Viscosity of liquid systems involving hydrogenated and fluorinated substances: Liquid mixtures of (hexane + perfluorohexane)

ContentslistsavailableatScienceDirect

Fluid

Phase

Equilibria

jo u r n al h om ep a g e :w w w . e l s e v i e r . c o m / l o c a t e / f l u i d

Viscosity

of

liquid

systems

involving

hydrogenated

and

fluorinated

substances:

Liquid

mixtures

of

(hexane

+

perfluorohexane)

Pedro

Morgado

a

_,

_Jana

_Black

b

_,

_J.

_Ben

_Lewis

b

_,

_Christopher

_R.

_Iacovella

b

_,

_Clare

_McCabe

b,c

_,

Luís

F.G.

Martins

d

_,

_Eduardo

_J.M.

_Filipe

a,∗

a_{CentrodeQuímicaEstrutural,InstitutoSuperiorTécnico,UniversidadeTécnicadeLisboa,1049-001Lisboa,Portugal} b_{DepartmentofChemicalandBiomolecularEngineering,VanderbiltUniversity,Nashville,TN37235,USA}

c_{DepartmentofChemistry,VanderbiltUniversity,Nashville,TN37235,USA}

d_{CentrodeQuímicadeÉvora,UniversidadedeÉvora,RuaRomãoRamalho,59,7000-671Évora,Portugal}

a

r

t

i

c

l

e

i

n

f

o

Articlehistory: Received21June2013

Receivedinrevisedform26July2013 Accepted30July2013

Available online 19 August 2013 Keywords: Viscosity Mixtures Perfluoroalkanes Alkanes Moleculardynamics

a

b

s

t

r

a

c

t

Theviscosityof(hexane+perfluorohexane)mixtureshasbeenmeasuredat298K,303Kand308K,over

thewholecompositionrange.Theresultsshowlargenegativedeviationsfromthearithmeticmeanofthe

viscositiesofthepurecomponents,reaching−17%atx(perfluorohexane)=0.7.Toobtainmolecularlevel

insightintothebehaviourofthesystem,all-atommoleculardynamicssimulationshavebeenperformed

andusedtocalculatetheviscosities,radialdistributionfunctions,androtationalrelaxationtimesof

thestudied(hexane+perfluorohexane)mixtures.Thisisthefirstefforttoassesstheeffectofmixing

hydrogenatedandfluorinatedmoleculesintheviscosity.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Itiswellknownthatalkane+perfluoroalkanemixturesdisplay

largedeviationsfromidealbehaviour(liquid–liquidimmiscibility,

largepositiveexcessvolumesandenthalpies,positivedeviationsto

Raoult’slaw,etc.).Inthecaseoftransportproperties,eventhough

theconceptofidealityisnotclearlydefined,atypicalbehaviour

couldalso be expected due to thealleged weak cross

interac-tionsbetweenhydrogenatedandperfluorinatedchains.However,

despiteitsobviousimportance,thesubjecthasnotbeenaddressed

intheliterature.Theonlymeasurementsofviscosityreportedfor

mixturesofalkanesandperfluoroalkanesarethoseofMcLureand

Clements[1],whofocusedonthediscontinuityinviscosityneara

liquid–liquidcriticaltemperature.Theauthorsmeasuredthe

vis-cosityof a single(hexane (H6)+perfluorohexane (F6))mixture,

atthecriticalcomposition(x(F6)=0.370),atseveraltemperatures

abovetheliquid–liquiduppercriticalsolutiontemperature(UCST).

BelowtheUCST,theviscosityofbothphasesinequilibriumwasalso

measuredatseveraltemperatures.

The knowledge of the viscosity behaviour of

alkane+perfluoroalkane mixtures is of special importance in

∗ Correspondingauthor.Tel.:+351218419261. E-mailaddress:efi[email protected](E.J.M.Filipe).

thefield ofbiphasicsynthesisand catalysisknownas“fluorous

chemistry”,sincethesolventsusedare(partiallymiscible)

mix-tures ofhydro-and fluorocarbons.Additionally, thisknowledge

is of great benefit to theinterpretation of perfluoroalkylalkane

(PFAA)viscosities,recentlymeasured[2]inourgroup.PFAAare

diblock compounds madeof alkyl and perfluoroalkyl segments

covalentlybondedtoformasinglechain.Theycanbepicturedas

chemicalmixturesoftwomutuallyphobicsegmentsthatinmost

caseswouldotherwisephaseseparate.

Structurally, thesubstitution of hydrogenfor the largerand

heavierfluorineatomresultsin alargercross-sectional areafor

perfluorinatedchains[3](0.283nm2 _{comparedto0.185nm}2 _for

n-alkanes) and also in higher densities and molar volumesfor

n-perfluoroalkanes,whencomparedton-alkanes withthesame

numberofcarbonatoms.Consequently,inamixtureofan-alkane

withthecorrespondingperfluoroalkane, thevolumefractionof

perfluoroalkaneisalwaysconsiderablyhigherthanitsmole

frac-tionandaprobeimmersedinsuchamixturewould“experience”a

muchmorefluorinatedenvironmentthanexpectedfromthemolar

composition. Anotherimportantdifference betweenfluorinated

andhydrogenatedchainsisconformational.Whilethen-alkanes

tendtobeinanall-transplanarform,n-perfluoroalkanesdisplay

ahelicalconformation[4];also,theflexible characterof

hydro-genatedchainscontrastswiththerigidityoftheirperfluorinated

counterparts.Itisbelievedthatoneoftheconsequencesofchain

0378-3812/$–seefrontmatter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.fluid.2013.07.060

stiffnessinliquidfluorocarbonsisalessefficientmolecularpacking

andtheexistenceof“holes”intheliquid.Thiscanexplain(atleast

inpart)theenhancedsolubilityofsimplegases(oxygen,nitrogen,

etc.)inliquidperfluoroalkanes.Ontheotherhand,

perfluoroalka-nestendtobemorevolatilethanthecorrespondingn-alkanesdue

totheirweakercohesiveforces,showinghoweverhigher

viscos-ity,which canalsobeattributedto theknownrigidityoftheir

molecularstructure.

Considerable changes in volume occur when hydrogenated

chains are mixed with fluorinated chains. We have recently

reportedpartialmolarvolumesatinfinitedilutionforaseriesof

n-perfluoroalkanesinn-octane[5,6].Forexample,whenamolecule

ofF6isimmersedinn-octane,atinfinitedilution,itsmolarvolume

increasesfrom202.4cm3_mol−1_to229.3cm3_mol−1_,i.e.,13%.

Lit-erally,alayerofemptyspaceiscreatedaroundtheF6molecule.

Theeffectisevenmorepronouncedwhenn-alkanesaredissolved

inn-perfluoroalkanes:theirmolarvolumesincreaseby20%[7]!It

shouldbeemphasizedthattheseareinfinitedilutiondata,i.e.,no

packingeffectsareincludedthatathigherconcentrationswould

contributetotheexcessvolume.Inviewoftheseresults,itcanbe

anticipatedthattheviscosityofmixturesofalkanesand

perfluo-roalkanesmightbelowerthanthatexpectedfromtheviscosityof

thepurecomponents.

In this work we have measured the viscosity of mixtures

of(H6+F6) atthree temperatures, over thewholecomposition

range.Additionally,all-atommoleculardynamicssimulationshave

been performed and used to calculate viscosity and rotational

relaxation timesfor the studied mixtures, as wellas to

exam-inethemolecularlevelstructureviaradialdistributionfunctions.

Webelieve this isthefirst studyofthetransport properties of

alkane+perfluoroalkanemixtures.

2. Experimental

H6andF6,bothwith99%claimedpurity,werepurchasedfrom

Sigma–Aldrichandusedasreceived.Themixtureswereprepared

byweightin20mlscrew-capvials.Toensureinternalconsistency

oftheresults,theviscosityofH6wasmeasuredinthesame

experi-mentalconditionsasthemixtures.TheviscosityofF6wasrecently

measuredbyourgroup[2]usingthesamemethod,brieflyoutlined

below.

The viscosities were measured using a Schott-Geräte Type

545–00/0Ubbelhodeviscometer,withtheAVS440measuringunit.

Thisautomaticsystemcomprisesaviscometerstand(AVS/S)with

opticalsensors,anautomaticpumpingsystemandacontroland

recordingunit;theviscometerstandisimmersedinathermostatic

bath,withtemperaturestabilitybetterthan0.01K.Eachviscosity

reportedisobtainedfromtheaverageoffiveindependent

mea-surementsofflowtime, witha scatteringof lessthan0.2%;the

uncertaintyofeachflowtimemeasurementusingthisunitis0.01s.

Thetemperatureofthethermostaticbathwasmeasuredwitha

calibratedplatinumresistanceprobeconnected toa 5½ digital

multimeter(Keithley191),withanaccuracyof0.05Kanda

pre-cisionof0.01K.

ThedensityofallsampleswasmeasuredwithanAntonPaar

DMA5000vibrating-tubedensimeter,forthedeterminationofthe

dynamicviscosity.Thisinstrumentwascalibratedwithwater

(dis-tilled,deionizedina Milli-Q185Pluswater purificationsystem

andfreshlyboiled)andairat20.000◦C,takingintoaccount

atmo-sphericpressure;itsinternaltemperaturecontrolsystemisstable

at±0.001K.Thedensityoftheliquid mixtureswasdetermined

beforeandaftertheviscositymeasurements,inordertoestimate

anycompositionchangeduetodifferentialevaporationduringthe

measuringprocess.Theviscositiesweremeasuredat298K,303K

and308K,andthemeasurementat298Kwasrepeatedattheend

Table1

Density,relativevolumeandsimulatedrotationalrelaxationtimesatT=298Kasa functionofF6molarfraction.

x(F6) Density(gcm−3) Volumerelative topureH6 H6(ps) F6(ps) 0 0.655 1.000 12 – 0.25 0.956 1.111 11 17 0.50 1.207 1.194 12 19 0.75 1.457 1.266 13 23 1.0 1.672 1.333 – 27

oftheseriestofurtherassesstheeffectofeventualevaporation. Whenthetwomeasurements,compositionand/orviscosity,were differenttheaveragevaluewastaken.Thechangesincomposition wereestimatedtobealwayslessthan0.015molefraction.

Eventhoughthetemperaturewasconstantwithin_±0.01K,it wasnotpossibletostabilizethetemperatureatexactlythesame valueforallthestudiedcompositions.Hence,reported temper-aturescorrespondtoan averagevalue of±0.1K.Thecombined uncertaintyofthemeasuredmixtureviscositiesisestimatedtobe lessthan1.5%.

3. Simulations

Tocompare withexperiment, atomistic moleculardynamics (MD)simulationshavebeenperformed.Theoptimizedpotentials for liquid simulations all-atom (OPLS-AA) forcefield, originally developedfor alkanes[8]andextended toperfluoroalkanes[9],

was used to model the molecules studied. The OPLS-AA force

fieldwasdevelopedtoaccuratelyreproduceexperimental

prop-ertiesofliquids,andrecentstudies[2,10]demonstrateexcellent

agreementwithexperimentforpressurevs.densityisothermsof

PFAAalkanes.SimulationswereperformedusingtheLAMMPS

sim-ulationengine[11],carried out in theNVTensemble (constant

numberofatoms,temperature,andvolume)atatemperatureof

298Ktomatchexperiment.Temperaturewascontrolledviathe

Nosé–Hoover thermostat; therRESPAmulti-timestep algorithm

wasusedwithtimestepsof0.1fsforbond,angle,anddihedral

interactions and time stepsof 1.0fsfor thenon-bonded

inter-actionsandelectrostaticinteractions,whicharecomputedusing

theparticle–particle,particle–meshalgorithm(PPPM).Simulations

wereconductedfor systemscomposedof256molecules.While

systemsizeNisconservedasthemolarratioofalkanesto

per-fluoroalkaneswaschanged,thesystemvolume wasadjustedto

theexperimentaldensityreportedbyBedfordandDunlap[12];the

densityandsystemvolumeratiosusedarereportedinTable1.

The rotational relaxation time of the individual H6 and F6

molecules was calculated for each composition. The rotational

relaxationtimeofeachmoleculewasestimatedusingthe

auto-correlationfunctionofthedotproductoftheend-to-endvectorof

themolecule: C(t)=



1 N



N i=1u



i.u



j



(1)

whereNisthetotalnumberofmoleculesinthesystemandûiis

thedirectorofmoleculei.Toobtaintherelaxationtime,,C(t)isfit

usingtheKohlrausch–Williams–Wattsfunction[13]:

C(t)=˛exp



−



t 



ˇ

(2)

wheretcorrespondstothetimeand˛andˇarefitting

param-eters.Fittedrotationalrelaxationtimesat298K,roundedtothe

nearest picosecond, are reportedin Table 1 for each

composi-tion.UsingthecriteriaproposedbyMondelloandGrest[14]that

0 50 100 150 200

time (ps)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

(cP)

x(100) x(75) x(50) x(25) x(0)

Fig.1. AverageviscositycalculatedasafunctionoftimefromsimulationatT=298K fortherangeofF6concentrations.

rotationalrelaxationtimeofthemoleculeinordertoobtain10%

statisticaluncertaintyinviscositycalculations,thisdatasuggests

a minimumof 5.4ns simulation time isneeded. Simulationsat

eachcompositionwereconductedinexcessof60ns(in3

indepen-dentsimulationsforeachcomposition),wellbeyondthisminimum

timeandofsufficientlengthtocapturefluctuationsinthespatial

arrangementofmoleculesinthemixtures.

ThezeroshearviscositywascalculatedusingtheGreen–Kubo

relation: = V kBT

_∞ 0



Pij(0)Pij(t)



dt (3)

Inthisequation,Visthevolumeofthesystem,kBisthe

Boltz-mannconstant,Tis temperature,andtis time.ThequantityPij

correspondstotheijoff-diagonalcomponentofthepressure

ten-sor.Theaverageautocorrelationfunctionofthepressuretensoris

calculatedwithina200pswindow;sincethisquantitydependson

therelativetimedifference(i.e.,t–t0),theaverageautocorrelation

functionwasconstructedviatimeshiftingoftheorigin,witha

cor-relationoffsetof10fs.Ingeneral,theviscosityasafunctionoftime

plateausby∼50psintothiswindow(seeFig.1);theaverage

vis-cosityreportedinthisworkwascalculatedfrom20evenlyspaced

datapointsfrom100to200pswithinthisplateauedregion.To

pro-videanestimateoftheerrorinthesemeasurements,eachofthe

three20nssimulationtrajectoriesareconsideredindependently

andthestandarddeviationofthesethreeviscositymeasurements

isreportedastheerror.

4. Results

TheexperimentallymeasureddynamicviscosityofH6is

pre-sentedinTable2.Theresultscomparefavourablywithliterature

data[15](deviationsarelessthan0.5%).Table3reportsthe

experi-mentalviscositiesoftheH6+F6mixtures,aswellasthecalculated

Table2

ExperimentaldynamicviscosityofH6.

T/K /mPa.s

298.20 0.2937

303.16 0.2798

308.13 0.2669

Uncertaintyinviscositylessthan0.8%.

0

0.2

0.4

0.6

0.8

1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

(cP)

sim, 298K exp, 298K exp, 303K exp, 308K

0

0.2

0.4

0.6

0.8

1

x(F6)

-20

-15

-10

-5

0

100

/

(a)

(b)

Fig.2.(a)Experimentalandsimulatedviscosityof(H6+F6)mixtures,.(b)Relative viscositydeviationsfromthearithmeticmeanofthepurecomponentviscosities, /,fromexperimentandsimulation;forclaritythishasbeenexpandedbyafactor of100togivepercentdeviation.Thelegendappliestobothpanels.

deviationsfromthearithmeticmeanofpurecomponent viscosi-ties:

=exp−(F6xF6+H6xH6) (4)

Theexperimentalviscosityofthepurecompoundsateach tem-peraturewasinterpolatedusingtheAndradeequation[17].Fig.2a

plotstheviscositiescalculatedfromexperiment.

Relative viscosity deviations, /, are plotted in Fig. 2b.

AlthoughtheabsolutedeviationsshowninTable3decreasewith

increasingtemperature,thisisnotthecasefortherelativevalues

plottedinFig.2b.Thetemperaturedependenceoftherelative

vis-cositydeviationsisprobablywithintheestimatedexperimental

uncertainty.Measurementsinaclosedviscometer(without

evap-oration)wouldallowalargerrangeoftemperaturestobestudied.

Ascanbeseen,thehighestrelativeviscositydeviationhasavalue

ofabout−17%andislocatedbetweenx(F6)=0.5andx(F6)=0.7.

Table3

Experimentaldynamicviscosityof(H6+F6)mixtures,anddeviationsfromthearithmeticmeanofpurecomponentviscosities(Eq.(4)),.

x(F6) 298.0K 303.0K 308.0K

/mPa.s /mPa.s /mPa.s /mPa.s /mPa.s /mPa.s

0.000 0.294 – 0.280 – 0.267 – 0.243 0.341 −0.036 0.319 −0.036 0.299 −0.036 0.500 0.408 −0.057 0.375 −0.059 0.350 −0.058 0.500 0.407 −0.057 – – 0.352 −0.056 0.705 0.458 −0.076 0.428 −0.070 0.401 −0.064 0.727 0.468 −0.074 0.438 −0.067 0.410 −0.061 0.902 0.556 −0.045 0.520 −0.039 0.483 −0.037 1.000 0.635 – 0.589 – 0.548 –

0

0.2

0.4

0.6

0.8

1

x(F6)

0.2

0.3

0.4

0.5

0.6

0.7

0.8

(cP)

H6+F6 H6+H9

Fig.3.ViscosityofH6+F6(298.0K)andH6+H9(298.15K)mixtures.Thelines representthearithmeticalmeanofpurecomponentviscosities.

Fig.2a alsoincludes viscositiescalculated fromthe

molecu-lardynamicssimulations,whichareadditionallysummarizedin

Table4.Thevaluescalculatedfromthesimulationsdemonstrate

closeagreementwithexperiment.Aspreviouslymentioned,the

simulationsweredoneattheexperimentaldensity.Itisknown

fromprevious work[16] that the forcefield usedis unableto

reproducetheexperimentalpositiveexcessvolume,even when

theH–Fcrossinteractionisreducedby25%,whichissufficientto

bringagreementbetweentheexperimentalandsimulatedexcess

enthalpy. It is interesting to note that simulations run at the

experimentaldensityproduceviscositiesthatagreecloselywith

experiment.Wealsonotethattherelativelylargeuncertaintyinthe

reportedviscosityforpureH6makesitdifficulttoprovidea

mean-ingfulcomparisonoftherelativeviscositydeviations,althoughthe

absoluteviscositydeviationsshowsimilarbehaviourtothatseen

experimentally.

5. Discussion

Itisknownthatbinarymixturesoftenshownegativeviscosity

deviations,asdefinedinEq.(4).Hence,mostpredictivemethods

fortheviscosityofmixturesproposetheuseofequationssimilarto

Eq.(4),butintermsofthelogarithmofviscosity,usingsubsequent

adjustableparameterswhicharefittedtoeachbinarymixture[17].

Toprovideadditionalinsightintothemixingbehaviour,Fig.3

comparestheexperimentalviscositiesofH6+F6mixturesat298K

withliteratureresults[18]forH6+nonane(H6+H9)at298.15K.

Thislatter mixturewaschosen becausetheviscosityofnonane

(0.66mPa.s) is close to that of F6 (0.635mPa.s) and because

bothcomponentshave,inthis case,verysimilarintermolecular

interactions,i.e.,dispersionforcesbetweenhydrogenatedchains

exclusively. As can be seen from Fig. 3, the (H6+H9) system

hasnegativedeviationsfromthelinearbehaviour,whichreacha

Table4

SimulateddynamicviscosityoftheH6+F6systemat298.0Kandthecalculated viscositydeviations. x(F6) /mPa.s /mPa.s 0.000 0.344±0.061 – 0.250 0.329±0.026 −0.085 0.500 0.380±0.006 −0.105 0.750 0.466±0.010 −0.089 1.000 0.625±0.034 –

maximumof5.5%attheequimolarcomposition.However,as pre-viouslyshown,thealkane+perfluoroalkanesystemshowsamuch largernegativeviscositydeviationwithamaximumof17%.The dif-ferenceintheseotherwisesimilarsystemsislikelyrelatedto(i)the significantvolumeincreaseassociatedwithH6+F6mixtures;(ii) therigidityofthefluorinatedmolecule;(iii)theweaknessofthe crossinteractionbetweenalkylicandfluorinatedchains.As pre-viouslydiscussed,themolarvolumeofperfluoroalkanesinfinitely dilutedinn-alkanesincreasesby13%,whereasthemolarvolume ofn-alkanesdissolvedinn-perfluoroalkanesincreasesby20%;also recallthatthevolumeofpureF6islargerthanthevolumeofpure H6byafactorof1.33.

Tofurtherexploretheeffectofvolumeofthesesystemsatthe molecularlevel,theradial distributionfunction(RDF)hasbeen calculatedfromthesimulation trajectories;thefirstpeaks, cor-respondingto thenearest neighbourdistances, are highlighted

inFig.4.TheRDFiscalculatedbetweenthecarbonatomsinthe

moleculebackbone(excludingdistances betweenatomsonthe

samechain) asa function ofmolecularspecies. Fig.4a–cshow

theRDFsfor H6–H6,H6–F6,and F6–F6,respectively. Thereis a

significant increase in the width and the intensity of the first

peak oftheH6–H6 RDFas F6isadded tothesystem(Fig.4a),

whereasboththepeakwidthandthepeakintensityremain

rel-ativelyunchangedintheothertwocases.In theH6–H6RDF,as

theF6molefractionisincreasedfrom0to0.75,thepeakwidth,

asestimatedfromthelocationofthefirstminimum,increasesby

∼1 ˚A.ThissuggeststhatthevolumeaccessibletoaH6molecule

thatisincontactwithotherH6moleculesincreasesasF6isadded

tothesystem.Recall thattheexcess volumefor this systemis

positiveandlarge,∼5cm3_mol−1_{,andthatthesimulationswere}

performed at the experimental density. The simulation results

seemtoindicatethataportionofthisincreasedvolumeismore

efficiently taken by H6, probably aidedby its higherflexibility

relativelytoF6(e.g.,intheequimolarmixture,theaverage

end-to-enddistanceofthecarbonbackboneis6.10±0.04 ˚AforH6and

6.38±0.03 ˚AforF6).Thus,asF6isaddedtothesystemthe

effec-tivedensityoftheH6fractiondecreases,whichshouldresultin

areductioninthecontributiontothetotalviscosityofthe

mix-ture.These findings are compatible withtheincrease of molar

4

6

8

10

0.8

1.0

1.2

x = 100 x = 75 x = 50 x = 25 x = 0

4

6

8

10

r (Å)

0.8

1.0

1.2

4

6

8

10

0.8

1.0

1.2

(a)

(b)

(c)

Fig.4.Radialdistributionfunction(RDF)calculatedbetweenthecarbonatomsofthecompoundsasafunctionofF6concentration,x,for:(a)H6–H6(b)H6–F6,and(c)F6–F6. Thelegendin(b)appliestoallpanels.NotetheRDFonlyincludesatomsonseparatechains.

volumeatinfinitedilutionofH6whendissolvedinF6.However,the

simulationsshownodirectexplanationfortheincreaseofpartial

molarvolumeatinfinitedilutionofF6whendissolvedinalkanes.

Thiscouldindicatethepresenceofamoresubtleeffect,solvophobic

inorigin.Theincreaseofthewidthandintensityofthefirstpeakof

theH6–H6RDFasF6isaddedtothesystemcouldalsobethesignof

anincipientsegregationbetweenthetwosubstancesandthe

cor-respondingformationofalkaneandperfluoroalkanerichdomains.

Thishypothesiswillbeinvestigatedinfuturework.

Re-examinationoftherotationalrelaxationtimesreportedin

Table1providesadditionalinsightintotheroleofmolecular

rigid-ity.WhiletherotationalrelaxationofH6isrelativelyunchangedat

allstatepoints,therotationalrelaxationtimeofF6increases

sig-nificantlyastheF6concentrationisincreased.Statepointswith

shorterrotationalrelaxationtimesshouldexhibitincreased

rota-tionalfreedom;withlessencumberedmolecularmotion,itwould

beexpectedthattheviscosityshouldbereduced,reducingthe

con-tributiontothetotalviscosityofthemixture.This,incombination

withtheincreasedaccessiblevolumeofH6asF6isaddedtothe

system,islikelythecauseofthesignificantdeviationsseeninboth

experimentandsimulation.

6. Conclusions

Theeffectofmixinghydrogenatedandfluorinatedmolecules

onthe viscosityof the mixture hasbeen assessedfor thefirst

time.TheviscosityofH6+F6hasbeenmeasuredoverthewhole

compositionrange,atthreetemperaturesandshowslarge

nega-tivedeviationsfromthearithmeticmeanoftheviscositiesofthe

purecomponents,reaching−17%atx(F6)=0.7.Theresultsobtained

frommoleculardynamicssimulationsfollowthetrendexhibited

bytheexperiments,andgiveadditionalinsightintothemolecular

levelbehaviour.Specifically,itisseenthatthevolumeaccessibleto

H6moleculesincreasesasF6isaddedandtherotationalrelaxation

timeofF6issignificantlyreducedasH6concentrationincreases;

thesetwofactorslikelycontributeforthestrongnon-linearmixing

behaviourseeninbothexperimentandsimulation.Theseresults

correlatewiththerecentlyreportedvolumeexpansionsobserved

inalkane+perfluoroalkanemixtures.

Acknowledgements

PM acknowledges funding from Fundac¸ão para a

Ciên-cia e Tecnologia, in the form of a Post-Doctoral Grant (Ref.:

SFRH/BPD/81748/2011).CMC,CRI,JBLandJBacknowledgepartial

supportfromtheNationalScienceFoundationundergrantnumber

OCI-1047828.JBalsoacknowledgessupportfromtheDepartment

ofEducationforaGraduateAssistanceinAreasofNationalNeed

(GAANN)FellowshipundergrantnumberP200A090323.

References

[1]I.A.McLure,P.J.Clements,Shearviscosityofhexane+tetradecafluorohexane neartheuppercriticalendpoint,Ber.Bunsenges.Phys.Chem.101(1997) 114–119.

[2]P.Morgado,C.M.C.Laginhas,J.B.Lewis,C.McCabe,L.F.G.Martins,E.J.M. Fil-ipe,Viscosityofliquidperfluoroalkanesandperfluoroalkylalkanesurfactants, J.Phys.Chem.B115(2011)9130–9139.

[3]P.Kirsch,ModernFluoroorganicChemistry,Wiley-VCH,Weinheim,2004. [4]S.S.Jang,M.Blanco,W.A.Goddard,G.Caldwell,R.B.Ross,Thesourceofhelicity

inperfluorinatedn-alkanes,Macromolecules36(2003)5331–5341. [5]P.Morgado,R.Tomás,H.Zhao,M.C.dosRamos,F.J.Blas,C.McCabe,E.J.M.Filipe,

Solutionbehaviorofperfluoroalkanesandperfluoroalkylalkanesurfactantsin n-octane,J.Phys.Chem.C111(2007)15962–15968.

[6]P.Morgado,H.Rodrigues,F.J.Blas,C.McCabe,E.J.M.Filipe,Perfluoroalkanesand perfluoroalkylalkanesurfactantsinsolution:partialmolarvolumesinn-octane andhetero-SAFT-VRmodelling,FluidPhaseEquilib.306(2011)76–81. [7]L.Lepori,E.Matteoli,A.Spanedda,C.Duce,M.R.Tiné,Volumechangesonmixing

perfluoroalkaneswithalkanesorethersat298.15K,FluidPhaseEquilib.201 (2002)119–134.

[8]W.L.Jorgensen, D.S.Maxwell,J.Tirado Rives,Developmentandtestingof theOPLSall-atomforcefieldonconformationalenergeticsandpropertiesof organicliquids,J.Am.Chem.Soc.118(1996)11225–11236.

[9]E.K.Watkins,W.L.Jorgensen,Perfluoroalkanes:conformationalanalysisand liquid-statepropertiesfromabinitioandMonteCarlocalculations,J.Phys. Chem.A105(2001)4118–4125.

[10]P.Morgado,J.B.Lewis,C.M.C.Laginhas,L.F.G.Martins,C.McCabe,F.J.Blas, E.J.M.Filipe,Systemsinvolvinghydrogenatedandfluorinatedchains: volumet-ricpropertiesofperfluoroalkanesandperfluoroalkylalkanesurfactants,J.Phys. Chem.B115(2011)15013–15023.

[11]S.Plimpton,Fastparallelalgorithmsforshort-rangemolecular-dynamics,J. Comput.Phys.117(1995)1–19.

[12]R.G.Bedford,R.D.Dunlap,Solubilitiesandvolumechangesattendingmixing forthesystem:perfluoro-n-hexane-n-Hexane,J.Am.Chem.Soc.80–2(1958) 282–285.

[13]G.Williams,D.C.Watts,Non-symmetricaldielectricrelaxationbehaviour aris-ingfromasimpleempiricaldecayfunction,Trans.FaradaySoc.(1970)80–86, 66-565P.

[14]M.Mondello,G.S.Grest,Viscositycalculationsofn-alkanesbyequilibrium moleculardynamics,J.Chem.Phys.106(1997)9327–9336.

[15]P.J.Linstrom,W.G.Mallard(Eds.),NISTChemistryWebBook,NISTStandard Ref-erenceDatabaseNumber69,NationalInstituteofStandardsandTechnology, GaithersburgMD,20899,http://webbook.nist.gov,(retrieved20.07.11). [16]W.Song,P.J.Rossky,M.Maroncelli,Modelingalkane+perfluoroalkane

interac-tionsusingall-atompotentials:failureoftheusualcombiningrules,J.Chem. Phys.119(2003)9145–9162.

[17]B.E.Poling,J.M.Prausnitz,J.P.O’Connell,ThePropertiesofGasesandLiquids, 5thed.,McGraw-Hill,NewYork,2001.

[18]A.Aucejo,M.C.Burguet,R.Munoz,J.L.Marques,Densities,viscosities,and refractiveindicesofsomen-alkanebinaryliquidsystemsat298.15K,J.Chem. Eng.Data.40(1995)141–147.