Silicone membrane contactor for selective volatile fatty acid and alcohol separation

(1)

ContentslistsavailableatScienceDirect

Process

Safety

and

Environmental

Protection

j o ur na l h o me pa g e : w w w . e l s e v i e r . c o m / l o c a t e / p s e p

Silicone

membrane

contactor

for

selective

volatile

fatty

acid

and

alcohol

separation

Harish

Ravishankar

a,∗

_,

_Paolo

_Dessì

a

_,

_Stefano

_Trudu

b

_,

_Fabiano

_Asunis

a,b

_,

_Piet

_N.L.

_Lens

a a_Department_of_{Microbiology,}_School_of_Natural_Sciences_and_Ryan_Institute,_National_University_of_Ireland_Galway_(NUIG),_University_Road,_Galway,_H91 TK33,Ireland

b_Department_of_Civil,_{Environmental}_and_{Architectural}_Engineering,_University_of_Cagliari,_Via_Marengo_2,_09123,_Cagliari,_Italy

a

r

t

i

c

l

e

i

n

f

o

Articlehistory: Received6July2020 Receivedinrevisedform 22September2020 Accepted24September2020 Availableonline28September2020 Keywords:

Cheesewhey

Volatilefattyacids(VFA) Silicone

Membranecontactor

a

b

s

t

r

a

c

t

TheeffectofpHandextractiontemperatureonﬂux,recovery,masstransfercoefﬁcientandseparation

factorofvolatilefattyacids(VFAs)andalcoholsfromsyntheticsolutionsandcheesewheyfermentate

wasinvestigatedusingasiliconemembranecontactorwithwaterasextractant.Thesiliconemembrane

allowedextractionofundissociatedacidsonly,resultinginsubstantiallyhigherrecoveryefﬁcienciesatpH

3thanatpH5.Furthermore,thenon-poroussiliconemembranefavouredextractionoflongerchainover

shorterchainacids.Caproicacidwasextractedwiththehighestﬂuxof1.30(±0.02)gm−2_h−1_in_short

time(32h),witha41.5%recoveryefﬁciencyatpH3and20◦C,indicatingthefeasibilityofitsselective

separationfromtheVFAmixture.Asimilartrendwasobservedforalcohols,withbutanolbeingextracted

witha39%recoveryefﬁciencyat40◦_C,_against₃₂_%_and₁₉_%_of_propanol_and_ethanol,_{respectively,}_while

themasstransfercoefﬁcientswerenotaffectedbytemperature.Whenapplyingthesiliconemembrane

contactortorealcheesewheyfermentateatpH3,butyricandaceticacidwereextractedwith21.5%and

7%recoveryefﬁciency,respectively,suggestingthefeasibilityofthecontactorforVFArecoveryfromreal

fermentate.

anopenaccessarticleundertheCCBYlicense(http://creativecommons.org/licenses/by/4.0/).

1. Introduction

Waste valorisationplays akeyrole incircular economythat reliesonthetransformationofvaluechainsfromlineartoclosed loop(Mainaetal.,2017).ThisisnecessarytoachieveEU’s long-termgoalofalowcarboneconomyby2050(Scarlatetal.,2015). Biologicalprocessessuchasdarkandphotofermentation(Yuan etal.,2019)havethepotentialtopartiallyreplacefossilfuel-based refineriestoproduceplatformchemicalssuchasvolatilefattyacids (VFAs)andalcohols.Hence,wasteprocessingismovingtowardsa biorefineryapproach,inwhichacombinationofphysico-chemical andbiologicalprocessesisusedtoobtainmarketableproductsfrom waste.Separationandrecoveryofthesevaluableproductsisstill amajorbottleneckduetolowconcentrationsandcomplex phys-iochemicalnatureofthefermentateanddigestate(Zacharofand Lovitt,2014).In-linerecoveryoftheseproductsfrombiological pro-cesseswouldbeadvantageousascontinuousharvestingofVFAs andalcoholswouldfacilitatetheirefficientandstableoperation (Tradetal.,2015).

∗ Correspondingauthor.

E-mailaddress:harish.ravishankar@nuigalway.ie(H.Ravishankar).

Severalseparationmethodshavebeenreportedinthe litera-ture,i.e.solventextraction,adsorption,andmembraneprocesses, including electrodialysis, reverse osmosis/nano-ﬁltration and membraneextraction(Atasoyetal.,2018), eachhavingitsown beneﬁtsand drawbacks.Solventextractionisa methodusedto separatecompoundsorcomplexesbasedontheirrelative solubili-tiesintwodifferentimmiscibleliquids.Differentextractantssuch astrioctylamine(TOA),trioctylphosphineoxide(TOPO),Alamine 336orN,N-didodecylpyridin-4-aminearereportedintheliterature withTOAbeingthemostused(Lietal.,2002;Alkayaetal.,2009; Reyhanitashetal.,2016).However,suchchemicalsareexpensive andrequirearegenerationstep.Furthermore,theextractantsare non-selective,andextractionofothercompoundsthanVFAs(e.g. saltsandalcohols)canresultinlowpurity.Theseextractantsare alsomostlytoxictomicroorganismsandcanthusnotbeapplied in-lineincombinationwiththebiologicalprocesses(Playneand Smith,1983).

Adsorptionisa surfacephenomenoninwhichthemolecules fromagaseousorliquidmediumadheretoa solidsurface. Ion-exchangeresinsareusedasadsorbentsforVFAadsorption(Bertin et al.,2016; Cabrera-Rodríguez et al., 2017; Reyhanitash et al., 2017)withamaximumreportedrecoveryyieldof85%from syn-theticmixtures(Rebecchietal.,2016).Adesorptionstepishowever https://doi.org/10.1016/j.psep.2020.09.052

(2)

H.Ravishankaretal. ProcessSafetyandEnvironmentalProtection148(2021)125–136 requiredtoobtaintheﬁnalproducts,whichresultsindeterioration

overtime(Aktijetal.,2020),therebyincreasingtheprocesscosts. Furthermore,adsorptionisnotselectivetowardsVFAs,resultingin productscontaminatedwithanionssuchasphosphates,sulphate andchloride,commonlyfoundinbiologicalprocesses.

Membrane processes are welldocumented in the literature and have beenpreviously used forVFA separation (Aktij et al., 2020).Membranesfacilitateinseparationwhileavoidingcontact betweenthebulksolutionandthepermeate.Typically,membrane processesinvolveapplicationofhighpressure(e.g.nanofiltration and reverseosmosis; pressurerange:3.5–20bar) oran electric field(e.g.electrodialysis)acrossasemi-permeableorionexchange membrane, respectively, that separates solutessuchas salts or organicmoleculesfromthesolventandothercompounds(Aktij etal.,2020).Nanofiltrationandreverseosmosismembraneshave beeninvestigatedforVFAseparationfromdifferentmatrices,and aseparationof75–90%havebeenreportedunderoperating con-ditionsofpH3.5and5bar,andpH2.93and50bar,respectively (Xiongetal.,2015;Zhouetal.,2013).Jonasetal.(2015)applied electrodialysis forVFAseparationfroma syntheticsolutionand showed99 %recoverywithin60minof operation(Joneset al., 2015),whereasZhangandAngelidaki(2015)recovered98.3%of VFAsfromdigestedpigmanureviabipolarmembrane electrodial-ysis.However,bothprocessesareenergyintensiveandstillrequire considerableresearchtomakethemcost-effective,particularlyfor recoveryofproductsfromwastestreams,duetoproblemssuchas inhomogeneityandfouling.

Another type ofmembrane based VFA separation processis thevapourpermeationmembranecontactorthatworksonvapour pressuredifferenceandconcentrationgradient(Aydinetal.,2018). Yesil et al. (2014) studied VFAs extraction from organic solid wasteleachatesolutionsusinghydrophobic polytetrafluoroethy-lene(PTFE)membraneswithNaOHasextractantforVFAdiffusion andprecipitationassodiumsalts.Integrationofamembrane con-tactorwithaleachbedreactordemonstratedseparationofacetic, butyric and caproicacid from the leachate(Yesil et al., 2014). Aydinetal.(2018)extendedtheideaandstudiedtheapplication ofextractantfilledmembranes(PTFEmembranewithairandtwo tertiaryamine(trioctylamine(TOA)andtridodecylamine(TDDA)) forVFAsseparationfromasyntheticmixtureandafermentation brothofmunicipalorganicsolidwaste,anaerobiclandfillleachate andchickenmanuredigestate.TOA-filledPTFEmembranesshowed highremovalefficienciesforallVFAspresentinthefermentation brothandlandfillleachate,withamaximumremovalefficiencyof 86–95%forpropionic,butyric,valericandcaproicacid(Aydinetal., 2018).

OutramandZhang(2018)reportedsolventfreeseparationof VFAsusingsiliconemembraneswithwaterasanextractant.The advantagesofusingwaterincludelowercost(thansolventssuch asNaOHorTOAwithoctanol)andrecoveryofVFAsinthe undisso-ciatedform,therebyeliminatingtherequirementforacounter-ion removalprocessasinthecaseofotherreportedstudies(Aydinetal., 2018;Yesiletal.,2014;Tugtas,2014).Furthermore,foulingdidnot occuronthesiliconemembraneandtheprocessrequiresalarge membranesurfaceareaduetothelowmasstransfercoefﬁcients ofVFAs.However,thestudydidnotlookatthediffusionofcaproic acidathighconcentrationsnoratorganicsolventssuchasalcohols. Thesearevitaltounderstandtheseparationcharacteristicswhile treatingamulticomponentsolutionsuchasfermentatewhichcan affecttheperformanceandeconomicscalabilityofthesystem.

The present study investigated the applicability of silicone membranesforVFAand alcoholseparationfromsynthetic solu-tionsand amodel anaerobicfermentate,i.e. cheesewhey,with waterastheextractant.AsyntheticmixtureofconcentratedVFAs andalcohols,andcheesewheyfermentatewereexaminedforVFA separation atdifferenttemperatures(20,30and 40◦C) andpH

(3and5).The VFAsand alcoholsdiffusionthroughthesilicone membranesandtheirflux,masstransfercoefficients,recoveryand separationfactorwereinvestigated.Thisstudyprovidesinsights inVFAseparationfromasyntheticsolutionandcheesewhey,and reportsforthefirst-timethealcoholextractionthroughasilicone membrane,inviewofutilisingthesiliconemembraneseparation processforin-lineVFAandalcoholextractionfromanaerobic fer-mentates.

2. Materialsandmethods 2.1. Sourceoffeedsolution

In viewof understandingthe separation of carboxylic acids andalcoholsinamulticomponentsolution,asyntheticVFA solu-tioncontainingequalamountsofacetic,propionic,butyric,valeric andcaproicacids (5gL−1 each)and alcoholsolutions contain-ingethanol,propanolandbutanol(5gL-1 _each)_were_prepared

in ROwater (resistivity 13 Mcm−1).Equal concentrations of VFAsandalcoholswerechosentoavoidtheconcentrationrelated changesinﬂuxandseparationfactor.Cheesewheyfromcowmilk wasobtainedfromthedairyindustry (Dairygold,Mitchelstown, Ireland)andthefermentate,richinVFAs,wasobtainedafter fer-mentationofcheesewheyat35◦CandpH5for7–8days(Dessì etal.,2020).PreliminaryanalysisoftheVFAcontentshowedthe predominanceof butyricand acetic acidwith an average con-centrationof,respectively,4.6and4.0gL−1inthecheesewhey fermentate.

2.2. Experimentalset-up

Asystemconsistingoftwobeakers,asiliconetubemembrane (peroxidecross-linked,withinternaldiameterof3mmand exter-naldiameterof5mm,and2mlength,VWRLtd),aperistalticpump (Masterﬂex)andasystemofnon-permeabletubes(MasterﬂexL/S TygonE-LabE-3603)connectingthefeedandthedrawwasusedfor theexperiments(Fig.1).Thefeedbeakercontained400mLof syn-theticVFA/alcoholsolutionorcheesewheyfermentate,whereas thedrawbeakercontained400mLROwater.Theperistalticpump wasoperatedat55mLmin−1.Thedrawsolutionwasstirredat 150rpmbya magneticstirrerwithtemperaturecontrol,inside whichthesiliconemembranewasimmersedforextractiontests. Anactiveinternalmembraneareaof0.0125m2_was_in_contact_with

theROwaterinthedrawbeaker.Allexperimentswereperformed induplicatesandmembraneswerechangedaftereachexperiment. ThedetailsofindividualexperimentsaresummarisedinTable1. Experimentswereconductedforatleast70hwithsamplestakenat periodicintervalsfrombothbeakersforVFA/alcoholanalysis.The temperatureofthedrawsolutionwasmaintainedat20,30or40◦C usingahotplate.TheexperimentswithsyntheticVFAfermentate wereperformedatpH3and5atthreedifferentextraction temper-atures(20,30and40◦C).ThepHvalueschosenwerebelowand slightlyabovethepKaoftheacids.Beforethestartofeach experi-ment,ifnecessary,thefeedpHwasadjustedusingH2SO4orNaOH.

ThesyntheticalcoholsolutionhadapHof2andwasusedassuchfor theexperiments,sincealcoholdoesnotdissociateatlowpH.For theexperimentsonVFArecoveryfromcheesewheyfermentate, theﬁrstsamplewascollectedafter10min.(toensurethecheese wheyfermentatewasabletoﬂowthroughthemembranewithout anyblockage/clogging).

2.3. Analyticalmethods

ThepHandconductivityforthedrawandfeedsolutionwas monitoredusinganaccumet®_pH_and_conductivity_meter_(AB200)

(3)

Fig.1. Schematicrepresentationoftheexperimentalset-up.

Table1

Overviewoftheexperimentalconditionsappliedtothesiliconemembranecontactor.

Experiment Feedsolution FeedsolutionpH Temperatureofdrawsolution(◦_C)

1 SyntheticVFAsolution 3and5 20,30and40

2 CheeseWheyfermentate 3and5 20,30and40

3 Syntheticalcoholsolution 2 20,30and40

throughoutthecourseoftheexperiment.Thechangeinmassofthe feedwasmonitoredusingaweighingscale(OhausScout®_SKX).

VFAconcentrationsweremeasuredusingaVarian450gas chro-matograph(GC)equippedwithaflameionisationdetectorandan SGEBP-21column(30mlong,internaldiameter0.25mmandfilm thickness0.25␮m).Heliumwasusedasacarriergasataflowrate of1mLmin−1.TheGCoventemperaturewasincreasedfrom60◦C to110◦Catarateof30◦Cmin-1_and_from₁₁₀◦_C_to₂₀₀◦_C_at_a

rateof10◦Cmin−1.Theinjectoranddetectortemperatureswere 250◦Cand300◦C(Nzeteuetal.,2018).PriortoGCanalysis,cheese wheyfermentatesampleswerecentrifugedat11,000rpmfor 6 min(Eppendorfnon-IVDCentrifuge5430G)andthesupernatant wasﬁltered(0.2␮m)anddilutedappropriately.Thealcohol con-centrationwasmeasuredusingliquidchromatography(LC)(1260 InﬁnityII,Agilent,USA)equippedwitharefractiveindexdetector (RID)andaHi-PlexHcolumn.ThemobilephasewasH2SO4(5mM)

ataﬂowrateof0.7mLmin−1. 2.4. Calculations

Theﬂux(J)orpermeationrateofindividualVFAoralcoholwas calculatedusingEq.1:

J= 1_Am_t (1)

wheremisthemassofVFAoralcoholpermeatedthroughthe membrane(g),Aisthemembranesurfacearea(m2₎_and_t_is_the

timeinterval(h).

Theoverallmasstransfercoefﬁcient,K,wasestimatedusingEqs. 2and3(OutramandZhang,2018):

Ji=AK

Ci,D−Ci,D∗

(2) ln

_C i,Dt−C∗i,D Ci,D0−Ci,D∗

= AKt VF (3) whereCi,D0istheinitialconcentrationatt=0,Ci,Dtisthe

concen-trationattime,C*i,Ddenotestheequilibriumconcentration,Vfis

theinitialvolumeoftheVFAsolution,tistime(h)andAisthe sur-facearea(m2_)._The_values_were_calculated_using_the_draw_VFA_or

alcoholconcentration.

Themembraneseparationfactor(␤VFA)wasestimatedusingEq.

4(Aydinetal.,2018):

ˇVFA/Water=

VFAoralcoholweightfractioninpermeate VFAoralcoholweightfractioninthefeed

Waterweightfractioninthepermeate Waterweightfractioninthefeed

(4)

Thewaterweightfractioninthepermeateandfeediscalculated usingEq.5(Aydinetal.,2018):

W_water,i₌ ixVi−(

mVFA,i)

ixVi

(5) where␳iisthedensityofthesolution(gL−1),Viisthevolumeof

thesolution(L)andmVFA,iisthemassofindividualVFA(g).

TherecoverypercentagewascalculatedusingEq.6: Recovery (%)

=

ConcentrationofVFAorAlcoholindrawsolution(tn)

ConcentrationofVFAorAlcoholinfeedsolution(t0)

∗100

(6) wheret0andtnrepresentthestartandtheendtime(h)of

exper-iment.Forfewexperiments,thefeedconcentrationofindividual VFAsatthestart wasobserved tobeless than5gL−1 andthe observedvalueswereusedtocalculatetherecoverypercentage.

ThevapourpressurewascalculatedusingEq.7:

Ps =PS,t ×Xs,t (7)

wherePsisthevapourpressureoftheVFAinsolution,Ps,tisthe

vapourpressureoftheindividualVFAatagiventemperatureand Xs,tisthemolefractionoftheindividualVFAatagiventemperature.

(4)

H.Ravishankaretal. ProcessSafetyandEnvironmentalProtection148(2021)125–136

Fig.2. DrawsolutionVFAconcentrationswithsyntheticVFAmixtureasfeedatpH3a)20◦_C,_b)₃₀◦_C,_c)₄₀◦_C,_and_pH5_d)₂₀◦_C,_e)₃₀◦_C_and_f)₄₀◦_C._(Note_the_different y-axisscalefordifferentpHconditions).

3. Resultsanddiscussion

3.1. VFArecoveryfromsyntheticVFAsolution

Fig.2showstheconcentrationproﬁleofVFAsinthedraw solu-tion.Attheendof 70hoperation,thedrawsolutionhad more valericacidascomparedtootherVFAsregardlessofpHand temper-ature.TheVFAconcentrationinthedrawsolutionincreasedwith time, exceptforcaproicacid,which initiallyincreasedand then decreased foralltheconditions(Fig.2).Analysisofthefeedside caproicacidconcentrationshoweditdidnotdiffusebacktothefeed compartment.AtpH3,thecaproicacidconcentrationincreased rapidlyinthedrawsolutioninashorttime(10−30hdepending

onthetemperature)andstartedtodecreaseovertimeafter32,24 and11hat20,30and40◦C,respectively,suggestingthe possibil-ityofevaporativelossorformationofanimmisciblelayerontop ofthedrawsolutionduetothelowsolubilityasopposedtoother VFAsinwater(Khoretal.,2017).ThistrendwasalsonoticedatpH 5foralltemperaturestested,withdecreasingcaproicacid concen-trationsinthedrawsolutionafter46,22and22hat20,30and 40◦C,respectively.Thepresentstudyusedthehighestcaproicacid concentration(5gL−1)inthesyntheticsolutioncomparedtoother modelsolutionsreportedintheliterature(Aydinetal.,2018;Yesil etal.,2014;OutramandZhang,2018;Tugtas,2014).Nonetheless, therapidincreaseincaproicacidconcentrationinashorttime(20

(5)

h)canbeusedforitsselectiveextractionfromaVFAmixtureeven atsuchhighinitialconcentrations.

Feedcomposition,pHandtemperatureaffectedmasstransfer fromthefeedtothedrawsolution,resultingindifferentoverall massfluxvalues(Table2).AtbothpH3and5,thefluxvaluesof thefattyacidsgenerallydecreasedwithincreaseinextraction tem-perature.Thiswasaresultofthehighernetvapourpressureinthe drawside,resultinginaresistancetodiffusionandhencereduction intheflux.AtpH3,thecalculatedvapourpressureofindividual acidsinthedrawsolutionwashighercomparedtothefeedside (TableS1),withanexceptionbeingaceticacid,wherethefluxdid notchange(Table2).Asthemajorityoftheacids(about60%)is dissociatedatpH5(pKa∼4.7–4.8),thevapourpressuredatafor individualacidsinfeedsidewasnotcalculated.However,the indi-vidualvapourpressureoftheacidsinthedrawsolution(wherepH <4)wascalculated(TableS1,FigS1)andcanbeexpectedtobe higherthanthefeedsidewithpH5(BandiniandGostoli,1992). Valericacidshowedthehighestoverallfluxafter70hofoperation (0.70(±0.07)gm−2h−1)atpH3and20◦C,followedbycaproicacid (0.52(±0.06)gm−2h−1).However,atthefirst32hofoperationat pH3and20◦C,caproicacidwasextractedwithamaximumfluxof 1.3(_±0.02)gm−2h−1,withanoverallrecoveryof41.5%.

Aydinetal.,(2018)conductedastudywithamicroporousPTFE andPTFE-TOAcompositemembraneforVFAsseparation(atpH3.9 for21,30and38◦C)fromasyntheticVFAsolutionusingNaOHas extractantfora7hperiod.Thestudyreportedthatthemassfluxof individualVFAswasnotsignificantlydifferentat21◦Cand30◦C, butslightlyincreasedat38◦CforseparationusingthePTFE mem-brane.ThemassfluxesobtainedforPTFE-TOAmembraneswere comparablewithtemperature(at21,30and38◦C)withslightly lowerfluxobservedforvalericandcaproicacidwhenthe tempera-turewasincreasedfrom21◦Cto30◦C.Yesiletal.(2014)conducted aVFAseparationstudyusinghydrophobicPTFEmembraneswith syntheticmixedVFAsolutionatpH3and30◦Cwith1NNaOHas drawsolutionfor30h.Propionicacidhadamaximumfluxof14.21 gm−2h−1followedbyaceticacidat13.12gm−2h-1_._Caproic_acid

hadthelowestﬂuxof2.13gm−2h−1(Yesiletal.,2014).Thiswas duetothehighaceticandpropionicacidconcentrations(6gL−1) comparedtothatofcaproicacid(1gL-1₎_in_the_synthetic_feed_VFA

mixture.

Thefluxvaluesobtainedinthepresentstudyundersimilar oper-atingconditions(pH3andtemperature30◦Cwithwaterasdraw solution) (Table2)wereratherlowcomparedtothosereported withsimilarextractionmethods(Aydinet al.,2018; Yesiletal., 2014).Thiscanbeattributedduetothenon-porouspropertyof thesiliconemembranethatresistsmasstransferasopposedtothe porousPTFEmembranes(Aydinetal.,2018;Yesiletal.,2014)which facilitatesolutetransport.Generally,non-porousmembranesare used toseparatemoleculeswithsizesinsame orderof magni-tude(Mulder,2012).Thefluxoforganicliquidsorvapoursthrough non-porousmembranesdependsonconcentrationgradientswith diffusivitiesincreasingwithconcentration(Mulder,2012).Higher massfluxesof7.34gm−2h-1_and_5.4_g_m−2_h-1_were_reported_for

aceticacidat25◦Cwhenusingrespectively,poroushydrophobic hollowﬁbremembranes(Qinetal.,2003)andTOAimpregnated hydrophobic membranes,in pervaporationunits (Thongsukmak andSirkar,2007).Thesestudiesreportinghighermassﬂux(Aydin etal.,2018;Yesiletal.,2014;Qinetal.,2003;Thongsukmakand Sirkar,2007)ofVFAshowever,requireeitherachemicalextractant orahighenergyinvestmentasopposedtothepresentwork.

Sincetheextractionprocessthroughsiliconeisdrivenbya con-centrationgradient,amaximumofhalftheinitialconcentrationin thefeedsolutioncanbetheoreticallyobtainedinthedrawsolution. BasedontheinitialVFAconcentrationinfeedandﬁnalVFA con-centrationindrawsolution,therecoveryofindividualVFAswas calculatedfortheexperimentalconditions(Fig.3.).Theincreasein

temperatureconsiderablydecreasedtheVFAsmasstransferfrom thesyntheticVFAsolutionandhencereducedtheVFArecoveryat bothpHvaluesinvestigated.AtpH3,therecoverytrendafter70 hwasvaleric>caproic>butyric>propionic>aceticacid. How-ever,caproicacidrecoverycouldbeenhancedwhenextractingat ashorterextractiontimeof32,24and11hat20,30and40◦C, respectively,whereitsconcentrationwashigherascomparedto 70h(Fig.2).Amaximumrecoveryof5,15,29,45and38%was obtainedatpH3and20◦Cforacetic,propionic,butyric,valeric andcaproicacid,respectively.AtpH5,theVFArecoveryfolloweda similartrendasobservedatpH3.Digestateswithhigh concentra-tionofcaproicacid,viz.3.2and4.5gL−1(valuescomparedclosely tothiswork)presentinleachateofafermentedorganicwasteand chickenmanuredigestateshowedlowrecoveriesof8.5and<10% foranexperimentaldurationof15daysand7h,respectively,when usingPTFEandPTFE-TOAmembranecontactors(Aydinetal.,2018; Yesiletal.,2014).Shorterextractiontimescouldhaveledtoabetter caproicacidrecoveryasobservedinthepresentstudy.

TherecoveryofVFAswaslowerthanthetheoretical equilib-riumconcentrationof2.5gL−1,whichcouldpossiblybedueto theadsorptionofVFAsonthesiliconemembraneduetoitshigh hydrophobicity.Asimilarobservationofconcentrationsbelowthe predictedequilibriumconcentrationisreportedforasyntheticVFA solutionandwasattributedtothedistributionofVFAsinthe mem-branes(OutramandZhang,2018).

TheoverallmasstransfercoefficientsofVFAsextractedat dif-ferenttemperaturesandpH(Table3)werecalculatedconsidering themaximumconcentrationobtainedinthedrawsolutionasthe equilibriumconcentration.Theoverallmasstransfercoefficients ofVFAsfollowedtheorderofthecarbonchainlength(caproic> valeric>butyric>propionic>acetic)for bothpHvalues inves-tigated,irrespectiveofthetemperature.Thelongerchainacids, indeed,haveahigheraffinitytothesiliconemembraneduetotheir higherhydrophobicity(Yesilet al.,2014).The coefficientswere loweratpH5ascomparedtopH3(Table3),althoughtheeffect oftemperatureontheoverallmasstransfercoefficientswasnot observed(Table3).

Themasstransfercoefficientvalueswerecomparabletothose reportedbyOutramandZhang (2018).Thestudyreportedthat iso-valericacidhadthehighestmasstransfercoefficient(0.14␮m s−1)followedbybutyric(0.082␮ms−1)andacetic(0.02␮ms−1) acidat25◦C.Yesiletal.(2014)alsoreportedahighermass trans-fercoefficientforcaproicacid(2.07␮ms−1)asopposedtoother fattyacids(acetic,propionic,butyricandvalerichavingmass trans-fercoefficientsof0.56,0.71,0.97and1.21␮ms−1,respectively) usingacross-flowmembranecontactorwithaPTFEmembrane.The masstransfercoefficientsreportedbyAydinetal.(2018)werealso higherinmagnitude,rangingbetween0.61–2.3,1.16–3.3,2.7–7.5, 4.4–16.3and5.8–19.7␮ms−1foracetic,propionic,butyric,valeric andcaproicacid,respectively.Thisisduetotheuseofporous mem-branesalong withNaOHasextractantwhichcreatesa stronger drivingforceforVFAtransfer.Themasstransfercoefficientstrongly dependsonthehydrodynamicsofthesystemandcantherefore bevariedandoptimised.Inthepresentsystem,themovementof VFAsthroughthesiliconemembraneisthroughabsorption, diffu-sion,anddesorptionintotheextractant,whichisaslowprocess (OutramandZhang,2018).Furtherresearchonthemasstransfer resistanceisneededtoelucidatewhichofthesecomponentsis lim-itingthetransfer.However,reducingthethicknessorincreasingthe flowvelocitycouldfurtherimprovethemasstransfercoefficient ofthesiliconemembranecontactorsystemstudied,though nega-tivelyaffectingtheselectivitytowardslongerchainacids(Outram andZhang,2018).

pHandelectricalconductivityweremonitoredforthefeedand drawsolution,whichinthisexperimentwasanindicatoroftheVFA migration.Fig.S1andS2showsthevariationsinpHand

(6)

conduc-H. Ravishankar et al. Process Safety and Environmental Protection 148 (2021) 125–136 Table2

FluxofVFAsacrossasiliconemembranefordifferentfeedcompositionandtemperatures. Flux(gm−2_h-1₎

VFAs SyntheticVFAsolution Cheesewheyfermentate

pH3 pH5 pH3 pH5 20◦_C ₃₀◦_C ₄₀◦_C ₂₀◦_C ₃₀◦_C ₄₀◦_C ₂₀◦_C ₃₀◦_C ₄₀◦_C ₂₀◦_C ₃₀◦_C ₄₀◦_C Aceticacid 0.04±0.00 0.04±0.00 0.03±0.02 0.01±0.00 0.02±0.02 0.01±0.01 0.04±0.02 0.06±0.04 0.09±0.00 0.02±0.00 0.02±0.00 0.03±0.00 Propionicacid 0.16±0.01 0.16±0.02 0.10±0.04 0.05±0.00 0.06±0.05 0.04±0.00 – – – – – – Butyricacid 0.40±0.03 0.38±0.03 0.20±0.08 0.10±0.00 0.09±0.00 0.05±0.00 0.25±0.08 0.28±0.02 0.25±0.02 0.02±0.00 0.07±0.00 0.09±0.00 Valericacid 0.70±0.07 0.47±0.07 0.23±0.10 0.32±0.03 0.20±0.00 0.12±0.16 – – – – – – Caproicacid 0.52±0.06 0.22±0.06 0.09±0.04 0.30±0.06 0.20±0.02 0.10±0.02 – – – – – – Table3

MasstransfercoefﬁcientvaluesofVFAsthroughasiliconemembranewithsyntheticandcheesewheyfermentateasthefeed. Masstransfercoefﬁcient(␮ms−1₎

VFAs Syntheticfermentate Cheesewheyfermentate

pH3 pH5 pH3 pH5 20◦C 30◦C 40◦C 20◦C 30◦C 40◦C 20◦C 30◦C 40◦C 20◦C 30◦C 40◦C Aceticacid 0.12±0.02 0.06±0.02 0.10±0.00 0.02±0.00 0.04±0.00 0.06±0.04 0.06±0.04 0.03±0.01 0.09±0.04 0.12±0.09 0.12±0.08 0.14±0.01 Propionicacid 0.15±0.01 0.08±0.03 0.13±0.00 0.05±0.03 0.04±0.01 0.08±0.07 – – – – – – Butyricacid 0.17±0.01 0.13±0.06 0.16±0.00 0.07±0.02 0.06±0.01 0.18±0.06 0.14±0.02 0.13±0.02 0.08±0.01 0.17±0.03 0.07±0.01 0.20±0.04 Valericacid 0.23±0.02 0.33±0.20 0.30±0.07 0.11±0.05 0.16±0.08 0.16±0.07 – – – – – – Caproicacid 0.49±0.17 0.92±0.76 0.73±0.49 0.35±0.23 0.46±0.35 0.55±0.32 – – – – – –

Acidswerenotpresentincheesewheyfermentate.

(7)

Fig.3.VFArecoverythroughthesiliconemembranefromsyntheticfeedata)pH3andb)pH5.ItshouldbenotedthatamaximumVFArecoveryof50%isachievablebecause theextractionisaconcentrationgradientdrivenprocess.

tivityofthefeedanddrawsolutionsatthedifferentexperimental conditionsofVFAandalcoholextraction.Thedrawsolutionshowed asharpdecreaseduringtheinitial2h(forallconditions),indicating arapiddiffusionofVFAoralcoholacrossthesiliconemembrane (Fig.S1).Theconductivityproﬁlesofthefeedsolutions(Fig.S2) decreased withtime, whilethatofthedrawsolutionincreased, furtherconﬁrmingtheVFAoralcoholmigration.

3.2. VFArecoveryfromcheesewheyfermentate

Fig.4showstheVFAconcentrationproﬁlesofdrawsolutions, extractedfromthecheesewheyfermentate.Forallexperimental conditionstested,butyricacidconcentrationsinthedrawsolution increasedfasterthantheaceticacidconcentration(Fig.4).A sub-stantiallyhigherVFAextractionwasachievedatpH3than5,andat bothpHvalues,ahighertemperaturefavouredVFAextraction.This effectoftemperatureonVFAextractionthroughthemembranecan beunderstoodthroughtherelationshipbetweenpenetrationand temperatureestablishedbytheVan’tHoff–Arrheniusrelationship: P= P0exp(−

Ep

RT ) (8)

wherePisthepenetration,Poisapre-exponentialfactor,Risthe

molargasconstant,TisthetemperatureandEpistheapparent

acti-vationenergyofpermeationrequiredforVFAintothemembrane and theopeningbetweenthepolymericchainsofmembraneto allowtheVFAtodiffuse(Hanetal.,2001).

Eq.8showsthatforagivenmembraneandapenetrant(VFA), penetration(P)increaseswithtemperatureresultinginahigher extraction efﬁciency (Han et al., 2001). Although the tempera-tureshouldimprovetheVFAextraction,forthesyntheticsolution (seesection3.1),thehighernetvapourpressureinthedrawside resulted in lower recovery for longer carbon chain acids with increasedtemperature(TableS1).

Thecheesewheyfermentatecontaineddifferentions(NO2−,

NO3−andSO42-)withpredominanceofCa2+,K+,Na+,PO43-,Cl−

andNH4+of210,2297,694,124,270and462mgL-1,respectively.

Thepresenceofproteinsincheesewheyfermentatewasmeasured tobeintherangeof0.15−0.65gL-1_._The_high_conductivity_of_around

10mScm-1_conﬁrmed_the_presence_of_these_ions_in_the_cheese_whey

fermentate,whichfurtherincreasedto16mScm-1_at_pH₃_upon

additionwithH2SO4.

Themaximumrecoveryefﬁciencyforaceticandbutyricacid after70hoperationat40◦Camountedto,respectively,7and21.5

%forpH3,and 3.5and7%forpH5(Fig.5).Thelowerremoval efﬁciencyobservedcomparedtothesyntheticVFAsolutioncan beattributedto thepresence of thesolidsand ions present in thecheesewheyfermentate(Aydinetal.,2018).Forexample,the presence ofcalcium and phosphate ionsin the fermentatecan resultintheirprecipitationwhentheCa3(PO4)2solubilityproduct

isexceeded,resultinginadeclineofthepermeationfluxand selec-tivityofthemembrane(Chandrapalaetal.,2015).Furthermore,the ionscanresultinconcentrationpolarisation,formingaboundary layerwhichcaninduceinprecipitationofsaltsandthusimpede theVFAtransferacrossthemembrane(Bellman,2012).Thesolids orothersuspended particles, includingproteins, present inthe cheesewheyfermentatemayhaveformedalayeronthesilicone membranes,limitingthetransferofVFAs,eventhoughfoulingwas notvisuallyobserved.Similarrecoveryefficienciesof3.3and7.2% for,respectively,acetic,andbutyricacidhavebeenreportedfrom organicwasteleachateatpH6.6usingacounter-currentflow mem-braneextractionsystemwithNaOHasextractant(Yesiletal.,2014). Incontrast,acetic acidwasrecovered witha greaterthan 45% efficiencyfromthreedifferentorganicwastes(fermentationbroth, landfillleachateandchickenmanuredigestate)(Aydinetal.,2018) usingaTOA-filledPTFEporousmembraneinacontactor.Plácido and Zhang(2018) lookedat VFArecovery fromslaughterhouse bloodanaerobicfermentateusinga poroushydrophobichollow membranesystemwithOctanol/TOAasextractsolution. Unacid-ifiedslaughterhousebroth(wherepHwasunmodified)showed aVFArecoveryof <5%,confirmingthenecessityoflow pHfor VFArecovery(PlácidoandZhang,2018).Uponacidification(exact valuenotreported)oftheslaughterhouseanaerobicfermentate, theoverallVFArecoveryincreasedto80%,withvaleric,butyric, propionicandaceticacidshowing100,94,80and42%recovery, respectively.

Table2showstheVFAsfluxthroughthesiliconemembraneat differentpHandtemperature.Butyricacidhadahigherfluxas com-paredtoaceticacidatallconditionstested.AtpH3,themaximum fluxobtainedforbothbutyricandaceticacidwas0.28(±0.02)g m−2h−1(at30◦C)and0.09(_±0.00)gm−2h−1(at40◦C), respec-tively.Yesiletal.(2014)conductedexperimentsforVFAseparation fromorganicwasteleachateusingaPTFEmembranecontactor.The leachatesolutionhadapHof6.6at30◦Candcontainedfattyacids includingacetic(14,277mgL−1),propionic(846mgL-1_),_butyric

(3926mg L-1_),_valeric ₍₄₂₈_mg _L−1₎_and_caproic₍₃₂₂₃ _mg_L−1₎

acid.Themaximumﬂuxwasobtainedforacetic(0.240gm−2h−1), followedbybutyric(0.150gm−2 h−1)andcaproic(0.140gm−2

(8)

Fig.4.DrawsolutionVFAconcentrationswithcheesewheyfermentateasfeedatpH3a)20◦_C,_b)₃₀◦_C,_c)₄₀◦_C_and_pH5_d)₂₀◦_C,_e)₃₀◦_C_and_f)₄₀◦_C_(Note_the_different y-axisscalefordifferentpHconditions).

h−1)acid,suggestinghigherVFAconcentrationsfavouredagreater ﬂux,whichwasfacilitatedbytheconcentrationgradientacrossthe porousPTFEmembranes.Inthepresentstudy,theaveragebutyric acidconcentration(4.6gL−1)wasslightlyhigherthanthatofacetic acid(4.0gL−1),andthelongerchainacidsmigratedfasterthrough thesiliconemembrane.

Themasstransfercoefficientofbutyricacidwascomparatively higherthanaceticacidformostconditionsinvestigated(Table3). Acetic andbutyricacidshowedmaximum masstransfer coeffi-cientsof0.14(±0.01)and0.20(±0.04)␮ms−1atpH5.Overall,the coefficientsobtainedfromcheesewheyfermentatearesimilarto thoseobtainedforthesyntheticfattyacidsolutionforthe major-ityoftheconditionstested,confirmingthenetdrivingforcefor separationisthefreeacidconcentration.Themasstransfer coef-ficientofbutyricacidfromafishfermentationbroth(pH7)using asiliconemembraneextractionsystemwasreportedtobe0.157 ␮ms−1 (OutramandZhang,2018).Yesiletal.(2014)reporteda slightlyhighermasstransfercoefficientof0.022␮ms−1forbutyric

acidextractedfromorganicleachateofaleachbedreactoratpH 6.6usingahydrophobicpolytetrafluoroethylene(PTFE)membrane withNaOHas extractant.Plácido and Zhang(2018) reported a butyricacidmasstransfercoefficientof0.291␮ms−1obtainedfrom slaughterhouseanaerobicfermentateunder acidifiedconditions witha porouspolypropylenemembrane and TOA+1-Octanol as extractant.Table6comparesVFAextractionusingdifferent mem-branecontactorsreportedintheliterature.

3.3. Alcoholrecoveryfromsyntheticalcoholsolution

DependingontheoperationpHorprevailingconditionsinthe fermenter,thecheesewheyfermentationpathwaycanshiftto sol-ventogenesis,producingalcoholsratherthanVFAs(Caleroetal., 2018).Therefore,preliminarytestsofalcoholextractionacrossthe siliconemembranewereperformed.Allthealcoholstestedmoved fromthefeedtothedrawsolutionacrossthesiliconemembrane, withbutanolshowingthemaximumconcentrationattheendof 132

(9)

Fig.5. VFArecoverybythesiliconemembranefromcheesewheyfermentateata)pH3andb)pH5.ItshouldbenotedthatamaximumVFArecoveryof50%isachievable becausetheextractionisaconcentrationgradientdrivenprocess.

operation,followedbypropanolandethanol(Fig.6).Similarly,to theobservationnotedwithVFAs,thiscanbeascribedtothehigher hydrophobicityoflongerchainalcoholsthathavemoreafﬁnityfor thesiliconemembrane.

Thealcoholrecoveryincreasedwithtemperature(Fig.6),with anexception forbutanolduetothedecreaseinitssolubilityin waterat40◦C(StephensonandStuart,1986).Themaximum recov-eryof42%wasobservedforbutanolat30◦C,whereasthehighest propanolandethanolrecoveriesof32and19%,respectively,were achievedat40◦C(Fig.7).Butanolhadahigherfluxascomparedto propanolandethanol(Table4).Highertemperatureimprovedthe fluxofallalcoholsresultinginamaximumfluxof0.25(±0.01), 0.33 (±0.03) and0.42 (±0.07)gm−2 h-1 _for_ethanol, _propanol

andbutanol,respectively,at40◦C.ThongsukmakandSirkar(2007) reportedbutanoland ethanolseparation fromthefeedsolution containing1.5wt%butanol,0.8wt%acetoneand0.5wt%ethanol usingpervaporationwithTOAimmobilisedonhollowﬁbre mem-branes.Butanolandethanol hadaﬂuxof11 and1.2gm−2h-1_,

respectivelyat54◦C.Thelowerﬂuxvaluesreportedinthepresent workareduetotheinherentnatureoftheprocess(non-pressure driven)wheretheconcentrationgradientisthedrivingforcefor alcoholseparation.However,theuseofanon-poroussilicone mem-branecontactorenablesselectiverecoveryoflongerchainalcohols andischeaperthanpervaporationprocesses.

Theoverallmasstransfercoefficientsofalcoholsfollowedthe orderbutanol>propanol>ethanol foralltemperatures investi-gated(Table4).AsobservedforVFAs,theincreaseintemperature didnotsubstantiallyaffectthealcoholmasstransfercoefficients. Butanolhadthehighestmasstransfercoefficientof0.16(±0.00) ␮ms−1at40◦C,whereaspropanolandethanolhadthehighest coefficientof0.13(±0.00)and0.12(±0.03)␮ms−1,respectively, at20◦C.Lietal.(2011)investigatedin-situbutanolseparationfrom acetone-butanol-ethanolfermentationbrothusingPDMS compos-itemembranesthroughapervaporationprocess.Theoverallmass transfercoefficientsofbutanolduringtheseparationfromabinary (butanol/water),amodelandafermentationculturesolutionwere 0.41,0.35,and0.3␮ms−1,respectively(Lietal.,2011).

3.4. Membraneseparationfactor

Theseparationfactorisdeﬁnedastheabilityofamembraneto separateatargetcompoundandisacrucialparameterwhen select-ingmembranes(Luis,2018).Theseparationfactorcalculatedinthe

Fig.6. Drawsolutionconcentrationofalcoholsextractedusingasiliconemembrane contactorfromasyntheticalcoholmixture(pH2)ata)20◦C,b)30◦Candc)40◦C.

(10)

Table4

FluxandmasstransfercoefﬁcientsofalcoholsacrossasiliconemembraneatpH2fordifferenttemperatures.

Alcohol Flux(gm−2h-1₎ _Mass_transfer_coefﬁcient_(␮m_s−1₎

20◦_C ₃₀◦_C ₄₀◦_C ₂₀◦_C ₃₀◦_C ₄₀◦_C

Ethanol 0.16±0.00 0.18±0.01 0.25±0.01 0.12±0.03 0.08±0.02 0.10±0.01

Propanol 0.24±0.012 0.27±0.02 0.33±0.03 0.13±0.00 0.10±0.00 0.12±0.00

Butanol 0.40±0.034 0.40±0.05 0.42±0.07 0.15±0.00 0.14±0.01 0.16±0.00

Table5

SeparationfactorofVFAsandalcoholsfromsyntheticsolutionsandcheesewheyfermentate. SyntheticVFAsolution

VFA pH3 pH5 20◦_C ₃₀◦_C ₄₀◦_C ₂₀◦_C ₃₀◦_C ₄₀◦_C Aceticacid 0.17±0.01 0.25±0.01 0.28±0.01 0.06±0.00 0.13±0.00 0.16±0.00 Propionicacid 0.53±0.02 0.74±0.01 0.83±0.01 0.38±0.00 0.47±0.00 0.62±0.00 Butyricacid 1.06±0.00 1.34±0.00 1.41±0.00 0.70±0.00 0.80±0.00 0.85±0.00 Valericacid 1.63±0.02 1.64±0.02 1.56±0.02 1.55±0.00 1.64±0.00 1.71±0.00 Caproicacid 1.37±0.00 0.94±0.00 0.74±0.01 1.78±0.01 1.87±0.01 1.58±0.00

Cheesewheyfermentate

VFA pH3 pH5

20◦_C ₃₀◦_C ₄₀◦_C ₂₀◦_C ₃₀◦_C ₄₀◦_C

Aceticacid 0.39±0.01 0.40±0.01 0.38±0.01 0.77±0.01 0.65±0.01 0.68±-0.01

Butyricacid 1.68±0.01 1.68±0.01 1.68±0.01 1.32±0.02 1.33±0.02 1.28±0.02

SyntheticAlcoholsolution

Alcohol pH2

20◦_C ₃₀◦_C ₄₀◦_C

Ethanol 0.61±0.00 0.61±0.00 0.65±0.00

Propanol 0.95±0.00 0.89±0.00 1.13±0.00

Butanol 1.41±0.00 1.66±0.00 1.34±0.00

Fig.7.Alcoholrecoverythroughasiliconemembranefromasyntheticsolutionat pH2.Itshouldbenotedthatamaximumalcoholrecoveryof50%isachievablebecause theextractionisaconcentrationgradientdrivenprocess.

presentworkshowstheselectivityofVFA/alcoholoverwater.The non-poroussiliconemembraneusedinthisstudydoesnotsupport watertransferandtheseparationoffattyacidsdependssolelyon theconcentrationgradientthatactsasthedrivingforce.Valericand caproicacidhadahigherseparationfactorfromthesynthetic solu-tionascomparedtootherfattyacidsatbothpH3and5(Tables5,6). Astheseparationfactorswerecalculatedbasedontheﬁnal concen-trationsinthedrawsolution,caproicacid,duetoitslowsolubility, showedalowerseparationfactorcomparedtovalericandbutyric acidatpH3and30aswellas40◦C.However,theseparation fac-torforcaproicacidincreasedatpH5,whichisduetothelower separationofacetic,butyricandpropionicacidatpH5asopposed topH3(Table5).Theseparationfactorincreasedforacetic, pro-pionicandbutyricacidwithincreaseintemperatureatbothpH valuesinvestigated.Highseparationfactorvalues(>1)indicatea betterselectivity,suggestingthesuitabilityofthemembrane.For

separationofvolatileorganiccarbons(VOCs),suchasaceticacid, ethyleneglycolanddimethylacetamide(DMAC)fromwater,the separationfactortypicallyrangesfrom1–5forsiliconemembranes (Luis,2018).Anincreaseinseparationfactorbeyond5providesvery littleadditionalbeneﬁtsforVFAseparation(Baker,2012).When cheesewheyfermentatewasusedasthefeed,butyricacidhada higherseparationthanaceticacidatbothpH3and5(Table5).The increaseintemperaturehadanegligibleeffectontheseparation factorofbothacids,indicatingbutyricacidhadabetterselectivity overaceticacidregardlessofthetemperature.

TheeffectoftemperatureontheVFAsseparationfactorusing PTFEandPTFE+TOAmembraneswasinvestigatedusingsynthetic solutions(Aydinetal.,2018).Thecarboxylicacidseparationfactor increasedwiththealkylchainlengthincreasedatalltemperatures assessed(Aydinetal.,2018).Theseparationfactorwashigherfor thePTFE+TOAmembrane at30 ◦C, being0.74,7.05, 171.2for acetic,propionic,butyric,respectivelyand>1250forbothvaleric andcaproicacid(Aydinetal.,2018).Thepresentstudyfounda sim-ilarselectivityorderaswell,showinghigherselectivityforlonger alkylchainlengthsfromboththesyntheticsolutionandcheese wheyfermentate(Table5).Similarly,theseparationfactorof alco-holsinthepresentworkindicatedthatbutanolhadthehighest separationfactorfollowedbypropanolandethanolforall temper-aturesinvestigated(Table5).Thetemperaturedidnotshowany effectonalcoholselectivity(Table5),signifyingthatalongercarbon chainhadabetterextractionselectivityregardlesstheextractant temperature.Typicallyforalcohols,theseparationfactorranges between5–20forsiliconemembranes(Lietal.,2011).

3.5. Practicalimplication

AlthoughtheresultsindicatethefeasibilityofextractingVFAs andalcoholsfromfermentatewithouttherequirementofchemical extractantsandwithlittleenergy(non-pressuredrivenprocess), onlyamaximumrecoveryof50%canbeobtainedwiththepresent technology(whereusingsolelytheconcentrationgradientasthe drivingforceforextraction).Integrationofthepresentsystemfor 134

(11)

Ravishankar et al. Process Safety and Environmental Protection 148 (2021) 125–136 Table6

ComparativestudiesonVFAextractionusingmembranecontactors.

Stocksolution Membrane Operatingconditions Parameter Reference

pH,Timeand Temperature

Extractant VFAconcentration(g L−1) Masstransfer coefﬁcient(␮ms−1) Flux(gm−2_h-1₎ _Selectivity /separationfactor Recovery(%) Organicmunicipal leachate

PTFE 6.6,15dand30◦C 1NNaOH AA-14.27,PA-0.84, BA-3.9,VA-0.42,and CA-3.2

AA-0.003, PA-0.004,BA-0.021, VA-0.020,and CA-0.021

AA-0.240, PA-0.008,BA-0.150, VA-0.023,and CA-0.140 AA-0.03,PA-0.02, BA-0.06,VA-0.08, andCA-0.11 AA-3.3,PA-1.8, BA-7.2,VA-10.8, andCA-8.5 21

Syntheticsolution PTFE 3,30hand30◦_C ₁_N_NaOH _AA-_6,_PA-_6,_BA-_2, VA-2,andCA-1 AA-0.56,PA-0.71, BA-0.97,VA-1.21, andCA-2.07 AA-13.12, PA-14.21,BA-5.25, VA-5.27,and CA-2.13

AA-0.9,PA-1.10, BA-1.70,VA-2.20, andCA-5.60

Notreported 21

Syntheticsolution PTFE 3,24hand25◦_C _0.5_N_NaOH _AA-_6,_PA-_6,_BA-_5.5, andVA-1,

AA-0.7,PA-0.91, BA-0.947,and VA-1.64 AA-11.65, PA-13.16, BA-12.015.25,and VA-2.99 AA-1.20,PA-1.75, BA-1.79,and VA-4.42

Notreported 23

Leachateof fermented organicwastes

PTFE 3.03,24hand25◦C 0.5NNaOH AA-6,PA-6,BA-2, VA-2,andCA-1 AA-0.64,PA-0.75, BA-1.4,VA-0.64, andCA-0.86 AA-12.93,PA-6.77, BA-12.94, VA-3.57,andCA-2.03 AA-1.23,PA-0.87, BA-0.93,VA-1.61, andCA-1.37 Notreported 23

Syntheticsolution PTFE 2.9,7hand38◦_C _0.5_N_NaOH _AA-_1.25,_PA-_1.25, BA-1.25,VA-1.25,and CA-1.25 AA-2.33,PA-3.33, BA-3.88,VA-4.44, andCA-5.83 ∼*AA-8,PA-11, BA-12.5,VA-13, andCA-14.5 ∼*AA-3.1, PA-6.51,BA-6.51, VA-15.05,andCA-21.5

AA-54,PA-64, BA-69,VA-72,and CA-74

20

Syntheticsolution PTFE 2.9,7hand38◦_C _0.5_N_NaOH _AA-_1.25,_PA-_1.25, BA-1.25,VA-1.25,and CA-1.25

AA-0.61,PA-2.55, BA-7.5,VA-16.4, andCA-19.7

∼*AA-3,PA-9, BA-17.5,VA-23,and CA-24

∼*AA-0.75,PA-4, BA-60,VA-1300, andCA-1350

AA-34,PA-85, BA-98,VA-99,and CA-99

20

Landﬁllleachate PTFE-TOA 4,7hand38◦_C _0.5_N_NaOH _AA-_4.3,_PA-_0.48, BA-7.1,VA-0.14,and CA-0.756

Notreported Notreported Notreported AA->45,PA->86, BA->86,VA->86, andCA->86

20

Chickenmanure digestate

PTFE-TOA 4,7hand38◦_C _0.5_N_NaOH _AA-_2.66,_PA-_1.98, VA-0.189,andCA-4.53

Notreported Notreported Notreported *AA->45,PA-78, VA-30,andCA-8

20 Fermentationbroth PTFE-TOA 4,7hand38◦C 0.5NNaOH AA-4.2,PA-1.0,

BA-7.8,VA-0.13,and CA-0.89

Notreported Notreported Notreported AA->45,PA->95, BA->95,VA->95, andCA->95

20

Slaughterhouse blood

Polypropylene Acidiﬁed,and6h TOA+1 -Octanol

Notreported AA-0.058, PA-0.170,BA-0.291, VA-0.00,and CA-0.145

Notreported Notreported AA-42,PA-80, BA-94,andVA-100

34

Slaughterhouse blood

Polypropylene Un-acidiﬁed,and2 h

TOA+1 -Octanol

Notreported AA-0.099, PA-0.121,BA-0.091, VA-0.027,and CA-0.091

Notreported Notreported TotalVFArecovery <5

34

Fishfermentate Siliconemembrane 3,200hand25◦_C _Water _**AA-_11.4,_PA-_3.5, BA-12.8,VA-0.4,and CA-0.3

AA-0.0,PA-0.114, BA-0.157, VA-0.209,and CA-0.144

Notreported Notreported Notreported 22

Fishfermentate Siliconemembrane 6.6,200hand25◦_C _Water _**AA-_11.8,_PA-_3.6, BA-14,VA-0.5,andCA-0.5

AA-0.00,PA-0.00, BA-0.00,VA-0.00, andCA-0.00

Syntheticsolution Siliconemembrane 2.5,200hand25◦_C _Water _AA-_6,_PA-_6,_BA-_2, VA-2,andCA-1

AA-0.0017, PA-0.0075,BA-0.016, VA-0.053,and CA-0.199

Syntheticsolution Siliconemembrane 3&5,70h,20,30& 40◦_C

Water AA-5,PA-5,BA-5, VA-5,andCA-5

ReferTable3 ReferTable2 ReferTable5 ReferFig.3 Thisstudy

Cheesewhey

fermentate

Siliconemembrane 3&5,70h,20,30& 40◦_C

Water AA-4,andBA-4.6 ReferTable3 ReferTable2 ReferTable5 ReferFig.5 Thisstudy

Note:d_-_days;_h-_hours;_AA-_Acetic_acid,_PA-_Propionic_acid,_BA-_Butyric_acid,_VA-_Valeric_acid_and_CA-_Caproic_acid;_*Approximate_values_obtained_from_bar_graphs_or_ratios_in_the_manuscript;_**_Averaged_from_two_ﬁsh_fermentates usedinthestudy.

(12)

H.Ravishankaretal. ProcessSafetyandEnvironmentalProtection148(2021)125–136 concurrenthydrogenproductionthroughfermentationandbutyric

acidextractionfromfermentatedemonstratedgoodseparationof butyricacidwithhighpurity(over90%oncarboncontentbasis) withoutaffectingthehydrogenproductionrates(Dessìetal.,2020). However,inviewofachievingasustainableextractionprocess, fur-therresearchisstillrequiredtoadvancethetechnologyreadiness levelofthesiliconemembranecontactor.Couplingthecontactor withamembranedistillationunitcanseparatetheindividualVFAs fromthemixtureatdifferenttemperatureswiththepotentialto obtainpureVFAs while maintainingtheconcentrationgradient forextraction(Aktijetal.,2020).Theselectivityofmembranefor VFAs/alcoholscanalsobeimprovedthroughmembrane modifi-cation (fillingextractantsin themembrane pores)(Aydinetal., 2018),therebyimprovingtheoverallrecoveryefficiencyfromthe fermentate.

4. Conclusion

VFAsand alcoholrecoveryfromconcentratedsynthetic solu-tionsandcheesewheyfermentatethroughasiliconemembrane contactorusingwaterastheextractantwasdemonstrated.A max-imum of 45 %and 41.5 %recoveryof valeric and caproicacid, respectively,aswellas42%recoveryofbutanol,wasachievedfrom syntheticsolutions.Maximumrecoveryofcaproicacidoccurred within 20 hof experimentaloperation. Acetic and butyricacid extractionfromcheesewheyfermentate(pH3;40◦C)wasachieved with7%and21.5%recoveryefficiency,respectively,showcasing thefeasibilityofthesiliconemembranecontactorforconcentration drivenVFAseparationfromcheesewheyfermentate.The separa-tionfactorvaluesindicatelongercarbonchainVFAs/alcoholshad betterselectivitythroughsiliconemembrane.Furtherresearchon membrane modificationordownstreamprocessingcouplingthe membranecontactorwithbetterselectivityofVFAs/alcoholscan improvetheoverallrecoveryefficienciesandevolveintoamature technologyforin-lineseparation ofVFAs/alcoholsfromreal fer-mentate.

DeclarationofCompetingInterest

Theauthorsdeclarethattheyhavenoknowncompeting ﬁnan-cialinterestsorpersonalrelationshipsthatcouldhaveappearedto inﬂuencetheworkreportedinthispaper.

Acknowledgements

TheauthorsthankScienceFoundationIreland(SFI)for support-ing theresearchworkthrough theirSFIResearchProfessorship ProgrammeentitledIETSBIO3 _(Innovative_Energy_Technologies_for

Biofuels,BioenergyandaSustainableIrishBioeconomy)[grant num-ber:15/RP/2763]andtheResearchInfrastructuregrantPlatformfor BiofuelAnalysis[grantnumber16/RI/3401].

AppendixA. Supplementarydata

Supplementarymaterialrelatedtothisarticlecanbefound,in theonlineversion,atdoi:https://doi.org/10.1016/j.psep.2020.09. 052.

References

Aktij,S.A.,Zirehpour,A.,Mollahosseini,A.,Taherzadeh,M.J.,Tiraferri,A.,Rahimpour, A.,2020.J.Ind.Eng.Chem.81,24–40.

Alkaya,E.,Kaptan,S.,Ozkan,L.,Uludag-Demirer,S.,Demirer,G.N.,2009. Chemo-sphere77,1137–1142.

Atasoy,M.,Owusu-Agyeman,I.,Plaza,E.,Cetecioglu,Z.,2018.Bioresour.Technol. 268,773–786.

Aydin,S.,Yesil,H.,Tugtas,A.E.,2018.Bioresour.Technol.250,548–555. Baker,R.W.,2012.MembraneTechnologyandApplications.JohnWiley&Sons. Bandini,S.,Gostoli,C.,1992.J.Membrane.Sci.70,119–127.

Bellman,K.L.,2012.TheOhioStateUniversity.

Bertin,L.,Martinez,G.,Domingos,J.M.B.,Rebecchi,S.,Fava,F.,2016.NewBioeth., S18.

Cabrera-Rodríguez,C.I.,Moreno-González,M.,deWeerd,F.A.,Viswanathan,V.,van derWielen,L.A.,Straathof,A.J.,2017.Bioresour.Technol.237,186–192. Calero,R.R.,Lagoa-Costa,B.,Fernandez-Feal,M.Md.C.,Kennes,C.,Veiga,M.C.,2018.

J.Chem.Technol.Biotechnol.93,1742–1747.

Chandrapala,J.,Duke,M.C.,Gray,S.R.,Zisu,B.,Weeks,M.,Palmer,M.,Vasiljevic,T., 2015.J.DairySci.98,4352–4363.

Dessì,P.,Asunis,F.,Ravishankar,H.,Cocco,F.G.,DeGioannis,G.,Muntoni,A.,Lens, P.N.,2020.Int.J.Hydrog.

Han,S.,Ferreira,F.C.,Livingston,A.,2001.J.Membrane.Sci.188,219–233. Jones,R.J.,Massanet-Nicolau,J.,Guwy,A.,Premier,G.C.,Dinsdale,R.M.,Reilly,M.,

2015.Bioresour.Technol.189,279–284.

Khor,W.C.,Andersen,S.,Vervaeren,H.,Rabaey,K.,2017.Biotechnol.Biofuels10, 180.

Li,Z.,Qin,W.,Dai,Y.,Chem,J.,2002.Eng.Data.47,843–848. Li,S.Y.,Srivastava,R.,Parnas,R.S.,2011.Biotechnol.Prog.27,111–120.

Luis,P.,2018.FundamentalModelingofMembraneSystems:MembraneandProcess Performance.Elsevier.

Maina,S.,Kachrimanidou,V.,Koutinas,A.,2017.Curr.Opin.GreenSustain.Chem.8, 18–23.

Mulder,M.,2012.BasicPrinciplesofMembraneTechnology.SpringerScience& BusinessMedia.

Nzeteu,C.O.,Trego,A.C.,Abram,F.,O’Flaherty,V.,2018.Biotechnol.Biofuels11,108. Outram,V.,Zhang,Y.,2018.Bioresour.Technol.270,400–408.

Plácido,J.,Zhang,Y.,2018.Waste.Biomass.Valori.9,1767–1777. Playne,M.,Smith,B.,1983.Biotechnol.Bioeng.25,1251–1265. Qin,Y.,Sheth,J.,Sirkar,K.,2003.Ind.Eng.Chem.Res.42,582–595.

Rebecchi,S.,Pinelli,D.,Bertin,L.,Zama,F.,Fava,F.,Frascari,D.,2016.Chem.Eng.J. 306,629–639.

Reyhanitash,E.,Zaalberg,B.,Kersten,S.R.,Schuur,B.,2016.Sep.Purif.Technol.161, 61–68.

Reyhanitash,E.,Kersten,S.R.,Schuur,B.,2017.ACSSustain.Chem.Eng.5,9176–9184. Scarlat,N.,Dallemand,J.-F.,Monforti-Ferrario,F.,Nita,V.,2015.Environ.Dev.15,

3–34.

Stephenson,R.,Stuart,J.,1986.J.Chem.Eng.Data31,56–70. Thongsukmak,A.,Sirkar,K.,2007.J.Membrane.Sci.302,45–58.

Trad,Z.,Akimbomi,J.,Vial,C.,Larroche,C.,Taherzadeh,M.J.,Fontaine,J.-P.,2015. Bioresour.Technol.196,290–300.

Tugtas,A.E.,2014.WasteManage.34,1171–1178.

Xiong,B.,Richard,T.L.,Kumar,M.,2015.J.Membrane.Sci.489,275–283. Yesil,H.,Tugtas,A.,Bayrakdar,A.,Calli,B.,2014.WaterSci.Technol.69,2132–2138. Yuan,Y.,Hu,X.,Chen,H.,Zhou,Y.,Zhou,Y.,Wang,D.,2019.Sci.TotalEnviron.,

133741.

Zacharof,M.-P.,Lovitt,R.,2014.WaterSci.Technol.69,495–503. Zhang,Y.,Angelidaki,I.,2015.WaterRes.81,188–195. Zhou,F.,Wang,C.,Wei,J.,2013.J.Membrane.Sci.429,243–251.