantum mechanical/molecular
(http://physics.pnas.org/)
mechanicsorselfconsistentreactionfieldtechniques,ispresented.Awiderangeofillustrativeapplications,
focusingonmaterialsscienceandbiology,arediscussedbriefly. Physics
(http://physics.pnas.org/)
Click here for the
coupledcluster(/search?fulltext=coupled+cluster&sortspec=date&submit=Submit&andorexactfulltext=phrase) latest Physics
densityfunctionaltheory(/search?fulltext=density+functional+theory&sortspec=date&submit=Submit&andorexactfulltext=phrase)content published
in PNAS.
secondorderModlerPlesserperturbationtheory(/search?fulltext=secondorder+Modler%E2%80%93Plesser+perturbation+theory&sortspec=date&submit=Submit&an
Article Tools
mixedquantum/molecularmechanics(/search?fulltext=mixed+quantum/molecular+mechanics&sortspec=date&submit=Submit&andorexactfulltext=phrase)
Article Alerts
Over the past three decades, ab initio quantum chemistry has become an essential tool in the study of Export Citation
atomsandmoleculesand,increasingly,inmodelingcomplexsystemssuchasthosearisinginbiologyand Save for Later
materials science. The underlying core technology is computational solution of the electronic Schrodinger Request Permission
equation given the positions of a collection of atomic nuclei, and the total number of electrons in the (/site/aboutpnas/rightperm.xhtml)
system,calculatetheelectronicenergy,electrondensity,andotherpropertiesbymeansofawelldefined,
automated approximation (a model chemistry). The ability to obtain goodenough solutions to the Share
electronicSchrodingerequationforsystemscontainingtens,orevenhundreds,ofatomshasrevolutionized
theabilityoftheoreticalchemistrytoaddressimportantproblemsinawiderangeofdisciplinestheNobel (/external-ref?
PrizeawardedtoJohnPopleandWalterKohnin1998isareflectionofthisobservation. tag_url=http://www.pnas.org/cgi/content/short/
-+Friesner%20102%20%2819%29%3A%206648+
In its exact form, the electronic Schrodinger equation is a manybody problem, whose computational
(/external-ref?
complexity grows exponentially with the number of electrons, and hence, a brute force solution is
tag_url=http://www.pnas.org/cgi/content/long/1
intractable.HartreeFocktheory,ameanfieldapproach,producesreasonableresultsformanyproperties -+Friesner%20102%20%2819%29%3A%206648+
but is incapable of providing a robust description of reactive chemical events in which electron correlation
hasamajorrole.Thus,akeyproblemhasbeenthedevelopmentoftreatmentsofelectroncorrelationthat
exhibitatractablescalingincomputationaleffortwiththesizeofthesystem.Theconsiderableprogressthat (/external-ref?
hasbeenmadealongtheselinesisoutlinedbelow. tag_url=http://www.pnas.org/cgi/content/long/1
-+Friesner%20102%20%2819%29%3A%206648+
Givenawelldefinedtheoreticalframeworkofapproximation,thenextrequirementisefficientcomputational
implementation.Considerablesophisticationisrequiredtoachieveacceptableaccuracyandefficiencythe (/external-ref?
leadingquantumchemistryprogramsaremillionsoflinesofcomputercode,andmathematicalalgorithmsto tag_url=http://www.pnas.org/cgi/content/long/1
reduceformalscalingofcomputationaleffortwithsystemsizehaveanincreasinglycrucialroleinmeeting -+Friesner%20102%20%2819%29%3A%206648+
the challenge of handling complex system relevant to practical applications. Below, the most important (/external-ref?
computationaladvancesaredescribedbriefly,andtheirimpactontheabilitytoaddresscriticalproblemsis tag_url=http://www.pnas.org/cgi/content/long/1
analyzed. -+Friesner%20102%20%2819%29%3A%206648+
Thetreatmentoflarge,condensedphasesystems(e.g.,proteinsinaqueoussolution)entirelybyab initio (/external-ref?

methodsisextremelyexpensivecomputationally.However,itisoftenthecasethatarelativelysmallregion tag_url=http://www.pnas.org/cgi/content/long/1
of the system can be modeled at the ab initio quantum chemical level, whereas the remainder can be -+Friesner%20102%20%2819%29%3A%206648+
treatedmoreapproximately[e.g.,bymeansofmolecularmechanics(MM)orcontinuumsolvationmodels]. Published online before print
The technologies for coupling quantum chemical methods to these alternative types of models [mixed May 3, 2005, doi:
quantum mechanics (QM)/MM and selfconsistent reaction field (SCRF) approaches] have become an 10.1073/pnas.0408036102
PNAS (Proceedings of the National
essential component of the theoretical arsenal, enabling realistic modeling of even the most complex Academy of Sciences) May 10,
molecular structures. The key issues in these types of calculations, as well as a discussion of recent 2005 vol. 102 no. 19 6648-6653
successfulimplementations,arepresentedbelow. Classications
Chemical Theory and
Computation Special Feature
Perspective (/search?
Last,aremarkablerangeofapplicationshaveappearedinthepastdecade,impactingnearlyeveryaspect Perspective (/search?
tocsectionid=Perspective&sortspec=date&submit=Subm
of chemistry, biology, and materials science. Space permits discussion of only a small fraction of these
calculations, which are presented below as examples of what is possible at the current state of the art.
Continued improvements in the theory and implementation, and reduction in the cost/performance of Other Articles
computing,ensurethatdramaticprogresswillcontinuetobemadeintheyearstocome,advancingtoward Citing This Article
theultimategoaloftheoryachievingfullpartnershipwithexperimentasbothanexplanatoryandpredictive Google Scholar
methodology.
PubMed
Similar to This Article
Quantum Chemical Theory SUBMITANARTICLE

(HTTP://WWW.PNASCENTRAL.ORG)
TwohighlyproductiveapproachestosolutionoftheelectronicSchrodingerequationhavearisenoverthe
past 50 years. Wavefunctionbased approaches expand the electronic wavefunction as a sum of Slater
determinants, the orbitals and coefficients of which are optimized by various numerical procedures.
HartreeFock theory is the simplest method of this type, involving optimization of a single determinant
however,asmentionedabove,itsusefulnessislimitedbecauseofcompleteneglectofelectroncorrelation.
We discuss three types of correlated wavefunctionbased approaches: secondorder MollerPlesset
perturbation theory (MP2), with an emphasis on the localized version of this methodology (1) methods
based on the coupledcluster ansatz, focusing on the widely used CCSD(T) (coupled cluster with single,
double, and triple perturbative excitations) variant (2) and multireference perturbation methods, such as
CASPT2(completeactivespacewithsecondorderperturbationtheory)(3).Eachoftheseapproximations
hasadifferentcomputationalscalingwiththenumberofelectronsandis(arguably)themethodofchoice
fordifferenttypesofproblems.
Thesecondclassoftheoreticalapproachesarebasedondensityfunctionaltheory(DFT),which,following
theHohenbergKohntheorem(4),mandatesexpressionofthetotalenergyofthesystemasafunctionalof
theelectrondensity.Becausetheelectrondensitydependsononlythreecoordinates(asopposedtothe3N
coordinatesofNelectrons),thecomputationaleffortrequiredtosolvetheequationsofDFTiscomparable
with that required for HartreeFock theory, thus rendering DFT highly attractive from the point of view of
computational implementation. However, the correct functional of the energy is unknown and has to be
constructedbyheuristicapproximation.Initialfunctionals,basedprincipallyonbehavioroftheelectrongas
(5), were lacking in the accuracy required for chemical applications. Breakthroughs over the past two
decades(610)haveledtothedevelopmentoffunctionalscapableofremarkableaccuracyandbreadthof
applicabilityacrosstheperiodictable,althoughitisimportanttonotethatthereremainlimitationsaswell.
Atpresent,therearetwoprincipalclassesoffunctionalsthathavebeenextensivelydeployedandtestedin
largescale applications as well as small molecule benchmarks: gradientcorrected (7,8, 10) (e.g., BLYP
ref. 7), and hybrid (7, 9) (e.g., B3LYP ref. 7), functionals. Gradientcorrected functionals begin with the
localdensityapproximationbutaddtermsinvolvingthegradientoftheelectrondensity.Hybridfunctionals
also incorporate gradient corrections but add an empirically fitted admixture of exact HartreeFock
exchange.Althoughtherearedifferencesbetweentheindividualfunctionalsineachclass,thedifferences
between classes are substantially larger hence, we focus below on one example from each class (BLYP
andB3LYP,ascitedabove)asrepresentativemembers.
Coupledclustermethodsarethemostcomputationallyexpensivebutalso,inprinciple,themostaccurateof
the approaches that we consider. By an exponential ansatz for the coupling of correlated electron pairs,
highly excited determinants are incorporated into the wavefunction without the combinatorial explosion of
effort implied by nave configurationinteraction approaches at present, the CCSD(T) approach (in which
tripleexcitationsaretreatedperturbatively)providesthebesttradeoffof accuracy and efficiency (2).The
formal scaling with number of electrons N of CCSD(T) is N7 for technical reasons, it has been difficult to
improve on this behavior in practice. Thus, calculations are limited to small to mediumsized molecules,
andparallelsupercomputersarerequiredattheupperendofthisrange.
In return for this high computational cost, the level of accuracy that one can expect from a benchmark
CCSD(T) calculation for atomization energies (and other less demanding thermochemical quantities) of
smallorganicmoleculesisontheorderofafewtenthsof1kcal/mol(1cal=4.18J)(11).Equilibriumbond
lengths and bond angles are typically highly accurate (12). Indeed, the level of precision in achieved in
geometryisunnecessaryforenergeticaccuracyhence,lessexpensivemethodsareoftenusedtocompute
structures.Excitedstatemethodshavebeendevelopedandhavedisplayedencouragingperformance(13).
Turning next to MP2 methods, the computational cost is reduced considerably. Whereas formal scaling is
N5,numericalmethodsdevelopedinthepastdecade(discussedfurtherbelow)yieldsubstantialreductions
inscalingbothformallyandasymptotically(aswellassignificantabsolutecentralprocessingunitreductions
for smaller systems), enabling systems containing a few hundred atoms to be treated routinely. The
performance of MP2 methods for properties involving making or breaking electron pairs is substantially
inferior to that of CCSD(T) and also not as good as the best DFT methods, although recent results have
comeclosertoDFTbenchmarks(14).However,formostsystems,MP2doesextremelywellfornonbonded
interactionsandinternalconformationalenergetics,typicallydeliveringaccuracywithin0.3kcal/molifone
extrapolatestothebasissetlimit(15,16,23). A recent interesting exception has been noted for stacking
interactionsofthebenzenedimer,whereCCSD(T)correctionsarerequiredtoavoiderrorsintherangeof
0.51.0kcal/mol(17)moreinvestigationwillberequiredtoenumerateotheroutliersofthistype.
Multireferenceperturbationmethods,suchasCASPT2(3),aremoredifficulttousethaneitherCCSD(T)or
MP2 in that they require a careful definition of the active space of electrons to be treated at the
multireference level to be effective (electrons not in the active space are treated perturbatively). The
computationalcostincreasesexponentiallywiththesizeoftheactivespace.Nevertheless,therearesome
problems where CCSD(T) methods have severe difficulties because of the fact that the electronic
wavefunction is no longer dominated by a single (HartreeFock) determinant an obvious example is
generating a complete potential energy curve in the process of breaking a bond (18). In this and other
cases, CASPT2 provides a reasonable alternative enabling incorporation of multireference character from
thestart,anditmaybethemethodofchoiceinmanysuchcases.
BothgradientcorrectandhybridDFTmethodsachieveexcellentresultsforequilibriumgeometries(19,20)
acrosstheperiodictable.Foratomizationenergies,thereisacleardifferenceinperformanceforexample,
theBLYPfunctionalhasanaverageerrorof7.09kcal/molfortheG2setof148smallmolecules,whereas
theB3LYPfunctionalhasanaverageerrorof3.11kcal/molforthesamedataset(21).Fortransitionmetal
containingsystems,theexperimentaldataarelessreliable,andtherearemanydifferenttypesofbonding
to consider. However, in typical situations, the performance of B3LYP is clearly respectable, with average
errorsforbondenergiesofsmalltransitionmetalcomplexesintherangeof35kcal/mol(82).Thus,hybrid
DFT provides a quantitative means of investigating reactive chemistry in large systems, albeit with large
errorsforasmallnumberofoutliers.
However, DFT is inferior to basis set limit MP2 in achieving high precision for energy differences in which
bondsarenotmadeorbroken,includinghydrogenbondingandconformationalenergetics(22,23).Itisnot
thecasethatDFTyieldsbadresultsfortheseproblems(indeed,itisclearlyanimprovementoverHartree
Fock)itsimplyislessaccurate(e.g.,by0.51kcal/molforhydrogenbondingenergies)thanMP2when
averagedovermanycases.Furthermore,theabilityofDFTtomodeldispersion(vanderWaals)interaction
has not yet been established (15) again, MP2 represents a qualitatively superior alternative, which can
itself be applied to large systems at a quantitatively (but not qualitatively) larger computational cost. Last,
thetimedependentversionofDFT(TDDFT)inmanycasesprovidesreasonableresultsforexcitedstates
(2426) and is considerably less expensive than alternatives such as CASPT2 and CCSD(T)based
excitedstatemethods.
Wecansummarizetheaboveresultsbyconsideringwhichquantumchemistrymethodologyisthecurrent
methodofchoiceforvarioustypesofcomputationalchemistrymodelingproblems.Foraccuratestudiesof
small molecules in the gas phase, CCSD(T) and related methods are the approach of choice, perhaps
combined with less expensive methods to carry out geometry optimization and/or obtain basis set
convergence.Forinvestigationofreactivechemistryinmediumtolargesystems,DFTisatpresentthebest
approach,insomecasesitispossibletocheckatruncatedmodeloftheactualsystembyusingcoupled
clusterbasedmethods(althoughtypicallynotatabenchmarklevel).MP2methodsaregenerallyusefulfor
studying nonbonded interactions in both small and large systems and in developing molecular mechanics
forcefields(discussedfurtherbelow)thesetypesofcalculationsarehighlyrelevantinbiologyandmaterials
science problems. DFT can be used also in the investigation of conformational and hydrogen bond
interactions,providedthattheaccuracylimitationsareunderstood.Last,thereareanumberofchoicesfor
excited states, none of which are completely satisfactory but several of which have yielded respectable
results.Thisareashouldbenefitsubstantiallyfromdevelopmentefforts.
Computational Implementation
Below, we briefly discuss stateoftheart computational implementation for each of the four quantum
chemicalmodelspresentedabove,beginningwiththeleastexpensive(DFT).Therearetwofundamentally
different numerical approaches that are used currently for solving the KohnSham equations (27) as is
requiredinlargescaleDFTcalculations.Thefirstapproach,arisingprimarilyfromthephysicscommunity,
represents the electron density orbitals by a plane wave basis set (25). To date, efficient use of such
methodshasbeenpossibleonlyforgradientcorrected(asopposedtohybrid)DFTfunctionals,becausethe
matrix elements of the exact exchange operator are difficult to evaluate in a plane wave basis set. The
approach is most naturally applied to condensedphase systems (periodic solids or liquids) in which the
imposition of periodic boundary conditions (implicit in the use of plane waves, which are periodic) is
appropriate.
Planewave methods have been implemented by two different algorithmic approaches. CarParinello
methods(2831) mandate performing abinitio molecular dynamics, evolving the system by means of an
extended Lagrangian formalism. This type of approach is most useful in studying dynamics in liquids and
moltensolids,aswellasinlocatingstationarypointsbymeansofsimulatedannealinginsystemsinwhichit
isdifficulttomakeagoodinitialguess.Conjugategradientbasedoptimizationapproaches,implementedin
programs such as CASTEP(30) and VASP(31) are typically used to study solid state and surface science
problems.
Thealternativetoplanewavemethodsistheuseoflocalized,atomcenteredGaussianbasissets,ashas
been used for many years in solving the HartreeFock equations. Indeed, most localized basis
implementations of DFT were developed by modifying existing HartreeFock codes, as was done, for
example,intheGaussianseriesofprograms(32).TheuseofGaussianbasisfunctionsallowshybridDFT
functionalstobetreatedinstraightforwardfashion.
Over the past decade, the following advances have increased the efficiency of Gaussianbased DFT
methodssubstantially:
1.Numerical quadrature methods were developed to perform integrals accurately over the exchange
correlationfunctional,whichcannotbehandledanalytically(33,34).
2.ThecomputationofcostoftheCoulombtermwithsystemsizecanbereducedfromN2toNbytheuse
ofmultipoleexpansions(3537).
3.Two numerical approaches to reducing the formal scaling (as opposed to asymptotic scaling) of the
Coulomb and exchange terms have been developed. The first approach, pseudospectral methods (37,
38),wereintroducedmorethantwodecadesago,andarerobustlyimplementedinthe JAGUARprogram
(39). The second approach, resolution of the identity, is technically different in its details but similar in
spirit. Resolutionoftheidentity methods (36) are implemented efficiently in the TURBOMOLE program
(40). Both approaches in their current form deal effectively with the Coulomb term pseudospectral
methods handle exact exchange (and, thus, hybrid DFT functionals) with equal effectiveness. Both
approaches provide substantial central processing unit reductions for medium to large systems as
comparedwithconventionaltwoelectronintegralmethodology.
Two fundamental advances have dramatically reduced the computational cost, and scaling with system
size,ofMP2calculationsascomparedwiththeoriginalformulations.Thefirstadvanceisthedevelopment
of localized MP2 (LMP2) theory several decades ago(1). In this approach, the HartreeFock orbitals are
localizedtobondsandlonepairs,andelectronsarecorrelatedstartingfromtheseorbitalsratherthanthe
delocalized canonical HartreeFock orbitals. In principle, this methodology should lead to substantial
reductionsincomputationaleffort,becauseanelectroninalocalizedorbitalcanbecorrelatedbyexcitation
intoalocalizedvirtualspace.Inpractice,therearetechnicaldifficultiesinrealizingthesetheoreticalgainsby
using standard transforms of analytical twoelectron integrals. However, the combination of localized
methodswithpseudospectral(23)(and,morerecently,resolutionoftheidentityref.41)methodsenables,
as in the case of DFT, a reduction of the formal scaling, in this case from N5 to N3 (asymptotically
approachingN2andevenNatverylargedistances),andtheconcomitantabilitytotreatlargesystemswith
computational effort that is a small multiple of DFT, at least for single point calculations. Efficient LMP2
methodsareimplementedinthe JAGUAR(39)and QCHEM(42)programs,aswellasincodesdescribedin
ref.41.
It has proven to be more difficult to reduce the formal, or asymptotic, scaling, of coupled cluster and
multireference perturbation methods, although this area is of considerable active research interest
(discussed further in Conclusion and Future Directions). Details of the technology in both cases have
improved substantially, and application of these methods to small to mediumsized molecules in the gas
phaseisnowroutine,althoughrequirementsforcomputationalpowerrisesteeplyasthesizeofthesystem
increases. Efficient coupledcluster methods are implemented in the GAUSSIAN (32) and QCHEM (42)
programstheMOLCASprogram(43)containsanoptimizedCASPT2methodology.
Allfourfundamentalapproacheshavebenefitedfromthetremendousadvancesincomputationalhardware
overthepastdecade.LocalizedbasisfunctionDFTandlocalMP2calculationscanbeperformedroutinely
for systems with hundreds of atoms on personal computers, which offer unparalleled cost/performance
savings.Planewavemethodsrequiretightlycoupledparallelmachinesiflargeunitcellsaretobehandled,
increasing their actual cost in dollar terms significantly beyond what would be assessed from reported
central processing unit times. Largescale coupledcluster and multireference calculations also require
parallelmachinesiftheupperendoftheaccessiblerangeofmoleculesistobeaddressed.
Last, a crucial issue in obtaining accurate quantum chemical results is convergence of desired quantities
(e.g.,atomizationenergies)withbasissetsize,particularlyforwavefunctionbasedmethods.Oneapproach
is to employ composite methods in which each challenging aspect of an ab initio calculation (geometry
optimization, basis set convergence, inclusion of highlevel correlation) can be individually optimized with
themostcosteffectivetechnology,andtheresultscombined(typicallyviaasimplelinearscheme,oftenwith
anadditivecorrectiontomimictheeffectofhighangularmomentumfunctions)toyieldalevelofaccuracy
notfarfromthatobtainablefromemployingthemostexpensivetypeofcalculationacrosstheboard.Typical
methods,suchasthewidelyusedG2(21) and G3 (44) theory, incorporate a coupledcluster component,
andarecapableofachieving12kcal/molaverageerrorsinatomizationenergiesatacostfarsmallerthan
benchmark CCSD (T) calculations with large basis sets. A number of alternative approaches of this type
havealsobeendeveloped,someofwhichappeartoyieldimpressivecost/performanceimprovements(45).
An alternative approach is to employ extrapolation methods, which enable the effects of very large basis
setstobeinferredfromaseriesofsmallerbasissetcalculations(46,47).
Condensed-Phase Environments: SCRF and Mixed

Quantum/Classical Methods
Theremarkableadvancesinabinitioquantumchemicalmethodologyandimplementationdescribedabove
shouldnotobscurethefactthatsuchcalculationsarestillveryexpensivecomputationallyascomparedwith
more approximate models. Evaluation of the energy of a configuration of a 50atom molecule by using a
molecular mechanics potential function requires substantially <1 ms on a modern personal computer, as
compared with the several minutes needed even at the DFT level for a singlepoint electronic structure
calculation. The huge difference in multiplicative coefficient persists, regardless of asymptotic scaling, as
one goes to larger systems this fact makes quantum chemical calculations for systems containing
thousands or tens of thousands of atoms impractical, particularly if significant numbers of configurations
need to be studied. Furthermore, such heroic efforts are also unnecessary in many applications if the
reactive event of interest is localized (as is always the case in biological systems and often the case in
materialsproblems),thentreatmentofallbutasmallreactiveregionviaaclassicalforcefieldis,ingeneral,
an excellent approximation, reproducing the structural and electrostatic effects of the environment on the
reactivechemistrywithareasonabledegreeoffidelity.
Mixed QM/MM methods (48) provide a means of combining quantum chemical and molecular mechanics
models in a way that preserves the integrity of the overall energy function. The key issue is the interface
betweentheQMandMMregions.IftheQMandMMregionsarenotcovalentlylinked,thentheproblemis
relativelystraightforwardintroductionofvanderWaalsparametersontotheQMatoms,andoptimizationof
theseparameterstofitfullyQMhydrogenbondingenergies(49),willyieldamodelthatdoesareasonable
job of reproducing the fully QM results over the entire phase space. However, when there are covalent
connectionsbetweenregions(forexample,whenpartofaproteinactivesiteneedstobetreatedattheQM
level),definitionoftheinterfacerequiresmoresophisticatedtechnology.
Overthepastdecade,twoapproacheshaveemergedwhichhaveenabledsignificantprogresstobemade
indevelopingaccurateQM/MMinterfaces.Thefirstapproachemployslinkatoms,whicharefictitiousatoms
thatareusedtocaptheQMandMMregions(5052).Thesecondusesfrozenlocalizedorbitals,derived
from model molecules, at the interface between a QM and MM atom (49,53). The quality of the results
dependsnotonlyonthefunctionalformoftheinterface(probably,bothmethodsarecapableofachieving
good results if appropriately optimized) (54) but also on the details of the parameterization. The most
accurate results are achieved by making the parameters explicitly depend on the local chemical
environmentforexample,inref.49, parameters are developed for each amino acid side chain, with cuts
between the carbon and carbon as well as in the peptide backbone. Such specific parametrization is
capableofyieldingarobustmodelwithaverageerrorsontheorderof0.51.0kcal/molascomparedwitha
fully quantum chemical results, for various tests (conformational energetics, deprotonation energies, and
morecomplexassessmentsinwhichthesizeoftheMMregionisincreasedsystematically).Forthemost
part, these errors are smaller than the intrinsic errors in the quantum chemical methods and, hence, are
unlikelytobethelimitingfactorinachievinghighaccuracy.
Inprinciple,thecomputationalcostofaQM/MMsinglepointcalculationshouldbenearlyidenticaltothatof
a QM calculation of the same size as the QM region because the MM region contributes negligibly. For
stationary point optimization, a problem arises because of the fact that a large system, containing
thousands of atoms, will ordinarily take many more gradient cycles to optimize than a system containing
hundreds of atoms. However, this problem can be solved by adiabatically minimizing the MM region after
everyQMgeometrystep(49,55,56).ThisstrategyrendersaQM/MMoptimizationthatisapproximatelyas
expensiveasthecorrespondingoptimizationoftheQMregionalone,makingsuchoptimizationsfeasiblefor
largesystems.
TheQM/MMapproachisideallysuitedtostudyingreactionsinproteins,disorderedsolids,largemolecules,
and other systems that have well defined structures and are heterogeneous in character. However, for
processesinsolution,itisnecessarytoaverageextensivelyoverthepositionsofthesolventmoleculesto
properly take into account environmental effects. Thus, QM/MM approaches are possible but
computationally very expensive. An alternative that delivers reasonable accuracy at a much lower
computational cost is the use of continuum models to describe the solvent. The coupling of continuum
modelswithquantumchemicalcalculationsusingSCRFapproaches(57,58) has been implemented over
the past decade in a number of widely available ab initio quantum chemistry programs (e.g., GAUSSIAN,
JAGUAR , and QCHEM). By alternating cycles of continuum PoissonBoltzmann (59) (or approximations of
suchcyclesref.60) and quantum chemical computations, the interacting reaction field of the solvent and
chargedistributionofthesoluteareiteratedtoconvergence.Geometryoptimizationofstructuresinsolution
isalsopossiblebyusinganalyticalgradientmethods(61,62).Toachieveaccurateresultsforsolvationfree
energies, parameters of the continuum model must be fit to small molecule experimental data (60, 63).
SCRFbased methods can also be used effectively for pKa prediction (64). Last, it is possible to combine
QM/MM and continuum solvation approaches, although few articles combining these approaches have
beenpublished.
Applications
Forsmallmoleculesinthegasphaseandinsolution,abinitioquantumchemicalcalculationscanprovide
results approaching benchmark accuracy (11), and they are used routinely to complement experimental
studies. A wide variety of properties, including structures (12), thermochemistry (11) (including activation
barriersref.65),spectroscopicquantitiesofvarioustypes(13,66),andresponsestoexternalperturbations
(67), can be computed effectively. As discussed above, SCRF methods (or simply ignoring the solvent
entirely,anapproximationthatissometimesacceptable,particularlyinnonpolarsolventsorwhenaquantity
thatisinsensitivetothedielectricoftheenvironmentisbeingcomputed)enablearelativelystraightforward
extensionofgasphasequantumchemicalmethodstoobtainingresultsformoleculesinsolution.
Materialsscienceapplicationshavegrownexponentiallyoverthepastdecadeandnowinvolveexploration
ofhighlycomplexstructuresandchemistries.Bothbulkandsurfacepropertiescanbecomputedforsolids
(68),inmanycasesyieldingexcellentagreementwithexperiment.Bandgapsandopticalpropertiesofsolids
have been addressed successfully by DFTbased methods (69). The interactions of small molecules with
surfaceshasalsobeenthesubjectofextensiveinvestigation,withthefocusonstructures,bindingenergies,
and catalytic chemistry (7072). Calculations of this type are central to practical applications involving
industrially relevant catalytic processes, semiconductor processing, and modeling of conductivity for
microelectronicsapplications.Anothersignificantareaofboththeoreticalandpractical interest is modeling
photochemicalapplications(e.g.,solarenergyconversionbydyesadsorbedontotitaniumdioxideref.20),
as implemented in the Graetzel cell (73), which has achieved a conversion yield very close to that of
amorphoussilicon.
Applicationstonanotechnologyarerelativelyrecentbutrepresentafocusofincreasinginterest.Amongthe
systems that have been investigated are carbon nanostructures (buckyballs and nanotubes) (25, 7476)
and semiconductor quantum dots (26, 77, 78). The effects of quantum confinement and structural
modificationimposedbyvariousnanostructuresonchemicalandelectricalpropertiescanbeinvestigatedin
arigorousfashion,whichwillbeessentialasattemptsaremadetousethesematerialsinvarioustypesof
devices.
Biological applications have focused principally on the modeling of enzymatic catalysis and activesite
chemistry (7988), although interesting investigations of other phenomena, such as cooperativity in
backbone hydrogen bonding (89) and modeling of sheet formation propensities by a periodic DFT
calculation(90), have been performed recently. The use of QM/MM methods permits incorporation of the
full protein environment, which is crucial in an important subset of cases, as has recently been shown in
cytochrome P450 (79), for example. A wide variety of enzymatic systems have been investigated, with
quantitativecomparisonwithexperimentalstructuralandthermodynamicdatayieldinganencouraginglevel
ofagreementinthemostaccuratemodelingefforts.Abinitiocalculationsprovideawealthofdetailthatis
not available from experiment and a degree of confidence in the results that is not available from more
empirical approaches. In systems such as methane monooxygenase, in which a substantial number of
carefulcalculationshavebeenperformedbyseveralgroups(81,9196)andclosecontactwithexperiment
hasbeenachievedformanyaspectsofthecatalyticprocess,acomprehensivepictureofthefunctioningof
theenzymeisbeginningtobeassembledbymeansoftheinteractionoftheoryandexperiment.Last,the
couplingofabinitioquantumQM/MMmethodswithsimulationandproteinstructurepredictiontechniques
permitsinvestigationofeventsinwhichreactivechemistryiscoupledtosubstantialconformationalchanges,
suchasthecatalyticloopmotionintriosephosphateisomerase(82).
Last,akeyapplicationofquantumchemicalmethodsisinthedevelopmentofmolecularmechanicsforce
fields(9799).Thequalityofforcefieldshasimprovedenormouslyinthepasttwodecades,andaprincipal
reasonhasbeentheabilitytofitparameterstotheresultsofhighlevelabinitiocalculationsforlargerand
morecomplexmodelsystems.Forfixedchargeforcefields,torsionalparameterscanbefittocomputations
of conformational energy differences of model molecules, whereas charge distributions can be fit to
quantum chemical electrostatic potentials. Dimerinteraction energies have typically been used as a
heuristicguidetomodelingnonbondedinteractionshowever,forapolarizableforcefield,thesequantities
canbefitdirectlyaswell,becausethemodelissupposedtorepresentnonbondedinteractionsaccuratelyin
both the gas phase and the condensed phase (100, 101). Polarizability parameters can also be fit to
quantum chemical data, although there are some issues associated with basis set size that must be
consideredwhendoingso(100).
At this point, it is useful to summarize how advances in theory and computational implementation have
enabledthetremendousgrowthintheimportantapplicationsdescribedabove.Smallmoleculesinthegas
phasearenowtypicallyaddressedbyhighlevelmethodssuchasCCSD(T),whichinmanycasesaremore
accuratethanexperiment.Assystemsincreaseinsize,onehastoatsomepointswitchovertoDFTand/or
MP2. Multireference perturbation methods are typically applied when there are particular difficulties
associatedwithneardegeneracies,orforpropertiessuchasexcitedstatesforwhichspecificmultireference
approaches (such as CASPT2) have proven to be superior to DFT in a wider subset of cases. The
overwhelmingmajorityoflargescalematerialscienceandbiologicalapplicationshavebeenperformedwith
DFTthisstateofaffairsismandatedbythelargesizeofthesystemsthatarebeingconsidered,theneedin
somecasesforperiodicboundaryconditions,andtheavailabilityofQM/MMandSCRFmethodsprincipally
for DFT based approaches. Last, for forcefield development, localized MP2 is the method of choice
because of its acceptable treatment of dispersion interaction and higher accuracy for conformational
energies and hydrogen bonding energies than DFT while maintaining a similar (if somewhat larger)
computationalcost.
Conclusion and Future Directions

Overthenextdecade,weexpecttoseesignificantincreasesinaccuracyinallfouroftheprincipalquantum
chemical directions described above. Progress in developing DFT functionals has been difficult since the
breakthroughs of the early 1990s however, a number of promising approaches (102105) and key
systematic difficulties, such as problems in predicting barrier heights of small molecule radical reactions
(106),havebeenidentified.Coupledclusterandmultireferencealgorithmsincludeattemptstouselocalized
reference states that potentially could lead to the same huge reductions in scaling with system size that
have been realized with localized MP2 (107110). Continued reductions in the cost/performance of
computing and improvements in algorithmic details should continue to yield shorter time to solution for
increasinglylargersystems.
Similarly,improvementcanbeexpectedintreatmentofthecondensedphaseenvironment.Optimizationof
the accuracy of continuum solvation methods is far from a solved problem furthermore, there is some
evidence that the inclusion of a small number of explicit water molecule can improve results (111), but
methodsofthistypemustbeformulatedverycarefullytoavoiddoublecounting.QM/MMmethodscanbe
made more accurate, robust, costeffective, and easy to use. Last, sampling algorithms in the condensed
phase are crucial for many largescale applications, and significant advances can be expected as more
ambitiousproblemsareaddressed.
Evenifthetechnologyweretostandstill,onewouldexpectalargenumberofimportantnewapplicationsto
becarriedoutoverthenextdecadewithadvancesintheory,software,andcomputationalhardware,larger
datasetsandsystemsofincreasingsize,willbeamenabletostudy.However,themostexcitingpossibilityis
thattheparalleladvancesintheoryandexperimentwillenablefullyexplanatoryandpredictivemodelstobe
Navigate This Article
constructed for the complex, condensedphase processes that govern most of the natural world. Ab initio
quantum chemical methods are not the only technology that will be a component of such a development, Top
buttheysurelyareanessentialone,asdemonstratedbytheprogressandprospectsoutlinedabove.
Abstract
Quantum Chemical Theory

Acknowledgments
Computational
ThisworkwassupportedinpartbyNationalInstitutesofHealthGrantGM40526andDepartmentofEnergy Implementation
GrantDEFG0290ER141.
Condensed-Phase
Environments: SCRF and
Mixed Quantum/Classical
Footnotes Methods
*Email:rich@chem.columbia.edu(mailto:rich@chem.columbia.edu). Applications
Thispaperwassubmitteddirectly(TrackII)tothePNASoffice. Conclusion and Future

Directions
Abbreviations:MM,molecularmechanicsQM,quantummechanicsSCRF,selfconsistentreactionfieldMP2,second
orderMollerPlessetperturbationtheoryDFT,densityfunctionaltheory. Acknowledgments
Copyright2005,TheNationalAcademyofSciences
Footnotes
References References
1. Sabeo,S.&Pulay,P.(1991)Annu.Rev.Phys.Chem. 44,213236.
CrossRef(/externalref?access_num=10.1146/annurev.pc.44.100193.001241&link_type=DOI)
2. Bartlett,R.J.(1989)J.Phys.Chem. 93,16971708.
CrossRef(/externalref?access_num=10.1021/j100342a008&link_type=DOI)
3. Roos,B.O.,Andersson,K.,Fulscher,M.P.,Malmqvist,P.A.,SerranoAndres,L.,Pierloot,K.&
Merchan,M.(1996)Adv.Chem.Phys. 93,219331.
4. Hohenberg,P.&Kohn,W.(1964)Phys.Rev. 1326,B864B871.
5. Sham,L.J.&Kohn,W.(1964)Phys.Rev. 145,561567.
6. Kohn,W.,Becke,A.D.&Parr,R.G.(1996)J.Phys.Chem. 100,1297412980.
CrossRef(/externalref?access_num=10.1021/jp960669l&link_type=DOI)
WebofScience(/externalref?access_num=A1996VA59500020&link_type=ISI)
7. Johnson,B.G.,Gill,P.M.W.&Pople,J.A.(1993)J.Chem.Phys. 98,56125626.
CrossRef(/externalref?access_num=10.1063/1.464906&link_type=DOI)
8. Becke,A.D.(1998)Phys.Rev.A38,30983100.
9. Becke,A.D.(1993)J.Chem.Phys. 98,56485652.
WebofScience(/externalref?access_num=A1993KV99700048&link_type=ISI)
10. Lee,C.T.,Yang,W.T.&Parr,R.G.(1988)Phys.Rev.B37,785789.
CrossRef(/externalref?access_num=10.1103/PhysRevB.37.785&link_type=DOI)
WebofScience(/externalref?access_num=A1988L976200011&link_type=ISI)
11. Martin,J.M.L.&deOliveira,G.(1999)J.Chem.Phys. 111,18431856.

WebofScience(/externalref?access_num=000081698700008&link_type=ISI)
12. Thomas,J.R.,Deeleeuw,B.J.Vacek,G.,Crawford,T.D.,Yamaguchi,Y.&Schaefer,H.F.(1993)J.
Chem.Phys. 99,403416. CrossRef(/externalref?access_num=10.1063/1.465764&link_type=DOI)
13. Stanton,J.F.&Bartlett,R.J.(1993)J.Chem.Phys. 98,70297039.
14. Grimme,S.(2003)J.Chem.Phys. 118,90959102.

15. Tsuzuki,S.,Uchimaru,T.&Tanabe,K.(1998)Chem.Phys.Lett. 287,202208.

CrossRef(/externalref?access_num=10.1016/S00092614(98)001596&link_type=DOI)
16. Tsuzuki,S.,Uchimaru,T.,Matsumura,K.,Mikami,M.&Tanabe,K.(1999)J.Chem.Phys. 110,

1190611910. CrossRef(/externalref?access_num=10.1063/1.479130&link_type=DOI)
17. Sinnokrot,M.O.,Valeev,E.F.&Sherrill,C.D.(2002)J.Am.Chem.Soc. 124,1088710893.
CrossRef(/externalref?access_num=10.1021/ja025896h&link_type=DOI)
Medline(/externalref?access_num=12207544&link_type=MED)
18. Dutta,A.&Sherrill,C.D.(2003)J.Chem.Phys. 118,16101619.

19. Murray,C.W.,Laming.G.J.,Handy,N.C.&Amos,R.D.(1992)Chem.Phys.Lett. 199,551556.

CrossRef(/externalref?access_num=10.1016/00092614(92)85008X&link_type=DOI)
20. Redfern,P.C.,Zapol,P.,Curtiss,L.A.,Rajh,T.&Thurnauer,M.C.(2003)J.Phys.Chem.B107,
1141911427. CrossRef(/externalref?access_num=10.1021/jp0303669&link_type=DOI)
21. Curtiss,L.A.,Raghavachari,K.,Redfern,P.C.&Pople,J.A.(1997)J.Chem.Phys. 106,1063
22. Tsuzuki,S.&Luthi,H.P.(2001)J.Chem.Phys. 114,39493957.
23. Murphy,R.B.,Beach,M.D.,Friesner,R.A.&Ringnalda,M.N.(1995)J.Chem.Phys. 103,1481

24. Stratmann,R.E.,Scuseria,G.E.&Frisch,M.J.(1998)J.Chem.Phys. 109,82188224.
25. Bauernschmitt,R.,Ahlrichs,R.,Hennrich,F.H.&Kappes,M.M.(1998)J.Am.Chem.Soc. 120,

50525059. CrossRef(/externalref?access_num=10.1021/ja9730167&link_type=DOI)
26. Baroufalis,C.S.,Zdetsis,D.&Grimme,S.(2001)Phys.Rev.Lett. 87,276402.
Medline(/externalref?access_num=11800900&link_type=MED) Navigate This Article
27. Kohn,W.&Sham,L.J.(1965)Phys.Rev. 140,1133. Top
CrossRef(/externalref?access_num=10.1103/PhysRev.140.A1133&link_type=DOI)
WebofScience(/externalref?access_num=A19657000000015&link_type=ISI) Abstract
28. Car,R.&Parrinello,M.(1985)Phys.Rev.Lett. 55,24712474.
CrossRef(/externalref?access_num=10.1103/PhysRevLett.55.2471&link_type=DOI)
Computational
WebofScience(/externalref?access_num=A1985AUN4900027&link_type=ISI)
Implementation
29. Payne,M.C.,Teter,M.P.,Allan,D.C.,Arias,T.A.&Joannopoulos,J.D.(1992)Rev.Mod.Phys. 64,
10451097. CrossRef(/externalref?access_num=10.1103/RevModPhys.64.1045&link_type=DOI) Condensed-Phase
WebofScience(/externalref?access_num=A1992JV36300004&link_type=ISI) Mixed Quantum/Classical
Methods
30. Segall,M.D.,Lindan,P.J.D.,Probert,M.J.,Pickard,C.J.,Hasnip,P.J.,Clark,S.J.&Payne,M.C.
(2002)J.Phys.Cond.Matt. 14,27172744. Applications
CrossRef(/externalref?access_num=10.1088/09538984/14/11/301&link_type=DOI)
Conclusion and Future
31. Kresse,G.&Furthmuller,J.(1996)Phys.Rev.B54,1116911186. Directions
WebofScience(/externalref?access_num=A1996VT67500040&link_type=ISI) Acknowledgments
32. Frisch,M.J.,Trucks,G.W.,Schlegel,H.B.,Scuseria,G.E.,Robb,M.A.,Cheeseman,J.R.,
Footnotes
Montgomery,J.A.,Jr.,Vreven,T.,Kudin,K.N.,Burant,J.M.,etal.(2003)GAUSSIAN03 (Gaussian,
Pittsburgh),RevisionB.02. References
33. Gill,P.M.W.,Johnson,B.G.&Pople,J.A.(1993)Chem.Phys.Lett. 209,506512.
CrossRef(/externalref?access_num=10.1016/00092614(93)801259&link_type=DOI)
34. Treutler,O.&Ahlrichs,R.(1995)J.Chem.Phys. 102,346354.

WebofScience(/externalref?access_num=A1995PZ71400037&link_type=ISI)
35. White,C.A.,Johnson,B.G.,Gill,P.M.W.&HeadGordon,M.(1996)Chem.Phys.Lett. 253,268

278. CrossRef(/externalref?access_num=10.1016/00092614(96)001753&link_type=DOI)
36. Sierka,M.,Hogekamp,A.&Ahlrichs,R.(2003)J.Chem.Phys. 118,91369148.
37. Murphy,R.B.,Cao,Y.,Beachy,M.D.,Ringnalda,M.N.&Friesner,R.A.(2000)J.Chem.Phys. 112,

38. Friesner,R.A.,Murphy,R.B.,Beachy,M.D.,Ringnalda,M.N.,Pollard,W.T.,Dunietz,B.D.&Cao,
Y.X.(1999)J.Phys.Chem.A103,19131928.
CrossRef(/externalref?access_num=10.1021/jp9825157&link_type=DOI)
39. Beachy,M.,Murphy,R.,Philipp,D.,Mainz,D.,Ringnalda,M.,Cao,Y.,Greeley,H.,Pollard,T.,
Braden,D.,Friesner,R.,etal.(2004)JAGUAR (Schrdinger,Portland,OR),Version4.0.
40. Ahlrichs,R.,Bar,M.,Haser,M.,Horn,H.&Kolmel,C.(1989)Chem.Phys.Lett. 162,165169.
CrossRef(/externalref?access_num=10.1016/00092614(89)851188&link_type=DOI)
41. Werner,H.J.,Manby,F.R.&Knowles,P.J.(2003)J.Chem.Phys. 118,81498160.

42. Kong,J.,White,C.A.,Krylov,A.I.,Sherrill,D.,Adamson,R.D.,Furlani,T.R.,Lee,M.S.,Lee,A.M.,
Gwaltney,S.R.,Adams,T.R.,etal.(2000)J.Comput.Chem. 21,15321548.
CrossRef(/externalref?access_num=10.1002/1096987X(200012)21:16%3C1532::AIDJCC10%3E3.0.CO2
W&link_type=DOI)
43. Karlstrom,G.,Lindh,R.,Malmqvist,P.A.,Roos,B.O.,Ryde,U.,Veryazov,V.,Widmark,P.O.,Cossi,
M.,Schimmelpfennig,B.,Neogrady,P.,etal.(2003)Comput.Mat.Sci. 28,222239.
44. Curtiss,L.A.,Raghavachari,K.,Redfern,P.C.,Rassolov,V.&Pople,J.A.(1998)J.Chem.Phys.
109,77647776. CrossRef(/externalref?access_num=10.1063/1.477422&link_type=DOI)
45. Lynch,B.&Truhlar,D.(2003)J.Phys.Chem.A107,38983906.
46. Truhlar,D.G.(1998)Chem.Phys.Lett. 294,4548.

47. Ochterski,J.W.,Petersson,G.A.&Montgomery,J.A.(1996)J.Chem.Phys. 104,25982619.

48. Field,M.J.,Bash,P.A.&Karplus,M.(1990)J.Comput.Chem. 11,700733.

CrossRef(/externalref?access_num=10.1002/jcc.540110605&link_type=DOI)
WebofScience(/externalref?access_num=A1990DK75400004&link_type=ISI)
49. Murphy,R.B.,Philipp,D.M.&Friesner,R.A.(2000)J.Comput.Chem. 21,14421457.

CrossRef(/externalref?access_num=10.1002/1096987X(200012)21:16%3C1442::AIDJCC3%3E3.0.CO2
O&link_type=DOI)
50. Das,D.,Eurenius,K.P.,Billings,E.M.,Sherwood,P.,Chatfield,D.C.,Hodoscek,M.&Brooks,B.R.
(2002)J.Chem.Phys. 117,1053410547.
51. Amara,P.&Field,M.J.(2003)Theor.Chem.Acc. 109,4352.
52. Maseras,F.&Morokuma,K.(1995)J.Comput.Chem. 16,11701179.

53. Gao,J.L.,Amara,P.,Alhambra,C.&Field,M.J.(1998)J.Phys.Chem.A102,47144721.
54. Reuter,N.,Dejaegere,A.,Maigret,B.&Karplus,M.(2000)J.Phys.Chem.A104,17201735.
55. Vreven,T.,Morokuma,K.,Farkas,O.,Schlegel,H.B.&Frisch,M.J.(2003)J.Comput.Chem. 24,

760769. CrossRef(/externalref?access_num=10.1002/jcc.10156&link_type=DOI)
56. Zhang,Y.K.,Liu,H.Y.&Yang,W.T.(2000)J.Chem.Phys. 112,34833492. Top
WebofScience(/externalref?access_num=000085345300003&link_type=ISI) Abstract
57. Tannor,D.J.Marten,B.,Murphy,R.,Friesner,R.A.,Sitkoff,D.,Nicholls,A.,Ringnalda,M.,Goddard,
W.A.&Honig,B.(1994)J.Am.Chem.Soc. 116,1187511882.
CrossRef(/externalref?access_num=10.1021/ja00105a030&link_type=DOI) Computational
WebofScience(/externalref?access_num=A1994QA28800030&link_type=ISI) Implementation
58. Tomasi,J.&Persico,M.(1994)Chem.Rev. 94,20272094.
Condensed-Phase
CrossRef(/externalref?access_num=10.1021/cr00031a013&link_type=DOI)
59. Honig,B.&Nicholls,A.(1995)Science 268,11441149.
Methods
Abstract/FREEFullText(/cgi/ijlink?linkType=ABST&journalCode=sci&resid=268/5214/1144)
60. Klamt,A.,Jonas,V.,Burger,T.&Lohrenz,J.C.W.(1998)J.Phys.Chem.A102,50745085. Applications

CrossRef(/externalref?access_num=10.1021/jp980017s&link_type=DOI)
61. Barone,V.,Cossi,M.&Tomasi,J.(1998)J.Comput.Chem. 19,404417. Directions
CrossRef(/externalref?access_num=10.1002/(SICI)1096987X(199803)19:4%3C404::AIDJCC3%3E3.0.CO2
W&link_type=DOI)
Acknowledgments
62. Cortis,C.M.,Langlois,J.M.,Beachy,M.D.&Friesner,R.A.(1996)J.Chem.Phys. 105,5472
5484. CrossRef(/externalref?access_num=10.1063/1.472388&link_type=DOI) Footnotes
63. Marten,B.,Kim,K.,Cortis,C.,Friesner,R.A.,Murphy,R.B.,Ringnalda,M.N.,Sitkoff,D.&Honig,B. References

(1996)J.Phys.Chem. 100,1177511788.
CrossRef(/externalref?access_num=10.1021/jp953087x&link_type=DOI)
64. Klicic,J.J.,Friesner,R.A.,Liu,S.Y.&Guida,W.C.(2002)J.Phys.Chem.A106,13271335.
CrossRef(/externalref?access_num=10.1021/jp012533f&link_type=DOI)
65. Guner,V.,Khuong,K.S.,Leach,A.G.,Lee,P.S.,Bartberger,M.D.&Houk,K.N.(2003)J.Phys.
Chem.A107,1144511459. CrossRef(/externalref?access_num=10.1021/jp035501w&link_type=DOI)
66. Gauss,J.&Stanton,J.F.(1996)J.Chem.Phys. 104,25742583.
67. Sekino,H.&Bartlett,R.J.(1993)J.Chem.Phys. 98,30223037.

68. Milman,V.,Winkler,B.,White,J.A.,Pickard,C.J.,Payne,M.C.,Akhmatskaya,E.V.&Nobes,R.H.
(2000)Int.J.QuantumChem. 77,895910.
CrossRef(/externalref?access_num=10.1002/(SICI)1097461X(2000)77:5%3C895::AIDQUA10%3E3.0.CO2
C&link_type=DOI)
69. Rohlfing,M.&Louie,S.G.(2000)Phys.Rev.B62,49274944.
70. Hammer,B.&Norskov,J.K.(2000)Adv.Catal. 45,71129.

71. Wu,C.J.&Carter,E.A.(1991)Chem.Phys.Lett. 185,172178.
CrossRef(/externalref?access_num=10.1016/00092614(91)80159U&link_type=DOI)
72. Raghavachari,K.&Halls,M.D.(2004)Mol.Phys. 102,381393.

CrossRef(/externalref?access_num=10.1080/00268970410001675590&link_type=DOI)
73. Oregan,B.&Gratzel,M.(1991)Nature 353,737740.

CrossRef(/externalref?access_num=10.1038/353737a0&link_type=DOI)
WebofScience(/externalref?access_num=A1991GL69600062&link_type=ISI)
74. Zhou,Z.Y.,Steigerwald,M.,Hybertsen,M.,Brus,L.&Friesner,R.A.(2004)J.Am.Chem.Soc. 126

,35973607. Medline(/externalref?access_num=15025489&link_type=MED)
75. Bettinger,H.F.,Yakobson,B.I.&Scuseria,G.E.(2003)J.Am.Chem.Soc. 125,55725580.
CrossRef(/externalref?access_num=10.1021/ja0288744&link_type=DOI)
76. Jhi,S.H.,Louie,S.G.&Cohen,M.L.(2000)Phys.Rev.Lett. 85,17101713.

77. Zhou,Z.Y.,Friesner,R.A.&Brus,L.(2003)J.Am.Chem.Soc. 125,1559915607.

78. Melnikov,D.V.&Chelikowsky,J.R.(2004)Phys.Rev.Lett. 92,046802.

79. Guallar,V.&Friesner,R.A.(2004)J.Am.Chem.Soc. 126,85018508.

CrossRef(/externalref?access_num=10.1021/ja036123b&link_type=DOI)
80. Gherman,B.F.,Goldberg,S.D.,Cornish,V.W.&Friesner,R.A.(2004)J.Am.Chem.Soc. 126,

76527664. CrossRef(/externalref?access_num=10.1021/ja036879a&link_type=DOI)
81. Gherman,B.F.,Baik,M.H.,Lippard,S.J.&Friesner,R.A.(2004)J.Am.Chem.Soc. 126,2978

2990. CrossRef(/externalref?access_num=10.1021/ja036506+&link_type=DOI)
82. Guallar,V.,Jacobson,M.,McDermott,A.&Friesner,R.A.(2004)J.Mol.Biol. 337,227239.

CrossRef(/externalref?access_num=10.1016/j.jmb.2003.11.016&link_type=DOI)
83. Siegbahn,P.E.M.&Blomberg,M.R.A.(2000)Chem.Rev. 100,421437.

CrossRef(/externalref?access_num=10.1021/cr980390w&link_type=DOI)
WebofScience(/externalref?access_num=000085319500004&link_type=ISI) Navigate This Article
84. Siegbahn,P.E.M.,Blomberg,M.R.A.&Blomberg,M.L.(2003)J.Phys.Chem.B107,10946 Top

10955. CrossRef(/externalref?access_num=10.1021/jp035486v&link_type=DOI)
Abstract
85. Cui,Q.&Karplus,M.(2001)J.Am.Chem.Soc. 123,22842290.
CrossRef(/externalref?access_num=10.1021/ja002886c&link_type=DOI) Quantum Chemical Theory
Computational
Implementation
86. Rothlisberger,U.,Carloni,P.,Doclo,K.&Parrinello,M.(2000)J.Biol.Inorg.Chem. 5,236250.

Condensed-Phase
CrossRef(/externalref?access_num=10.1007/s007750050368&link_type=DOI)
Medline(/externalref?access_num=10819469&link_type=MED) Mixed Quantum/Classical
WebofScience(/externalref?access_num=000086969600010&link_type=ISI) Methods
87. Wu,N.,Mo,Y.R.,Gao,J.L.&Pai,E.F.(2000)Proc.Natl.Acad.Sci.USA97,20172022. Applications

Abstract/FREEFullText(/cgi/ijlink?linkType=ABST&journalCode=pnas&resid=97/5/2017)
Directions
88. Amara,P.,Volbeda,A.,FontecillaCamps,J.C.&Field,M.J.(1999)J.Am.Chem.Soc. 121,4468
4477. CrossRef(/externalref?access_num=10.1021/ja983971b&link_type=DOI)
Acknowledgments
89. Wieczorek,R.&Dannenberg,J.J.(2003)J.Am.Chem.Soc. 125,81248129.
Medline(/externalref?access_num=12837081&link_type=MED) Footnotes
90. Rossmeisl,J.,Kristensen,T.,Gregersen,M.,Jacobsen,K.W.&Norskov,J.K.(2003)J.Am.Chem. References

Soc. 125,1638316386. CrossRef(/externalref?access_num=10.1021/ja0359658&link_type=DOI)
91. Siegbahn,P.E.M.(2001)J.Biol.Inorg.Chem. 6,2745.

CrossRef(/externalref?access_num=10.1007/s007750000184&link_type=DOI)
92. Basch,H.,Mogi,K.,Musaev,D.G.&Morokuma,K.(1999)J.Am.Chem.Soc. 121,72497256.

93. Hoffmann,M.,Khavrutskii,I.V.,Musaev,D.G.&Morokuma,K.(2004)Int.J.QuantumChem. 99,

972980. CrossRef(/externalref?access_num=10.1002/qua.20141&link_type=DOI)
94. Guallar,V.,Gherman,B.F.,Miller,W.H.,Lippard,S.J.&Friesner,R.A.(2002)J.Am.Chem.Soc.
24,33773384. CrossRef(/externalref?access_num=10.1021/ja0167248&link_type=DOI)
95. Dunietz,B.D.,Beachy,M.D.,Cao,Y.X.,Whittington,D.A.,Lippard,S.J.&Friesner,R.A.(2000)J.
Am.Chem.Soc. 122,28282839.
96. Siegbahn,P.E.M.&Crabtree,R.H.(1997)J.Am.Chem.Soc. 119,31033113.

CrossRef(/externalref?access_num=10.1021/ja963939m&link_type=DOI)
97. Kaminski,G.A.,Friesner,R.A.,TiradoRives,J.&Jorgensen,W.L.(2001)J.Phys.Chem.B105,
64746487. CrossRef(/externalref?access_num=10.1021/jp003919d&link_type=DOI)
98. MacKerell,A.D.,Bashford,D.,Bellott,M.,Dunbrack,R.L.,Evanseck,J.D.,Field,M.J.,Fischer,S.,
Gao,J.,Guo,H.,Ha,S.,etal.(1998)J.Phys.Chem.B102,35863616.
CrossRef(/externalref?access_num=10.1021/jp973084f&link_type=DOI)
99. Wang,J.M.,Wolf,R.M.,Caldwell,J.W.,Kollman,P.A.&Case,D.A.(2004)J.Comput.Chem. 25,

11571174. CrossRef(/externalref?access_num=10.1002/jcc.20035&link_type=DOI)
100. Kaminski,G.A.,Stern,H.A.,Berne,B.J.&Friesner,R.A.(2004)J.Phys.Chem.A108,621627.
101. Kaminski,G.A.,Stern,H.A.,Berne,B.J.,Friesner,R.A.,Cao,Y.X.,Murphy,R.B.,Zhou,R.H.&
Halgren,T.A.(2002)J.Comput.Chem. 23,15151531.
102. Becke,A.D.(2003)J.Chem.Phys. 119,29722977.

103. Tozer,D.J.&Handy,N.C.(1998)J.Chem.Phys. 109,1018010189.

104. Tao,J.M.,Perdew,J.P.,Staroverov,V.N.&Scuseria,G.E.(2003)Phys.Rev.Lett. 91,146401.

105. Zhao,Y.,Lynch,B.J.&Truhlar,D.G.(2004)J.Phys.Chem.A108,47864791.
CrossRef(/externalref?access_num=10.1021/jp049253v&link_type=DOI)
106. Lynch,B.J.,Fast,P.L.,Harris,M.&Truhlar,D.G.(2000)J.Phys.Chem.A104,48114815.
CrossRef(/externalref?access_num=10.1021/jp000497z&link_type=DOI)
107. VanVoorhis,T.&HeadGordon,M.(2002)J.Chem.Phys. 117,91909201.

108. Flocke,N.&Bartlett,R.J.(2003)Chem.Phys.Lett. 367,8089.

109. Schutz,M.&Werner,H.J.(2001)J.Chem.Phys. 114,661681.

110. Maslen,P.E.,Lee,M.S.&HeadGordon,M.(2000)Chem.Phys.Lett. 319,205212.
CrossRef(/externalref?access_num=10.1016/S00092614(00)001135&link_type=DOI) Top
111. Rempe,S.B.,Asthagiri,D.&Pratt,L.R.(2004)Chem.Phys. 6,19661969.
Abstract
CrossRef(/externalref?access_num=10.1039/b313756b&link_type=DOI)

We recommend
Computational
Thecolorofrhodopsinsattheabinitio Abinitioderivationofmodelenergydensity Implementation
multiconfigurationalperturbationtheoryresolution. functionals
Condensed-Phase
PedroBCotoetal.,ProcNatlAcadSciUSA, JacekDobaczewski,JPhysGNuclPartPhys, Mixed Quantum/Classical
2006 2016 Methods
Structure,initialexcitedstaterelaxation,andenergy Theoryofoneandtwodonorsinsilicon Applications

storageofrhodopsinresolvedatthe ALSaraivaetal.,JPhysCondensMatter,2015
multiconfigurationalperturbationtheorylevel. Conclusion and Future
TadeuszAndruniwetal.,ProcNatlAcadSciUS ComputationalActinideChemistry:AreWeThere Directions
A,2004 Yet?
DOE/PacificNorthwestNationalLaboratory, Acknowledgments
Progresstowardchemicalaccuracyinthecomputer ScienceDaily,2007
simulationofcondensedphasereactions. Footnotes
PABashetal.,ProcNatlAcadSciUSA,1996 Germanene:thegermaniumanalogueofgraphene
References
Thekernelenergymethodofquantummechanical AAcunetal.,JPhysCondensMatter,2015
approximationcarriedtofourthorderterms.
LuluHuangetal.,ProcNatlAcadSciUSA,2008 Superoperatorcoupledclustermethodfor
nonequilibriumdensitymatrix
Evidenceforafirstorderliquidliquidtransitionin AlanADzhioevetal.,JPhysAMathTheor,2014
highpressurehydrogenfromabinitiosimulations.
MiguelAMoralesetal.,ProcNatlAcadSciUSA,
2010
Poweredby
HighWire Press-hosted articles citing this article

Understandinghydrogenbondpatternsinproteinsusingnetworkmotifs
Bioinformatics(Bioinformatics)200925(22)29212928
Abstract(http://bioinformatics.oxfordjournals.org/cgi/content/abstract/25/22/2921) FullText(HTML)
(http://bioinformatics.oxfordjournals.org/cgi/content/full/25/22/2921) FullText(PDF)
(http://bioinformatics.oxfordjournals.org/cgi/reprint/25/22/2921)
Theoreticalstudiesonbimolecularreactiondynamics
Proc.Natl.Acad.Sci.USA(PNAS)2008105(35)1264912653
Abstract(/cgi/content/abstract/105/35/12649) FullText(HTML)(/cgi/content/full/105/35/12649/F6) FullText
(PDF)(/cgi/reprint/105/35/12649)
QuantumDynamicsofChemicalReactions
Science(Science)2008321(5890)789791
Abstract(http://www.sciencemag.org/cgi/content/abstract/321/5890/789) FullText(HTML)
(http://www.sciencemag.org/cgi/content/full/321/5890/789) FullText(PDF)
(http://www.sciencemag.org/cgi/reprint/321/5890/789)

Top
Abstract
Computational
Implementation
Condensed-Phase

Ab Initio Quantum Chemistry - Methodology and A

Cargado por

Información del documento

Título original

Derechos de autor

Formatos disponibles

Compartir este documento

Compartir o incrustar documentos

Opciones para compartir

¿Le pareció útil este documento?

¿Este contenido es inapropiado?

Copyright:

Formatos disponibles

Ab Initio Quantum Chemistry - Methodology and A

Cargado por

Copyright:

Formatos disponibles

CURRENT ISSUE (/CONTENT/CURRENT) ARCHIVE (/CONTENT) NEWS & MULTIMEDIA (/MULTIMEDIA) AUTHORS (/SITE/AUTHORS/INDEX.

ABOUT (/SITE/ABOUTPNAS/INDEX.XHTML) COLLECTED ARTICLES (/SITE/MISC/COLLECTEDPAPERS.XHTML) BROWSE BY TOPIC (/SEARCH)

EARLY EDITION (/CONTENT/EARLY/RECENT) FRONT MATTER (HTTP://FRONTMATTER.PNAS.ORG)

ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica This Issue

Ab initio quantum chemistry: Methodology and applications

Thetreatmentoflarge,condensedphasesystems(e.g.,proteinsinaqueoussolution)entirelybyab initio (/external-ref?

Quantum Chemical Theory SUBMITANARTICLE

Condensed-Phase Environments: SCRF and Mixed

Conclusion and Future Directions

Quantum Chemical Theory

Thispaperwassubmitteddirectly(TrackII)tothePNASoffice. Conclusion and Future

11. Martin,J.M.L.&deOliveira,G.(1999)J.Chem.Phys. 111,18431856.

14. Grimme,S.(2003)J.Chem.Phys. 118,90959102.

15. Tsuzuki,S.,Uchimaru,T.&Tanabe,K.(1998)Chem.Phys.Lett. 287,202208.

16. Tsuzuki,S.,Uchimaru,T.,Matsumura,K.,Mikami,M.&Tanabe,K.(1999)J.Chem.Phys. 110,

18. Dutta,A.&Sherrill,C.D.(2003)J.Chem.Phys. 118,16101619.

19. Murray,C.W.,Laming.G.J.,Handy,N.C.&Amos,R.D.(1992)Chem.Phys.Lett. 199,551556.

23. Murphy,R.B.,Beach,M.D.,Friesner,R.A.&Ringnalda,M.N.(1995)J.Chem.Phys. 103,1481

25. Bauernschmitt,R.,Ahlrichs,R.,Hennrich,F.H.&Kappes,M.M.(1998)J.Am.Chem.Soc. 120,

34. Treutler,O.&Ahlrichs,R.(1995)J.Chem.Phys. 102,346354.

35. White,C.A.,Johnson,B.G.,Gill,P.M.W.&HeadGordon,M.(1996)Chem.Phys.Lett. 253,268

37. Murphy,R.B.,Cao,Y.,Beachy,M.D.,Ringnalda,M.N.&Friesner,R.A.(2000)J.Chem.Phys. 112,

41. Werner,H.J.,Manby,F.R.&Knowles,P.J.(2003)J.Chem.Phys. 118,81498160.

46. Truhlar,D.G.(1998)Chem.Phys.Lett. 294,4548.

47. Ochterski,J.W.,Petersson,G.A.&Montgomery,J.A.(1996)J.Chem.Phys. 104,25982619.

48. Field,M.J.,Bash,P.A.&Karplus,M.(1990)J.Comput.Chem. 11,700733.

49. Murphy,R.B.,Philipp,D.M.&Friesner,R.A.(2000)J.Comput.Chem. 21,14421457.

51. Amara,P.&Field,M.J.(2003)Theor.Chem.Acc. 109,4352.

52. Maseras,F.&Morokuma,K.(1995)J.Comput.Chem. 16,11701179.

55. Vreven,T.,Morokuma,K.,Farkas,O.,Schlegel,H.B.&Frisch,M.J.(2003)J.Comput.Chem. 24,

60. Klamt,A.,Jonas,V.,Burger,T.&Lohrenz,J.C.W.(1998)J.Phys.Chem.A102,50745085. Applications

63. Marten,B.,Kim,K.,Cortis,C.,Friesner,R.A.,Murphy,R.B.,Ringnalda,M.N.,Sitkoff,D.&Honig,B. References

67. Sekino,H.&Bartlett,R.J.(1993)J.Chem.Phys. 98,30223037.

70. Hammer,B.&Norskov,J.K.(2000)Adv.Catal. 45,71129.

72. Raghavachari,K.&Halls,M.D.(2004)Mol.Phys. 102,381393.

73. Oregan,B.&Gratzel,M.(1991)Nature 353,737740.

74. Zhou,Z.Y.,Steigerwald,M.,Hybertsen,M.,Brus,L.&Friesner,R.A.(2004)J.Am.Chem.Soc. 126

76. Jhi,S.H.,Louie,S.G.&Cohen,M.L.(2000)Phys.Rev.Lett. 85,17101713.

77. Zhou,Z.Y.,Friesner,R.A.&Brus,L.(2003)J.Am.Chem.Soc. 125,1559915607.

78. Melnikov,D.V.&Chelikowsky,J.R.(2004)Phys.Rev.Lett. 92,046802.

79. Guallar,V.&Friesner,R.A.(2004)J.Am.Chem.Soc. 126,85018508.

80. Gherman,B.F.,Goldberg,S.D.,Cornish,V.W.&Friesner,R.A.(2004)J.Am.Chem.Soc. 126,

81. Gherman,B.F.,Baik,M.H.,Lippard,S.J.&Friesner,R.A.(2004)J.Am.Chem.Soc. 126,2978

82. Guallar,V.,Jacobson,M.,McDermott,A.&Friesner,R.A.(2004)J.Mol.Biol. 337,227239.

83. Siegbahn,P.E.M.&Blomberg,M.R.A.(2000)Chem.Rev. 100,421437.

84. Siegbahn,P.E.M.,Blomberg,M.R.A.&Blomberg,M.L.(2003)J.Phys.Chem.B107,10946 Top

86. Rothlisberger,U.,Carloni,P.,Doclo,K.&Parrinello,M.(2000)J.Biol.Inorg.Chem. 5,236250.

87. Wu,N.,Mo,Y.R.,Gao,J.L.&Pai,E.F.(2000)Proc.Natl.Acad.Sci.USA97,20172022. Applications

90. Rossmeisl,J.,Kristensen,T.,Gregersen,M.,Jacobsen,K.W.&Norskov,J.K.(2003)J.Am.Chem. References

91. Siegbahn,P.E.M.(2001)J.Biol.Inorg.Chem. 6,2745.

92. Basch,H.,Mogi,K.,Musaev,D.G.&Morokuma,K.(1999)J.Am.Chem.Soc. 121,72497256.

93. Hoffmann,M.,Khavrutskii,I.V.,Musaev,D.G.&Morokuma,K.(2004)Int.J.QuantumChem. 99,

96. Siegbahn,P.E.M.&Crabtree,R.H.(1997)J.Am.Chem.Soc. 119,31033113.

99. Wang,J.M.,Wolf,R.M.,Caldwell,J.W.,Kollman,P.A.&Case,D.A.(2004)J.Comput.Chem. 25,

102. Becke,A.D.(2003)J.Chem.Phys. 119,29722977.

103. Tozer,D.J.&Handy,N.C.(1998)J.Chem.Phys. 109,1018010189.

104. Tao,J.M.,Perdew,J.P.,Staroverov,V.N.&Scuseria,G.E.(2003)Phys.Rev.Lett. 91,146401.

107. VanVoorhis,T.&HeadGordon,M.(2002)J.Chem.Phys. 117,91909201.

108. Flocke,N.&Bartlett,R.J.(2003)Chem.Phys.Lett. 367,8089.

109. Schutz,M.&Werner,H.J.(2001)J.Chem.Phys. 114,661681.

Quantum Chemical Theory

Structure,initialexcitedstaterelaxation,andenergy Theoryofoneandtwodonorsinsilicon Applications

HighWire Press-hosted articles citing this article