• Aucun résultat trouvé

of partial

N/A
N/A
Protected

Academic year: 2022

Partager "of partial"

Copied!
95
0
0

Texte intégral

(1)
(2)
(3)
(4)
(5)

LIGHT SCATrERINGIN BI NARYGASHIXnJRE S:

EVIDENCE OF FAST AND SLOW' SOUNDHODES

By

LuoHongda, B,Sc

Athesis submitted in partial fulfillment of therequirements for thedegreeof

HasterofScience

Departmentof Physics HellorialUniversity of Newfoundland

August1990

St.John's Newfoundland

(6)

1+1

The authorhasgranted en Irrevocablenon- exctusiveIioenceallowingtheNaliooalLJbrary ofCanadato reproduce,loan.cfcstributeOfSEll copies of hislherthesisby any meansandin any formor format.makingthistfcstsavailable

to Interested persons,' .

The al.tthor retains ownership ofthecopyright in hisfher thesis.Neither the thesis00(

substantial extracts fromit maybeprinted(l(

otherwise reproduced withouthislherper.

mission.

L'auteuraaooon:Seunalicenceirrevocableet non exclusive permettanl A Ia BlbIIotMque nationale duCanadade reprodulre, prater, cflStribuer au vendeedes ooplesdesa lMoo de quetquemaniere at sous quelque fonne que ceeottpour mettre des exemplairesde cette these

a

Iadisposition des personoes interessees.

L'auteurCOOSM'eIa proprietedudroitd'auteur quiprotegesathese.NiIathesenides extralts substanliels de

ceue-cr

ne delvenl 6tre imprim6sau autrernentreproduits sans son autortsatoo.

(7)

In this theah Brillouinscatte r ingdata are presented for th reedifferentbinary gas mixturescomposedof specieswith wide l ydi f fe r en tMasses,i.e.AI'+H:dmu / mllz...20),SF,+eH.

<tnsFs/lllcll. "9),andSFe+HZ(tlls F 6 /~z..73).Underthe expe rime nta l co ndi ti o ns predictedby the theoryafast sound modecontr ibut io n to theli gh t scatteringspectra of AI'+H2. mixtures has been detected. Unlikethe ordinary soundmode the fast soundmode pr ops g a t e s onlyinthe lightcomponent with a velocity higherthan tha tobtained usinghy dr odyna mi e the o ry. However , theatte mpts to observe thesame eff e c t in IdxtureaofSFe ... eH. undersimila r condi tions vere unsuccess ful.An an a logousslowsound mode con tr i b ution tothe spectr aofSF6+H1mixtures ha s been clear l y identi f i ed andit ....as found thbmodewas, ingeneral,much easi e r tode te c t .

Anev parameter calledtheeffect ivemeanfreepa t hva s also in t r od u c edinorder to obtain II more consistentcharacte riza t ion ofthedynanlicalbeha vi or in binar ynlix.tu r e swithdis pa rate

i i

(8)

ACKN01lLEDGE/'lENTS

I wouldliketo expressIllysinceregutitudes to my supervisor Dr . H. J. cteucer for guidan ceat everystep of this expe x faenc,His experience has been of great help to me throughout th isproject and inthe prepa rationof thisthe sis.

I wishto thankmy supervisor Dr. H,Kieftefor his careful reading of this thesis,his encouragement andvalu ab l ehe l pin the finalstage of thethesis,

Inpar t icula r, I am gratefu lto Dr.A.Campaof th e Rockefel le rUnivereity, and Dr,J. A.Zollwagof the~choolof Chemi calEngi ne e ring , Cornell UniVersi ty for send i ng lIIe the theoret icalcalcul a tions conce r ni ng ourexper imen t,

Thanks goto Dr,J.K.C, Lewis and Dr . N,Richfor thei r helpful di sc ussion snd adviceand toIllyfriends V.Askerpour and Dr. Ri ch ardIt,J. Gouldingforwh a t thevhave donefo r me during Illy stay at Memori al University of Newfoundland.

Ialsowishto gratefully acknowledge the financi al support of my work from Memorial Uni ve r si t y ofNewfoundland.

iii

(9)

2.1 Dens it)' Corre l ati onFunctionl&

BrillouinScattering Chap t er1.

1.1 1.2 1.3 1.4

Chapte r2.

2.2

In troduc t ion

Fluc t ua t i on.&Sril louinScat te r i ng ThaK. anFreePaths:inSinnyHixtura . Hydr odyn a,lIic&Kinetic Regllllu llyna mic s of Dlsparate-Hass Gas Hhltures

TheFast&510..,Sound Hoda .

Tha or y

Hydrodyn a lllicTheo ry

12

12 22 2.3 KineticThe ory&Predictionof a FaItSound

Hodein Dispara te-Han Mixture s 25

Chaptor 3. Apparatua and£xperil'llen t a l Procedur e

OpticalSy stem

iv

33

33

(10)

J.2 Data Acquis i tionand Stab il izatio n

syseee(MS-I) 19

J.J Gas HandlingSystem 4l

Chapter 4. Experimenu.l Res ul tsandDiscussio n

"

4.1 GeneralRe mar ks

"

4.2 TheFu t Sound HodeInHz+Ar Mixtures 47

4.3 CH4+SFaMixtures 54

4.4 TheEff ect iv e Mean Free Path

"

4. ' The Slow SoundMode in SF, +Hz Hhture!l

4.' Conclu5ions 66

Referenc ee 72

Appendix Pub licat i onRepr i nt 76

(11)

LISTOFFIGURES

Fig.1.1 Classical pic ture ofI1ragg reflection .The incidentlightwaveis scattered by the densit.ywave1nt.he medium.

Fig. 1.2 Schematicrepresen tationof the lOcatt e r e d light epeecrua fraila fluid.

Fig.2.1 The scatteringangles definedin relation 18 (2.10)

Fig.3.1 Schematicd1s.gralllof theexperimenta lsetup . 35

Fig. 3.2 Typical sp e c trumre cor de d in I1rillouln 42 sc atteringexpe rimen t.

Fig .3.3 Schematic d1s.gr.ll1ll of the gashan d li ng syseee, 44

Fig. 4.1 Spect raforAr +Ha lIIixtures with a H2ba s e SO pr ess u reof 9.0ba rs,

Fi g.4.2 Spectra for Ar+H2mixtureswith a "2ba se 51

(12)

prassure of "1.7 bars.

Fig.4.3 Spectra for Ar+H1 lIIixtures wi t h at!1base sa pressure of6.3 bars.

Fig.4.4 Spectrafo r SF6+CH,mixtureswithII CH,

"

bas e pressureof 6.Jbars.

Fig .4.5 SpeetraforSF6+C;H~ mixtures'11t h IICH~ 57 base pressureof 3.7bars.

Fig.4.6 SpRct ra for SF6+H1 mixtureswithIl SF6 68 basepressure of 1.7bars.

vii

(13)

Chapter 1. In troduction

i.1.1. FLtJCTlJATIONS "IIRlu.ooI NSCAlTEll ING

The scatteringofligh t generallyarheaalitberesult or opt ' c al inhomogeneitie sin the scatteringmedium.The physical reasonsfor thelen er8tlonof opticalinhomog eneitiesin the scatte r i n g lIIediUlll are vari o u s.ForeX/I.IIIpl.,contalilna t i o nby for eign sub s t a nce. 11 one possibility.However,even insub stance s whi charecOCIpla t l l yfr••fr~anykind of fo uigncontamin . nt a. su t t ered Hshtcan baob••eveealltheresultof theata t btlcal cha ra c t e r of theraalmot i on .In thi sca . ..til,opt i cal inho.llKlgen e i t i e.l arecaus edbythe local fluctuation.of the optical dhlactricconstant,which are directlybrough tabout, in tum,bythefluctuatloruIn loc aldensityofthesubsta nc e.In contrastwith theimpurity sc atter ing and R...nIca t te ri n g ,this kin dof lightBcatU ri ng,arising from randollthermalfluctuat ions is called Rrillouinscatteri ng !l].

According to Brillouin,thespontaneOUlildensity£luctuationlll in themed i um can bedeco .posedintodifferentplane -waveFourier

.1.

(14)

ceeponents,These plane-wave apropag atinginthe med i Wl caus",a periodicdenslty changealong thepropagat iondirec tions. Scatte r ed lightisbr ought abou t bythe diffractionof the inciden t plane monochro mat i c llghtwavewhichis cOJ1si dl!lredto be diffractedby thedens ity maxi lllaofthe thermal501.1(, .1lO'ave in the same wayas X- r sys ar e sca ttered bya crys tal. Ifleoisthewavevec t or ofthe inci dentlight,kris the wavevecto r of the scatte red liGht andK i.thewave vectorof theKth Fourier ecepe neneof thesound wav e caus ed by the spontaneousloc aldend ty fluc tuatiofUI of the med i um in direc ti on

i ,

thenthecond it io nfor sc a tte ri ng is givenby the Rraggrelati on [Fi g.I.l )

K-±<0, •

k,> .

{I.1)

where the ehoice of slgna referstoeit he rof two oppositely directedsound wa ves .Becausethedensl ty max i llla ar e travelling withtheveloci t y ofsound inthefluid (fl ui d refersto both gas andliquid her e),the frequencyof the scattere d lightwUI eonsequently baDoppler sh i f ted_If1010 istheanguhrfrequen cy of theincidentlight wave,IoIf isthatofthesca tte red wave,v. is thesound vel oci t y,,isth e sc a t te ringangle. n is th e refractive index ofthe lIedi um,anl1c is theveloc i t y of ligh tin vacuum, thentoahigh degreeof approximation,thefreq uencyshift is gi venbythe woll-knolmBrillouin re lation

~_ "If• "10_±2wo:.n sln(~).

.2.

(1.2)

(15)

'Jr.J.J Claninl picture of Ira" nn.ctton. Ttle inddant.

Ulht. waveIelcect..n'

bJ'

t.h.'.ndtJ weve in t.he . .di_•. 1r.e1athewev.nctor of che incident. U&ht.It,h t.hewe.,.

dendtJ vlve. eneS ,1a t.heleatt.erll\l ancle.

3.

(16)

Although sound waves of many differentwavelen gthsand any propagatio ndirection ar e present in the medium,inan actual llghtsca t te r i ng exp erteencfor eachsc at te rin g angle(Jother than zero only ehesoun d wav eswithwavevectorgivenby(1. :~can be observed,Accordingto (1.2),the scatteringspectrumconsists of on l ytwo sharp l1ne.~with frequency sh Hlillof ±.:1<.l,respectively , frolll the centralfrequen c ywu'In reality, the spect rumofthe sc a t t er e d light alsoincludesa central linewith no frequency .IIh i l t. Lan dec. and Placzek(2) gavethe followingexpl a na t i on for the origin ofthe centr alpeak .They pointedou t thatt.:"t:h erm al fluc tuati ons inafluidwerecOPlposedof two par t s,namely, isoba ric fluctuationscause dbyent r opy fluctuat i onsatcons t a n t press ur e,and adiabaticf1uctuationa cau s e dby pressure fluctuatiQnsat cons ta nt en t r opy. Fluetuationsofthe fintty pe are unorganizedin ti me,90they are not propagal;edintheform of wavesand eons equentlythe centralline eanbee)'~lain..ttas elastic Rayl eighscattering . On the otherhand, thepre s surefluctuations repres entrandollloeal eompre s sionsor rarefa ctionswhich, as a cons equen ce of theelas t i c properties of the mediUDl, do not remaIn n"edbut travel through the fluid and give rise to'"shifred doublet. Because ofrhe ex1steneeof dissipativeproces seswhich da mp out the ela s t icwavesin the med ium,the doubl etccep cnenes haveneaaere Une widths.The nonzerowidth of the centr.alline ca n be explainedin termllof hlUllt conduction and diffusion.A typiea.l Brillouinspec t rum issh own in Fig.1.2 .

•4.

(17)

o 100 Frequency shift

800 I---~

.~ .. c .5 !

Fig.1.2 Sehemat i c rep re s enta t Ion of theBcet te re dligh t apec t r\Jlll frop;,•fluid: long- da sh edlIn erepresentsthe und isplacedRlI.yle1ghCOlllpone n t iahor t -da shed lineeep r e s enes th e St oke sandan t I - Sto k es &rilloulncomponents ;dotte d 11ne indicates.thenon-p r opllgHt l ne; fourth compone nt whichIs much leu intense thanoth en[I];solid line in dic ate sthe aggregat e intenaitydistributioninthe sca tter e d,pect rUlll.

•5.

(18)

§.1.2.TIlE MEANFREE PATIIS IN BINARYfUXTUltES

Theoret.icallythe spontaneousde nd t y fluctuat.ionsin t.he mediU/ll canbe decomposedinto densit.y plane-wavesof different wave lengths.In reality, however,onlyt.hose density waves with wave lengths larger thanor eorepareble to the eeen fre e patht existin the mediUll.Usuall y che product. of K and mean free patht is used to characterizethe conditions of thes8lllples(see§.1.3). The meanfreepath in a low-densityone-componentfluid is given by

(1.3)

where(I isthe har d sphere diamete r of the part lc leandnisthe nUlDber density.

For the binarymixtures t.he sit.uat.ion ia somewhat.eoepHeaced because thereare several possibledefinitions for themean free path. For example ,i tispossiblet.o define alISaa fre e path with respect to collisionsbetween the specific types ofparticles. Here we int roducetwo",eanfreepathsl1 and(1definedbyP]

(I -(t;~+t;~)-I ,

t

1-

(t;t ...

t;~>"l,

(1.4a) (1,4b )

where llJ(l,J-1, 2 ),theaveragedis t a nce travelled by a

.6.

(19)

part i c le of.pe c ie . i betweenit.two.ucce s. iv e co1lllions with par tic le.~f.pe ci. .j. i. . .Iv e nby thefollow ingrela..:on

where nl,171 and_.Ire p r es en t thenW'llberden!li ty,herl!.aphere

If aparti cla of spac ie' 1travelsa dh ta nc eL,acc ordi ng to thedefi n itio nofll J ' th o averagenUlDbera ofcol lision.it expe ri en c es with parti c les of species1 and2aregi venbyLi lli andLillare . p ec tiv el y .The.v erage totalnumber-ofcollision.ia given by Lllll+LIlli' HenceL1 ,whichequalstoL!(L/ll1+L/ll~)' canbe understoodasth e average dis t a ncebet we entwosuccessive col11.10 n .exp e rie nced bypa rt.ic l e s of.pee h. 1.However , as we viII seein Chapter 4, forthebinary .. btur. .wi t h large lila••

differenceL1andL,are in.uf f1 c h nt to characteriz.ethe conditions of binary fluid •.A possible .ol ut i on is viathe -.ffectiveICe anfre e path-whi chw11 1 be definedanddiscussed later.

§. 1.3. HYDRODYNAKIC~KINET IC REGlKES

In asimp lefluid, ..cihlen.i onlen p.u8meterKL(Lis th e llIeanfre epathofparti cle.in thefluid) iaus edto char8 ccerl ze

.7.

(20)

thenatur e of thedensityfluctu ationsto beprobe d{4].The in d i v i dual particles in the flu i d travela finite distance.(

be t we entneer-par ticl e co l lis i ons ...bic h ca use loc a ldensi t y fluctuat i o ns .SinceKisthewav e vect o r of the fluctua tions to be prob ed, a smallvaluefor Klmearvi!thatwi thinthe leng t h scale IjK(thewave-le ngth oftherma lsound wave) many c.Jllisionsoccur.

Sointhe limi tofK.t«l(hyd rod ynam ic reg ille ) . che deviationfrOlll local thermodynamic equll i briUIIIis verysmalland hyd ro dynamic theorycanbe used.On theotherhand, for Kt -l(k inetic regi me ), particles under gofewcol l h io ns over th e11Kle n gt h scale and tihe loc al equilibriumassumption cea s e s tobevalid.One mustuse kinet icchearyto obta i n aDor e ge nerelBnd more lDicroscop i c des c ript i on.Theregi on Kt»l is thefree -s t r eatllingre gionan d yieldsthe well-known Dop p ler profile, which 18not of inte rest here.

Ina bins ry flu i d, the situationIs 1II0recomplica te dsinc e th e r e ar eseveralpossible defini t i on softhelIIE1a n fr e epath.Mor e de taileddis cus sions wil l be give nin Cha pt er4.

§. 1.4 . DYNAMICSOF DI SPARATE-MASSGAS KIXTURES TheFastandSlow SoundKol'.es

Theinvestigat io nof binarydisp arat e ' lIIa ss mixtur e s,wh ere thetwocOlllponen ts hav e verydiffe rentma SS6S ,be gan In1960.At

.8.

(21)

thattime CUd ( 5)potnted outthatin a dilute dlsparllte·.ass .ixture theslow.."ch angeofki ne t i cene r&), be t we en two eeep en enee shoul dleadtotva te.peratu res:one is a..ocla te d with thelIgh t componentandano t he rh a. .ocle te d withthe he. vy coaponent.In th eae ee 1970a,wh.n Huck and Joh ns on (6 ) studied thebeha v i or of forc edsoundIIlOdu indilut ebinary dispa rat e-INII ga. lIi xtu re.of

their stud1e. ,theypredictedtheexla te nc:.of two diffe re n t forcedsound lIIode a, n. .el y a fastsound IDOde and a slovsound ec d e.the fas t sound mode b associatedonlywiththe light component and propag a t e svlth ave l ocit y grea te rthan that expectedfor hydrody nami chehaviorof the _ix t ura, whilethe sl ow .oun dlIode Sa Ju. ttheoppoaite: it1&aa.oc iated withthe heavy co mpo nen t andpropagat. awit h a lower va loc:i ty.In the expe r i_nta con du c t e d by !owlarand Johns on (7) in dilute bina ry. b tur e. of H.

and X" a ahArpincr•••e in aou ndvel oc itywa .fou ndund eran ex te rna l dSatu rb a nc ewithfr equenci u w>lO' H:z endHe .ale fractions hlghe rthan0.45.Hore rec ently ,in1986. Boase.J.cucc1 and Ronc:hen t t 1(8)found a fast propagating ao und1II0deineomputer aimula t i onsof LiPb liqu i dal loys withla rgeatOld c. . . . differenc e(.'1l/~1.l .30) .Theypredi c ted. forhigh fre quenc::ie sand la r g e waveItWllbenbeyond the hydrodynaJllicregi. a ,andunderthe co nd iti on ofhI ghconee n t rat i on of theligh tcomponent , thatthe fast sound couldbeobs ervedby tbe usinginela sticne ut r on sc a tte r i n gtech ni que.ThSa was eonfitllled inexpe r i llen u by Hontifrcoi.1,\lesterhulj andHa. n ( 9 ).In theirlataa t paper,Call1pa

. . .

(22)

andCohenllOJconsidered light scatteri ngin dilutedisp arate -mas s bi na r y gaslIIbttures with hi gh concentrationsof the light componentand predictedthata fastsound mode shouldalso be obs e rv e d as the pres enceofs1depea ksor extendedsh ou l de rswh ich ar e onlyass ociated withthe lighter compone n tin the scattering crossse c ti on for visible light.For a binary mixture, the scetteredlight intensitycanbe expressedasa weightedave rage of partialdynamicstructurefactors, (see detaileddiscussion in Chap.2)

where subscripts1and 2 refer to thelighterandtheheavier spe c ie srup. c t i v.l )' .'"I and Xl deno te th.lII018cularpolarizability and thelIIole fractionof component1. K and ",arewave numberan d angu l ar frequenc),to be probed,SlJ(K,w)are part i al dynamic struc t u r e factors.The first and thethird termsin(1 .6)are associatedwiththe lighterend heavierspeciesalone and the secondis amixed term.As we cansee from expression (1.6),the polari z1bilitiesoftwo componentspl ay tepereene rolesin ligh t sca t t e r i ng. In orderto observe the fast 1D0dewhi chis only associated withthe lighter component ,01/01should be aalar geas pOll8ib le(atleas tcomparable ).Conside ringbothmassand polar izabi1ityratios, the mbtturecombinationwhichCampa and Coh e n suggestedto studywasargonan dhydrogenwitha lIIasaratio of 1lI...~/ma 1"20and polarizability rat i o ofa-8/QA~"'O .5.

.10.

(23)

Althoughseveral llght-scattering expariments had previously beenconduetedon binarydisparate-masagas mixtures by Cornall fUJ, ClarkI12,13)and Le.tamendia(l4 ) ,the experimentalconditions and the mixtures theychose we re unfavorable[01"thaobservation o[ thefastsoun d mode. For eXlIlllple.inClar kand Letamend ia's exper i ments, th eyus ed xenon and heliwa("'x./~ ."'33)llIixturl'!s With a pola rh:at.lonratio of(1 . .1°110"19.6,muchhigherthan thatof Ar and H2Besidesthis,accordingto Campa'scaleulations , in order to ebs e'rve the fast sound 1Il0de the gas density sh ou l d be IllUch lower and ala r ger scattering angle isalsopreferred. In this thesis,wa willpresent recent;light scatteringresultsfor dilute Hz and Ar mixtures,which provideevidence fo rth e existenceof a fast Plods.Our sttelllptsto observeth e sallie effectinmix turesof CH, andSFI (tIIg'I/IllcIl,",9, Qsrll/ocll,-1.8)wereunsuccessful, Howeve r,for th e gas lIixtu resSFI and Hz (tllar/lIla

z"73), an analog ousslO'IlIode was observed.Compared wi t h the fast mode, i t seellls thst:the dow IlIOde iamucheasierto detect.

Before presenting theexperimenta lresults,itill necessaryto giv e a mor e detaileddescrip tion ofthetheo ry1nvolved in Br il l oui n scattering.Inth efollowi n g chap ter,we will out li ne thetheoreticalbackgroundpertaining to the therlilal ligh t scattering.The experillle\lta laapects willbe contained1n Chapter3 (Apparatus&Expe r 1mante l Proccdu t"e)and Chapt e r4 (Experiment alResults&Discussion).

.11.

(24)

Chapt er 2. Th eory

J.2.1. DENSITY CORRELATIONI'UNCTIOIS 6r,lULLOUlNSCo\TTERING

Thedensi t yfluc tu ationof fluids 15 usua l lydllScl:'l be d In thelanguageof spa ce·t l l1le correlation functlon aI15 ] .A space- tll11ecorrela ti on funet i on11 defineda. the thermodynamicave r age of theproductoftwo dyn.lllli calvariable .,eachofwhi ch -xp resstl.

theinsta n taneousc!ev..",t !onofIIfluidpropa r tyfrollIt a equ i l l · brllJJ11 value.ta pa r ti c u larpoin tinspac.andtilDe.Ceneral ly

C...(r,t; r',to)_C,<'A (r',t')"6(r,t»

- cu c l r · r· l.

t·to), (2.1)

where

U(r,t) ..A(r,t)•<A(r ,t» . 6B( r , t )..lI(r,t)•<8(r , t» .

are thefluct ua t i on.in the d)'TI&lllcalvariable .A{r,t)and8(r,t ).

•12.

(25)

Thefa ctor Coisa con.\ltantdefined fo r theconv en hnc eof the spe cific cor r elat io nfunct ions. Theangle bracke ts represent av erage s over the phas e coor d i na tesof allthe lllolecules Inthe fluidwithsn equi l i ht:i ume naeebkeas the wei ght ingfunc ti on,i.•.

<A( r, t)~

-

Z(JJ~V,N) -I

er ACr,t) · e· jlH( r) , (2.2)

wherercollec tivelyindicates all thephase spacevariab les , H(r) i_theHami l t onianoft~ee.ye t efl withN par t i cle llin volUllleV, ,B"C k!T)"1 ,andZ(P,N, V) is thepar t ition func t ion of theSY.\Itern.A space-ti mecorre latio n func ti onisth ere f ore a func t i on ofspace andtime , and it describesth ethe rma l flu c t ua t i onswhich edst spo nt ane ous l y In the equ ll I b r IUIIIsyst em.

Using cla u icalel ec t ro dynam i c theo ry the relatIonbetwe en the approp r ia teco r r elation function s an d the int ensity of Brillouin sCBtte r i n g can be obt ained.In order tosi lllp li f ythe dh;cuu i on we assumethBtth efluidsarecompos e d of optI ca lly ls otropic1l101ecule s , e.g.thein ertga se s.In this case the fluctuations of thedielectricconstan t causedbyorientationsof tho 1lO1e cui e s in the medIum do not need to be considered .

ConsIderinga fluid withan av erag edielectric cons ta n t co' thedIele ctric const ant w111flu ctua t efro lll place topla c e and fromti lleto timegIvIngrhe toinholJlogeneities.The instantaneousdi e l ec tric cons t an t t(r ,t)can beas sWIledtobe

.13.

(26)

equal!:> theavenge(0 pl usafluctuationpart Ac(r,t ), I.e.

I(r ,t) ..It+AI(r, t) , (2.3)

when rh the podtlon In the Ica t te ri ns .e41 \111.'nilvol Ullleof the lledlUIIIh ...-d tohe largesothat su r fat eeff. c tslIlaybe neg l ected.Be c au 5Ithe -.ollcu le. areis otropic ,( andA(are dmpl i f1 ed toheIcalars inlteadof tensora (16 ).

UsLng thlalocaldl.lectriccons tantin theHa x\lellequ a t i on s for anonconducting,noruaagn etlcmedlUJI,wehave

{ VVxVD)(.. 0E....(1/cZ)(a 'D/8 t ' ) (2.4)

",h. n Eh the elect ric fh14 vec t o r Inth e _dl UIII,.ndD11the Hnvell4i . placeDen tveeeer "hlchs.tisfies thefo110\11nl nlat10n

D...(r,t)•E.

'nI.equati ons tha tneedtoIHtlolve dare, there f o re (2.5)

In cla1ll 1c alelectrodyn4lD1e pertu r ba t i on theory , the sca t tered eteecrtefield E c.nbe expande d 1n termsofthe Inci den tf1eldKo plush11he rorderte nuwhlchonlyIncl udethelighttca t te re d In

.14.

(27)

oth erthan thedi rect ionoftheincidant l1r,ht,l.e .

E-!o + E1+ t; · ·· · ·· · · · ·

He rewe assUllHl that the inte ns ityof. ut t e re d Ur,htvit h (2 . 7)

Beatt eringans la,..011 alllallc:oraparedto thein tena i t y of the incide nt llsht;thh conditionunbegen eraUy real b ed inguea

Sublltitut ingequ a ti ons(2.3)and(2.7) Intoequa tion s (2,6) and setting aWIIIof th e aame or de rof~gnil:udeequa l tozero,we ob ta i naseriesofequa tions:

VZ",-

3- . ~ -

0

c "

...• • .etc.,

(2.8a)

(2.8b)

I!o,whichrep re aentatheincidentli ghtwavewill (acco rd in gto theapprox!Dlation)betrans lIit t .d throughthafluid wit hou t attenuation.£1 can b. consld.:!red asthere .ultofaalngle acattering proce..:I.e.In th a quantUII Dlechanl cald. a crl ptlon . photons represen t ed by £1 are sca t t eredon lyonce byphonon sll7].

Th.temof Dlost intereat In (2.7)h t;" th eling le scattering eera .The contributions ofEaand ot her hi gh er order sClltt.ring

.1 5.

(28)

terms areverysmall exc ept inthe neigbljourhoodofa critical point,

The incident fieldis assUlDedto be a planemonochromatic wave of the fQrm

Eo-Aoexp(iko'I'-iWot), (2.9)

when koisthe prQpaga tionvactor ofthe incident light,andWo isits angularfrequenc)' ,SinceEosatisfies (2,8a),we abtdn the relationleo-(I:O) I/2·("'o/ c )

E" in equation (2.8b)eanbe solv edfor by doubleFourier

transformationin spaee andti. e,lJhenR,the distan<:efro.tho origin toth e pointofobservation,is muchlar gerthan the la r ges t dimens ionofthescatter i ngmediUIII. the space-ti_

Fourie r tl'llnsforlllofE1(K,ltlt)is givenby[18]

xJ.,dr3 re XP(u ::'r) ]

~2

[bC(r,t) a XP(- iwot )].(2.10s) 21(It,t)...-

~

R ,

~

4cI(.exp(ikcR)

,16.

(29)

wh e n theSpliceIn t e gr ation1. overthe veh.eofthe sutte rlng

Incidentandscattere dli ght,• 11 theangl ebe tween the deetrlc ve ctorofthe IncidentwaveEoend theprope g3 t lonve ctorof the scattered"Ive~, 1M (,Ir11 the angu lar frequencyofthe lJcet t ered "ave(F1S.2.1).Inprsctleethe vari ationofl&(r,t). which 11 causedby fluctuAt i ons In the_dh~,11 Mleh 110\18r compare d to the fre que n c yof theInciden t l1&ht(Ibout IOUHz) . t t lder ivatives with re.peet ....tl.e can eon. equ e ntlybe Ignor ed compare d totheti lDe.der lva t lvelof theIncidentfield,Ind (2.10b) bec ome '

(2.11)

The Intensityof th ese e t t ere dlightcanbedefinedby the re latlon(19),

I(K, wpt)-

~ I <~(It, t) ' ~(It,t'+

t»exp(lw,t ') dt',(2.12 )

'Where th eangularbracketa denotetheaverar;"with re sp ectto a .ta tlona ryequil i br ium.ense mb le.E~isthe con j uge te of£1 'lJe ah o not efrotl theprop erties of thecanonicalens emb le,th a tth e ensembleave ra s eof • dynamicalvar i ableact ually do e . notdepe nd

.17.

(30)

i- k.

Incident.

light

SC.1ttered light

·f

FJg.2.1 The angle. usedIn (2.10) are defIned her..tbI electricfield in the IncIdent be. 18 along ther.

direction,andtheInc I de nt be_ propagat e .in thex direction.Iand' are the angl. . betweenthe prop_guion direction of the scatteredbeall. aOO the zandJI:axe . respectively .

18.

(31)

one, that la,

(2.1 )

Ualngthe !;"eeult(2.ll). (2.U)and(2.13 ),we thueobta in the frequencydistribu t ion of thescatter e d llr,ht al

Conal deringabi nary ..btu!;"e whichIs composedof Nt 1I01eculeaof.pee les 1 andNzlIol ecul e s of.pe c le e 2 Invol ume V (1fNzooO,'Weean eaaUyobtainthe resul tfora daplefluid) , the av era ge partillnWllb er densities.lire~-Nt/V,n:-N1/Vand the tota lrlI.lIlIberofmo l.tul ea 11N-N1+Nz •'Letr~and r:denot e the posi tions ofaole eu le numberi ofepec l e s1and 2,re s pe cttvely.

Then the die le c tr i c con s t.n tc.n be expanded Ina tayl orser IesIn the loea lden.tty aboutthe.verag'densi tyn&(i ..l, 2 ) so th a tlit In (2. 3) beeo....

.19.

(32)

vherei•1.2,... . ,NI;j-1,2, ·· · ·,Np

Bysubs t i t uti on of (2.15)into(2.14),'lieobtain

··•• ..(2.1 6 ) Afc er fureherdlllj)l1 !lu tion (2.1 6) c:anb.vrltcen . .

1(1.,"')orNI"~Sll(I., w)+ N!"~Szz(I.,w)+(N1M! )1 / 1"t " ! SU( I.,,,,,) +(N1N!)I /!(lI"!Su( J::,w) (2.17)

wherew..Iolt·ltIt ,andell(wh ichb propor t io nalto (8c/ant )n l_nA (19]) b thepol a rl u bil1tyofspe des1.TheSIJ(i,j -1,2),which are ueua11ycalledthe·pa rti al dynamicst ru c t u refactou·,are definedby:

.20.

(33)

where

(2.19 )

Sincefor clas :Ji calfluidsinequ llib riUlllallthe Slj (K,w)are real , and are evenfunctionsofw, we have

SlJ (R, w) - SjI (K,w),(l,j - 1, 2) (2.20 )

and ass\llIIing thedllu t e fluidis isotropic ,SlJ(K ,w)dependa only on

K-IKI .

SO all partIaldyn ami cstr uc t ure fa c tor s also sllt i s f y

SiJ(R,w)-SlJ(K,w). (i,j- 1,2) (2.21)

Using (2. 20 ) and(2. 21 ) we canell511 y see (2. 17) is jus t (1.3) us edin Chapt er1.

Itisclearlysh olom in(2,17) that the intensity of scatteredlight for binaryfluidsisproportional to theweight ed SW'II of the partial dynamicst ru c t u r e factors givenby the space-tilleFourier transformof the densi tycorrel ,ationfunctions. From the defLnltionof dynamic structure factor(2. 18 ) we can also see that 511 (orSa) isonly relatedto themicroscopicpr ope r t i es

.21.

(34)

ofthe l1gh ter(or theh'avh r)IIp e des Ina binaryIIb t urtlvh ile thecro. .te rti 5u 11 nlatedto thebo t h.Ho,tof thetheoretical atudiesconce rning th.rmAl ligh tsca tter i ngconc entrateon using di f fe re nt .... thods to calc u latethe partialdyulllic st ru c t u re factors .thereafterthetheoreticalre s ul t.c.n becompa re dvl th experbent.A.vu Dentlon edin Chapter 1. In theliliit vhe re Kt <<1thedendty fluctuations Ine flui dere gov.rnedby hydrodynamictheory.vtllle for thecuevh e re Kt_l.ontheoth e r ha nd, kineti ctheoryh nec e ss ary .Inthefollowing twosectio ns vewill IIho w how thepart i a ldyn amic struc tu re raetor s are obt a i nedinthes,two di ffe r en t re g illes.

I.2.2.HYDRODYNAM IC'nfEORY

In the bydrod)'f'Qlcde .cr l p t ion. the fluctua t i onsofthe loeeldie lectricCONtenter e reletedtothefl llCtv.atlon s ofa compl e t eseeof loedthenaodynamlcqua l'ltl th satsuch as pre ssure,tellperat u re and conce:ttre t l on,I.e.

(2.22)

.22.

(35)

Any th r eein d epende n tvar i3 b lea willauf f i c e for binary flu i ds.

butcertal nchoices ofvar i a b l es may provide.lIloreconv enie nce for thecalculationthan others [ 20 ].

Thespac eendtilllores pons e ofthe systemto a deviation from equlli b rlUllliscalculated by usingthelinear h e d hydr odynll.mi c equationsand initi alvalue sfor corr elationfuncti onsprovlde d by thermodynamicfluctuation theory.If (T,p,c)ar e chose n,then thes e equat i onsar e thecont i nui t yequat i on,

[~]+PoV·v-O.

thelongitudinalpart ofthe Navi er-S tokes equa tion, (2 . 2 3)

the diffusion equa tion,

andthe energytran sport equat i on,

(2.25)

Inthese equations,all the equilibrium values are denotedbya subscrip tzero .pis the den3ity.visthe ma ss vel oci t y ,jJisthe

.23,

(36)

che.icalpotentiA lof the. h tu rel2 1l,'I.and'I...arethe shea rand volu.e vIsc ositie.respec tively, Dillthe dlffu.!il1oncoefficient,S istheentr opy,Il'11thethenalconductivity,kt iatheth e r_ l diff usionratio ,andIt,11 • the t".-odyna-i eqU8ntitydefi ne d by

Hencei fwerewriteeqU8tions (2.23)- (2.26)intermsof the apetta! Fouri ertran.f orm.,we ca n obtai np(K, t),T(K,t) and e(K,t) inter lllsof init iAl fluc t uat io nsp(K),T(K)ande( I:). Meanwhile thespa tia lFourie rcr ans f o rlllof thelocaldielectr ic co ns ta n t canb• • •pr. . . .da.

c(K,t)-

[ ~]. ,tP(K, t)+( ~ ]p,..

T(K,t) +[

~

]",C(E.,t )(2.27 )

Usin g(1.17)and(1.14) , th efinal expruaioncanb.obta in ed in thefo n.

where 1.01'

Ao" ". ,

101'lU'1. andWwiare coe f ficients dep end i n g onK.TheRaylei gh peokcanno tsimply be consider. ds.the supe rp os i tion of th.flu ttwoLorentziansshown 1n (2.28) .In fac t , as the resu ltof thecouplingeffec tsbetw sendiffusionlind hutflow1n abinaryfluid, theact ua lcentra l puk isthe re sultantof• .uchIl101'.compl ica te dsuperposi t ion whichconsLsts

.14.

(37)

ofsixLorent ziaoa ( a u re f.20 for more deta il) .

In thehydr odynamic des c ri p t i on thelocalequilibrium assumption isus e d, which.ea ns tha tateach point in aflu i d the samethermodynlllDic rela t i ons can beus e d torelat edifferen t themodynamicvariab les,andthefluidisconsideredtobehave li ke a continuUlll.So th ehydrodynamic theory 18prope r only for sllIaUKandIIIunderth iscond it io n ,and themi c ro s c op i c deteilof local structur ecanbe igno red .

s,

2.3. KINETICTllEORY&PREDICT I ONOF A FAST·SOutmKODE INDIS PARATE· HASS KIX11JRES

In adilutefluidwhereKt",I,hyd r odynami cthe ory CellS1l.1Ito be anappl icablethe ory . Kine ti c theory,which invo lvesamor e de ta ile d microscop ic de s cription, isnee ded. Aa will beBeen later,kinet ictheoryis 1II0re genera land includes hydrodynamic theoryas a spec ificcase . Inthis sect ion,usingkinetic theor y, the dynamic st ruc t u refactor for simpl eflu i disded uc ed .

Ki net i ctheor y is bas...~ontheBol tz ma nn equat i?n whi ch involves tha diet r ib ut io nfunc t ion f( r,v,t) .Inaflu id this distributi on func ti on is define d insuch way . th atf(r,v ,t) t1r tr.v isthe averageflUIIIberof parti c lesatth et inside thevolumeAr aroundposition r,and with velocities wit hin thecange!J.V around

.25.

(38)

v.Because the interllolec ularinte r a ctionrange iswchslIIa lle r than theave r ag eintermol e cu lardis u. nc ein a di l ute flui d ,itis unli ke ly th a t1I0rethsn two particl e awillcollideat the. .me time. Thereforeunder the molecu l a r - cha o s assUlllptionthe Bol t zm&nn equation(2 .29)givesus the ti meevolution off(r, v , t ) :

~f(r,v,t)+V' Vf(r,v , t)-Jd31'Jdn ,,(I, lv ' lI'j) .Iv . v l x(f (r,v',t)f(r,'"tt )- f(r,v,t ) f ( r , v , t )} (2.29)

wherev' and v' (whic hare theveloci ti e s of two partic les,after abin a r y col11sion, within i ti a l velocitiesv andv,respectively) depend onth ei r initialve l oci t i esand thesc a t t e ring angle (I,f1) in polar coordina tes(po la raxis para lle,i. tov -").0 in(2 . 29) isthesolidangle and I1(I, lv - ..

I)

isthe differential collision cross section.

ThelIlost importancresul t of theBol cz mann equat i onh that for anygiven initial dis t r i butionfunct ion,f(r , v,to).the snlu t i onof(2.29)willapp r oac hthe absoluteequllibriUlll Maxwell dbtr ibution functi on:

(2.30) loIh e re IIisthe av era genUlllber densi tyof thefluid. Thl.res ul t is also known as ".Bol tzma nn 'sH· t heo re m".The asswoptionwhichfol l ows isChatequa t io n(2.29)can bel inea riz ed because the dis t rib u ti on fun:: t i onf(r,v,t)15 close toitsequllibr1 U/1lfo rlll, fM(v).Ifwe

.26.

(39)

defin e a small perturba ti on of the distributi onfunc tion as bf(r , v , t). thenf(v , r, t) -bf( v. r.t)+fll(v).Substitutionof the latter fortllint o (2.29).whilecon side ri ngonlythe first order of bf ,yields thelinea rized equatio n for bf(r,v ,t )i

t-lbf( r,v, t)]+v·Vl bf(r.v,t»)- ....B[bf(r .v.t)l. (2,31)

where~isaline ar integ ra lope rator whichactaonly onv.

Nowweint r oduc ethe dynll.llicvar i ablei(r, v, t)whiehisgiven by

(2, 32)

It ca n b. see n tha t i(r.v,t) de sc r ibestheparticlea' position and veLoc Ley dls tributi onat ti me t,and that itaensemb le aver ageis justequ altotheequ ilibri UIIIdis tributi onfunctionf ll(v ),while itspe rtur bationbi (r ,v,t)- i(r,v.t)• fH(v )sa tis fi esthe equation(2 . 31).Accordi n gtotheOn.sager re gr es sion hypot hesisin ste t i sticalmech anicsIl 9 ],thedecay of mic r osc opic spontaneous fluctu at i on s of a dynamic variable,andth edecay of its correlationfunction ,on theaverage, followthe 5aflle li nea rized equation . Therefor ewe can replac ebf (r,v,t ) in(2.31) withthe correlationfunction

C(lr·r'r.v.v '.t) -v<t.i( r ,v , t) t.t( r ' ,v',0». (2.33) end obta i n

.27.

(40)

Ftc(lr.r'I,v,v',t)+v'VC( lr · r' l ,v, v',t)

-ABC(lr. r' l,v, v', t ). (2,34)

Because of the 1sotropiccond itionin dilute fluids,the correlationfunctionin(2.3 3 ) dependsonly on the modulusof the relative position ,and the factor Vintroduc e dthe r e is for calculationconvenience, The n thespatialFouriertrsnsform of (2.33) ca nbe wri t t en88

C(K,v,v',t)- Ji lr.r' l exp{-iK·(r.r ')}C(lr-r'I,v,v' ,t)

where

ll.tu~.,v, t)

-l' t 6 (V-vp ( t ») ex PI . i K'rp (t) ] . (2.36)

and becauseof hotropyat equilibriUbl ,C(K,v,v ',t )depends on K - ltel.The equationfor C(K,v,v ',t) 18 ob tained froll theFourier transfom of equation(2.34)

~C(K,v,v ',t)+ite'vC(K,v,v' ,t)- B( K) C(K,v, V',t), (2.37) wher e IHK)is aUnearinte gral operato r.Us i ng (2. 18),(2.19) (consid eri ngcee pen en es l,j arethe slllIle), (2.32),and(2.33)the Btucturefactorcan iWDediatelybe obtainedas

.28.

(41)

S(K,w)..~Id teXP ( iwt)[Id' Vd3v'C(K,V,V',t )] , (2. 38 ) soth e keyproblemrellla1ningis tofind4'Way of calculating C( K,v , v ',t).

Since the BoltzlIlannequa tionisva lidonl yin the limit of dilute gases, based onahard- sphereecd eICohe nand Campaused a mod i f iedEnsKogthe ory in order toobt a i n the moregen eral resul t s whicharevalid for both dil u t e and dense fluids.And ins t e a d oftryingto obtainC(K,v ,v ' ,t ) directly frollltheequa t i on :r1l11ih .r to(2.37)which provedto be quit e difficult,they attemptedan indirect waybycal c ulat in gan innnita set of correla tionfunc tionsI22,23).ThecOlllpu tat i onsof partial struc ture factonSlj (K,w) areperfo1"llledby aspect r a l decompositionof a time-evolutionoper atorl.z(It ) 1n temsof db creteeigenmodell.The finalresultfor the partialstructure fac t orcanbeexpr esse d asaSUIIIof Lorentzian.:

(2.39)

where the sum run. over theei ge nvalues z"(K)of

Lz

(K) ,and i,j..1,2 denot etwodifferent eompor'lOm tGof the mixture,AI.l ,n (K) ar ethe amplit ude s correspon d i ng tothe eigenvaluez..(l),Nisthe numberofei genva luesus ed,whichischos enon the basis ofthe Bhatnagar-Cro.s-Krooklnethod[23].These dtffeI'ilntelgeRlllodes ca n be interpretedasth e differ entehannels by which the fluctuations deea yinctee .

.29.

(42)

InClolllpa' .calc ulat io n theeigenvalue s of theki n e ti c operator4(K)hll in totwotypes.Onetyp e 15 realandanother is complex.Bec ause ofthe exis tence ofdampin g processes ,th e real parts of both types arene ga tiv•.Thecomplex eigenvalue s alwayscomein conjugatepairs and the i rre al parts re pr e s e nt dampingpr oce s s e s whUe imagi nary par t srep r e s e ntpr opaga ti on pr oc ess e aintwoopposite direc tIonI'.

In tho hydrodynamic limit . (La .for K of suffieientsmall value)all th e 8IIIplitudea Ail ,e(K) are real and contt:ibut lo nsto

the SUlll(2.39 )come onlyfromfour eigenmode swI t h followIng

eIgenvalues

and

~)(K)- _OIKz,

~.(K)__ OzKz,

(2.404)

(2.40b) (2.40c)

wher ec. isthe adiabat icvel ocity ofsoundwhich is independent ofK, and

°

1,D2, rare positivecoefficientsassoclatedwI ththe damping pr oc e ss. Thereforefor K ..0,substit u t io nof(2.40)into (2.39)shows tha tthe kinetic resultisinagreementwith hydrodynlllllic result:(2.28).The descrlp t i on Ineeresof discr e t e lIIodes isthus a generalizationofthehydrodyn8lllic theory.

.30.

(43)

When the conditions arebeyond the hydrodynalllicregi me, the :;;itustionbecomes Illore complicated.Contribut ionsfrom higher propagating modes must be takenint o account, andsome of the amplitudes Al.j,n(K) canbe complex.The eigenvalues fordiffusive modes are stUI real, but theeige nva luesof the propagatingaode s come in complex conjugatepnira and eigenvalues Ilhould be written in the generalfOB

(2.41)

wherethene gs t i v e real part'Tn(K) represents the damping, and the imaginarypartlola (K), whichisthe dispersion relation for the n-thmode,determines thepropagation velocityof the mode assoicated with:tn(K) (thegr oup veIocd ey of the n-thmode c.Ia givenbyclwa(K)/cIK ).Contrary tothe hydrodyn8lllic lildt,w(K) can no longerbe expressedIn simple formas thepr oduc t of K andt. whi chillindependentof K.

Inth e calculationfordUute ,dispa rate-mass ,binary mlxtu>:>es ofAr+HiI,Campafoundtha t underthecondit ionsofhi gh concentrationof theligh t e rcomponen t and K larger than a certain value, one of the eigenval ueshe obtained had an imaginary part wn (K)for wh ic h the groupvelo.;:1tydld(K)/dKwa sconsiderab l y large rthanthatof other1I0des,Calculation ofthe part ial dynamic scrceeure fa c t ors showedthat thisfas t 1I10de was only associated withtheUght ceeponentand wouldgivean observable

.31.

(44)

shoul d eror pe a konlyinthepa r t i a l dynallllic .t ruc t u refacto r Sll(K.w).Nosh oul dero~peakwould appear inSu( K.w)and Su(K....).Thla.houlde~or peakin Sll (K.w)l. .d•• fora proper choiceof thetwoco-POnenU(m.cOllbl na t lon ofAr +H2 va•

• 'Ju e s t ed by CaIIpa).to a.hou l der in the diff are ntialcross sect i on for !r Ulou lnsca tt ad ng. Thisll1lplle.th.t thefastIIOde isassocia tedonlywi t h thelight ceep onent;•

•32.

(45)

Chapt er 3 . App ar at us and Exper i ment al Procedure

The overallarran gementof apper a tulusedtost udy Bri lloui n scat t e rin g frolll thegas mixture sIs schemat i callydepIcted11'1.

Fig.3.1.An Ar+laserwa s used8S the incident ligh t sewree.The spectrometerconsi:>tedof a piezoebctrieal ly.scannedFabry -Pero t inter f"tolleter, ,pllo t olllui tipiter tubeandaDat a Acquh i tionsnd Stablizatlon System(DAS-l).Thespectral datawe refirst ac cumulated for severs1hour s In theDASlZIemoryand then trans mitted toapersonalco.-purerfor furth erproteuing.Mo r e detailisgiven Infollowingsections.

I.3.1. OPTICALSYSTEII

Rec au s e ehs fr e que n c y shi ftsobs e rv edin Br i llou i n sca tt ering are verysmall(le ss thanlOGHzinour experiment)Ithe incid e nt rad i ation mus t:behighlymonochromati c.Theexcitingrad i a t i o n use dinour experimentwa sprov i dedbyan Ar+ laser (Coher en t

.3 3,

(46)

FIg.J.l Sehelllat1cdla gr AII of theexperimentalsetup.

SllIIIple cell:

B5 belllll-s pll t ter:

AT par tially pen e t ratingglass attenuater : electromecha nical shut te r: IU' mic roprocessori GF glassfiber :

nar row-bandgrat i ngfilt er : F·P Fabry·PerotInt erferollleter; PH pinhole :

PM thenaoele etr ice lly - c oo h d photollUltlplhr;

DAS dat aac qu b l t l on and atabUzatlon .yate.; H1,H2,H3.85,an d 86 lIlirrou;

H4 unco atedrefle ct or; APIandAP2 ape r t uru ; L1,1.2,13. and 1A len ses.

34.

(47)
(48)

Radiation,HodelInnova90-5)whieh WIlS operatedin a single cavitylllode witha nominalwavelengthof514. 5nmand output power of 30.°.500111"". Selectionof the outputlight frequency was made possible by using an intracavityprism.Anintrac avi t y etalonwas installedin side the resonatorfor obtaining8.narrow laser linewidth. The tUt of the etalon could be adjustedto obtaina singlemode.After severalhourswa rm-u p,boththe frequencyand power outputof thela lla rwe s general ly quite steble.No furthe r attemptwas lIIade for la s e r stabilization in thisexperiment.

Sinceli ght scatteringexperimentsat low or moderate pressure(less than 10 atm.inou r exper:iment)involved measu rementsofli gh tat verylow intendty levels,long data aqu is ition ti mes (from 5to24hou r s )we r erequire d.In orderto minbdzlItheeff ectsoflongte rm drif t.in the interf eromet.e r alignment,it wasconseque nt lynecessarytoutilize thefe e db a c k con t ro l capabilityof the DAS-lwh erea refere ncela s er algn a lwas provided as follows:The laller be am wasfirs t. dhidadinto two par ta by usi ng a beamap litt.e r (BS),ofuncoa t edglass. The nthe reflectedlightwas attenuate dandtransmi tte dthroughan opt.ica l glaaa -fiber (GF)whenth e shutter (8)W,U Iopen.The outp utof glass-fiberwa, accurata lydi rectedalong theop ticalaxisof spectrometerbya second uncoa te d re flec t or (K4),and fInally collec tedinthe OM-lfor feedbackcontrol(seesection3.2).

The ma i nlaserbeam was focusedbylensLI, and then

.36.

(49)

reflec ted bymir r or H2 andaprism into thecenterofscatter i ng cellC.Thescat tering angl e formed bytheinciden t beam andthe scattered lightwa s about 157.5°.This eonfigurat1.onalso permittedusto observe the spectrWII cor respondingto a smaller scatter ingangle of 22.5°byaddingtwomirrors H5,H6 as shown inFi g . 3.1,whilekee ping the distancefromLlto thecentre of th ecellalmos t unchan ged .The sca tte r ed ligh t passing throu gh apertu re APIwa.collec tedby thelen s L2and focu se d atapoint wh i ch was justcoincidentwith the virtua l image of thebri ght head oftheglass - fiberproduced byH4.Mi rror KJ wasus ed to directthesca tteringlighttoaFab r y- P e r o t (FP)interferometer (Burle ig h,Hodel RC-llO).r wa s anadju stablenarrow-ba nd monoch romat icgrati ngfilter(KRATOS,Mo del GM100-2 ),whic h wa s adjusted only to transmi t the light aigna lnear th e wavelength 5l4.5l'\lllandtorejectunwanted Raman ra to....ti on frOID theeaepj ee. Be caus a of th elowint en si ty of thesc a t tere d ligh t,it wa s neces sa rytolI.ininliz ethe intens i ty of the strayligh t.The aperturesAPI andAP2wereus e d to preventunwant ed stray light fromgetting in to the spectrome te r.This factwasalso considered inthedes i gn of the scatte ringcell .

Ape r ture AP2,pinho lePH,le ns e s L3and lA,theFabry-Pero t interfe rODle t e r FP,the ph?tOlllUltipliertubePKanda

·thermos taticallycontrolled heaterwereallcontained in II styr ofoam box coveredwith blac kpolyethyl ene. Inadd ition ,e large pieceofhe avy black clo t h wasdra ped over thebox .The

.37.

(50)

purpos e of thiswasfirsttoexc l ud est ra y l1gh':andsecond lyto provideprotec t ion fr omtemperatu r e flu c t uations ofthe surrounding s110as to mainta inthetempera tureinsidethe box ar ound20°C. A 4cIn hole was cut in thestyrofoambox$0that the len s L3 coul ddi rect bo th thescat tere d lightfrolll cel l andthe refere nc e las e r signa l toth eFab r y-Pero t int e r f e r olilete r.

In our exper i ment l , the Fabr y-Perotinterfe r omet erconsi s ted of twonatmirr on (with flat ne s s es of A/2 00and reflection ra tios of 98, )mounte d para lle lto each ot he r inanad j ustab le su pe r inv ar asselllb ly.If thecav ity betwe entheIlli r r orsis illUII.ina t edwitha be amof mono ch roma ticUghc,it willtra ns mit thebe/UII onlywhenthe relation (3.1) issat isfied, I.e.

ex- 2ndcos' (3.1)

where IIIis the orderof interference ,A.iswavelengthof the incidentlight,n is the ref ract ive indexof mediUIIIbetwe en the two mirrora,d is the mirrorspa c i n g , and'istheangle fOr1lled by theinciden tbeamand thenormal Hneof thepletes (in the present expe rimen ts, n...1,,..0).Thefront mirro rvasmounte d onanadj us tablemount usingthre eextremelyfine differen t i al micrometer adjustmentassemblie s.There a r mirrorwas supported by threeItacks of pie zo ele ctric tr ansduce r s placed ar oundits cirCUIIlferenc e .Theape c t r um of the scatteredlightwas obtained by cha ngi ng the voltageappli edto thepiezoelectric tra ns ducers ,

.38.

(51)

wbich consequentlychangedth eIpaci ng bet "'eent"'o plat e...rrrees• ThefringesproducedbytheInter f eren cebe cveentheparal lel .i r r orswerefocundonapi nh o le(PH )in. t a nedInfr ontof the photorault ip Ue r.Thedi _ terof pinh oleva. illpOrta nttothe over a llfine n eof the 'pect ro_ ter . If too lergeitbroa dened the Uneandi ftoo IlUl l it cutdownth etnnslllitte d intendty(24). Theapprop riat e sin ofthepinho l e.",bleh"'. .generallyeeteeeee by trielander r or , ra ngadfrolllOOJlUlto20 0,..11end the ob served finea sewas abo u t4).ThephotollUltipl hr tube(ITTHodel,FII130 ) was mou n te d ineth e r moele t ric ally-cooled,RF·lb ie l de d chlllllber, wh i chmaint ained the cath odetellperatureat ·20oCand reducedthe clarkco un tof thephotomul t i plierto ebout 1eoun tperaeecnd,

I.~.2.DAn AQUISl"l'IOMANDSTABILIZATIONSYSUX (DAS-l)

The longexperl llle nu lruNneededfor lowintel\ll1tyspectra required hi&h ata bUityof boththeInterfer_ t er and las er.

Howe v er,change s of inter f erOllete r alignmen tendfre que ncy dr ifting ofthe lalerlIbich could signi fi cantlybroadenthe instrw:llental l inewid t h arelargelyuna voidab le, it vas nec es sary to compensat e forsu c h drlf t:lngaf fe cts.This .... .madepossib l eby usingthe DIlS-I1nconj u nc t i o n ...ith a referenc e.i gna l whichwas lIIent i on el!pr e v i ously.

TheMsterclockof theDAS·lprtJue e.pulae....hieh areused

.3 9.

(52)

to stepthe data accumula t i onthr oughsuccess ivechannel s ofa 1024 channe lscale r.Thesepulsesarethe ms elve s accumula te dand subsequent lyconverted throug ha DAC to producethescanning ramp voltage whichisappliedsimultaneou slytoall threeofthe piezoelectrictransducers onwhiClh one of theFab r y ·P e rot mirrors ismounted.Asi n gle cycle in the repe t i tiv escanningprocessvas usuallycomple t edin lessth an5secon ds,andthe ampli tud eof therampwascho s e n to coversl ight ly lQorethan two fullorde rs of inter f eren c e (Le.so that th reesuccessive Ra yl ei gh peaks coul d beobs e rv e d ). The high lylinea rrelationbetwe e n appliedvoltsge and piczoelec t ..t e extensionresultsin acor re !lpond i ng relation betwean channel number andfrequ e ncy (seeEq.3.1for 8_0).

COlllF"nsat io nfor ins t rumen t aldri ft s Isnedepo ssibleby the pr ov isi on forextern a ladjustmentandcontrol of the zerole vel of the voltage ramp.Itisthuspo s d bleto selecta re f erenc epeak in theobserved spe ctrum and , thr ough feedbac k con t ro l,force the peakto maintai nttlpos i t i onatapre s e ltc t ed channel nUllbe r . In thopresent.eas e,however,the scatte redlightintensity wa s not sufficient for thepurpose,andit wasnec e ssa ry to pr ov ide a referencesignalof high e r int ensi ty thr ou&h dir e c t samplingof thela s etbeae,

The elec t r omechan i c a l sh u t te r S wa scontrolledbyaseparate etercpeecessoe,MP,which uti lizedthe di gi t al clockof the DAS·l 8Sti meba s e.It va sprogrammedtoopen the shutte rat a po i nt (usual lynearth e endof eachsweep)wherere sonant eraneet.ss ten

(53)

at thelaser freque neyv••• bou t tooc cu r,and tocl oa. It i_ d hte ly af terscann ingthelaserprofUe10 that noneo[ th b hi ghInt e ns it y lignal coul d con t...,l nateth e(10\1Int en sl ty)

!iri llou i n spectI1Ja. Any ten4e ncyfor thl. re f er encepeak todrl£t avay[ro mit.pre. allcted locetionve. dete ctedbytheDM·lwhen oper.lt l ng in its fe edback con t r ollIIO<le,andenepproprhte adj ustmen t to theze rolevelof the Icann i n,r&llpva.

autolll!lt i c al l y ..adetocompansa te for thedrift.Aal.nar but somewha t morecomp l icated. .th odisal so provid edfor opt lllllzation ofthe instrumentalf!nesu.Detailsof how the feedba ck controls areach i ev edcanbe found elsewh ere [ 25 , 2 6 ].

Ano tl.er import antfeat ureof »AS-lh sepantedscsnni ng Whichpertl,i U dif fenn tac annin g speed. fOTseeumula ting eounCsin theselecce d.nSi o ns.In our exper i_ntetwose gJHntsweresetfor theDAS·l.The _in..pent for the lpect na wever ein t er es tedin va•.at betweencha nnelnwlbera310-730 ,andthela lllraepan t , which va. used topr ovideenough countafor thereference leser lignal,va aset betweenchann elnumbers925-955.The spee d retio betw eenthefaitan4sl owpor t i on of therPlpva .99,.0 that for eachsca nalllOat90'of tiaevas usedintheiliaIn sepent.The spectr UIII v.aaccumul.ce d in a totaltime of betwe en5 to 24 hou r s. A typical spectrum obtianedfrom the DAS·l11eh ollO in FlS'3.2.

.41.

(54)

1200

,-- ---T...,

~

c

"

o o

O+---=::...--r--....:::::'---...L~

o 500

Channel number

1000

Fl.•3.2 Typical Ipect:rwrol'8co rded Inarl11 0ul n IC.l:tur ln gexperi llent (CH. +SF. lIixt urawtch CIt,ba..

pl:elS8ur e of 6.)bar and x".-0.08).

'2.

(55)

s.

3. 3.GAS HANDLING SYSTDI

Ablock diagr Ul ofthegas handli n gsystelllisshownin Fig.3. 3 .The highpuri ty(>99.99,>gasesus ed in ourexpe rime nts we r e prov i d e d by Ha the son GasProduc t s.ABour don -tube gauge wi th an estimatedaccur a cyof to.OSbar wa sus ed to measurethegl9 pressures.

The sarapl egas e s, foreXlllDpleSFeandC~,wer emixed usi n g th e followingproc e dur e s . Referi ng toFig .3.3 ,fint weclose d valves V3 ,V4,V5 ,V6,andV7, open ed valvesVi,V2 ,and va , turned on thero t ary vecuuapump.The scatteringcellandthe connected tube scoul dbe evacuatedtoapre s sureofabout10.2torr.

Thenw.clo ••d valveva , ope ned V4,an d.IIlowly filledthe scat te rin g cellwithCH_ inorde r to reduc ethe gasturbulenc e wh i ch coulddisturbany dust insi deth e cell.Whenthe gauge reach edthebase pre s surene eded,we cl osedva lves VIandV4, opened WJ , and turned on the rota ryvacuumpueptoevacuatethe conne ctingtubes,The procedurefollowedWilltocloseva , open V3 and fillth e connectingtubes wit h SFauntllthereadingonth e gaug ewa. alittlebithi gher thanth epre s sureinddethe cel l, the n closeV3 , ope ne dVIandadded Sfsinto the ctll.~yrepe ating thisproc e ss we ob taine dthe gealIbt.t ure a ofdi f fe r en t parti al pr essure ratio s.Weneeded tobecar efulduring thewholepr oc e ss to keep th e pressure insi de the connectingtube sa littlehigh er thanthe presureinsidethecell.

,43.

(56)

f

f , - i

,

I

. !

I ..

" Il

.; ~

I f :i

I a !

!

~

~ ~

i

!

J

i

I !

~

"

~

(57)

The enti re experilllent we re perforlHdinan air-conditioned rooll.No ••traprecautions wer e tllken to keeptheseatteri n& eell atcon. tent temper atu r e_

.45.

(58)

Ch apter Ii. Experimental R esult s and Discussion

§,4.1. GENERAL REMARKS

The methodused.todetect the fn tOJ:slow modeco n t ribut i o n toa give nspe ctr um wa sto compareth eobserved so un dvel oc i t y

v :

and the sou ndspeed".Cla lculatedusingthehyd J:odyn aml ctheor y (K ... 0). ,,;wasdeterminedbythe BrU lou inequation,

v ; ..

..,,>./[251n(I/2) } , (4. 1)

where"Bistheobserved freq uencyIIMft, ),isth ewav e l engt h of the inci den tligh t in the mediUlll,and 'isthe sca tteringangle.

In i t ia lly,the lawof partLa l pr e ss ur eswa s essueee to beva lid for all mix tures st udied,and valuesof".were ca l culated viathe ideal~asr..1IIt10n11'. . .(..,p/p)I/Z.wbere,. ..tp/c., . pis the pressure,andpIsthe massdensl t:y(forII;givenmixture) . Su bs e quent l yIDOr eaccuratevaluesfor ". were obtainedthr ou gh col labor ationwithDr.J.A.ZollwegoftheSch oolof Chemical

.46.

(59)

Engineering ,CornellUniver.!l1ty(2 7 ].Zol l weg's tec hniqu e utlliz;ed the bastc relation[28] ,

(4.2)

whereV.istheIlola r volumeof chemixture,H isth e average molecular weight ,and/c.is the adiabatic compr e a s i b ll i t y.Because the actual average molecularwe ightfo r adh pa r a t e - mas s mixture couldbe s1gnificantlydi fferent fro m that cal culatedonthe basis of partialpressures,the measured prusuraswere first used to eei.eutetethe concentrat ionof eac hspecies viathe virial sea ee equation. Then the sound speedisobtainedusi n gthe virial

(4.3)

wherep is aga i n th eto t al pressu re,C:is theideal - g a shea t capacity, and S, thesecondvirialcoeffic ien tforth e mixture, is obtainfld byusing Hayden-O 'Conne llmethod(29J. Asit turned out, th e valuesof"'.obta ine d by thesecondmetho d wer enot:

si g nificantlydifferentfrolllthe in itia lcalculations,wi t h one or two except ions.

§. 4.2.TIlE FASTSOUNDKeDE IN Hz+ArMIXTURES

A dilute gas mixture which is compoaedof two species with a

.47 .

(60)

larg e_!ISdi ffere nc ere sp ond s toather _ lfluctuationwit h propaga t i ng dendtr",.ves.Un der cond i ti on s of h1ghconcent r. tio n ofth elight erCOllpOnent, •densityvaveofhighfraquency .nd ahor t",.ve leng th_i shtonly be supported bythe lightar pa r ti c les

"'ithou c thep. rt i cipationofthe heav ierp.r tidell1nthe eerteec t ve~t ion.In thb easethe group veloc i t yof the lIixture isve ry do.etotha soundvelodtyin thesinSh-co_pone nt flui d whi ch is obtainedby relaOvd of allhea vierpsrticlea .nae dynamicsofthe tvoCOlPponentsu'e thu llpartid lyaepa r a te d. The eontr ib u t i on oftheheavier componentto thepr opagationmode is negligiblebecausethe he av i e rpar ti c les areunable tofolloW' the fast sound IlIOd.whi eh11ce ueedby the rapidosc illatioM ...ociatedwith theLiShterceepenene,

Thew. vevect orKofthelocaldensityper turbationto be detec t e d,correllpondinseoascat ter i ngangle" sa thlie s the relati on(1. 1).Sotheleng t hsu leof theeheraalfluct....ationto beprob edb give nby "..2..jX-.1/ (2.1n (l/2 ») ,whe r eAis the wsvelensthof inc id entlishtinthelIedi ...he... . ther. stsCll:n d mode, wh i ch correspond.to a short wavelengthperturbationin the fluid ,cu se.to prop. Sate ifIth slianerchanacertai n vtl ue ( 21), a largeaca t te ri .,g angle(157.5°)vas select e din our exper i ments. Foreachaarie s ofexperimentstha118lJIpiecel l ",a . firs tchargfldwith enoughoftheligh terIIpe ele. 10 that well- defLnedBrilloui npe a k. eou l d beob s e rv e d inthespeet rUlll.This spec t rull wa ssubaequentlymon i t or e d al luc ee. dve inerementsof

.48.

(61)

the he avier spe c ieswere adde d into the call.Be cau s eofthe fas t soundacdeeffect,it wa sexpec t ed th a tthepodtion ofth e BrU lo uin peakwouldremain substant ially unthsnged, or shi f tIlluch le s s thanexpe ctedon the ba sh of the hyd r ody nBlIli ctheory.

Threegroups of spe ctra were collected fordiff e r e nt Hzbase pr e s sur e s.Themixtur e s,cond itionsarelis t e dasfollows:

O:sxz:SO.23, O. 2lgtl:SO.27,Kt z..O.0 7,PI-9.0bars, lnsxz:SO.18 ,O.27sKt1:sO.3 2 , Ktl-0 .09,Pl-7.7bars, O.sXz.so.17,O.33 sKt1:s0.39. Ktz-O .lO,Pl-6.3bara,

wh ereXlistheconc e nt r at i on ofspe cieai (1 and 2 refer tothe ligh tand heavy ape c ies , re sp ectively) and tl isthe mean frea path of species i obt ainedusingtheconve n t io nal waywh ich has beendiscusse d in Chapter1.The f1 value sthat wereusedinth e calcul a t i ons were(71-0.297nm andf1z- 0. 340nm for Hzar." Ar respectiv ely. The cont ribut ion of the fa st soun dmod e to the spectrUllwasobserveliin all threecasesfor lowcon c en tra t i ons of Ar(Fig. 4.1-4.3). As the concentrationof th e heavyspe c ies inc reased, th eBrillouin peak.sweredecidedlybroadened due to the severeat tenua t ionofsound. Forxz~O.lthe Brillouinpeak s became unr ecog niz able.However,forxz:sO.l.it is clearthatthepeak pollition s on thescaleofnor mali zed frequencyshi f ts(thesolid trtangl e s)differincraa singlyfrom positions (t he opentriangles) obtain edus in ghydrodynamic the ory(K"O)astheconcen tra t i onof

.49 .

(62)

150 , - - - -- - - . . . ,

"j ..

t Normalized shift

2

nor-Uudto uniqoat thepolltcloQ oft:2wBr ill ou i npea ktOI' purelIa<uch horb ont a l dlvlllon cou••pond.tIto9.2CHz).Only the (Stok. . ) rallon

of

down-.hUtdfreque-Myh .hawn.TM opentrtenll. . inclte.t.theftOmdbed 'hif ta obtainedba..d

Thearrov indlcat•• thenomaUzedpodtlon of the Brillouin plllal:;for punlor.Th.ac a n adasansI. ••,157. 5·.

50.

(63)

I"

III

:l c

" o

o

150

r---~

2 Normalized shift

'1••

".J

Spect ra lorAr+Ha _latun_'11th. fbadHI pn..ura

of7.7ban,S. .Fla.4.1for additIonal let and.

51.

(64)

150

i .. ... ..

C :J

o

o

1 Normalized shift

2

FII_ 4.J Spectra forAr+~ab.tur. .with.fbed Ha preuur.

of6.3ban,SeeFlS'4.1foraddltlonalle&end..

52,

(65)

the heavy compone nt is increasod.

Based on ourexperimental re s u l tsPO),Campa us e d thehard sphere lllodel and parforlDedcalculations [31 ]topr odu c e theoretical Brillouinspec t r afor Ar+Hz mix t u r e swi t h aHzba se pressur eof 6.3bar s. He fou nd.thattherelative podtio nllof theobserved peaks, withre sp ec tto the peakpositionscalculatedon the basis of hydr odynamics(the rat i o ofthesol i dtr ianglepositi onstothe open tr iangle posi t i on s in Fig.4.3),agreedwith his calculated spectra (Table4.2).Healsofound forx2..0. l4,as observed experimentally,that there isnovisi b lecontributionfromthe fast soundI10de inthe Brillou i n spectrum.

Table 4.2 Compa risonofthePre.en t ExperbnentalRe eul u lind ClIIJlpa's Calcuilltionllforan Ar+HzKixtur ewithaH2.

Ba sePre s sure of6. 3 bars

x,

0.05 0.10

Observed Reaul t

1.30 1.51

Campa's Cal cu latio n

1.20 1,46

As showninFi g . 4.1-4.3,considerabladifficulty wa s experi encedindet ecti ng the fastmode contr ibution as a we ll·

definedspectra l fe a t u r e . Thismightbe due to(1)a relatively highpolarh:ab ili tyof Ar(IIA ~"'1.64xIO ·2.'/cm3 ,O'&z..0.8 2)(10·24

/ c1ll3 O'A~/O'iZ"'2)wh i ch fa vors the second andthi r d terlllS in (1.3),

.53.

(66)

(i i) broad e ningof thespe ctral featurebeca us eofthe$t r ong attenuation ofthefa s taoundmodecause dby the heavierpattic l e s , and (iii)the loW' abso lu tepolariza b il i t yof H2.whi chle ads to a loW'signa l lev e lintheden sityrangeof int ere st . Thereh.ti v e l y high int e nsi ty of the central pe a kalsoindicat edtha t the data

"lightbe contami na t edby $t rey light.

s.

4.3. CH,+SF, KIXTURES

Bea rin gthe fore goi n gpointsinmi nd,mixtures ofCH,+SF, were selectedforfur th erinvestigation s.The po la rb.ab i l1tyof CH,isabo ut threetimesgreate r thantha t ofH2. (acll,/a8

z" 3.2 ). In addi t ion,the ,",ola r izab il i ty rati o of SF,andCH~iaab ou t the samef,Sthat ofA r: andHz(Qsr,/Oc. ~ "'1.8 ).However,the lUISra ti o fo r the new combination(IIIsr/ IIlcI,"9. 1 )is lowere droughlybya factorof 2. Twogroupsofspect r a vet o col lectedfor eH,bas e prassu tesof 6,3ba ts and3. 7bar s ,with conditions asfollows:

Theuvaluesusedinthecslc ul a tions oft 1 we reul...O.3B9rua,

"2",O.46i1nm.Compa r i ng th e condi t i ons forCH ~+SFswith thosefo r

"2+Ar, itshouldbenoted that theywerequit e simila r. Beca u s e of thelargeincrea s einpolarizab ili ty.asexp e c t ed , the

.54 .

(67)

flrillou i npeaksof therecorded sp e c t ra werewell·defined. Neve r t he less,nosigni fi c an t fastmode con tr ibution.. asdet ec t able in eithe rcas e .Assholrit\ inFig.4. 4 and Fig. 4.S, forthespe ctr a correspond ingto th e CH, ba s epre s sure of 6.3ba rs ,the identified fr e qu ency sh iftsofth e spe c t r al peaksagr e edwiththe calcul ated re s ult sbasedon hy dr odyn a mi c the o r y.For spectr awith theCM,ba s e pres sureof 3.7bars,the obs erv e d freq uen cy sht f t sdidnotdiffe r sign i fica n t ly fr om thehy dr odyn8.llli c prediction.Usi ngthe hard- sphe relllode l , ClUIlpa agai ncon firme d that, under theexp e rillent a l cond itions cho s en above, no fastmode shou ldbe observedIn these cas e s [3 1].Furt h ermor e,itwas conc l udedth atthecond1t ions requ i r ed forde te c t io n ofth efas t modecontr i butio n forSFe+CH, mi x tu r eswere un l i kelyto beachie v ab leinpractice.

A pos siblepoin tof impor t ancehar aisthatthest~cture.of the SFe and eH,mol ec ules are lIIor e complica t ed thanthoseofAr andH1, and thevalidityof thehard·sphe re1Il0de l lIlaybe que s tionable.Howev e r ,becau seIIIsre /fllA ~"3.7andlIlee./mp, 1"7.9,the ave r a geve l oc i t y of boththeSFI andCH. molecules are conside r ab l y lower(VA ~/vS'I"2, vFJ/ vn; -2.8) .At the !lame etee, thet.va l ues for SFe+CH••which dep endonthe den si ti e s andth e diBllletersof theparticle s, remain almos tunchanged.Itwas consequen tlyexpected th a t devia tion sfrofllthe hard-spheremodel sho u l dno tbesign i fic an t (Le,because ofthe low densityand slowerspee d of the part i cle sin themed i um, thecollisiontimeis much lowar tha n theaeanfree time).Theobs erved resul ts also

.55.

(68)

~C :::I

o

o

1000,---~

Normalized sh ift

'1.8... Sto\•• spec t n forSFI~.btun awit h . flxadCIt.

p.utlalpr . ..ureof 6.3bar senc! fractionalconee ntrationsof SF, ••Ulte d.tachhoriz ontal diYb lonrepru .nUfre quency .hlf t of1.l CMz.Thautanc l a rI'Il rk,rahave the1_'. . .ni n l I' inF1S.4.1.

56.

(69)

1000 , - - - -- - -- - -- - - - ,

~C

-= o o

No rmalized shift

FJ,.4.' Stoke.spect ralor S',.aI.wi th.fixed!~partial pn uuraof3. 1bu•. S.. Fil.4.4for additional h ,and.

57.

(70)

provedthis.Hencethe negativeresultfor theSFs +CH~mi x t u r e rDightbecause d bya smalle rIDa/ 1IIl1rae te.

In th eabov,two cases( Hz+Ar&SF,+<:H. )themethodusedto charact e rizetheeltpe rim e ntalcondit i on s for the gas milttureswith dispa r at e ma sses (I. e .in temsof thet.valu esobtainedusing the convention alway ) providesno insightinto the distinctly differentruults.It wasconsequentlydecided to investigate alter n a t echar a c te r i z a t io n criteria.The following discussion intro duc e sanef f ec t i vemes n freepathwhich provide s alDor e cons is te ntdes c ription .

I.4.4 TIUtEFFECTIVE KEAN FREEPADIS DISCUSSIOIf

Firs titisnece ssarytoin t rodu c e the lIleanpersistenceratio 0iJ.which is: definedas the ae en ratio of theve l oc i t y component ofpart i clei along itsincidentdirectionto its originalvel oc i ty aftera colUsion wit hSoparticlej.01Jisgiven by(3]

whereHi -IId( lIIi+IlIJ)'(i,j -1,2). 1111 ,lIIJare the massesof two differentparticles,e.g.the H: molecules and Ar atoms in a binarygaseous sys t em.Using(4.4), 0iJvaluesweracalculatedfor thest ud i edmix tures. The re sul tsarelistedIn Table4.3whare it

.58.

(71)

shou ld be nohd that ,duetothelargll lila••differe nces .Ozl.. 1 aoo012 - O.1. ••th eve1 oe it,.vec torof aheav lerparti cleal_st does not cha nge after it collld••wi th aUlh t erpa~·t1cl e,and the vel odt,. vactorof a l1pter~r tlcleIe totall,. ch.nsadafta rit collideswithaheavi er partic le .

Tab le 4.3 The HeanPers is tenc eIIatio fo r theStudh d Mixtures

0"

", "

tH, SF,

" ,

0.406 2. 64xlO·a 7,01xI 0'3

0.937 0. 406

CR, 0.406 5.89xlO· a

SF, 0.982 0.871 0.406

Given apart i c le of .peeles i !DOvin ginthe ga.ofspecies j, af te ra tI\IlIberofcollisions wit h particles of.pe ele sJ.the ve l ocit y victorofpartic le iis rando. iz ad.The averegeflUllber of eolUslons required forrandomization,i.e.nlJ'canbe obtained usingthe Poh El on dl.atri butio n:

(i,J-1,2) (4.5)

The calculatediil.J for thestudied mixture . angivenin Table 4.4.

As shown in Table 4.4,i fthe coll isions occurbetwee n

,59.

(72)

identical par U cles , thl!velocityvector is randomize dafter les s than twocollisions(nll-1,68).I fthe collisionscecur between unlike partic leswithlsrgemass diffe rence ,theveloe i t yof the lighter pertic lesisgreatlychangedillll.llediatelyafter a co11idon with a heavier one, while tho:! velocityvecto rof a heavier particlecanberandOllli z e d ;)illyaftera long series of collisions

Table4.4 TheAverageRandomi za t i on Co llis io n for theStudied Mixtures

nlj H, A, CH, SF,

M, 1,68 1,03 1,01

15.9 1.68

CH. 1,68 1,.:16

SF, 54 . 6 7.73 1.68

with light e rpa rtic les . It sh ou lda1.5obe notedthatunlik enu, nZi isve r y senllitive to the change oflIIa s s ratio m2/ml'Compared to the Ar +H2mix t ur e , thelIlass rati oof theSFe.CH~mix t ure decr e a s e s byroughl y afact or of2;n21 also dec r e a s e s bya factor

For the pictu retobe comple te,it mustbare co gn ized thatIn abinarymixturethecollt~ionscanocc u rnotonlybetve enth e differentspecLea bu talsobe tween the same specIes . SoIn s t.ea dof using 0ijin fon.ula(4.8) it is moreappr op r i a t e touse a

.60.

Références

Documents relatifs

This compensation is also present when one consider climate change scenarios, it should be nice to analyse this compensations between models (e.g. Does the total drags are

Application of Database Management to Geographical Information Systems (GIS); Geospatial Databases; spatial data types [point, line, polygon; OGC data types], spatially-extended SQL

To register for the social evening at Les Brasseurs du Temps on October 12, check one of the two options on the registration form — meal and drinks (30 euros) or drinks only

Dr Adrian Reynolds, Regional Adviser in Health Promotion and Mental Health, welcomed all participants to the Philippines and to the Workshop on Problems Relating to the Use

Please find enclosed the program and the abstracts of communications to be presented to the XI th European Multicolloquium of Parasitology (EMOP XI, Cluj-Napoca, Romania, July 25

This is why, in the present study, the relationship between male factors, especially mass sperm motility, and fertility were the focus while other factors of variation were used

According to Canale and Swain (revised in 1983 by Canale) testing communicative competence is based on testing the learners 'abilities to use the target language through the

In Berg and Valent [3] appear explicit formulas for the quadruple associated with a birth and death process admitting quartic rates, and the order and type can be