0% found this document useful (0 votes)
37 views

Characteristics of Energy Surfaces: 3.1 Strategies For Geometry Optimization

This document discusses strategies for optimizing the geometry of molecules to locate energy minima on potential energy surfaces. It describes approximating the energy surface using Taylor series expansions and taking steps along eigenvectors of the Hessian matrix to minimize the energy in each vibrational mode independently. Steps are determined by the gradient and Hessian eigenvalues, with the step moving in the direction opposite the gradient divided by the eigenvalue if the eigenvalue is positive, or in the opposite direction if the eigenvalue is negative.

Uploaded by

tun lykoung
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
37 views

Characteristics of Energy Surfaces: 3.1 Strategies For Geometry Optimization

This document discusses strategies for optimizing the geometry of molecules to locate energy minima on potential energy surfaces. It describes approximating the energy surface using Taylor series expansions and taking steps along eigenvectors of the Hessian matrix to minimize the energy in each vibrational mode independently. Steps are determined by the gradient and Hessian eigenvalues, with the step moving in the direction opposite the gradient divided by the eigenvalue if the eigenvalue is positive, or in the opposite direction if the eigenvalue is negative.

Uploaded by

tun lykoung
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 12

Chapte3

r
Characteristicsof energy surfaces

3.1 Strategiesfor geometry optimization


The extension of the harmonic and Morse vibrational models to poly-
atomic moleculesrequiresthat the multidimensionalenergy surfacebe
analyzedin a mannerthat allows one to approximatethe molecule,smo-
tions in terms of many nearly independentvibrations.In this section,we
will explore the tools that one uses to carry out such an analysisof the
surface.

Many strategiesthat attemptto locateminima on nrolecularpotentialen-


ergy landscapesbegin by approximatingthc pcltentialener-syV for gcornetries
(collectivelydenotedin termsof 3A cartesiancoordinares
{qr})in a Taylorse-
riesexpansionaboutsonre"startingpoint" geometry(i.e..the currentnrolecular
geometryin an iterativeprocess):

I ' ( q r ) : I 1 ( 0 ) +t ( a r l o q ^ ) q+t t l 2 l o i H i . * 4 ** . . . . (3.1)

H e r e .i ' ( 0 ) i s t h e e n e r g y u * , nl u. r r " n r r e o n . ] e t rtyU,r, , t O , ^ l : 9 1 i s r h e - e r a d i e n t o f


t h e e n e r g y a l o n g t h e cqoao r d i n a t eH,1 . s= e 2 I / l d q i a q k )i s t h e s e c o n d d e r i v a t i v e
or Hessianmatrix.andqais thelengthof the"step"to be takenalongthiscartesian
direction.An exampleof an energysurfacein only two dinrensions is given
in Fig. 3.1 wherevariousspecialaspectsare illustrated.For example.minima
correspondingto stablemolecularstructures,transitionstates(first ordersaddle
points)connectingsuchminima.andhigherordersaddlepointsareclisplayed.
If the only knowled-qethat is availableis l'(0) and the gradientcoruDonents
(e.g..computation ofthesecondderivatives is usuallyr.nuch morecomputationally
taxing than is evaluationof the gradient),the linear approxirnation

I ' ( q i ) : r t t l )+ S^qi (3.2)


f

suggests thatoneshouldchoose"step"elementsq1 thatareoppositein signfrom


that of the corresponding
gradientelementsgr : (a vloqr). The magnitudeof

104
S t r a t e g i e sf o r g e o m e t r yo p t i m i z a t i o n 1n6

Secondorder saddlepoint
TransitionstructureA

Minimum for
productA

-0.5
Sccondorder imensional
0.5
saddlepoint potentialsurface
(r.) Valley-ridge s h o w i n gm i n i m a ,
- tr-) inflectionpoint
Minimum lor reactant transitionstates,and
_ - l
p a t h sc o n n e c t i n gt h e m .

thestepelementsis usuallykeptsmallin orderto remainwithin the "trust radius"


within which the linearapproximarionto z is valid to someprederermined
desired
precision.
when secondderivativedata is available,there are diIl-erentapproachesto
predictingwhat step lqrl to take in searchof a minimum. we first write the
quadraticTaylor expansion

I ' ( , t t =t I ' r 0 r+
f g ^ V -r l , 2 l l , H,.+,tt (3 . 3 )
k r."

in matrix-vectornotation

t t ( q ) =L , ( o ) + q r . +g t l 2 q r . H . q (3.4)

rvith the elements{q1


} collectedinto the column vectorq whosetransposers
denotedqr. Introducingthe unitarymatrix u that diagonalizesthe symmetricH
matrix,the aboveequationbecomes

rG): I r ( 0 1 +g r u u r q + t T 2 q r u u r H u u r q . (3.5)

BecauseUrHU is diagonal,

( u r H u ) r 1: 8 r . r i r (3.6)

and haseigenvalues )"a.For non-linearmolecules,3r/ - 6 of theseeigenvalues


rvill be non-zero;for linear molecules,3,A/- 5 will be non-zero.The 5 or 6 zero
eigenvalues of H haveeigenvectorsthat describetranslationand rotationof the
entiremolecule;they are zero becausethe energysurfacez doesnot changeif
the moleculeis rotatedor translated.The eigenvectorsof H form the columnsof
tuo C h a r a c t e r i s t i cosf e n e r g ys u r f a c e s

the array U that brings H to diagonallorm:

1-1.7)

Therefore.ifrve define

Q,:lL,[,,,qr and G,:lL' , ! ^, g r (i 8)

to be the componentof ,n.*r,"p {qp} and of tn. glual.nt along the rrth ergen-
vectorolH. the quadraticexpansionof l,'canbe u'rittenin termsof stepsalong
the 3,\ - 5 or 3t/ - 6 {Q,,,} directionsthat correspondto non-zeroHessian
eigenvalues:

v(qt): r'(0)+ t cl,e,,+ 112Le,,,)",,,e,,,. (3.r.r1

The advantageto transformingtlie gradient,step.unOH.rrlun to the eigenmode


basisis that eachsuchmode (labelednr) appearsin an independent uncoupled
form in the expansionof /. This allou's us to take stepsalong eachof the Q,,,
directionsin an independentmannerwith eachstepdesignedto lowerthepotential
ener-qy(aswe searchfor rninima).
For eacheigenmodedirection,one can askfor u&at stepp u'ouldthe quantity
G Q + I12),Q2 be a minimum.Differentiatingthis quadraticform with respect
to Q andsettingthe resultequalto zero gives

Q r ,= - G , l ) ' r , , (3 10)

but with a rnagnitudegiven


that is, or.reshouldtake a stepoppositethe _eradient
by the gradientdivided by the eigenvalueof the Hessianmatrix. lf the current
moleculargeometryis one that hasall positive/.,,,values,
this indicatesthatone
may be "close" to a minimum on the energysurface(becauseall i,,, arepositive
at minima).In sucha case,the step Q,,: -Gu,/i,,, is opposedto the gradient
along ali 3l/ - 5 or 31y'- 6 directions.The energychangethat is expectedto
occurif the step{p,,,} is takencanbe computedby substitutingp,, : -G,,/L,,
into the quadraticequationfor /:

/(afterstep): I,'(0)+ | Cl1-C,, l\,,) * 1 l2LL,,,(-G,,, 1I,,,)2


"'
: r(o)- r)z)- ).n,(--G,,,f),.,,)2. ,r.,,,

"downhill"
This clearlysuggeststhat the stepwill lead in energyas long asall of
the ).,,,valuesare positive.
However,if one or more of the ),n,arenegativeat the currentgeometry'. one is
in a regionof the energysurfacethatis not closeto a minimum. In fact,if only one
),,, is negative,onq anticipatesbeing neara transitionstate(at which all gradient
componentsvanishandali but one.i.n,arepositivewith one i.o,negative).In such
S t r a t e g i e sf o r g e o m e t r yo p t i m i z a t i o n
107

a case.theabor,'e
analysissuggests a step0/,r : -G,,/Ln, aloneall of the
takin_q
modes havingpositive).,,,,
but takin-e
a stepof oppositedirectione,, J +G,
1).,
alongthedirectionhavingnegarivel,?.
In anyevent.oncea stephasbeensuggested within the eigenmodebasis,one
needsto expressthat step in terms of the original cartesian coordinatesq1 so
thatthesecartesianvaluescan be alteredwithin the softwareprogramto effect
the predictedstep. Given valuesfor the 3N - 5 or 3l/ - 6 stef components
Q,, @.b..the stepcomponenrsQ,,, alongthe 5 or 6 modeshavinj zero Hessian
eigenvalues canbe takento be zerobecausethey would simply translate
or rotate
themolecule),onemustcomputethe{q1,}.To do so,we usethe relationshin

Q,:\ui, a1r (3.r2)


and"vrireits inverse(usingthe unitary natureof the u matrix):

qr:L,L1t.,,,Q,,, (3.1i)

to computethe desiredCartesianstepcomponents.
In usi'g the Hessian-based
approaches
outlinedabove.onehasto takespecial
carewhenone or moreof the Hessianei-senvaluesis small.This oftenhappens
wnen:

(i) onehasa molecuie "softrnodes"


containing (i.c.,de_urees
of tieedomalonglvhich
theenergyr.aries
little).or
(ii) onemovesfrorna regionofnegatir,ccur\arureintoa regionofpositivecur'ature
(orviceversa)
- in suchcases,
thecu.naturenlustmovetrrrough or nearzero.
For thesesituations.the expressiofl
e,,: -G,,/n,, can producea yerv lar-9e
stepalong the mode havingsmall curvature.care
must be takento not alrow
suchincorrectartiliciallylargestepsto be taken.
Beforeclosin-e thissection.I shouldnotethatthereareotheri'portant
regions
of potentialener-qy surf-acesthat one must be able to locateand characterize.
Above""vefbcusedon locarrninimaa'cl transition
states.In Chapterg, we will
discusshow to follow so-cailedreactionpathsthat
connectthesetwo krndsof
statlonarypoints using the type of gradienta'd
Hessianinformationthat ,"ve
introduced earlierin this chapter.
Finally.it is sometirnesimportanrto find
_qeomet'es at which two Born-
oppenheimerener-qysurfacesI"1(q) and z1(q.yintersect.
First. ret'sspenda fer.v
minutesthinkingaboutwhethersuchsurfacescan
rndeedintersect, becausestu-
dentsofte' hear that surf'acesdo not intersect
but. insteaclundergo..avoideci
crossings".To understandthe issue,let us assume
that we havetwo w.ve func_
ttons@1and @2bothof rvhichdependon
3,v - 6 coordinates {q}. Thesetwo
ftrnctionsare not assumedto be exactei-eenfunctions
of the HamiltonianF1.
but likely are chosento approximate sucheigentunctions. To find the improved
108 C h a r a c t e r i s t i cosf e n e r g ys u r f a c e s

functionsV1 and W2that more accuratelyrepresentthe eigenstates.


one usually
forms linear combinationsof (D1 and (D2,

V,r:Cx.rOr *C,r'.:O: ( 3r. 4 )


from which a 2 x 2 matrix eigenvalue problem arises:

Ht: - E Ht.:
(3.r5)
Htt Hz2 - E

This quadraticequationhas rwo solutions

2En : 1.11r.'
+ H2.2)
+ ( 3 r.6 )
Thesetwo solutionscanbe equal(i.e.,thetwo stateenergies cancross)only if the
squareroot factor vanishes.Becausethis factor is a sum of two squares(each
thus being positivequantities),this can only happenif two identitieshold:

Htt: H:.: ( 3 r.7 )


and

H r . :: o . (3.18)

The main point then is that in the 31/ - 6 dimensionalspace,the two statesn.ill
generallynot have equal energy.However.in a spaceof two lower dimensions
(because therearetwo conditionsthat must simultaneousll'be obeyed- Hr.t :
H22and Ht.2 :0), theirenergies maybe equal.Theydo not haveto be equal,but
it is possiblethat they are.It is basedupon suchan analysisthat one usuallysays
that potentialenergysurfacesin 31/ - 6 dimensionsmay undergointersections
in spacesof dimension3N - 8. If the two statesare of different symmetry.the
off-diagonalelement111.2 vanishesautomatically,so only one otherconditionis
neededto realizecrossing.So, we saythat two statesof differentsymmetrycan
crossin a spaceof dimension3N - 7.
To find the lower-dimensionalspacein which two surfacescross,one must
have availableinformation about the gradientsand Hessiansof both functions
V1 and,V2. One then usesthis information to locate a geometryat which the
differencefunction F : f\ - vil2 passesthrough zero by using conventional
"root
finding" methodsdesignedto locatewhereF : 0. once onesuchgeomerry
(qe) has been located,one subsequentlytries to follow the .,seam"along which
the function F remainszero. This is done by parameterizingstepsaway from
(qo) in a manner that constrainssuch stepsto have no componentalong the
gradientof F (i.e.,to lie in the tangentplanewhere F is constant).For a system
rvith 3N - 6 geometricaldegreesof freedom,this seamwill be a sub-surface of
lower dimension(31/ - 8 or 31/ - 7 as noted earlier).such intersectionseam
locationproceduresarebecomingmore commonlyemployed,but are still under
N o r m a lm o d e so f v i b r a t i o n 109

very activedevelopment.Locatingtheseintersectionsis an importantingredient


rvhenoneis interestedin studying,for example,photochemicalreactions
in rvhich
the reactantsand products may move from one electronicsurfaceto another.

3.2 Normal modes of vibration


H a v i n gs e e nh o w o n e c a n u s e i n f o r m a t i o na b o u t t h e g r a d i e n t sa n d H e s -
sianson a Born-oppenheimersurfaceto locategeometriescorrespond-
ing to stablespecies,let us now move on to see how this same data are
usedto treat vibrationson this surface.

For a polyatomic molecule whose electronic energy dependson the 31y'


cartesiancoordinatesof its N atoms,the potentialenergy z can be expressed
(approximately)in terms of a Taylor seriesexpansionabout any of the local
minima.of course,difrerentlocal minima (i.e., differentisomers)will have
difl-erentvaluesfbr the equilibrium coordinatesand for the derivativesof the
energywith respectto thesecoordinates.The Taylor seriesexpansionof the
electronicenergyis written as

v ( q * ) = l , 1 o l + f t a I l i ) q 1 ) q+pt l 2 L q , H 1 r . Q t * . . . , (3.19)
k j.k

ivherez(0) is thevalueof theelectronic energyat thestablegeometryunderstudy,


q1 is the displacement of the kth cartesiancoordinateawayfrom rhis starting
position.('dI"laqk) is the gradientof the elecrronicenergyalongthis direction,
andthe H 1.t;arethe secondderivatileor Hessianmatrixelements alongthesedi-
rections,H i.r : G2V lAq iAqk).If the geometrycorresponds to a stablespecies,
thegradienttermsrvill all vanish(meaningthis geometrycorresponds to a min-
tmum,maximum.or saddlepoint).and the Hessianmatrix will possess 3rv - 5
(for linearspecies)or 3N - 6 (for non-linearmolecules)positiveeigenvalues
and5 or 6 zeroeigenvalues (corresponding to 3 translational
and2 or 3 rotational
motionsof,themolecule).If the Hessianhasonenegativeeigenvalue, the geom-
etry correspondsto a transitionstate.From now on, we assumethat the
_qeometry
under sfudy correspondsto that of a stableminimum about which vibrational
motronoccurs.The treatmentof unstablegeometriesis of great importanceto
chemistry. bLrtthis materialwill be limitedto vibrationsof stablespecies.

3 . 2 . 1T h e N e w t o ne q u a t i o n so f m o t i o n f o r v i b r a t i o n
Thekineticand potentialenergymatrices
Truncatingthe Taylor seriesat the quadraticterms (assumingtheseterms dom-
inatebecauseonly small displacements from the equilibriumgeometryare of
110 C h a r a c t e r i s t i cosf e n e r g ys u r f a c e s

interest),onehasthe so-calledharmonicpotential:

t' (qr) = t' ((\ + \ l2lt1 i H i r,c/t (3 . 2 0 )


1.i

The classicalrnechanicalequationsof motionfor the 3N{r77,}


coordinatescanbe
written in termsof the abovepotentialenergyand the follou'ingkineticenergy
function:

T - 1 /) \-,,, ' '. t; t: ( 3 . 2t1


/-"'

where ri7 denotesthe tirnerateof changeof the coordinateq i andtn, is the mass
of the atomon whichthe.Tth Cartesiancoordinateresides.
The Neu,tonequations
thus obtainedare

nt,i-*lHrr,4t i l 1 ) t

wherethe forcealongtheTthcoordinateis givenby rninusthe derir.'ative of the


potentialI,' along this coordinarc(AI/ldq i): l* H 1.rtlri.r,ithinthe harntonic
approxin.ration.
Theseclassical equations canmorecompactlvbeexpressed in termsofthetime
evolutionof a setof so-calledmassweightedCartesiancoordinates definedas

xi:tlil.ntr)12. r l l l r

in terms of which the aboveNeu'tonequationsbecome

i',:-Lr',.^r^ (3.24)
k

and the mass-weighted


Hessianmatrixelementsare

H l . 1: H i . r \ t , , rtt) - t ' 2. (3.25


)

The harmonic vibrational energies and normal mode eigenvectors


Assumingthat the rl undergosomeform of sinusoidaltirneevolution:

) it ( t ) : - r , ( 0c) o s ( o / ) , (3.26)

and substitutingthis into the Newton equationsproducesa matrix eigenvalue


equation:

-2xy:lu',.rxt \ 3 . 2|7

in which the eigenvalues


arethe squaresof the so-callednonnal modevibrational
frequenciesandthe eigenvectors give the amplitudesof motion alongeachof the
31y'mass-weightedCartesiancoordinatesthat belong to eachmode. Hence,to
perform a normal-modeanalysisof a molecule,one forms the mass-weighted
tr
ffi'
N o r m a lm o d e s o f v i b r a t i o n 111
,

- -
Hessianmatrix and thenfindsthe 3N 5 or 3N 6 non-zeroeigenvalues cor2
as
rrellas the corresponding eigenuectors
-tf'l).
\\'ithin this harmonictreatmentof vibrationalmotion.the total vibrational
energl'ofthe rroleculeis givenas
l.\ -5 or 6
E { r ' 1 . r ' . . . . . . v :i o\ r r ) = t/ h' *t t,\ t' ,1 l vt l ' 2 t (3 . 2 8)

- -
a sum of 3,V 5 or 31y' 6 independent contributions.
one for eachnormal
mode.The corresponding
total vibrational
wave function

qr: ./,,(,,,,) (3.29)


ll
/=l.l\,5 or6

isa productof 3N - 5 or 3N - 6 harmonicoscillatorfunctionsly',.(.r{l)),onefor


eachnormalmode.The energygapbetweenone vibrationallevelandanotherin
whichoneof the vl quantumnurnbers by unity(i.e..for fundamental
is increased
t
r ibrationalr a n s i t i o n
) s
i s

(3.30)

Theharmonicmodel thus predictsthat the "fundamental"(rr - 0 --->r.': I ) and


"hotband"(u : I + v : 2) transitions
shouldoccurat thesameenergy.andthe
o!ertone(t, : 0 + l : 2) transitions
shouldoccurat exactlytwice this energy.

3 . 2 . 2T h e u s e o f s y m m e t r y
Symmetryadaptedmodes
Water
It is often possibleto simplity the calculationof the normal mode frequencres m o l e c u l es h o w i n gi t s
andeigenvectors by exploitingrnolecularpoint groupsymmetry.For molecules t w o b o n d l e n g t h sa n d
thatpossesssymmetrvat a particularstablegeometry.the electronicpotential b o n d a n g l e .
tr1(r7r)displayssymmetrvwith respectto displacements of symmetryequir.a-
lentCartesiancoordinates. For exantple.considerthe watermoleculeat its C2u
equilibriumgeornetryas illustratedin Fig. 3.2. A very small movementof the
H2Omolccule'slefi H atomin the positive-r direction(A-t1-)producesthe same
changein the potentiali/ as a correspondingly smalldisplacernent of the right
H atomin the negativc,tdirection(-A-rR).Similarly.movementof the left H in
thepositive_r'direction(A.v1)produces an energychangeidenticalto movement
of the right H in the positive,r'direction(A_r,n).
Theequivalcnce of thepairsof Cartesiancoordinatedisplacements is a resultof
thetact that the displacementvectorsareconnectedby the point groupoperations
of the C:, gror-rp. In particular.reflectionof Arlthrough the r,: plane(the two Two planes
planesare depictedin Fig. 3.3) produces-A.rn, and reflectionof Ars through of symmetry of the
thissameplaneyields A.r'R. w a t e rm o l e c u l e .
I tz C h ar a c t e r i s t i cosf e n e r g ys u r f a c e s

Moregenerallr''. it is possibleto combinesetsof cartesiandisplacernent coordi-


n a t e s{ r 7i1i n t o s o - c a l l e d s y m m e r r y a d a p t e d c o o r d i ' a t e s
{ e r . i ) , u , h e r e t h e i n d efx
labelstheirreducible representation in theappropriate poinlgroupand.7labelsthe
particularcombinationof that symmetrv.Thesesvmmetry
adaptedcoor<Jinates
can be formed by applyingthe point group prqectron operators(that
are treated
in detailin chapter4) to the indi'idual cartesiandisplacenient
coordinates.
To illustrate.againconsiderthe H2o nroleculein the coordinate
svsterncle_
scribedabove.The 3N:9 massu'eightedCartesian<.lisplacement
coordinates
( X r _ .l ' r _Z
. t . X o . f ' o . Z o . . l n , l n . Z { c a n b e s y m n r e t r y a d a p t ebcl l, a p p l l , r n g t h e
following4 projectionoperators:

Pu,:)*o,.+o,,*C;. ( 3 . 3I t
Pt,, : I f 6r,. -o.., - C':. (3.31r
Pt,,: l-o,.* o., -C:. ( 3 . 3)3
% : : 1 - o , , - { , ,* C . . ( 3 . 3)4

to eachof the 9 originalcoordinates (the symbolo denotesreflectionthrough


a plane and c2 meansrotationaboutthe molecules C2 axis).
of course.one
v'ill not obtain 9 x 4 : 36 independentsymmetry adapted
coordinatesin this
manner;man1,identicalcombinations will arise.andonly 9 will be independent.
Theindependent combinationsof c 1srvtnrctn.(normalizedto produce'ectors
of unit len_eth)
are

Qut.r=2r":Jf;-)fp1. (3.3
5)
Qo'.:.=Z'"Iii+ln]. (3.36)
? . , . - :: [ f o ] . 13171

Thoseof b2 symmerryare

Qbrr=2'/t[Xy+Xp.l. (3.8
3)
Q v z : 2 - t i 2 7 Y y- Y p l , (3.39)
0u,: : [Xo]. (3.40)

and the combinations

9 b , . ,= 2 - ' / ' l z t + Z p . l (3.1r)


Qa,t = lZol (.3.42)
are of b1 symmetry,whereas

Q o r . ,: 2 - r / 2 1 7 y - Z u ) (3.43)

is of a2 symmetry.
Normalmodesof vibration 113

Pointgroup symmetry of the harmonic potential


Thesenine Qr.i areexpressed
asunitarytransformations
of theoriginalmass-
Cartesian
rveighted coordinates:

Q r . ,: L C r . , . , , ' Y r (3.44)

Thesetranstbrmationcoefficients{Cr.ir} can be usedto carry out a unitary


transformationof the 9 x 9 mass-weighted
Hessianmatrix. In so doing,we need
only form blocks

Hi., : y. (nt r, )'''t C r., r (3.15)


L, C 11 H1,.1,'1,m

rvithin which the symmetriesof the two modes are identical.The off-diagonal
elements

pt y )-t i' Cy.1.y


Hi.l' = L Cr.,.rHr.r,(m (3.46)

vanishbecausethe potentialV(qi) (and the full vibrationalHamiltonianH :


T + V) commuteswith thc C2, pointgroupsymmetryoperations.
As a result,the 9 x 9 mass-weighted Hessianeigenvalue probler.n
canbe sub-
dividedintotwo 3 x 3 matrixproblems(of ar andb: symmetry),one2 x 2 matrix
of b1 symmetryand one I x I matrixof a1 symmetry.The eigenvalues of each
of theseblocksprovidethe squares of the harmonicvibrationalfrequencies. the
eigenvectorsprovidethe normalmodedisplacements as linearcombinations of
thesymmetryadapted{Qr. il.
Regardlessof whether symmetry is used to block diagonalizethe mass-
weightedHessian,six (for non-linearmolecules)or five (for linearspecies)of
theei-eenvalues will equalzero.The eigenvectorsbelongingto thesezeroeigen-
valuesdescribethe thrcetranslations and two orthree rotationsof the molecule.
Forexample,

I
(3.,17)
*lS1*Xp*-{ol.
I
(3.,18
)
*[Yt+]h*rol'
I
(3.,+9)
*tzt*ZP.*Zo)
arethreetranslationeigenvectors
of h2.ctl and bl Symmetry,and

I
- (/ aZ r - Zr<) (3.50)

ts a rotation(aboutthe _r,-axis
in Fig. 3.2) of a2 svmmetry.This rotationvector
canbe generatedby applyingthe a1projectionoperatorto ZL or to Zp. The other
C h a r a c t e r i s t i cosf e n e r g ys u r f a c e s

Symmetric
and asymmetricstretch / tn.'.
m o d e sa n d b e n d i n g
mode of water.
O
,/"

Symmetries
of vibrationsof
methane.

two rotationsare of b1 and bt symmetry and involve spinningof the molecule


aboutthe,t- and:-axesof Fig. 3.2,respectively.
So,of the nine Cartesiandisplacements,threeareof a1symmetry,threeof b2.
two of br. and one of a2.Of these,thereare threetranslations(a1,b2, and b1)
and threerotations(bz,bt, and a2).This leavestwo vibrationsofal and one of
b2 synmetry. For the H2O exarnpletreatedhere,the threenon-zeroeigenvalues
of the mass-weightedHessianare thereforeof a1, b2, and a1 symrnetry.They
describethe symmetricandasymmetricstretchvibrationsandthe bendingrnode,
respectively,as illustratedin Fig. 3.4.
N o r m a l m o d e so f v i b r a t i o n 1't5

Themethodofvibrationalanalysispresentedherecanwork for anypolyatomic


molecule.One knows the mass-weightedHessianand then computesthe non-
zeroeigenvalues. which thenprovidethe squaresof the normal modevibrational
frequencies. Point
groupsymmetrycanbe usedto blockdiagonalize thisHessian
and to the
label vibrationalmodesaccordingto symmetryas we showin Fig. 3.5
for the CHa moleculein tetrahedralsymmetry.

You might also like