byi

 

 

 

 

 

Scientif

Prof.AmDr.AlesDr.FabiProf.Fr

Rem

nnov

C

Di

deiM

Unive

ficComm

medeoLssandroioMontarancisco

mova

vativ

Chemic

ipartim

Materia

ersitàd

mittee

anciaErtoagnaroRodrígu

PhD

alofC

vefun

calEng

mentod

liedel

degliSt

uez‐Rein

Dissert

CO2f

nctio 

 

 

 

 

gineeri

diIngeg

llaProd

tudidi

noso

tation

from

onaliz

ng–Cyc

gneria

duzione

Napol

flue‐

zeds

cleXXV

Chimic

eIndus

iFeder

‐gas

sorbe

V

ca,

striale

ricoII

Thesisb

MarcoB

ents

by

Balsamo

o

2

Tableofcontents

 

CHAPTER1INTRODUCTION 4

CHAPTER2LITERATUREOVERVIEW 8

2.1UtilizationofactivatedcarbonsinCCSfield 8

2.1.1CO2adsorptionontoas‐synthesizedactivatedcarbons 8

2.1.2CO2adsorptiononfunctionalizedactivatedcarbons 10

2.2OtherclassesofCO2sorbentsinvestigated 12

2.2.1Zeolites 12

2.2.2Orderedmesoporoussilicas(OMS) 12

2.2.3Calciumoxide 13

2.2.4Metal‐organicFrameworks(MOFs) 14

2.3Ionicliquids(ILs) 15

2.3.1ILsapplicationsforCO2capture 16

2.3.2Functionalizationofsolidswithionicliquids(ILs) 17

CHAPTER3THEORETHICALFRAMEWORKOFTHEADSORPTIONPROCESS 19

3.1Adsorptionequilibria 19

3.2Dynamicsofadsorptioncolumns 21

CHAPTER4MATERIALSANDMETHODS 28

4.1Activatedcarbons,ionicliquidsandimpregnationprotocols 28

4.2Solidscharacterizationtechniques 29

4.2.1CO2/N2porosimetricanalyses 29

4.2.2Thermogravimetricanalyses(TGA) 30

4.3Lab‐scaleplantforCO2adsorption/regenerationtests 31

4.4Experimentalprotocolsforfixedbeddynamicexperiments 33

4.4.1ContinuousCO2adsorptiontests 33

4.4.2Adsorption/desorptioncyclesandregenerationtests 35

CHAPTER5RESULTSANDDISCUSSION 37

5.1Adsorbentscharacterization 37

5.1.1PorosimetricanalysesforF600‐900rawandimpregnatedwith[Hmim][BF4]/[Emim][Gly] 37

5.1.2TGAanalysesforF600‐900rawandimpregnatedwith[Hmim][BF4]/[Emim][Gly] 41

5.1.3PorosimetricanalysesforN.RGC30rawandimpregnatedwith[Hmim][BF4]/[Emim][Gly] 43

5.1.4TGAanalysesforN.RGC30rawandimpregnatedwith[Hmim][BF4]/[Emim][Gly] 46

3

5.1.5ComparisonbetweenTGAandporosimetricanalysesfortheinvestigatedsorbents 48

5.2CO2captureperformancesontheinvestigatedsorbents 51

5.2.1CO2adsorptiontestsontorawF600‐900andN.RGC30 51

5.2.2CO2adsorptiontestsontoF600‐900rawandfunctionalizedwith[Hmim][BF4]/[Emim][Gly] 56

5.2.3CO2adsorptiontestsontoN.RGC30rawandfunctionalizedwith[Hmim][BF4]/[Emim][Gly] 63

5.2.4IntertwiningamongsolidspropertiesandCO2captureperformances 69

5.3Adsorption/desorptioncyclesandregenerationexperimentsforF600‐900raw 73

5.4Adsorptionthermodynamicsandkineticsmodelling 79

5.4.1Thermodynamicaspects 79

5.4.2Kineticaspects 86

CHAPTER6CONCLUSIONSANDFUTUREPERSPECTIVES 92

Bibliography 95

 

 

 

 

 

 

 

 

 

 

 

 

 

 

4

CHAPTER1

INTRODUCTION

TheincreaseinworldwideCO2emissions,mainlyderivingfromfossilfuels, iscommonly

believedtobeamongthemaincontributorstoglobalwarming(Metzetal.,2005;Figueroaet

al., 2008). Currently, 85% of the global energy demand is supplied by fossil‐fueled power

plants accounting for about 40% of total CO2 emissions (Yang et al., 2008). Nowadays,

different options are available to mitigate CO2 emissions deriving from power sector,

including higher power generation efficiency, use of non‐carbon fuels (hydrogen and

renewableenergy),developmentofnewenergyproductionsystems,suchasoxy‐combustion

andchemical‐loopingcombustion,andtheadoptionofefficienttechnologiesforCO2capture

andstorage(CCS)(Metzetal.,2005;LiandFan,2008).

The CCS approach has the potential to reduce overall mitigation costs and increase

flexibilityinachievingareductioningreenhousegas(GHG)emissions:accordingtotheBLUE

MapScenariooftheInternationalEnergyAgency(IEA),thisroutecouldcontributetoa19%

cut inCO2 emissionsby2050 (International EnergyAgency (IEA), 2010).More specifically,

CCS technologies involve the separation and concentration of CO2 produced in large point

sources, the transportof thegas toasuitablestorage locationand long termisolation from

atmosphere (Metz et al., 2005).Main CO2 sequestration routes include geological injection,

oceandumpandmineralcarbonation(Metzetal.,2005).

Three main technological pathways can be pursued for CO2 capture from fossil‐fueled

power plants: post‐combustion capture, pre‐combustion capture and oxy‐combustion

(Figueroaetal.,2008;LiandFan,2008;Kannicheetal.,2010).Amongthem,post‐combustion

systemhas the greatest near‐termpotential for reducingGHG emissions, because it canbe

retrofittedtoexistingunitsthusprovidingaquickersolutiontomitigateCO2environmental

impacts(Figueroaetal.,2008;Leeetal.,2012). Themainbarrier to the implementationof

thistechnologyonindustrialscaleisrelatedtothelowthermodynamicdrivingforceforCO2

capturefromflue‐gas(gaspartialpressureisusuallylessthan0.15bar).Moreover,applying

current state‐of‐the‐art CO2 separation processes (absorption, adsorption, membrane

purification and cryogenic distillation) to existing coal‐fired power plantwould reduce the

power generation capacity by roughly one‐third (Figueroa et al., 2008). Post‐combustion

chemicalabsorptionofCO2inaqueousaminesolutions(mainlymonoethanolamine,MEA)is

the most widely used purification technology (Strube and Manfrida, 2011; Brúder and

5

Svendsen, 2012). The MEA process suffers many drawbacks related to the considerable

amounts of thermal energy required for absorbent regeneration, the high equipment

corrosion rate causedby contactwithMEAsolutionand the solventdegradationcausedby

oxygenandoxygen‐basedcompoundssuchasSO2andNOxpresentinatypicalflue‐gas(Kittel

et al., 2009; Strube and Manfrida, 2011). As a consequence of the aforementioned issues,

several research groups are making great efforts to develop high‐performance and cost‐

effective CO2 advanced separation processes in order to accelerate the techno‐economic

feasibilityofpost‐combustioncapturesystems.

Inthisscenario,adsorptionseemstobeaverypromisingtechnology,widelyusedforthe

treatment of gaseous and liquid effluents due to its potentially high removal efficiency and

operating flexibility, general low maintenance costs and, if coupled with an effective

regenerationprocess,fortheabsenceofby‐products(Abanadesetal.,2004;Choietal.,2009;

Balsamoetal.,2010;Sayarietal.,2011;Sjostormetal.,2011;Samantaetal.,2012;Balsamoet

al., 2013). Many sorbents can be used on purpose either raw or functionalized. Activated

carbonsshowhighpotentialityforapplicationinCO2capturebecausetheyaregenerallyless

costlythanothermaterials(e.g.,orderedmesoporoussilicas,metalorganicframeworks,etc.)

and have a complex structure characterized by high surface area and tunable porosity and

surface properties (Marsh and Rodrίguez‐Reinoso, 2006; Whaby et al., 2010). In addition,

carbon‐basedsorbentsareeasilyregenerableallowingtheiruseinprocessessuchaspressure

swingadsorption(PSA),temperatureswingadsorption(TSA)andvacuumswingadsorption

(VSA) (Gomes and Yee, 2002; Tlili et al., 2009). Despite these advantages, CO2 removal

performancesand long‐termstabilityof activatedcarbonsunder typical flue‐gas conditions

(CO21‐15%byvol.andatmosphericpressure)havebeenpoorlyinvestigated.

Anotherwidespreadresearchlineinthecontextofpost‐combustionpurificationsystems

concerns the investigationof ionic liquids (ILs) as innovative solvents forCO2 capture.The

ever‐increasing interest for this class of compounds in different fields is justified by their

uniquecharacteristicssuchasextremelylowvaporpressure,highthermo‐chemicalstability

andtunablechemico‐physicalproperties(Zhangetal.,2006a;Bourbigouetal.,2010;Hasib‐

ur‐Rahmanetal.,2010).Inparticular,thepossibilityoffunctionalizingILswithbasicgroups

(like amines) makes them very attractive for CO2 capture processes (Zhang et al., 2011).

Numerous literature studies are now focusing on the use of ILs supported on porous

membranes inorder toovercomethemain limits in the industrial‐scaleapplicationof ionic

liquidsforCO2capture,whicharerelatedtotheirhighcostandviscosity(Hasib‐ur‐Rahmanet

al., 2010; Lemus et al., 2011; Kolding et al., 2012). Notwithstanding a huge number of

6

scientificpapersdealswiththeutilizationofdifferentclassesofILsinCCSfield,thefollowing

criticalaspectscanbehighlighted:

CO2absorptiontestsaregenerallycarriedoutathighpressureandroomtemperature

whichareexperimentalconditionsnotrepresentativeofarealflue‐gas;

There is fragmentary information concerning the effect of confining ILs into

nanoporous substrates, particularly activated carbons, on CO2 capture performances

withrespecttotheirbulksolventproperties.

On the basis of the above‐mentioned analysis, the aim of this work is to provide a

contributioninelucidatingCO2captureperformancesofionicliquidssupportedonactivated

carbons characterized by different porosimetric properties. Specific thermodynamic and

kineticadsorptiontestshavebeencarriedoutonselectedactivatedcarbons,bothasrawand

impregnatedwithILsatdifferentconcentrations.Experimentaltestshavebeenperformedin

alab‐scalereactorandunderrealisticoperatingconditions(e.g.typicalflue‐gascompositions

and temperatures). Preliminary regeneration studies have been conducted on the sorbent

which displayed the highest CO2 capture capacity in order to determine its performances

underconsecutiveadsorption‐desorptioncyclesandassesstheoptimaloperatingconditions

forCO2storageafterdesorption.Theintertwiningamongrawsolidsproperties‐impregnation

conditions‐properties of the functionalized materials‐solids capture capacity has been

investigated by comparing CO2 adsorption results with outcomes obtained from sorbents

CO2/N2 porosimetric and thermogravimetric analyses. Adsorption isotherms have been

interpreted in the lightof theoreticalmodels fora comprehensionof themainmechanisms

involved in the capture of CO2 by the investigated solids. Breakthrough data have been

modelledalsoinordertoidentifytherate‐determiningstepoftheadsorptionprocessandthe

effectofoperatingparametersonmasstransferphenomena.

ThisPhDDissertationisorganizedasfollows.InChapter2aliteraturesurveyisreported

in order to analyse themain classes of sorbents employed for CO2 capturewith particular

emphasis on activated carbons together with themain applications of ionic liquids in CCS

field.Chapter3providesthemaintheoreticalaspectsconcerningboththermodynamicsand

kinetics of the adsorption phenomenon. Chapter 4 describes the experimental protocols

adopted for activated carbons impregnationwith ionic liquids, solids characterizations and

adsorption/regenerationexperiments;adescriptionofthelab‐scaleplantdesigned,builtand

optimizedfortheexecutionofadsorptionexperimentsisalsoprovided.InChapter5themain

results obtained from adsorbents characterizations and CO2 capture/regeneration tests are

7

discussed together with the main aspects derived from thermodynamics and kinetics

modelling.Finally,conclusionsandfuturedevelopmentsarereportedinChapter6.

8

CHAPTER2

LITERATUREOVERVIEW

In this Chapter the main adsorbent materials employed in CCS field are analysed with

particularemphasison theapplicationsofactivatedcarbons forCO2capture,as theyareof

majorinterestforthisPhDproject.Themainapplicationsofionicliquidsinthiscontextare

alsodiscussed.

2.1 UtilizationofactivatedcarbonsinCCSfield

Activated carbons are carbonaceousmaterialswith a common structuremade up of an

assemblyofdefectivegraphenelayersthathavehighpotentialityforCO2capturethankstoa

complex structure characterized by micropores that determine high surface area for

adsorption,butalsomeso‐andmacroporeswhichcanfacilitatethediffusion(fastkinetics)of

theadsorbatetotheinnerporosity(MarshandRodrίguez‐Reinoso,2006;Whabyetal.,2010;

Sayarietal.,2011).ActivatedcarbonsactasphysisorbentstowardsCO2,thustheiradsorption

capacity decreases rapidly as temperature increases (Choi et al., 2009; Sayari et al., 2011).

Moreover, the mild adsorption strength in the low‐pressure regime (<0.5 bar) makes

activated carbons easily regenerable. In the following, main retrieved results for CO2

adsorption on raw activated carbonswill be discussed (Section 2.1.1);moreover, themain

activationtreatmentsaimedattheintroductionofhighlyCO2‐affinefunctionalgroupsonthe

carbonsurfacewillbeanalysed(Section2.1.2).

2.1.1 CO2adsorptionontoas‐synthesizedactivatedcarbons

TheuseofactivatedcarbonsforCO2captureisnowadaysconsideredaviableroutemainly

for storage purposes because they can be efficiently used in pure CO2 streams and at high

pressures(Sayarietal.,2011).Nevertheless,thesynthesisoftailoredmicroporousstructures

canextendtheirusealsoforseparationandpurificationfieldsbydiscriminatingmoleculeson

shapeand/orsizebasis(Whabyetal.,2010).

Table2.1reportsmainliteraturedataconcerningCO2equilibriumadsorptioncapacityeq

onrawactivatedcarbonsandinpureCO2streamsatdifferentpressuresandtemperatures;it

isunderlinedthatthesorbentCO2/N2selectivity(S)isevaluatedastheratiooftheadsorbed

amountsofthetwogasesobtainedinsinglecompoundequilibriumtests.

9

Table2.1CO2equilibriumadsorptioncapacityinpurestreamsonas‐synthesizedactivatedcarbonsatdifferentpressuresandtemperatures

SorbentT[°C]

P[bar]

eq[mmolg‐1]

S(CO2/N2)

Reference

F30/470† 24† 0.16† 0.65† n.a.† (BerlierandFrère,1996)

Ajax† 25† 0.2† 0.75† n.a.† (DoandWang,1998)

Salnchunri†25†

55†

0.1†

0.1†

0.60†

0.25†

1.2†

1.3†(Naetal.,2001)

Filtrasorb400† 25† 0.1† 0.57† n.a.† (Luetal.,2008)

pitch‐basedactivated†carbon

30†

90†

1†

0.2†

1.9†

0.1†

5†

4†(Shenetal.,2010)

pitch‐basedVR‐5‐M†molecularsieve

0†

25†

50†

1†

1†

1†

8.6†

4.2†

2.3†

2.8†

n.a.†

n.a.†

(Whabyetal.,2010)

MaxsorbIII†30†

50†

2.86†

2.57†

5.4†

3.4†

n.a.†

n.a.†(Sahaetal.,2011)

†notavailable

As expected, it can be evidenced a decrease of CO2 adsorption capacity at higher

temperatures. Whaby et al. (2010) compared zeolites 13X and 5A with carbon molecular

sieves and observed that the latter show higher CO2 adsorption capacity at 1 bar and 0°C.

Theyalsoinferredthatthepresenceofnarrowmicropores(diameter<0.7nm)playsamajor

role indeterminingthesolidadsorptioncapacity.Noteworthy,Silvestre‐Alberoetal. (2011)

showedthatCO2adsorptiononcarbonmolecularsievemonolithsishighlyreversible,withno

loss of adsorption capacity under three consecutive adsorption/desorption cycles, making

themexcellentcandidatesforpressureswingadsorptionunits.

AninterestingexperimentalcampaignconcerningCO2captureinCO2/H2/N2(20/70/10%

by vol.) mixtures on raw activated carbon was carried out by García et al. (2011). In

particular,theystudiedtheremovalofCO2inafixedbedapparatusandanalysedtheeffectof

the temperatureandCO2partialpressureson the systemdynamicperformances.Themain

resultsshowedthatateachtemperatureandfixedcarbondioxideconcentration,higherCO2

partial pressures (obtained by increasing the system total gas pressure) determine longer

breakpointtime,andthisbehaviourwasimputedtoaslowerconcentrationfrontinthebed;

moreover, at higher temperatures the process was faster but a parallel reduction in

adsorptioncapacitywasobserved.Finally,Shenetal.(2011)studiedtherecoveryofCO2from

saturatedpitch‐basedactivatedcarbonbymeansofVacuumPressureSwingAdsorptionafter

fixed‐bedadsorptioninaCO2/N2mixture(15/85%byvol.):resultsshowedthatforaN2feed

pressureof2bar,aCO2purityof94%and78%recoverycouldbeobtained.

2.1.2

Them

are the im

amination

Impre

The d

activated

CO2 beca

carbonate

Choietal.,

Aroua

ofmolecu

showedth

beobtaine

increase in

andmicro

chemisorp

increases

molecules

usedmicr

amino‐2‐m

temperatu

CO2adso

main functio

mpregnatio

nwithgaseo

egnationw

dispersion

carbonsis

ause amin

e/bicarbona

,2009).

Figure

etal.(200

ularbasket

hatanincre

edatanop

nthesolid

oporesderi

ption effec

in the ch

s.Recently,

roporousac

methyl‐1‐pr

ure and atm

orptionon

onalization

on with am

ousammon

withamine

of compo

averyattr

nes are a

ate (in hum

2.1Reaction

8)usedpa

via impreg

easeofpur

ptimal0.26

dcapturep

ivingfrom

ct induced

harged‐amo

,an interes

ctivatedca

ropanol (A

mospheric

nfunction

treatment

mine‐based

nia.Thema

‐basedcom

unds cont

ractiveopt

able to f

mid condit

nschemeso

lmshell‐ba

gnationwi

eCO2adso

6%wt.PEI

performanc

polymerco

by nitrog

ount of PE

stingstudy

rbonbeds

AMP) for tr

pressure).

nalizedac

tsofactiva

d compou

ainresults

mpounds

taining am

tiontoincr

form eith

tion), as ev

ofaminegrou

asedactiva

ithpolyeth

orptioncap

loading.A

ceswasasc

onstriction

gen group

EI, likely c

yhasbeen

impregnat

reatinga 1

. The autho

ctivatedca

atedcarbon

nds or alk

arereporte

mine functi

reasetheir

er carbam

videnced in

upswithCO2

atedcarbon

yleneimine

acitywith

Asamatter

cribedtoth

noflargerp

ps present

constricted

carriedou

tedwithm

5%byvol

ors observ

arbons

ns investiga

kali/alkalin

edinthefo

ionalities o

capturepe

mate (in

n Figure 2.

2(Grayetal.

nasprecurs

e(PEI).Ver

respectto

offact,at

hecreation

poresasw

in the ac

the pore

utbyKhali

onoethano

l. CO2 gase

ved an incr

ated in the

ne earth m

ollowing.

over the

erformance

dry cond

.1 (Gray et

 

.,2005)

sorinthep

ry interest

thevirgins

lowPEIlo

nofadditio

wellastoan

ctive phas

accessibil

il etal. (20

olamine(M

ous stream

rease of CO

10

e literature

metals and

surface of

estowards

dition) or

t al., 2005;

production

tingly, they

solidcould

oadingsthe

onalmeso‐

nenhanced

se; further

ity to CO2

012)which

EA)and2‐

m (at room

O2 removal

0

e

d

f

s

r

;

n

y

d

e

‐

d

r

2

h

‐

m

l

11

capacityforbothfunctionalizedsorbents(1.11and0.70mmolg‐1forMEA‐andAMP‐carbon,

respectively)with respect to the rawmaterial (0.41mmol g‐1); the higher performances of

MEA‐carbon was ascribed to a more efficient dispersion of less sterically hindered MEA

moleculesoverthesupportsurfacewithrespecttoAMP,thuscreatingmoreaccessiblesites

forCO2capture.

Impregnationwithalkali/alkalineearthmetals

Impregnationofactivatedcarbonswithcalciumandmagnesiumoxidewasinvestigatedby

Yongetal. (2001), since thesemetalshaveahighbasicnaturewhich favors the interaction

withCO2acidicmolecule.Inparticular,theyobservedthatatlowtemperature(28°C)theraw

materials exhibited higher pure‐CO2 adsorption capacity with respect to metal‐doped

materials: thiswasascribedtothereductioninsurfaceareaoccurringduringtheactivation

process.Conversely,athighertemperature(300°C)theactivatedmaterialsshowedimproved

performancesbecauseoftheprevailingofchemisorptioneffectsoverphysisorption,thelatter

beingsurfacedependentanddominantatlowtemperatures.

Amination

Amination is a treatment usually referred to the reaction of gaseous ammoniawith the

surfaceof activatedcarbons,performedathigh temperatures (ranging from400 to900°C),

aimedatincreasingthesolidnitrogencontent(Plazaetal.,2009;Shafeeyanetal.,2010;Plaza

etal.,2011).Ammoniacanreactwithsurfaceoxidesandactivesitespresentattheedgesof

thegraphenelayerstoformamines,amides,imides,lactams,nitriles,pyridine‐orpyrrole‐like

functionalities. As an example of amination effect, Plaza et al. (2009) observed an

enhancement of pureCO2 adsorption capacities for almond shell‐derived activated carbons

aminated at temperatures greater than600°Cwith respect to theparent carbon;moreover

the authors observed that in the range 400‐900°C the sample aminated at 800°C had the

greatest CO2 adsorption capacity due to a maximum nitrogen content, as confirmed by

ultimateanalyses.Plazaetal.(2011)investigatedtheuseofaminatedbiomass‐basedcarbon

to capture CO2 in a 17% by vol. gaseous stream (balance N2) at 20°C and atmospheric

pressure;inparticular,theyfocusedonthesolidregenerabilityshowingthatThermalSwing

Adsorption carried out at 100°C allows an easy recovery of thepollutant and that after40

cyclestheadsorbentdidnotdisplayrelevantdeactivation.

12

2.2 OtherclassesofCO2sorbentsinvestigated

2.2.1 Zeolites

ZeolitesaremicroporouscrystallinealuminosilicatesbuiltofaperiodicarrayofSiO4and

AlO4tetrahedra(Ruthven,1984).Theiruniformporesizegrantsauniqueabilityasmolecular

sieves. CO2 is captured by zeolites mainly via electrostatic interactions generated by the

exchangeablecationsintheporesandbyhydrogenbondswithsurfacesilanolgroups(Choiet

al., 2009). In particular, zeolites characterized by a low Si/Al ratio have a high content of

extra‐framework cationswhich favourably interactwith CO2molecule (Sayari et al., 2011).

Generally,CO2adsorptiononzeolitesisnegativelyaffectedbyatemperatureincrease(Sayari

etal.,2011).Tlilietal. (2009)observedasix‐timesreductioninCO2adsorptioncapacityon

5Azeolitebyvarying theoperating temperature from25 to200°C. Inaddition, thesesolids

showlowerCO2adsorptioncapacityunderhumidconditions.RegeandYang(2001)showed

bymeans of FTIR analyses that there is a competition betweenwater vapour and CO2 for

adsorptionsitesonNaXzeolitesurface.Ithasbeenhighlightedthatzeoliticadsorbentshavea

strongerphysicalinteractionwithCO2andhigherheatsofadsorptioncomparedtoactivated

carbons, thusrenderingthedesorptionprocessmoreenergyintensive(Whabyetal.,2010).

Moreover, the hydrophobic nature ofmost activated carbonsmakes them less sensitive to

competitiveadsorptioneffectsbetweenCO2andwatervapourwithrespecttozeolites(Choi

etal.,2009).Ingeneral,itisunderlinedthatequilibriumadsorptionexperimentsonzeolites

arecarriedoutinpureCO2inmostofthecasesinvestigatedintheliterature:typicalreported

adsorptioncapacityvariesbetween0.2‐1.6mmolg‐1inthepressurerange0.1‐0.4barandata

temperatureof60°C(Sayarietal.,2011).Finally,CO2adsorptionkineticsonzeolitescanbe

rankedamongthefastestknown,reachinganequilibriumconditionwithinafewminutesin

mostcases(Choietal.,2009).

2.2.2 Orderedmesoporoussilicas(OMS)

Orderedmesoporous silicas (e.g.,MCM‐41, SBA‐15, TUD‐1, HMM‐33 and FSM‐16) are a

classof silicamaterials characterizedbydifferent cagestructures (suchashexagonal, cubic

andlamellar)thathaveattractedattentionincatalysisandseparationduetotheirextremely

highsurfaceareaandprecisetuningofporesizes(Chewetal.,2010).

Generally,fewstudiesareretrievableinthepertinentliteratureforCO2adsorptiononas‐

synthesized ordered mesoporous silicas, but many concern the removal of carbon dioxide

ontoamine‐modifiedOMS (BelmabkhoutandSayari,2009; Jangetal., 2009;Devadaset al.,

13

2010).Asamatteroffact,puresilicasurfacescontainresidualhydroxylgroupsthatarenot

abletointeractstronglywithCO2(Chewetal.,2010).TypicalCO2adsorptioncapacityinflue‐

gas conditions (temperature 75°C, CO2 5‐10% by vol.) reported for PEI‐impregnated OMS

varies in the range 2.1‐3.8 mmol g‐1 (Sayari et al., 2011). Post synthesis grafting is a

functionalizationtechniquewidelyappliedforthemodificationoforderedmesoporoussilicas

andinvolvesareactionbetweensurfacehydroxylgroupsofOMSandthealkoxyligandsofan

aminosilane,determiningalayeroftetheredaminegroupsonthesupportsurface(Changet

al.,2009;Serna‐GuerreroandSayari,2010;Sayarietal.,2011).Theseadsorbentshaveaclear

advantageoveramine‐impregnatedsilicasas theydonotshowanyamines leaching(unless

conditions are strong enough to break covalent bonds), thus determining potentially less

problems of equipment corrosion, usually associated to liquid amines (Choi et al., 2009).

FinallygraftedOMShaveshowngreat stabilityunder thousandsCO2adsorption‐desorption

cycles(Sayarietal.,2011).

2.2.3 Calciumoxide

Calcium minerals are the most abundant in nature among alkaline earth metal oxides,

commonly found in the formof carbonates suchas limestoneordolomite.When treatedat

hightemperatures,calciumcarbonatesliberateCO2andgeneratecalciumoxides.

TheremovalofCO2fromfluegasbycalciumoxidecanbeaccomplishedintwosteps(see

eqs.(2.1)and(2.2)):

Carbonation:CaO(s)+CO2(g)→CaCO3(s),exothermic(2.1)

Calcination:CaCO3(s)→CaO(s)+CO2(g),endothermic(2.2)

afirstreactionoftheoxidewithCO2toformcalciumcarbonate,performedinacarbonatorat

temperatures in the range 650‐700°C, and a subsequent heating of the carbonate at

temperatures higher than those of the carbonation step (calcination) to regenerate the

calciumoxideandreleaseconcentratedCO2(AbanadesandAlvarez,2003;Choietal.,2009;

Blameyetal.,2010).

Hughesetal.(2005)exploredtheinsituCO2capture,at700°Candatmosphericpressure,

inadual fluidizedbedcombustionsystem.TheadsorptionkineticsofCO2oncalciumoxide

adsorbents is much slower than on physisorbents such as zeolites and activated carbons,

sometimesrequiringseveralhourstoachieveca.70%ofthetotaladsorptioncapacity(Choiet

al.,2009).Moreover,calciumoxide‐basedadsorbentssufferfromarapiddegradationofCO2

capturecapabilityduringtherepetitionofcarbonation/calcinationcycles:thisreductionhas

14

mainlybeenascribedtoporeblockingandadsorbentsintering(AbanadesandAlvarez,2003).

Finally, material loss due to attrition and fragmentation in fluidized bed systems together

withthesorbentdeactivationproducedbythesulphationreactionwithSO2(alwayspresent

in a typical flue‐gas) are important issues related to Ca‐based sorbents (Montagnaro et al.,

2010;Coppolaetal.,2012aand2012b).

2.2.4 Metal‐organicFrameworks(MOFs)

Anemergingnewclassofcrystallinesolidscalledmetal‐organicframeworks(MOFs)has

recentlybeen investigatedassorbents forCO2capture.Thesematerialsgenerallyconsistof

three‐dimensional organic–inorganic hybrid networks formed by multiple metal–ligand

bonds(Eddaoudietal.,2002;Choietal.,2009;AnandRosi,2010;Sahaetal.,2010;Sayariet

al.,2011).ThesesolidsarehighlyversatilebecausetheporespacesofMOFsaretuneableover

asubstantialrangebyusingligandswithdifferentmoleculardimensions(AnandRosi,2010):

with some of the larger ligands the materials even became mesoporous. MOFs have been

developedforuseasCO2physisorbentsorstoragematerials,byoptimizingtheporessizefor

the carbon dioxidemolecule. Even if they show good adsorption capacities towards CO2 at

highpressures(greaterthan10bar),ithasbeenhighlightedthatforlowpressurerange(of

practical interest for post‐combustion capture) MOFs exhibit unfavourable adsorption

isotherms(Sayarietal.,2011).Additionally,MOFsareusuallyunstable inhumidconditions

andhigh temperatures showing lowCO2 selectivitywith respect toN2 (Sayari et al., 2011).

Finally,theirperformancesovermultipleadsorptionanddesorptioncycleshavetobetested

(Choietal.,2009).

InadditiontoMOFs,newtypeofsolidsarecurrentlybeinginvestigatedasCO2adsorbents

but are still at early stage of development: poly(ionic liquid)s (see Section 2.3.1), zeolitic

imidazolateframeworks(ZIFs)andcarbonnanotubes(Choietal.,2009;Herzogetal.,2009).

Fromtheanalysisoftheaforementionedsolids(eitherassynthesizedorfunctionalized),

recently reviewedbyChoi et al. (2009), Sayari et al. (2011) andSamanta et al. (2012), the

following aspects canbehighlighted: i) zeolites andactivated carbons are characterizedby

very fast CO2 adsorption kinetics but their performances decrease at temperatures greater

than100°Candinthepresenceofmoisture(alwayspresent ina fluegas); ii)calciumoxide

provides high CO2 adsorption capacities but requires high temperatures for regeneration,

which determine structural changes with loss of activity during several

carbonation/calcinations cycles; iii) amine‐functionalized solids usually display an

15

enhancementoftheCO2capturecapacityinpresenceofmoisture(relatedtothepossibilityof

forming carbonate/bicarbonate), but their application at high temperatures is limited by

aminesdegradability.

Asafurtherremark,itshouldbenotedthatevenifgreateffortsarecurrentlydevotedto

thedevelopmentofhigh‐performancesorbents forCO2capture, theiruse in typical flue‐gas

conditionsandformanyadsorption/desorptioncyclesisstilllimited.Moreover,thereisalack

of exhaustive information concerning their dynamic performances in different reactor

configurations(e.g., fixedbed, fluidized‐bed,circulating‐bedetc.), thisbeingakeyaspectfor

thedesignofindustrial‐scalepost‐combustionpurificationsystems.

In this scenario, the use of ionic liquids for porous solids functionalization is an

interestingalthough limitedlyexploredresearch topic, inorder todevelophighlyCO2‐affine

sorbents.Inthefollowing,abriefoverviewonionicliquids,theirapplicationandperspectives

inCO2capturetechnologiesispresented.

2.3 Ionicliquids(ILs)

Ionicliquids(ILs)areorganicsaltscomposedentirelybyionswithmeltingpointusually

lower than 100°C; many ILs are liquids at room temperature and, for this reason, are

commonly referred as Room Temperature Ionic Liquids (RTILs) (Zhang et al., 2006a;

Bourbigou et al., 2010; Hasib‐ur‐Rahman et al., 2010). ILs are characterized by negligible

vapourpressureatroomtemperature,abroadtemperaturerangeofliquidstate(depending

on the anionic/cationic couple), excellent thermal and chemical stabilities: these unique

propertiesmake themas optimal candidates as solvents and catalysts (Zhang et al., 2006a;

Boschettietal.,2007;Bourbigouetal.,2010).Theirtypicalviscosityrangesfrom50to1000

cP at room temperature (Figueroa et al., 2008). Some typical ILs cations and anions are

reported in Figure 2.2: cations are usually organics such as imidazolium, pyridinium or

ammoniumwhileanionsincludehalidesandfluoro‐borate/phosphate/sulphonate(Hasib‐ur‐

Rahman,etal.,2010).

F

ILspro

specific p

Liquids(T

2.3.1

Use of

overcome

degradatio

itcouldbe

significant

ManyR

thefollow

solubility

benzene)

strongeri

interaction

of the pu

absorption

typically

absorption

2011).

Ionicli

withbasic

Figure2.2Ty

opertiesca

roperties:

TSILs)(Zha

ILsappli

f ILs for C

the prob

onandequ

epossiblet

tlythecapt

RTILshave

wingaspects

inRTILs is

atthesam

nfluenceo

nbetween

ublished re

n capacity

0.05 mol

ncapacity

iquidabso

cgroups (l

ypicalcation

anbeadequ

this class

ngetal.,20

icationsfo

O2 capture

blems asso

uipmentcor

torecover

turecosts(

ebeenstud

sarehere

sgreater t

mepressure

nCO2solu

theanion

esearch pap

in imidaz

/mol .

stronglyd

rptionprop

like amines

nsandanion

uatelytune

of ionic li

006a;Bour

orCO2cap

e is gainin

ociated wi

rrosion(Zh

CO2fromf

(Figueroae

diedforCO2

highlighted

than thato

eandtemp

bilitythan

andtheac

pers repor

zolium‐base

Moreover

decreasesw

pertiestow

s) toobtain

nsconstitutin

edbyintrod

iquids is c

rbigouetal

pture

g interest

ith traditio

hang,etal.

fluegaswi

etal.,2008)

2absorptio

d(Zhanget

observed in

peratureco

thatofthe

idicCO2m

rts CO2 so

ed RTILs

r, RTILs p

with tempe

wardsCO2

n task‐spe

ngILs(Hasib

ducingpro

ommonly

.,2010).

due to th

onally‐emp

2006a;Bou

thouthavin

).

n(inpartic

tal.,2006a

ncommon

onditions;i

ecation,be

olecule.It

lubility in

at lower p

physically

erature (So

canbeenh

cific ionic

b‐ur‐Rahman

perfunctio

referred as

eir unique

ployed liqu

urbigoueta

ngtocooli

cularimida

a;Bourbigo

organic so

i)thenatu

ecauseofw

shouldbeu

RTILs at

pressures

interact w

orianoet al

hancedbyf

liquid (Zha

netal.,2010

onalgroup

s Task‐Spe

e propertie

uid amine

al.,2010):

itfirst,thu

azolium‐typ

ouetal.,20

olvents (e.g

ureofthea

weakLewis

underlined

high pres

(2 bar) an

with CO2,

l., 2009;Zh

functionali

anget al.,

16

0)

stoobtain

ecific Ionic

es that can

s such as

potentially

usreducing

peILs)and

010):i)CO2

g., heptane,

anionhasa

sacid‐base

dthatmost

sures; CO2

nd 25°C is

thus their

hanget al.,

zingtheIL

2006aand

6

n

c

n

s

y

g

d

2

,

a

e

t

2

s

r

,

L

d

2006b; Zh

([Pabim][B

ILisablet

foramine

Figu

Recent

showingt

than in ca

explored t

anionasC

andaneas

Anothe

synthesizi

2008); thi

imidazoliu

RTILs.Bar

asnitrile‐c

polymeric

2.3.2

Impreg

which exh

industrial

determine

coefficient

Phase(SIL

(smaller t

hang et a

BF4])conta

toformcar

solvents(F

ure2.3Propo

tly,Zhang

that inpres

ase of abse

the use of

CO2solvent

sythermal

er interest

ingsolidm

is class of

um‐basedp

raetal.(20

containing

cmembran

Function

gnation of

hibit high v

scale abso

ehigh liqui

ts (Hasib‐u

LP)materia

than the so

al., 2011).

ainingana

rbamatewi

Figure2.3)

osedreactio

etal.(2006

senceofsm

ence of wa

ionic liqui

t,showinga

regenerati

ing approa

membranes

f solids is t

poly(ionicl

008)showe

alkylgrou

es.

nalization

ionic liqu

viscosity (

orbers. Mo

id‐gas cont

ur‐Rahman

alsisalsoj

orbent por

Bates et

aminegrou

ithCO2follo

.

nmechanism

6b)synthe

mallamoun

ater due to

idswith am

absorption

ionofthes

ach related

usingpoly

termed po

liquid)sha

edthationi

ups)canen

nofsolids

ids onto s

mainly TSI

oreover, th

tact surfac

n et al., 20

ustifiedby

re volume)

al. (2002

upcovalent

owingare

mbetween[

esizedane

ntofwater

o bicarbon

mino‐funct

ncapacities

olvent.

d to ILs ap

ymerizable

oly(ionic liq

aveenhance

icliquidmo

nhancethe

withioni

olid suppo

ILs), thus

he IL dispe

eand, con

10). The g

ythepossib

), providin

2) were th

tlytethered

actionmec

[Pabim][BF4

wclassof

r theCO2a

nate forma

ionalized i

supto0.9m

plication in

ionicliquid

quid)s. Tan

edCO2ads

onomersm

separation

icliquids

orts is a su

determinin

ersion on a

sequently,

great inter

bilityofmi

ng importa

he first to

dtoanimi

chanismsim

]andCO2(B

amino‐acid

absorption

tion. Finall

midazolium

mol /mo

n CCS field

ds(Tanget

ng et al. (2

orptionca

modifiedwit

nselectivity

(ILs)

uitable tech

ng problem

a solid sup

greaterex

est for Sup

nimizingth

nt econom

o synthesiz

dazoliumc

milartotha

Batesetal.,2

dbasedion

capacityw

ly, Xue et

m cation a

ol at25°C

d is the po

tal.,2005;

2005) obs

apacitiesco

thpolargr

tytowards

hnique for

ms of appli

pport can

xternalma

pported Io

heamount

mic benefit

17

ze a TSIL

cation;this

atreported

2002)

nic liquids,

wasdouble

al. (2011)

nd taurine

Cand1bar

ssibility of

Baraetal.,

erved that

omparedto

oups(such

CO2ofthe

r those ILs

icability to

potentially

ss transfer

onic Liquid

tofILused

s for their

7

L

s

d

,

e

)

e

r

f

,

t

o

h

e

s

o

y

r

d

d

r

18

application(Lemusetal.,2011).Nevertheless,thebehaviourofILsconfinedinnanospacesin

thecontextofCO2captureprocessesisonlyatapreliminarystageofinvestigation(Tanakaet

al.,2011;Koldingetal.,2012).Forexample,Tanakaetal.(2011)observedthatthedispersion

of 1‐hexadecyl‐3‐methylimidazolium chloride ionic liquid into nanoporous silica

microspheres determines an enhancement of its CO2 capture performanceswith respect to

thebulksolventproperties.Theauthorsascribedthisbehaviourtotheformationofordered

moleculardomains,promotedbysilicasurface‐ILinteractions,inwhichCO2occupiesspecific

positions.Zhanget al. (2006b)observed that forCO2adsorptionbyamino‐acidbased ionic

liquids supported on porous silica gel, the equilibrium is reached faster than bubbling CO2

through bulk ILs. In addition, ILs can be supported on porous alumina membranes or

adsorbedonpolymericmembranestoincreasetheirselectivitywithrespecttoCO2(Hasib‐ur‐

Rahman et al., 2010). Finally, immobilization of imidazolium‐type ionic liquids onto silica

supports is gaining great interest also because ILs act as high‐activity catalysts in the

cycloaddition reaction between epoxides and CO2 to produce five‐membered cyclic

carbonates (Shim et al., 2009; Udayakumar et al., 2009 and 2010). Cyclic carbonates are

excellent aprotic polar solvents and intermediates commonly applied in the production of

pharmaceuticalsandfinechemicals.

 

 

 

 

 

 

 

19

CHAPTER3

THEORETHICALFRAMEWORKOFTHEADSORPTIONPROCESS

InthisChapterthemaintheoreticalaspectsconcerningboththermodynamicsandkinetics

oftheadsorptionphenomenonwillbediscussedinordertoprovideausefulbasisforadeep

comprehensionof themainmechanisms involved in the captureof CO2by the investigated

solids.

3.1 Adsorptionequilibria

Thethermodynamicstudyofadsorptionprocessesallowsobtainingadsorptionisotherms,

i.e.experimentalcurveswhichestablisharelationshipbetweenthesolidspecificadsorption

capacitytowardatargetpollutantandthecontaminantpartialpressureinthegasphaseata

fixed temperature and under equilibrium conditions. The evaluation of equilibrium

adsorption capacity for a given gas‐solid system is of great importancenot only because it

providesinformationabouttheamountofpollutantthatcanbeloadedonthesorbentandthe

natureof interactionsgoverning theprocess: thesystem thermodynamicbehaviour, in fact,

affects thedynamicperformancesofanadsorptionprocessandconsequently thesizeofan

adsorberunit.Theinterpretationofexperimentalequilibriumadsorptiondatabymeansofan

adequatetheoreticalmodelallowstodefinethesolidaffinitytowardaspecificpollutant(by

evaluating the interaction energy between the sorbent and each gaseous species) and the

capturemechanisminvolvedintheprocess.

Asimplemodelusedintheliteraturefortheinterpretationofequilibriumadsorptiondata

isLangmuirmodel,which isbasedonthe followingassumptions: i)eachsitecanguestone

adsorbatemolecule(monolayer); ii)thereisnomobilityofadsorbedspeciesonthesurface;

iii) theheatof adsorption is constantwith loading; iv) all sites are energetically equivalent

(Ruthven,1984).ThegeneralformoftheLangmuirisothermcanbeexpressedas:

ω (3.1)

whereeq[mmolg‐1]andPeq[bar]aretheadsorbentspecificmolaradsorptioncapacityand

theequilibriumgaspartialpressureoftheadsorbaterespectively;KL[bar‐1]andmax[mmol

g‐1] represent the Langmuir equilibrium constant and themaximumadsorption capacity of

theadsorbedspeciesrespectively.

20

Alternatively, for lowadsorbatepartialpressures, the solid‐fluid adsorptionequilibrium

canbeexpressedbytheHenry’slaw:

ω K P (3.2)

inwhichKH[mmolg‐1bar‐1]istheHenryequilibriumconstant.

TheFreundlichisothermcanbeappliedtoaccountforsurfaceheterogeneity:

ω K P (3.3)

In eq. (3.3), KF [mmol g‐1 bar‐1/n] and 1/n [‐] are the Freundlich constant and the

heterogeneityparameterrespectively(bothgenerallytemperature‐dependent)(Do,1998). 

Thepotential theoryhasbeendevelopedmainlybyDubininandPolanyi to characterize

the adsorption process in microporous solids (such as activated carbons) (Polanyi, 1932;

Dubinin and Radushkevich, 1947; Ruthven, 1984; Do, 1998). In such solids, the pore

dimensioniscomparabletothatoftheadsorbatemoleculeandtheadsorptionmechanismis

duetofillingbecausetheadsorptionfieldencompassestheentiremicroporevolume. Inthe

microporefillingtheoryanadsorptionpotentialAisdefinedas:

A RTln (3.4)

where P0 and Peq are the liquid sorbate vapour pressure and the pressure of the gas in

equilibriumwiththeadsorbatephaserespectively,atthesametemperatureT.Animportant

featureofthemicroporefillingtheoryisthatforagivenadsorbent‐adsorbatesystemthereis

a unique temperature‐independent relationship between the adsorption potential and the

adsorbatefractionalloadingreferredasthecharacteristiccurve(Ruthven,1984;Do,1998).

Dubinin and Radushkevich suggested the following Gaussian expression to relate the

degreeofmicroporefillingandtheadsorptionpotential(DubininandRadushkevich.,1947):

exp (3.5)

inwhichVisthevolumeofadsorbateinthemicroporesperunitmassofthesolid,V0isthe

maximum specific volume that the adsorbate can occupy (obtainable from porosimetric

analyses), and E is a characteristic energy (related to the adsorption strength between

adsorbateandadsorbent).OnceVandV0areknown, thesolidmolaradsorptioncapacity

21

and the saturation uptake max can be calculated assuming a liquid‐like adsorbed phase

accordingtoGurvitschas(Do,1998):

ω (3.6)

ω (3.7)

whereVmistheliquidmolarvolume.

Itshouldbeunderlinedthatabovetheadsorbatecriticaltemperaturetheconceptofliquid

ceases to exist. In this context, different methods have been proposed to evaluate Vm and

replacethevapourpressurewithapseudo‐vapourpressure;someoftheseexpressionshave

beencollectedin(Do,1998)andarenotreportedhereforthesakeofbrevity.

Finally, the determination of the adsorption isotherms at different temperatures allows

theestimationoftheisostericheatofadsorption(qst),whichisausefulparameterproviding

information on the degree of energetic heterogeneity of gas‐solid interactions. It can be

computedbyapplyingthewell‐knownClausius‐Clapeyronequation(Do,1998):

q RT (3.8)

3.2 Dynamicsofadsorptioncolumns

Mostoftheadsorptionprocessesarecarriedoutinfixedbedadsorbers,typicallytubular

reactors packed with the adsorbent material which is contacted with a gaseous stream

containing thepollutant tobe removed. In suchsystems, thecompositionsofboth the fluid

and solid phases changewith time as well as with the position in the bed (McCabe et al.,

1993). In the commonpracticeof adsorptionexperiments it isdifficult tomeasure internal

compositionprofilesforevaluatingthedynamicperformancesofanadsorptioncolumn;thus,

itismoreconvenienttomonitortheconcentrationoftheadsorbateatthecolumnoutletasa

functionof time, obtaining the so‐calledbreakthrough curve as reported inFigure3.1. It is

highlightedthatthetimeevolutionofthecompositionprofilecanbeconvenientlyexpressed

intermsoftheratiobetweenthevolumetricflowrateoftheadsorbedispeciesatthecolumn

outlet and the correspondent value of the feed (Qiout(t)/Qiin), if percentage volumetric

concentrationscanbeexperimentallyobtained(asinthecaseoftheNDIRanalyzeradoptedin

thisproject,cf.Chapter4).

22

t

Qiou

t (t)/

Qiin

, -

0.0

0.2

0.4

0.6

0.8

1.0

Figure3.1Characteristicbreakthroughcurveobtainedfromfixed‐bedadsorptionexperiments

It can be observed that the curve is S‐shaped: initially the pollutant is completely

adsorbed,thenitsconcentrationattheoutletincreasesuptotheinletlevelwhenthesolidis

completely saturated. In general, for the breakthrough curve it is possible to define a

characteristic breakpoint time tb for which Qiout(t)/Qiin=0.05; in the industrial practice, it

usuallyrepresentsalimitingworkingconditionfortheadsorbercorrespondingtoregulation

emissionlimit(McCabeetal.,1993).

Itcanbedemonstratedthattheareaabovethebreakthroughcurveisproportionaltothe

totalamountofpollutantcapturedbytheadsorbent;asamatteroffact,inordertoevaluate

thesolidadsorptioncapacity,amaterialbalance[mg]onispeciesovertheadsorptioncolumn

isrequiredasreportedineq.(3.9):

Q ρ dt Q t ρ dt mdω(3.9)

in which Qiin and Qiout(t) [L s‐1] represent again the column inlet and outlet i species

volumetric flow rates respectively, i [mg L‐1] is the pollutant density at the operative

temperatureandpressure,m[g]istheadsorbentmassandd[mgg‐1]thedifferentialsolid

adsorption capacity. The mass balance equation (3.9) takes into account that, in the

differentialtimedt,thepollutantmassadsorbedonthesolid(md)equalsthesamequantity

lostbythegaseousphase(LHSineq.(3.9)).

Byrearrangingeq.(3.9)andintegratingbetweenzeroandtimet*forwhichQiout(t)/Qiinis

practically unitary (complete solid saturation), it is possible to evaluate the solid total

tb

0.05

23

adsorptioncapacityeq,asreportedineq.(3.10):

ω

1 dt∗

(3.10)

The correct interpretationof the effectsof the adsorptionkineticsand theextentof the

axialmixinginthecolumnontheadsorbentdynamicperformancesrequiresanappropriate

modelthatgivesatheoreticalbreakthroughcurvematchingtheexperimentaldata.Moreover,

the design of an adsorption column can be realized without the recourse to extensive

experimentationbypredictingapriorithekineticresponsecurvefromequilibriumdataand

byestimatingthemasstransfercoefficientsoncethesolid‐sorbatepropertiesareknownand

thesystemfluiddynamicsisfixed.Therigorousmathematicalmodelrequiresthesolutionof

equationsderiving frommass,momentumandenergybalances.Thehypothesesadopted in

this work to describe the fixed bed dynamics are (Ruthven, 1984; Ding and Alpay, 2000;

Delgadoetal.,2006;Serna‐GuerreroandSayari,2010;Shenetal.,2010):

theflowpatternisdescribedwiththeaxiallydispersedflowmodel;

themasstransferrateisgovernedbyalineardrivingforce(LDF);

thegasphasebehavesasanidealgasmixture;

radialconcentrationandtemperaturegradientsarenegligible;

thesystemisisothermal.

It should be noted that even if a typical industrial‐scale adsorption column is operated

adiabatically, lab‐scale experiments are usually carried out in a temperature‐controlled

environmentwhich leads to the useful approximation of an isothermal fixed‐bed adsorber

that simplifies the system mathematical modelling (Serna‐Guerrero and Sayari, 2010). In

addition,typicallyemployedphysisorbents(suchasactivatedcarbons)showlowadsorption

heat which determines a negligible effect on the gas temperature, thus supporting the

hypothesisofanisothermalprocess.

Massandmomentumbalances

On thebasis of the above‐mentioned assumptions, themassbalance for the adsorbate i

speciesinadifferentialelementofthecolumndz(totallengthL)isgivenby:

εD ε 1 ε ρ 0(3.11)

24

whereCi istheadsorbateconcentrationinthegaseousstream,tthetime, thebedvoidage

fraction,uthegassuperficialvelocity,pistheadsorbentparticledensityandDaxrepresents

the axial dispersion coefficient. The resolution of themass balance equation (3.11) can be

obtainedbyfixingthefollowinginitialandboundaryconditions:

t=0i=0z(3.12a)

t=0Ci=00<z≤L(3.12b)

z=0Ci=Ciint(3.12c)

=0t(3.12d)

inwhichCiinisthepollutantconcentrationinthegaseousphaseatthecolumninlet.Itshould

be noted that in eq. (3.12a) it has been assumed that the adsorbent is initially free of

adsorbate(i=0).

Therateofadsorptionfortheadsorbateisexpressedas:

1 ε ρ 1 ε ρ k , ω∗ ω (3.13)

where ks,i is a lumpedmass transfer coefficient andi* the solid adsorption capacity for i

componentwhichwouldbeinequilibriumwithitsconcentrationinthegaseousphase(Ci).In

this context, it is underlined that the resolution of eq. (3.13) requires an appropriate

equilibrium expression i*=f(Ci,T) which can be obtained from one of the adsorption

isothermsdiscussedinSection3.1.

Therelationshipbetweenthetotalpressuregradientandthegassuperficialvelocitycan

bederivedfromtheErgun’sequation:

u.

u (3.14)

whereandgaretheviscosityanddensityofthegasrespectively,whiledprepresentsthe

meanSauterparticlediameter.

Evaluationofaxialdispersionandmasstransfercoefficients

Theresolutionof theequationsrelativeto therateofadsorptionandmassbalance for i

component of the system expressed in eqs. (3.11 and 3.13), requires an estimation of the

effectoftheaxialdispersion(Dax)andtheglobalmasstransfercoefficientks,i.

25

In general for aporousadsorbentmaterial the adsorptionprocess is characterizedbya

complexmechanismwhichinvolves(Ruthven,1984;PerryandGreen,1997):

externalmasstransferofthepollutantinthefluidfilmlayersurroundingthesolid

particle;

macroporeandmicroporediffusionoftheadsorbatewithintheadsorbent;

adsorptiononthesolidactivesites.

The pseudo‐reaction between the pollutant and the solid sorbent is usually fast for

physical adsorption, thus the evaluation of diffusion resistances allows the identification of

therate‐determiningstepoftheprocess.

Theglobalmasstransferresistanceiscommonlyexpressedasalinearcombinationofthe

film, macropore and micropore diffusion resistances as (Ruthven, 1984; Perry and Green,

1997):

, , , ,(3.15)

where kext,i is the external fluid film mass transfer coefficient for i, Dmacro,i and Dmicro,i its

macropore and micropore diffusivities, p represents the particle porosity, Hi is the

dimensionless Henry constant for i obtained from the slope of the linear part of the

adsorption isotherm (by expressing the solid adsorption capacity in terms of volumetric

concentrationasafunctionoftheconcentrationofiinthegaseousphase).

In order to evaluate the film mass transfer coefficient, it is useful to define the

dimensionlessReynolds(Re),Sherwood(Sh)andSchmidt(Sc)numbersas:

Re ; Sh , ;Sc

in which Dij is the i molecular diffusivity into the gas matrix (i.e. CO2/N2 mixtures in this

project), which can be evaluated according to the Chapman‐Enskog equation (Perry and

Green,1997):

D 1.858 ∗ 10.

.

(3.16)

where Mi and Mj are the molecular weights for i and j species, ij the average collision

26

diameterandDatemperature‐dependentcollisionintegral(tabulated).

Thevalueofkext,icanbeobtainedaccordingtotheWakaoandFunazkricorrelation(Perry

andGreen,1997;Shenetal.,2010):

Sh 2 1.1Re . Sc (3.17)

ThemacroporediffusivityDmacro,icanbeevaluatedas(Shenetal.,2010):

,

τ,(3.18)

wherepistheporetortuosity.TheKnudsendiffusivityDk,iisgivenby(Ruthven,1984):

D , 48.50d.(3.19)

withdpore[m]representingthemeanporediameter.

Themicropore diffusion is an activated process and exhibits an Arrhenius dependence

fromtemperature(Ruthven,1984):

D , D , exp (3.20)

inwhichD0micro,iisthelimitingdiffusivityatinfinitetemperatureandEatheactivationenergy;

Dmicro,i is usually evaluated from chromatographic andNMR studies or from separate batch

adsorptionexperiments(Ruthven,1984).

The axial dispersion in packed beds usually derives from two main mechanisms:

molecular diffusion and turbulentmixing arising from splitting and recombination of flows

aroundtheadsorbentparticle(Ruthven,1984).Theseeffectscanbeconsideredadditive,thus

theaxialdispersioncoefficientcanbeexpressedas:

D γ D (3.21)

with and constants; the values proposed for 1=0.73 and 20.5 1 by

EdwardsandRichardsonhavebeenusedinthiswork(PerryandGreen,1997).

Finally, theeffectofaxialdispersioncanbeevaluatedbycomputingthefixed‐bedPéclet

numberdefinedas:

27

PeuLεD

Typically, for Pe>100 it is possible to consider an ideal plug‐flow for the system (i.e.

negligibleaxialdispersion)(InglezakisandPoulopoulos,2006).

The numerical resolution ofmass andmomentum balance equations (3.11), (3.13) and

(3.14)wasobtained in thisworkwithAspenAdsimTMmodelling environment adopting the

methodoflines:aTaylor‐basedUpwindDifferencingSchemewasusedforthediscretization

of first‐order spatial derivatives and a second‐order Central Differencing Scheme for the

discretization of the second‐order term (axial dispersion in eq. (3.11)). The aim of the

mathematicalmodellingwastoprovideanestimationofthemicroporediffusivityDmicro,i(for

each investigated gas‐solid adsorption system) which is the only parameter not directly

computable (kext,i and Dmacro,i values can be determined once the adsorbent properties are

knownandthesystemfluiddynamicsisfixed).Tothisend,AspenAdsimTMsoftwareenables

the evaluation of Dmicro,i as a fitting parameter by minimizing the sum of the squared

differences between numerically calculated and experimentally observed values of the

gaseousphasecompositionatthefixed‐bedoutlet(leastsquaresmethod).Thecomparisonof

masstransferresistancesineq.(3.15)allowedthedeterminationoftherate‐limitingstepof

theadsorptionprocess.

 

 

 

 

 

 

 

 

In this

with ionic

discussed.

experimen

4.1 A

Two a

work in o

efficiency

size600‐9

1000m,

Two d

(supplied

(molecula

methylimi

ILwhichi

2012).The

Thead

batch stir

differenta

was separ

order to

methanol

[Hmim][B

s Chapter t

c liquids, s

. A descrip

ntsisalsop

Activated

ctivated ca

order to a

forbothra

900m,F6

N.RGC30).

different io

by Sigma

rweight:2

idazoliumg

isabletofo

eILsmolec

Figure4

dsorbentsi

red system

activephas

rated from

completely

for [Emim

F4] impreg

(a)

MA

the experim

solids char

ption of th

provided.

dcarbons

arbonswer

assess the

awandILs

600‐900)a

onic liquids

Aldrich):

254gmol‐1

glycine[Em

ormcarbam

cularstruct

4.1Molecula

mpregnati

m with an

seinitialco

m the liquid

y remove

m][Gly], a

gnated acti

CH

ATERIAL

mental pro

racterizatio

he lab‐scal

s,ionicliq

re selected

effect of

impregnat

andMeadW

s were tes

1‐hexyl‐3‐

)whichisa

mim][Gly](

matewith

turesarere

arstructures

onwascar

IL solution

oncentratio

d solution b

the solven

further e

ivated carb

HAPTER

LSAND

otocols app

ons and a

le plant de

quidsand

d for the ex

different

tedmateria

WestvacoN

sted as im

‐methylimi

atypicalph

(molecular

carbondio

eportedin

sof(a)[Hmi

rriedoutas

n (liquid to

ons(5.6×10

by filtratio

nt, ethyl a

evaporation

bons, the s

R4

METHO

plied for a

adsorption/

esigned fo

dimpreg

xperimenta

porosimetr

als:Calgon

NucharRG

mpregnating

idazolium

hysicalsolv

weight:18

oxide(Zhan

Figure4.1.

im][BF4]and

sfollows:t

o solid rat

0‐3and2.2×

on and ove

acetate in

n step wa

solvent wa

(b)

ODS

ctivated ca

/regenerati

r the exec

gnationp

al campaig

ric structu

CarbonFil

C30(granu

g agents of

tetrafluoro

venttoward

85gmol‐1)

ngetal.,20

d(b)[Emim]

hesubstrat

tio equal to

×10‐2M).A

en dried at

the case

as required

s removed

arbons imp

ion experi

cution of a

protocols

gn carried

ures on CO

ltrasorb40

ulometric r

f the selec

oborate [H

dsCO2and

anamino

006a;Kasa

][Gly]ILs

tewascon

o 5.4 mL g

After1wee

t 100°C ov

of [Hmim]

d. In part

d by feedin

28

pregnation

ments are

adsorption

s

out in this

O2 capture

00(particle

range600‐

cted solids

Hmim][BF4]

d1‐ethyl‐3‐

acid‐based

aharaetal.,

ntactedina

g‐1) at two

ekthesolid

ernight. In

][BF4] and

ticular, for

ng pure N2

8

n

e

n

s

e

e

‐

s

‐

d

,

a

o

d

n

d

r

2

29

(flowrateequal to60NLh‐1, for1h)at180°C intoacolumnpackedwith the impregnated

solid while in the case of [Emim][Gly] functionalized sorbents, methanol removal was

achieved in an oven at 100°C under vacuum for 5 h. It is underlined that the adopted

operating conditions for the evaporation stage were chosen on the basis of the different

thermal stabilities of the two ILs (lower decomposition temperature for [Emim][Gly], cf.

Chapter5).Moreover,theeffectivenessoftheevaporationstepwasvalidatedfromprevious

experiments inwhichthesolidswereimpregnatedonlywiththesolvents:completesolvent

removalwasverifiedbyweighingthesorbentbeforeandafterthermaltreatment.Foraneasy

identification, the functionalized solids were labelled according to the adopted activated

carbon,ILandimpregnationconcentrationas:

F600‐900andN.RGC30forrawactivatedcarbons;

F600‐900[Hmim][BF4]10‐3Mand10‐2M,N.RGC30[Hmim][BF4]10‐3Mand10‐2M

for activated carbons impregnated with [Hmim][BF4] and adopting IL initial

impregnationconcentrationsof5.6×10‐3and2.2×10‐2M;

F600‐900 [Emim][Gly]10‐3Mand10‐2M,N.RGC30[Emim][Gly]10‐3Mand10‐2M

foractivatedcarbonsfunctionalizedwith[Emim][Gly]underILinitialimpregnation

conditionsof5.6×10‐3and2.2×10‐2M.

4.2 Solidscharacterizationtechniques

Thesolids tested in thiswork forCO2captureexperimentswerecharacterizedadopting

the following techniques: i) CO2/N2 porosimetric analyses to determine the solids textural

parameters; ii) thermogravimetric analyses (TGA) to evaluate the amount of ionic liquid

loaded on each sorbent after the impregnation treatment and for assessing the thermal

stability of the active phase confined in the porous substrates. All the analyses have been

carried out at the Laboratorio de Materiales Avanzados (LMA), Department of Inorganic

ChemistryofUniversidaddeAlicante(Spain).

4.2.1 CO2/N2porosimetricanalyses

Porosimetric analyses were carried out in a home‐made fully automated equipment

designed and constructedby theAdvancedMaterials group (LMA), now commercialized as

N2Gsorb‐6(GastoMaterialsTechnology;www.g2mtech.com),workingat‐196and0°CforN2

and CO2 respectively. Adsorption measurements were recorded in the relative pressure

(P/P0)rangeof10‐7‐1fornitrogenand10‐7‐0.03forcarbondioxide.Priortoadsorptionruns

30

each sample was degassed under vacuum at 100°C in order to remove humidity or other

volatileimpurities.Noteworthy,thedegassingproceduredidnotdeterminedesorptionofthe

ionic liquid fromthesubstrate, as confirmedbyTGAanalysescarriedouton functionalized

sorbents before and after vacuum application. Moreover, it is highlighted that CO2

measurementswereconductedonly for rawF600‐900andN.RGC30activatedcarbons.The

reason for this choice is related to the specific interactions establishingbetween theprobe

CO2 gaseousmolecules and the ionic liquid dispersed onto the substrate (with a chemical

conversion into carbamate occurring in the case of [Emim][Gly]), which could modify the

distributionoftheactivephaseinsidethesorbentpores,thusleadingtoapossibleincorrect

evaluationofthenarrowmicroporosityoftheimpregnatedsamples.

The raw N2/CO2 adsorption data were processed according to the common models

retrievablefromliterature inordertoevaluatethesolidmicrostructuralparameters,andin

particular:

thesorbentstotalporevolumewasderivedfromN2adsorptionisothermsbyapplying

theGurvitsch rule for the volumeofnitrogen adsorbed atP/P0=0.97 (Leofanti et al.,

1998);

thetotalmicroporevolumewasevaluatedwithDubinin‐Radushkevich(DR)equation

(appliedintheN2isothermregionP/P0=10‐4‐10‐2)(MorlayandJoly,2010);

the apparent surface area was obtained from N2 adsorption data by means of BET

equationappliedintherelativepressurerangeP/P0=0.01‐0.15,whichbestagreedwith

criteria proposed by Roquerol et al. (2007) for the applicability of BET method to

microporoussorbents;

thevolumeofnarrowmicropores (porediameterup to0.7nm)wasevaluated from

CO2adsorptionisothermat0°CusingtheDRequation;

the absolute pore size distributions were obtained by applying the Quenched‐Solid

Density Functional Theory (QSDFT) to N2 adsorption data (Neimark et al., 2009;

Silvestre‐Alberoetal.,2012).

4.2.2 Thermogravimetricanalyses(TGA)

Thermogravimetric analyses on both raw and IL‐impregnated solids were performed

using a TA Instrument SDT 2960 operated in the temperature range 25‐400°C under a N2

inertatmosphere(flowrate95mLmin‐1)at5°Cmin‐1scanrate.TGAmeasurementsallowed

tofollowthesamplemass%evolutionasa functionofthetemperature.Thecomparisonof

the therm

amount o

adoptedfo

adsorbed

%wt.

in which

wt.%impr

impregnat

ionic liqu

particular

determine

4.3 L

Figure

apparatus

adsorption

F.C.massfMpressuVvalveB.H.bandb

CO2

F

mogramsob

of IL loade

ortheimpr

onthesub

∆wt

%wt.IL‐ads

(dec‐IL) a

ted and ra

id decomp

, the mass

edbyvolati

Lab‐scale

s 4.2 and

s designed

n/regenera

flowcontrolleuregauge

barrelheaters

N2

.C.

F.C. 

btained for

d on the a

regnationp

stratewas

t.% ∆

s is the m

and wt.%

awmateria

poses de

s loss of th

ilemattero

plantfor

4.3 show a

d and buil

ationruns.

Figure

er

s

A

r rawand i

activated c

procedure.

derivedac

∆T

mass perc

%raw(dec‐I

ls, respect

c‐IL (that c

he raw ma

orhumidity

rCO2ads

a schemati

lt during

 

e4.2Layout

BfixedbeT.C.temperTtemperVtvent

Adsorberli

 

impregnate

carbons, at

Morespec

ccordingto

∆wt.%

centage of

IL) represe

tively, in t

can be rou

aterial is s

ydesorbing

sorption/

ic represen

this PhD

oftheexper

edadsorberraturecontrolratureprobe

B

ine

T

V1

T.C.

T

edmateria

t each init

cifically,aq

thefollowi

% ∆T

f the IL

ent the m

the temper

ughly ident

subtracted

gfromthe

/regener

ntation and

project fo

rimentalapp

llerP.I.D.

By‐passline

T

B.H.B

M

M

T

lsprovided

tial active

quantitative

ingexpress

in the im

mass perce

rature rang

tified from

to elimina

substrated

rationtes

d a picture

or the ex

paratus

FfloCcoIRinPCpe

e

F

Vt

PC

V

d informat

phase con

eestimatio

sion:

mpregnated

entage loss

ge inwhich

m literature

ate any co

duringthe

sts

e of the exp

xecution o

owmeterompressornfraredCO2anersonalcomp

t

IR

V2

F

Vt

31

tionon the

ncentration

onoftheIL

(4.1)

d sorbent,

ses of the

h the pure

e data). In

ontribution

test.

perimental

f dynamic

nalyzeruter

C 

M

V3

1

e

n

L

,

e

e

n

n

l

c

The fe

FlowBron

gascompo

CO2ad

bedcolum

45 μm po

operations

arrangedc

bandheat

PMPIDc

a relation

granularb

was defin

interspace

CO2co

by a cont

acquisition

LabView™

monitored

exit of the

acquisition

massflcontroll

Figure4.3

edgas com

nkhorst20

ositions(1‐

dsorption/r

mn(length=

orous sept

s.Thefixed

coaxiallyw

ters,envelo

controllers

nship amo

bedtemper

ned during

etemperatu

oncentratio

tinuous AB

n and elab

™ software.

dbymeans

e adsorptio

n.

owlers

3Pictureof

mposition (

01‐CV),whi

‐15%CO2b

regeneratio

=0.13m;in

tum and c

dbedtemp

withtheads

opedinath

(Watlow).

ng band

ratures.On

adsorption

urebymea

onmeasure

BB NDIR (n

borationw

. Gas volum

sofamass

on column,

fixedbeadsorbe

+heatingsy

thelab‐scale

(N2+CO2)w

ichallow t

byvol.).

ontestson

nerdiamet

composed

peraturew

sorberunit

hermalinsu

Beforedyn

heater sur

ncethethe

n tests by

ansoftype

ementsdur

non‐disper

ere perform

metric flow

s flowcont

, and digita

eder

ystem

eplantforC

wasdeterm

togenerate

theinvesti

ter=0.02m

of two u

wascontroll

t:itconsist

ulatinglay

namictests

rface‐band

ermalprofi

setting an

Jthermoco

ringadsorp

rsive infrar

med by in

w rate var

trollerseri

ally interfa

flowme 

O2adsorptio

minedviam

egaseouss

igatedsorb

m)madeup

nits for ad

ledbymea

tsofthree

erofceram

s,acalibrat

d heater/fi

lewaskno

nd controll

ouples.

ption/regen

red) AO202

terfacing t

iations, oc

iesElFlow

acedwith t

eter

NDIRanaly

on/regenera

mass flowc

streamssim

bentswere

ofPyrexg

dsorbent c

nsofanad

500Wcyli

micfibers,a

tioncurvew

xed bed i

own,thefix

ing the ba

nerationte

20 Uras 26

he analyze

curring in

wBronkhor

the PC uni

yzer

ationtests

controllers

mulating ty

carriedou

glass,equip

charging/d

dhocheati

indricalsh

andconnec

wasbuiltt

interspace‐

xedbedte

and heater

estswerec

6 gas anal

erwith a P

the fixed

rst201‐CV,

it for an on

PC

32

s (seriesEl

ypical flue‐

utinafixed

pedwitha

discharging

ngsystem,

ellWatlow

ctedtoEZ‐

oestablish

‐adsorbent

mperature

/fixed bed

carriedout

lyzer. Data

PC unit via

bed, were

, setat the

n‐line data

2

l

‐

d

a

g

,

w

‐

h

t

e

d

t

a

a

e

e

a

33

Aby‐passlinewasalsoimplementedinthelab‐scaleapparatusinordertoverifythefeed

compositionpriortoadsorptionexperiments:athreewayballvalve(V1)allowstoadequately

switchthegasflowwhiletwoballvalves(V2andV3)aredevotedtoavoidbackflowtowards

theexcludedline.

4.4 Experimentalprotocolsforfixedbeddynamicexperiments

InthisSectionadescriptionoftheexperimentalprotocolsandconditionsadoptedforCO2

adsorption experiments, adsorption/desorption cycles and regeneration tests will be

provided.

4.4.1 ContinuousCO2adsorptiontests

CO2continuousadsorptionrunsrequiredtwodifferentsteps:a)plantpreparation;b)test

execution.Aschematicdescriptionoftheexperimentalprotocolisprovidedinthefollowing.

a)Plantpreparation

Charge of reactor (column) with a known adsorbent amount previously heated

overnightat105°Ctoremovehumidity;

Filloftheremainingpartofthecolumnwithinertglassbeadstouniformthegaseous

flowatthecolumninlet;

CheckofthedefinedCO2concentrationbyNDIRanalyzer,directingthegaseousstream

throughtheby‐passlinewithvalveV3closedtoavoidbackflowintheadsorberline;

Flushofallplantpipelineswithnitrogentoeliminatethepresenceofatmosphericair;

Checkofgasleakageforallpipelinesconnectionsinpresenceofanitrogengasstream;

Check of column leakage by closing the solenoid valve of the mass flow controller

placedattheadsorberoutlet,pressurizingwithN2andmonitoringthepressurewith

pressuregauges.

b)Testexecution

Injection of the defined N2 and CO2 volumetric flow rates viamass flow controllers

throughtheby‐passlinefor30sandventingtotheatmospherewithvalveV2closedto

reachastationaryvalueofconcentration;

SwitchingofthegaseousstreamtowardtheadsorberlineviathreewayvalveV1and

simultaneouslyacquiringtheNDIRconcentrationsignal.

34

The sorbents CO2 capture performances have been investigated under the following

experimentalconditionsadoptedforthefixed‐bedreactor:

Sorbent dose: 15 g for F600‐900 derived sorbents and 13 g for N.RGC30 based

materials;

Totalgasflowrate:2.5×10‐2Ls‐1(evaluatedatT=20°CandP=1bar);

CO2initialconcentration:1‐30%byvol.,balanceN2;

Temperature:30,50and80°C;

Totalgaspressure:1bar.

It isunderlinedthatforN.RGC30sorbents(rawandILs‐impregnated)experimentswere

conductedusingalowersolidamountbecauseofthelowerdensityofthissubstrate(13gof

solid completely filled the column). In addition, CO2 capture tests at initial pollutant

concentrations greater than typical 15% flue‐gas (namely 25 and30%)wereperformed in

ordertobetterinterpretthequalitativetrendoftheadsorptionisotherms.

Thedynamicbehaviorofthegas‐solidadsorptionsystemwasfollowedbymonitoringthe

concentration of the adsorbate at the column outlet as a function of time, obtaining the

breakthrough curves. In particular the time evolution of the composition profile was

expressed in termsof theratioof thevolumetric flowratesofCO2speciesat thebedoutlet

relativetothatinthefeedQ t /Q .

CO2 kinetic adsorption results at 30, 50 and 80°C were processed to obtain the

corresponding adsorption isotherms. The material balance on CO2 over the adsorption

column,leadstothefollowingexpressionfortheequilibriumCO2specificadsorptioncapacity 

eq [mmolg‐1](seeSection3.2):

ω

1 dt∗

(4.2) 

whereCO2[mgL‐1]representsthepollutantdensity(evaluatedat20°Cand1bar)whileMCO2

[mgmmol‐1]isitsmolecularweight;m[g]isthesorbentdoseandt*[s]representsthetime

requiredforreachingcompletesolidsaturation.ThetimeevolutionoftheCO2volumetricflow

rate at the column outlet Q t was obtained from NDIR analyzer which provides this

measurebyfollowingthetemporalvariationsofthegaseouscompositionasreportedineq.

(4.3):

35

Q t

(4.3)

inwhichC t representstheCO2time‐dependentpercentagevolumetricconcentrationin

thegaseousstreamatthecolumnoutlet.

Theresolutionoftheintegralineq.(4.2)foreachinvestigatedinletCO2concentration,was

obtainedbyapplyingthetrapezoidalruletotheexperimentalkineticdata.

Itshouldbeobservedthatineq.(4.3)itisassumedthattheN2volumetricflowrateQ is

practicallyconstantduringthetest:thismeansthatN2adsorptiononallthetestedsorbentsis

negligible. This hypothesis was experimentally verified, prior to CO2 adsorption tests, by

injectingpureN2(2.5×10‐2Ls‐1)throughtheadsorberchargedwitheachinvestigatedsorbent

andmeasuringthegasflowrateatthecolumnoutletbymeansofamassflowmeter.

4.4.2 Adsorption/desorptioncyclesandregenerationtests

Adsorption/desorptioncyclesandpreliminaryregenerationtestswerecarriedoutonthe

sorbent which displayed the highest CO2 capture capacity among the investigated

experimentalconditions(namelyrawF600‐900,cf.Chapter5).

For adsorption/desorption cycles realization, F600‐900 raw was first saturated with a

15%CO2 inN2 gasmixture under the same conditions of total gas flow rate, pressure and

sorbentdosagepreviouslydescribed(adsorptionstep).Subsequently,thepackedcolumnwas

flushed with pure N2 to remove adsorbed CO2 (desorption step), until its concentration

reachedtheNDIRanalyzerlowdetectionlimit(0.1%byvol.).Thisprocedurewasreiterated

for10consecutiveadsorption/desorptioncyclesat30°C.

Regenerationtestswereconductedinordertodefinetheoptimaloperatingconditionsfor

CO2recoveryfromthespentactivatedcarbon.Tothisend,agaseousstreamcontainingCO2at

15%by vol. (total flow rate=2.5×10‐2 L s‐1, P=1 bar)was continuously fed to the adsorber

(m=15 g, T=30°C) until equilibrium conditions were reached. After, the packed bed was

heateduptothedefinedtemperatureforthedesorptionstep(approximately1hrequiredto

reachthermalequilibriumconditions),afixedamountofpureN2asdesorbingagentwassent

through the columnandCO2 concentrationwasmonitoredbymeansof theNDIRanalyzer.

Thisdesorptionstepwascarriedoutat60and100°Cand,foreachtemperaturelevel,three

N2 flow rateswere tested (6.95×10‐3 L s‐1, 1.11×10‐2 L s‐1 and 1.39×10‐2 L s‐1, evaluated at

T=20°CandP=1bar).

36

The regeneration profiles were elaborated to obtain the total specific amount of CO2

desorbedfromtheactivatedcarbondes[mmolg‐1],throughamaterialbalancesimilartothat

reportedinEq.(4.2):

ω

Q t dt. (4.4)

in which t0.1 is the time required to complete the desorption process, assumed as the one

correspondingtotheNDIRlowdetectionlimit(0.1%CO2byvol.).

CO2 concentration profiles were also quantitatively analysed to define the best

experimental conditions for an efficient regeneration process in terms of CO2 recovery

amount, time required for the desorption process (at a fixed regeneration level) and CO2

concentrationinthedesorbingflow,thelatterbeingacriticalaspectforCO2storagepurposes.

To this aim, four different characteristic desorption times (t50, t70, t80, t90) have been

considered,eachcorrespondingtoadifferentCO2recoverypercentageofthetotaladsorbed

amount (e.g. t50 corresponds toa50%of totalCO2 recoveredbydesorption).Consequently,

themeanCO2concentrationinthedesorbingflow(i

2COC )uptothetimeticanbeexpressed

as:

C

(4.5)

inwhich i2COV and i

2NV represent theCO2 totalvolumedesorbedandthepurgegasvolume

fedtothecolumnuptotimeti,respectively.Thevaluesof i2NV and

i

2COC werecomputedfor

variable CO2 recovery percentage i.e. i=50%, 70%, 80% and 90% for each investigated N2

purgeflowrate( des2NQ )anddesorptiontemperature( des

C60T and desC100T ).

37

CHAPTER5

RESULTSANDDISCUSSION

In this Chapter the main results obtained from solids characterization techniques and

adsorption/regeneration experiments will be discussed with a particular emphasis on the

intertwining among adsorbents microstructural properties and their CO2 capture

performances.Finally,theresultsderivedfromboththermodynamicandkineticmodellingof

adsorptiondatawillbeanalysed.

5.1 Adsorbentscharacterization

5.1.1 Porosimetric analyses for F600‐900 raw and impregnated with

[Hmim][BF4]/[Emim][Gly]

Figures5.1and5.2showtheN2adsorption‐desorptionisothermsat‐196°Cobtainedfor

the sorbents F600‐900 raw and impregnated with [Hmim][BF4]/[Emim][Gly] ILs at initial

activephaseconcentrationsC°=5.6×10‐3and2.2×10‐2M.Asageneralconsideration,itcanbe

observedthattheisothermsforallthesamplesareverysimilarandtheyaretypeIaccording

to IUPAC classification (Patrick, 1995) and the strong adsorption observed at very low

relative pressures (i.e. P/P0<10‐3) testifies the presence of a highly developedmicroporous

structure (Morlay and Joly, 2010). A comparison between the adsorption and desorption

branches testifies the non‐negligible presence of mesopores: the narrow hysteresis loop

observed from isotherms is of typeH4 according to IUPAC, commonly associatedwith the

presenceofslit‐shapedpores(MorlayandJoly,2010).Inaddition,itcanbeobservedthatthe

volumeofN2adsorbedreducesastheILconcentrationincreasesforbothactivephaseswith

respect to the parent material above all in the “knee” region of the adsorption isotherms

(P/P0<0.1): this is a clue of the preferential adsorption of both ILs in the substrate

micropores.

38

P/P0, -

0.0 0.2 0.4 0.6 0.8 1.0

Vad

s ST

P, c

m3

g-1

0

100

200

300

400

F600-900

F600-900 [Hmim][BF4] 10-3 M

F600-900 [Hmim][BF4] 10-2 M

 Figure5.1N2adsorption‐desorptionisothermsat‐196°CforF600‐900rawandimpregnatedwith

[Hmim][BF4]atC°=5.6×10‐3and2.2×10‐2M

P/P0, -

0.0 0.2 0.4 0.6 0.8 1.0

Vad

s ST

P, c

m3

g-1

0

100

200

300

400

F600-900

F600-900 [Emim][Gly] 10-3 M

F600-900 [Emim][Gly] 10-2 M

Figure5.2N2adsorption‐desorptionisothermsat‐196°CforF600‐900rawandimpregnatedwith

[Emim][Gly]atC°=5.6×10‐3and2.2×10‐2M

Table5.1reportstheapparentsurfacearea(SBET),thetotalspecificporevolume(Vt),the

specificmicroporevolume(V0)andthespecificmesoporevolume(Vmeso=Vt−V0)determined

for both raw and [Hmim][BF4]/[Emim][Gly] impregnated sorbents by applying themodels

39

indicatedinSection4.2.1toN2adsorptiondataat‐196°C.Thevolumeofnarrowmicropores

(Vn)derivedforrawF600‐900fromCO2adsorptionisothermat0°Cisalsoincluded.

Table5.1TexturalparametersobtainedforF600‐900rawandimpregnatedwith[Hmim][BF4]/[Emim][Gly]ILs

SampleSBET

[m2g‐1]Vt

[cm3g‐1]V0

[cm3g‐1]Vn

[cm3g‐1]Vmeso

[cm3g‐1]

F600‐900 1076 0.58 0.41 0.32† 0.17

F600‐900[Hmim][BF4]10‐3M 1018 0.55 0.39 n.a.† 0.16

F600‐900[Hmim][BF4]10‐2M 961 0.52 0.36 n.a.† 0.16

F600‐900[Emim][Gly]10‐3M 1029 0.55 0.39 n.a.† 0.16

F600‐900[Emim][Gly]10‐2M 971 0.52 0.36 n.a.† 0.16†notavailable

Results highlight a prevailingmicroporous nature for all the investigated sorbents (for

F600‐900 raw micropores contribute to a nearly 70% of the total pore volume), thus

confirming the observations derived from the analysis of the N2 adsorption isotherms.

Moreover, a comparison between the microstructural parameters of raw and ILs

functionalized F600‐900 shows a decrease of both SBET and V0 values when the initial

concentration of each ionic liquid increases, while the mesopore volume reduction is not

significant.TheseobservationsconfirmthatbothILspartiallyobstructonlythemicroporesof

therawmaterial.Finally,at fixed impregnationcondition [Hmim][BF4]and [Emim][Gly] ILs

determinethesamemicroporevolumereduction.

Figures5.3and5.4depict theabsoluteporesizedistributions (PSD)obtained forF600‐

900 raw and functionalized with [Hmim][BF4]/[Emim][Gly] ionic liquids by applying the

QSDFTmethodtoN2adsorptiondataat‐196°Candassumingaslit‐shapegeometryforpores.

Theplotsareexpressedintermsoftheratiobetweenthedifferentialvariationofthespecific

porevolumerelativetothatoftheporediameter(dVp(d))asafunctionoftheporediameter

(d). Results confirm the highly microporous nature of all the samples analysed with a

prevailing contribution of pores smaller than 10 Å. The pore size distributions for ILs‐

impregnatedF600‐900adsorbentsshowsimilarqualitativepatternswithrespecttotheraw

material except for a slight reduction of dVp(d) values in the pore diameter region smaller

than10Åandduetothealreadydescribedpartialmicroporeclogging.

40

pore diameter, Å

0 10 20 30 40 50

dV

p(d

), c

m3

g-1 Å

-1

0.00

0.02

0.04

0.06

0.08

0.10F600-900

F600-900 [Hmim][BF4] 10-3 M

F600-900 [Hmim][BF4] 10-2 M

 

Figure5.3AbsoluteporesizedistributionsforF600‐900rawandimpregnatedwith[Hmim][BF4]atC°=5.6×10‐3and2.2×10‐2M

pore diameter, Å

0 10 20 30 40 50

dV

p(d

), c

m3

g-1 Å

-1

0.00

0.02

0.04

0.06

0.08

0.10F600-900

F600-900 [Emim][Gly] 10-3 M

F600-900 [Emim][Gly] 10-2 M

Figure5.4AbsoluteporesizedistributionsforF600‐900rawandimpregnatedwith[Emim][Gly]at

C°=5.6×10‐3and2.2×10‐2M

41

5.1.2 TGA analyses for F600‐900 raw and impregnated with

[Hmim][BF4]/[Emim][Gly]

Figures 5.5 and5.6 show the results obtained from thermogravimetric analyses carried

outonrawandILs‐impregnatedF600‐900intermsofsamplemasspercentageasafunction

of the temperature. Thermograms show for all materials a weight loss (1‐3%) for

temperatureslowerthan100°Cwhichcanbelikelyascribedtothedesorptionofhumidityor

othervolatileimpurities.Moreover,forF600‐900[Hmim][BF4]solidsthemasslossdetectable

in the temperaturerange280‐380°C(clearly identifiable for thesample impregnatedunder

more concentrated conditions, Figure 5.5) can be ascribed to the IL decomposition, in

accordance with published research findings concerning TGA analyses for the pure ionic

liquid (Crosthwaite et al., 2005). Finally, for [Emim][Gly] impregnated adsorbents the ionic

liquiddecompositioncanbeapproximatelylocatedintheT‐range170‐330°C(Muhammadet

al.,2011),asconfirmedbythedifferentslopesof thethermogramsforthesematerialswith

respecttorawF600‐900inthesametemperaturerange.

T, °C

0 100 200 300 400

% m

ass

93

94

95

96

97

98

99

100

F600-900

F600-900 [Hmim][BF4] 10-3 M

F600-900 [Hmim][BF4] 10-2 M

 

Figure5.5ThermogravimetricanalysesforF600‐900rawandimpregnatedwith[Hmim][BF4]atC°=5.6×10‐3and2.2×10‐2M

42

T, °C

0 100 200 300 400

% m

ass

93

94

95

96

97

98

99

100

F600-900

F600-900 [Emim][Gly] 10-3 M

F600-900 [Emim][Gly] 10-2 M

Figure5.6ThermogravimetricanalysesforF600‐900rawandimpregnatedwith[Emim][Gly]at

C°=5.6×10‐3and2.2×10‐2M

The comparison of themass loss profiles for each impregnated sorbentwith respect to

raw F600‐900 allowed to estimate the amount of ionic liquid adsorbed on the substrate

accordingtothenumericalproceduredescribedinSection4.2.2(eq.(4.1)).Table5.2reports

acomparisonoftheactivephasemasspercentageadsorbedonthesubstrate(%wt.IL‐ads)with

respect to the one initially used for the impregnation procedure (%wt.IL‐load) for both

[Hmim][BF4] and [Emim][Gly] ILs. Quantitative results are also conveniently expressed in

terms of both IL specificmolar amount initially loaded (mmolIL‐load g‐1AC) and the ILmolar

specificadsorptioncapacityofF600‐900(AC)(mmolIL‐adsg‐1AC).

Table5.2QuantitativeparametersderivedfromTGAanalysesforF600‐900impregnatedwith[Hmim][BF4]and[Emim][Gly]ILs

Sample %wt.IL‐load %wt.IL‐ads mmolIL‐loadg‐1AC mmolIL‐adsg‐1AC

F600‐900[Hmim][BF4]10‐3M 0.76 0.38 0.030 0.015

F600‐900[Hmim][BF4]10‐2M 2.98 1.29 0.121 0.052

F600‐900[Emim][Gly]10‐3M 0.56 0.51 0.030 0.028

F600‐900[Emim][Gly]10‐2M 2.19 1.04 0.121 0.057

Resultsevidencethatfor[Hmim][BF4]‐adsorbentsnearly50%and43%oftheinitialamount

ofILusedfortheimpregnationisadsorbedonthesubstrateforC°=5.6×10‐3and2.2×10‐2M,

43

respectively. The estimated adsorption efficiencies are 93% and 47% for F600‐900

[Emim][Gly]10‐3Mand10‐2Mrespectively.Finally,undermoreconcentratedconditionsthe

amountof[Emim][Gly]adsorbedisslightlyhigherthaninthecaseof[Hmim][BF4](0.057vs.

0.052mmolg‐1)whileitisalmostdoubleforC°=5.6×10‐3M(0.028vs.0.015mmolg‐1).

5.1.3 PorosimetricanalysesforN.RGC30 raw and impregnated with

[Hmim][BF4]/[Emim][Gly]

N2porosimetricresultsat ‐196°CforN.RGC30rawandfunctionalizedwith[Hmim][BF4]

and[Emim][Gly]ionicliquidsaredepictedinFigures5.7and5.8.

P/P0, -

0.0 0.2 0.4 0.6 0.8 1.0

Vad

s ST

P, c

m3

g-1

0

200

400

600

800

N.RGC30

N.RGC30 [Hmim][BF44

] 10-3 M

N.RGC30 [Hmim][BF44

] 10-2 M

Figure5.7N2adsorption‐desorptionisothermsat‐196°CforN.RGC30rawandimpregnatedwith

[Hmim][BF4]atC°=5.6×10‐3and2.2×10‐2M

44

P/P0, -

0.0 0.2 0.4 0.6 0.8 1.0

Vad

s ST

P, c

m3

g-1

0

200

400

600

800

N.RGC30

N.RGC30 [Emim][Gly4

] 10-3 M

N.RGC30 [Emim][Gly4

] 10-2 M

Figure5.8N2adsorption‐desorptionisothermsat‐196°CforN.RGC30rawandimpregnatedwith

[Emim][Gly]atC°=5.6×10‐3and2.2×10‐2M

IsothermsaremixedtypeIandIVandinallthecaseswithahighadsorptioncapacityat

low relative pressures which testifies the significant presence of micropores, while the

observed hysteresis loops indicate a well‐developed mesoporosity for all the investigated

samples. Inaddition, forboth [Hmim][BF4]adsorbents it canbeobservedareductionofN2

adsorbedvolumewithrespecttotheparentactivatedcarbon,andtheiradsorptionisotherms

almostoverlap(aslightlyhigheradsorptioncapacityforthesampleimpregnatedundermore

diluted conditions is detected for P/P0<10‐3). In the case of [Emim][Gly]materials, amore

marked reduction of N2 adsorption capacity is observed under more concentrated

impregnationconditions,whileforthesampleobtainedatC°=5.6×10‐3Mdifferenceswithraw

N.RGC30canbedetectedonlyatlowrelativepressures(withaslightdecreaseofN2adsorbed

volumefortheimpregnatedmaterialatP/P0<10‐2).

The main microstructural parameters obtained from CO2 porosimetric analysis at 0°C

(onlyrawN.RGC30)andN2adsorptiondataat ‐196°Caresummarized inTable5.3.Results

underline a prevailing mesoporous nature for N.RGC30 activated carbon with a 57%

contribution to theoverallporosity;moreover, thesignificantdifferencebetweenV0andVn

valuessuggeststhepresenceofabroadmicroporesizedistributionforthissample(Krutyeva

etal.,2009).AcomparisonbetweenV0andVmesovaluesindicatesthepreferentialmicropore

occlusioninducedbyeachIL,thisbeingmoreimportantastheinitialconcentrationofactive

45

phase used for the impregnation process increases. Finally, it can be observed that under

more diluted conditions [Hmim][BF4] IL determines a highermicropore volume reduction,

whilethiseffectiscomparableforbothionicliquidsatC°=2.2x10‐2M.

Table5.3TexturalparametersobtainedforN.RGC30rawandimpregnatedwith[Hmim][BF4]/[Emim][Gly]ILs

SampleSBET

[m2g‐1]Vt

[cm3g‐1]V0

[cm3g‐1]Vn

[cm3g‐1]Vmeso

[cm3g‐1]

N.RGC30 1427 1.15 0.50 0.32 0.65

N.RGC30[Hmim][BF4]10‐3M 1350 1.12 0.47 n.a.† 0.65

N.RGC30[Hmim][BF4]10‐2M 1318 1.11 0.46 n.a.† 0.65

N.RGC30[Emim][Gly]10‐3M 1418 1.14 0.49 n.a.† 0.65

N.RGC30[Emim][Gly]10‐2M 1307 1.11 0.46 n.a.† 0.65†notavailable

Pore size distributions obtained for N.RGC30 raw and functionalized with

[Hmim][BF4]/[Emim][Gly]ionicliquidsarereportedinFigures5.9and5.10.Resultsconfirm

the presence of a broad micropore size distribution and a significant contribution of

mesopores with a mesopores‐peak centred approximately at 35 Å. In addition, both ionic

liquidsdonotaffectremarkablytherawmaterialPSD,apart fromlittledifferences forpore

diameters lower than 10 Å which can be ascribed to the already observed pore occlusion

effect(Table5.3).

pore diameter, Å

0 10 20 30 40 50

dV

p(d

), c

m3

g-1 Å

-1

0.00

0.01

0.02

0.03

0.04

0.05

0.06N.RGC30

N.RGC30 [Hmim][BF44

] 10-3 M

N.RGC30 [Hmim][BF44

] 10-2 M

 Figure5.9AbsoluteporesizedistributionsforN.RGC30rawandimpregnatedwith[Hmim][BF4]at

C°=5.6×10‐3and2.2×10‐2M

46

pore diameter, Å

0 10 20 30 40 50

dV

p(d

), c

m3

g-1 Å

-1

0.00

0.01

0.02

0.03

0.04

0.05

0.06N.RGC30

N.RGC30 [Emim][Gly4

] 10-3 M

N.RGC30 [Emim][Gly4

] 10-2 M

 

Figure5.10AbsoluteporesizedistributionsforN.RGC30rawandimpregnatedwith[Emim][Gly]at

C°=5.6×10‐3and2.2×10‐2M

5.1.4 TGA analyses for N.RGC30 raw and impregnated with

[Hmim][BF4]/[Emim][Gly]

Thermogravimetric patterns derived for N.RGC30 both raw and impregnated with

[Hmim][BF4]and[Emim][Gly]ILsat C°=5.6×10‐3and2.2×10‐2MareshowninFigures5.11

and5.12.Experimentalresultsshowthatapartfromthedesorptionofhumidityandvolatile

matterobservedfortemperaturelowerthan100°C,thethermaldecompositionrangeforeach

ionicliquidpracticallycoincideswiththatobservedforF600‐900functionalizedsorbents(cf.

Figures 5.5 and 5.6): 280‐380°C and 170‐330°C for [Hmim][BF4] and [Emim][Gly]

respectively. This observation leads to the conclusion that the different microstructural

propertiesofF600‐900andN.RGC30activatedcarbonsdonotaffectremarkablythethermal

stabilityoftheionicliquidsinvestigated.

47

T, °C

0 100 200 300 400

% m

ass

96

97

98

99

100

N.RGC30

N.RGC30 [Hmim][BF4] 10-3 M

N.RGC30 [Hmim][BF4] 10-2 M

Figure5.11ThermogravimetricanalysesforN.RGC30rawandimpregnatedwith[Hmim][BF4]at

C°=5.6×10‐3and2.2×10‐2M

T, °C

0 100 200 300 400

% m

ass

96

97

98

99

100

N.RGC30

N.RGC30 [Emim][Gly] 10-3 M

N.RGC30 [Emim][Gly] 10-2 M

 Figure5.12ThermogravimetricanalysesforN.RGC30rawandimpregnatedwith[Emim][Gly]at

C°=5.6×10‐3and2.2×10‐2M

Thecomparisonofthemasslossprofilesbetweentherawmaterialandtheimpregnated

ones allowed an estimation of the amounts of each ionic liquid adsorbed on N.RGC30

sorbents;themainderivedparametersareschematicallyreportedinTable5.4.Ingeneral,it

48

canbeobserved that theadsorptionefficienciesarehigher in thecaseof [Emim][Gly] ionic

liquid(90%and47%forC°=5.6×10‐3and2.2×10‐2M,respectively),whilesimilaractivephase

uptakeefficiencieswererecordedforN.RGC30[Hmim][BF4]10‐3and10‐2M(43%and42%

undermoredilutedandconcentratedimpregnationconditions,respectively).

Table5.4QuantitativeparametersderivedfromTGAanalysesforN.RGC30impregnatedwith[Hmim][BF4]and[Emim][Gly]ILs

Sample %wt.IL‐load %wt.IL‐ads mmolIL‐loadg‐1AC mmolIL‐adsg‐1AC

N.RGC30[Hmim][BF4]10‐3M 0.76 0.32 0.030 0.013

N.RGC30[Hmim][BF4]10‐2M 2.98 1.28 0.121 0.051

N.RGC30[Emim][Gly]10‐3M 0.56 0.49 0.030 0.027

N.RGC30[Emim][Gly]10‐2M 2.19 1.04 0.121 0.057

5.1.5 Comparison between TGA and porosimetric analyses for the

investigatedsorbents

The main parameters derived from thermogravimetric and porosimetric analyses for

F600‐900 and N.RGC30 both raw and impregnated with [Hmim][BF4]/[Emim][Gly] ionic

liquidsaresummarizedinTable5.5.Thepercentagereductionofthetotalmicroporevolume

derivedforeachimpregnatedmaterialwithrespecttotheparentone(V0‐raw−V0‐impr)/V0‐raw

(with V0‐raw and V0‐impr representing the total micropore volume of the parent and

functionalizedsorbent,respectively)isalsoreportedforabetterunderstandingoftheeffect

of the impregnation treatment conditions on the microstructural properties of each

investigated sorbent. The most interesting aspects deduced from the data analysis are

highlightedinthefollowing.

RawN.RGC30ischaracterizedbyabroadermicroporesizedistributionwithrespectto

F600‐900,asconfirmedbythehigherdifferencebetweenV0andVnvalues(Krutyeva

et al., 2009), and displays a remarkably higher contribution of mesopores (Vmeso is

nearlyfourfoldthevalueobtainedforF600‐900).

[Hmim][BF4] and [Emim][Gly] ILs preferentially adsorb in micropores for both

substrates and in general the higher the IL loading, the greater the micropore

occlusion.

For F600‐900 impregnated sorbents, even if the specific amount of [Emim][Gly]

adsorbedon the substrate is higher than in the case of [Hmim][BF4] (almost double

undermorediluted impregnationconditions) themicroporevolumereduction is the

49

sameatfixedinitialactivephaseconcentration.AsimilarsituationoccursforN.RGC30

materials functionalized at C°=2.2×10‐2 M, while at lower initial active phase

concentration(5.6×10‐3M)[Emim][Gly]producesamicroporevolumereductionthat

isone‐thirdtheoneobservedinthecaseof[Hmim][BF4],despiteofadoublespecific

amountadsorbed.Thedescribedtrendscanberelatedtothemorebranchedshapeof

[Hmim][BF4] imidazolium ring which is characterized by a C6 side chain (C2 for

[Emim][Gly], cf. Section4.1): the largercationic sizeof this ionic liquiddeterminesa

similar (or even higher) pore occlusion effectwith respect to [Emim][Gly] for lower

adsorbedamount.

A comparison between F600‐900 and N.RGC30 impregnated with [Hmim][BF4]

revealsthatsimilarspecificamountsofionicliquidareadsorbedatfixedinitialactive

phase concentration: under more diluted impregnation conditions the micropore

volume reduction is comparable for both substrates (4.9 and 6% for F600‐900 and

N.RGC30respectively),whiletheporecloggingeffectismoresignificantforF600‐900

athigherILconcentration.Likewisefor[Emim][Gly]adsorbents,asimilarILadsorbed

amount for both substrates at fixed impregnation condition determines a higher

microporevolume reduction forF600‐900activated carbon.These results shouldbe

interpreted in the lightof thedifferent texturalpropertiesofF600‐900andN.RGC30

activated carbons. As a matter of fact, the presence of a narrower micropore size

distributionwithaprevailingcontributionofverysmallporediameters(<10Å,cf.also

Figures5.3and5.4)forF600‐900determinesamorerelevantocclusioneffectforeach

ILandat fixed impregnationcondition than in thecaseofN.RGC30adsorbentwhich

exhibitsagreatercontributionofwiderporesinthemicroporesrange(cf.Figures5.9

and5.10).

The aforementioned analysis is considered to be an important reference for the

subsequent discussion concerning the effects of the different adsorbents microstructural

propertiesontheirCO2captureperformances.

 

50

Table5.5MainparametersderivedfromporosimetricandTGAanalysesforF600‐900andN.RGC30rawandimpregnatedwith[Hmim][BF4]and[Emim][Gly]ILs

SampleSBET

[m2g‐1]V0

[cm3g‐1]Vn

[cm3g‐1]Vmeso

[cm3g‐1](V0‐raw‐V0‐impr)/V0‐raw

%mmolIL‐adsg‐1AC

(TGA)

F600‐900 1076 0.41 0.32 0.17

F600‐900[Hmim][BF4]10‐3M 1018 0.39 n.a.† 0.16 4.9 0.015

F600‐900[Hmim][BF4]10‐2M 961 0.36 n.a.† 0.16 12.2 0.052

F600‐900[Emim][Gly]10‐3M 1029 0.39 n.a.† 0.16 4.9 0.028

F600‐900[Emim][Gly]10‐2M 971 0.36 n.a.† 0.16 12.2 0.057

N.RGC30 1427 0.50 0.32 0.65

N.RGC30[Hmim][BF4]10‐3M 1350 0.47 n.a.† 0.65 6 0.013

N.RGC30[Hmim][BF4]10‐2M 1318 0.46 n.a.† 0.65 8 0.051

N.RGC30[Emim][Gly]10‐3M 1418 0.49 n.a.† 0.65 2 0.027

N.RGC30[Emim][Gly]10‐2M 1307 0.46 n.a.† 0.65 8 0.057

†notavailable

 

51

5.2 CO2captureperformancesontheinvestigatedsorbents

5.2.1 CO2adsorptiontestsontorawF600‐900andN.RGC30

Figure5.13depictsCO2adsorptionisothermsobtainedforrawF600‐900andN.RGC30at

(a)30,(b)50and(c)80°Cintermsofthesolidmolaradsorptioncapacityeqasafunctionof

thepollutantequilibriumpartialpressure inthegaseousphase(Peq). It ishererecalledthat

eqvalueswerederivedbyintegrationofthebreakthroughcurvesobtainedfromkineticruns

at different CO2 concentrations (1‐30% by vol.) according to the procedure described in

Section4.4.1.ResultsevidenceareductionofCO2adsorptioncapacitywhenthetemperature

increases for both activated carbons, as a consequence of the exothermic character of

physisorption (Plaza et al., 2007). In particular, under typical flue gas conditions (Peq=0.15

bar) eq is 0.575 mmol g‐1 at 30°C for F600‐900 which is 1.8‐ and 4.2‐times the values

obtainedat50and80°C,respectively.Similarly,eqforN.RGC30atPeq=0.15barreducesfrom

0.499mmolg‐1at30°Cto0.276and0.139mmolg‐1at50and80°C,respectively.Asageneral

considerationCO2adsorptionperformancesobtainedinthisworkforF600‐900andN.RGC30

arecomparablewiththosereportedintheliteratureforrawactivatedcarbonstestedunder

similarexperimentalconditions(cf.Table2.1inSection2.1.1).

AcomparisonofF600‐900andN.RGC30adsorption isothermsat thedifferentoperating

temperatures reveals that at 30°C the former sorbent displays higher CO2 removal

performances for all investigated pollutant equilibrium partial pressures, while differences

reduceat50°Ctobecomepracticallynegligibleat80°C.Theseresultsshouldbe interpreted

considering that at lower temperatures physisorption plays a major role in the pollutant

capture and its contribution decreases at higher temperatures (Plaza et al., 2007). In this

context, on the basis of the sorbents textural properties reported in Section 5.1 (with

particularreferencetoTable5.5)onewouldhaveexpectedhigherCO2capturecapacitiesfor

N.RGC30samplebecauseof itshigher surfaceareaandmicroporevolume.Nevertheless, as

recentlyhighlightedbyWhabyet al. (2010and2012), thepresenceof anarrowmicropore

sizedistributionwithwell‐definedporesizeentrances(mainlyporediameters<5Å)seemsto

beakeyfactorindeterminingCO2adsorption,becauseinnarrowmicroporestheoverlapping

potentialproducesamoreeffectivepackingofCO2molecules.Asaconsequence,thenarrower

microporesizedistributionwithaprevailingcontributionofverysmallporediameters(<10

Å)observedforF600‐900solidshouldberesponsibleforitshighercaptureperformancesat

30°C. At intermediate temperature (50°C) physisorption contribution is less important and

 

52

consequentlydifferencesinCO2sorptivepropertiesbetweenthetwoactivatedcarbonstend

toreduceandbecomepracticallynegligibleat80°C.

Peq, bar

0.0 0.1 0.2 0.3 0.4

eqm

mol

g-1

0.0

0.2

0.4

0.6

0.8

1.0F600-900N.RGC30

(a)

Peq, bar

0.0 0.1 0.2 0.3 0.4

eqm

mol

g-1

0.0

0.2

0.4

0.6

0.8

1.0F600-900N.RGC30

(b)

 

 

53

Peq, bar

0.0 0.1 0.2 0.3 0.4

eqm

mol

g-1

0.0

0.2

0.4

0.6

0.8

1.0F600-900N.RGC30

(c)

Figure5.13CO2adsorptionisothermsobtainedforrawF600‐900andN.RGC30adsorbentsat(a)30,(b)

50and(c)80°C

TherawadsorbentsdynamicadsorptionperformancesarecomparedinFigures5.14and

5.15whichreportthebreakthroughcurvesofCO2onF600‐900andN.RGC30,respectively,at

30,50and80°Cadoptinga15%byvol.CO2‐concentratedgaseousstream,representativeofa

typicalflue‐gascomposition.

t, s0 20 40 60 80 100 120

QC

O2ou

t (t)/

QC

O2in

, -

0.0

0.2

0.4

0.6

0.8

1.0

T=30°CT=50°CT=80°C

 Figure5.14BreakthroughcurvesofCO2onrawF600‐900at30,50and80°Cfora15%byvol.CO2

gaseousstream

 

54

t, s0 20 40 60 80 100 120

QC

O2ou

t (t)/

QC

O2in

, -

0.0

0.2

0.4

0.6

0.8

1.0

T=30°CT=50°CT=80°C

 Figure5.15BreakthroughcurvesofCO2onrawN.RGC30at30,50and80°Cfora15%byvol.CO2gaseous

stream

As a general consideration, it can be observed that for each activated carbon the

breakthroughcurveshowsamonotonicshift towards lowertimesasthetemperaturerises.

Forinstance,thebreakpointtimetb(Q t /Q =0.05)is43sat30°CforF600‐900,avalue

4.8‐timesgreaterthantheoneobtainedat80°C(9s).InthecaseofN.RGC30tb=32,18and8s

for adsorption temperaturesof 30, 50 and80°C.Moreover, higher temperaturesdetermine

fasteradsorptionkineticsastestifiedbytheincreasingslopeofthesigmoidandthereduction

of the equilibrium time t* (for which CO2 concentrations at the bed inlet and outlet are

practicallyequal).Asamatterof fact, t* isapproximately2.7‐and2.2‐timesgreaterat30°C

withrespectto80°CforN.RGC30(195vs.71s)andF600‐900(293vs.131s),respectively.

These results can be interpreted considering that two factors determine faster adsorption

kineticsathighertemperatures:i)thereductionoftheadsorptioncapacity(lowernumberof

active sites to be occupied)which allows higher diffusion rates; ii) an intrinsic increase in

intraparticle diffusivitywith temperature (mainly inmicropores, cf. Sections 3.2 and 5.4.2)

(Ruthven,1984).

ConsideringthatdifferentamountsofF600‐900andN.RGC30wereusedforadsorption

experiments (15 and 13 g for F600‐900 and N.RGC30 respectively, cf. Section 4.4.1), a

comparisonofthedynamicperformancesofthetwosorbentsonthebasisoftbisnotfeasible

because breakthrough curves translate along time axis by varying the sorbent dose under

 

55

constantpatternconditions(Ruthven,1984;McCabeetal.,1993).Inthiscontext,differences

in mass transfer rates can be better evaluated by introducing a time parametert0.9−tb

(with t0.9 being the time for which Q t /Q =0.9) which is related to the slope of the

linearpartofthesigmoid:thesmallerthisparameterthesteeperthebreakthroughcurveand,

consequently, the faster adsorption kinetics. The values of derived from the kinetic

patterns forF600‐900andN.RGC30rawat30,50and80°Care listed inTable5.6 together

withthesorbentsmainmicrostructuralparameters.

Table5.6Comparisonbetweenvaluesobtainedfromkineticadsorptiontestsat30,50and80°CandthemaintexturalparametersderivedforF600‐900andN.RGC30activatedcarbons; =15%byvol.

Sample

[s]

Microstructuralparameters

T=30°C T=50°C T=80°C V0[cm3g‐1]

Vn[cm3g‐1]

Vmeso[cm3g‐1]

F600‐900 22 10 5 0.41 0.32 0.17

N.RGC30 15 8 5 0.50 0.32 0.65

Results show that for each adsorbentvalue decreases from 30 to 80°C confirming

fasteradsorptionkineticsathighertemperature;moreover,isgenerallyhigherforF600‐

900sorbentwithrespecttoN.RGC30anddifferencesreduceastheadsorptiontemperature

rises (at80°C thevaluespractically coincide).Theseoutcomes suggest fastermass transfer

rates for N.RGC30 sorbent which can be ascribed to the larger contribution of mesopores

(Vmeso=0.65and0.17cm3g‐1forN.RGC30andF600‐900,respectively)andtothepresenceof

widermicropores(cf.V0andVnvalues)withrespecttoF600‐900allowingaquickerdiffusion

ofCO2moleculesinsidetheadsorbentpores.Theeffectofthedifferenttexturalparameterson

the adsorbents dynamic adsorptionperformances becomes less important from30 to 80°C

because of the already‐mentioned increase in intraparticle diffusion rates at higher

temperatures.

 

56

5.2.2 CO2 adsorption tests onto F600‐900 raw and functionalized with

[Hmim][BF4]/[Emim][Gly]

Figure 5.16 reports CO2 adsorption isotherms on F600‐900 raw and impregnated with

[Hmim][BF4]/[Emim][Gly] ILs atdifferent activephase concentrations andat (a)30, (b)50

and(c)80°C.

Peq, bar

0.0 0.1 0.2 0.3 0.4

eq

, mm

ol g

-1

0.0

0.2

0.4

0.6

0.8

1.0

F600-900

F600-900 [Hmim][BF4] 10-3

M

F600-900 [Hmim][BF4] 10-2

M

F600-900 [Emim][Gly] 10-3

M

F600-900 [Emim][Gly] 10-2

M

(a)

 

Peq, bar

0.0 0.1 0.2 0.3 0.4

eq

, mm

ol g

-1

0.0

0.1

0.2

0.3

0.4

0.5

0.6

F600-900

F600-900 [Hmim][BF4] 10-3

M

F600-900 [Hmim][BF4] 10-2

M

F600-900 [Emim][Gly] 10-3

M

F600-900 [Emim][Gly] 10-2

M

(b)

 

57

Peq, bar

0.0 0.1 0.2 0.3 0.4

eq

, mm

ol g

-1

0.0

0.1

0.2

0.3

0.4 F600-900

F600-900 [Hmim][BF4] 10-3

M

F600-900 [Hmim][BF4] 10-2

M

F600-900 [Emim][Gly] 10-3

M

F600-900 [Emim][Gly] 10-2

M

(c)

Figure5.16CO2adsorptionisothermsonF600‐900rawandimpregnatedwith[Hmim][BF4]/[Emim][Gly]

ILsatC°=5.6×10‐3and2.2×10‐2Mandata)30,(b)50and(c)80°C

Aqualitativeanalysisoftheadsorptionisothermsdenotesthefollowingaspects:

At30°CF600‐900rawshowsthehighestCO2captureperformancesamongthetested

adsorbents in the whole Peq range investigated. Moreover, [Emim][Gly] sorbents

display higher eq values with respect to [Hmim][BF4] materials and adsorption

isothermsatdifferentimpregnationconditionsoverlapforeachionicliquidexamined.

At50°C[Emim][Gly] functionalizedsorbentsshowsimilaradsorptioncapacitieswith

respect to the raw material up to Peq=0.15 bar, while an improving of the parent

adsorbent capture performances are observed for higher pollutant concentrations;

once again, [Hmim][BF4] materials are characterized by worse adsorption

performanceswhichismoreevidentforthesampleimpregnatedatC°=2.2×10‐2M.

At 80°C the eq ranking is F600‐900 [Emim][Gly] 10‐3 M>F600‐900 raw≈F600‐900

[Hmim][BF4]10‐3M>F600‐900[Emim][Gly]10‐2M>F600‐900[Hmim][BF4]10‐2M.

Before providing a quantitative correlation between the sorbents microstructural

properties and their sorption capacities, it is fundamental to underline that many factors

contribute in determining a complex phenomenology in CO2 removal for the investigated

systems.Inparticular,inthecaseoftherawactivatedcarbonphysisorptionisthemainactive

mechanism for CO2 capture and its contribution decreases with temperature (Plaza et al.,

 

58

2007).Differently,dealingwith impregnatedsamples,theCO2capturedeterminedbytheIL

and the occlusion of the substrate pores act in opposite directions, the latter leading to a

reduction in the contribution of the rawmaterial adsorption. In addition, for both ILs the

increaseintemperatureisdetrimentalforthecaptureprocess:[Hmim][BF4]actsasaphysical

solvent towardsCO2 (solubility is typically equal to 0.05mol /mol at roomT andnear

atmosphericpressure),while[Emim][Gly] includesanamino‐groupinitsstructurewhichis

potentiallyabletoformcarbamatewithcarbondioxideviaareversibleexothermicreaction

following the same scheme described for commonly employed amine‐based solvents

(theoreticalcapturecapacityunderdryconditions0.5mol /mol )(Kimetal.,2005;Plaza

etal.,2007;Krumdiecketal.,2008;Zhangetal.2011;Kasaharaetal.,2012).Moreover, the

less sterically‐hindered [Emim][Gly] molecule makes the access of CO2 to its active site

(amino‐group) and in the sorbentporespotentially easier than in the caseof [Hmim][BF4].

Finally,itshouldbealsoconsideredthatthedispersionoftheactivephaseintothesubstrate

porescouldbealteredbothbytemperaturevariations(influencingforexampletheILsurface

tension and viscosity) and by interactions establishingwith CO2molecules (with a further

dependenceontheamountofpollutantadsorbed).

In light of the aforementioned observations, Table 5.7 reports a comparison of the

adsorptioncapacitiesobtainedfortheinvestigatedsorbentsundertypicalflue‐gasconditions

(CO215%byvol.,ω %)andthemainquantitativeparametersderivedfromporosimetricand

TGAanalyses(porevolumereductionandthespecificamountofionicliquidadsorbedonthe

substrate,seealsoTable5.5).

Table5.7Comparisonamong %valuesandthemainparametersderivedfromporosimetricandTGA

analysesforF600‐900rawandimpregnatedwith[Hmim][BF4]and[Emim][Gly]ILs

Sample

%[mmolg‐1] (V0‐raw‐V0‐impr)/V0‐raw

%mmolIL‐adsg‐1AC

(TGA)T=30°C T=50°C T=80°C

F600‐900 0.575 0.315 0.137

F600‐900[Hmim][BF4]10‐3M

0.437 0.264 0.139 4.9 0.015

F600‐900[Hmim][BF4]10‐2M

0.443 0.230 0.101 12.2 0.052

F600‐900[Emim][Gly]10‐3M 0.502 0.324 0.157 4.9 0.028

F600‐900[Emim][Gly]10‐2M 0.502 0.317 0.125 12.2 0.057

 

59

Results suggest that at T=30°C the pores blocking induced by the presence of the ILs

prevailsontheircontributiontoCO2capture,thusdeterminingareductionoftheadsorption

performancesof the functionalized sorbentswith respect to theparentmaterial.Moreover,

for each IL, the higher pore volume reduction experienced under more concentrated

impregnationconditioniscounterbalancedbythecontributiontoCO2capturedeterminedby

ahigheramountofILloaded.Thehigherω %valuesobtainedat30°Cfor[Emim][Gly]solids

with respect to [Hmim][BF4] ones could be explained considering that, even if at fixed

impregnation condition thepore clogging is the same forboth ILs, a greateramountof the

aminoacid‐basedionicliquidisadsorbedonthesubstrate(almostdoubleatC°=5.6×10‐3M)

and its capture capacity towards CO2 is one order magnitude higher than in the case of

[Hmim][BF4] (0.05and0.5mol /mol for [Hmim][BF4]and[Emim][Gly]respectively).At

50°Ctheporeblockingeffectonthefunctionalizedsorbentsadsorptioncapacitiesisgenerally

reduced(duetothedecreasingphysisorptionof therawmaterial)and it isbalancedbythe

active phase CO2‐capture contribution in the case of [Emim][Gly], while it persists for less

CO2‐affine[Hmim][BF4](toagreaterextentforthesampleobtainedundermoreconcentrated

condition for which the pore volume reduction is higher). Finally, at 80°C pore clogging

becomesnegligibleforF600‐900[Hmim][BF4]10‐3Msamplewhileitisstillimportantforthe

sample obtained at C°=2.2×10‐2 M. Differences between [Hmim][BF4] adsorbents could be

ascribedtoavery lowCO2solubility in the ILat this temperature(even if solubilitydataat

this temperaturearenotavailable in the literature), thusundermorediluted impregnation

conditiontheloweramountofILchargedandthehigherspreadingoftheILonthesupport

surface(inducedbyareductionofitssurfacetension)wouldmakeeasiertheaccesstoF600‐

900 active siteswith respect to F600‐900 [Hmim][BF4] 10‐2M sample. On the other hand,

[Emim][Gly]isabletoproduceanincreaseinF600‐900CO2capturecapacityonlyundermore

diluted impregnation condition (ω %is nearly 15% greater with respect to the parent

material), thanks to a good compromise between IL loading, CO2 affinity and pore

accessibility, while the pore volume reduction prevails for the sorbent impregnated at

C°=2.2×10‐2M.

The breakthrough curves obtained for F600‐900 raw and impregnated with

[Hmim][BF4]/[Emim][Gly] ILs at different temperatures and for a 15%CO2 gas stream are

reportedinFigure5.17((a)30°C,(b)50°Cand(c)80°C).

 

60

As a general consideration, it can be observed that, as expected, the increase in the

operatingtemperaturedeterminesfasteradsorptionkineticsforalltheimpregnatedsamples

with a shift of the breakthrough curves towards lower times as a consequence of higher

diffusion rates of CO2molecules in the sorbent pores.Moreover, the equilibrium times are

approximately 10‐ and 3‐times lower at 80°C with respect to 30°C for [Hmim][BF4] and

[Emim][Gly]sorbents,respectively.

t, s0 20 40 60 80 100 120

QC

O2ou

t (t)/

QC

O2in

, -

0.0

0.2

0.4

0.6

0.8

1.0

F600-900

F600-900 [Hmim][BF4] 10-3

M

F600-900 [Hmim][BF4] 10-2

M

F600-900 [Emim][Gly] 10-3

M

F600-900 [Emim][Gly] 10-2

M

(a)

 

t, s0 20 40 60 80

QC

O2ou

t (t)/

QC

O2in

, -

0.0

0.2

0.4

0.6

0.8

1.0

F600-900

F600-900 [Hmim][BF4] 10-3

M

F600-900 [Hmim][BF4] 10-2

M

F600-900 [Emim][Gly] 10-3

M

F600-900 [Emim][Gly] 10-2

M

(b)

 

61

t, s0 10 20 30 40

QC

O2ou

t (t)/

QC

O2in

, -

0.0

0.2

0.4

0.6

0.8

1.0

F600-900

F600-900 [Hmim][BF4] 10-3

M

F600-900 [Hmim][BF4] 10-2

M

F600-900 [Emim][Gly] 10-3

M

F600-900 [Emim][Gly] 10-2

M

(c)

 Figure5.17BreakthroughcurvesofCO2onF600‐900rawandimpregnatedwith

[Hmim][BF4]/[Emim][Gly]fora15%byvol.CO2gaseousstreamat(a)30,(b)50and(c)80°C

AcomparisonoftheadsorbentsdynamicperformancesisprovidedinTable5.8whichlists

breakpointtimestbandvalues(cf.Section5.2.1)obtainedfrom15%CO2adsorptiontests

at30,50and80°C.

Table5.8Comparisonbetweenandtbvaluesobtainedfromkineticadsorptiontestsat30,50and80°CforF600‐900rawand[Hmim][BF4]/[Emim][Gly]‐impregnated; =15%byvol.

Sample

[s]

tb[s]

T=30°C T=50°C T=80°C T=30°C T=50°C T=80°C

F600‐900 22 10 5 43 21 9

F600‐900[Hmim][BF4]10‐3M 16 9 6 31 20 10

F600‐900[Hmim][BF4]10‐2M 17 9 5 32 16 7

F600‐900[Emim][Gly]10‐3M 17 10 6 38 24 11

F600‐900[Emim][Gly]10‐2M 17 10 6 39 23 10

Results show that the ranking observed for breakpoint times at each temperature

generally coincides with the one shown in Table 5.7 in terms of equilibrium adsorption

capacities (1 s differences fall within the experimental error). In the case of F600‐900

 

62

[Emim][Gly]10‐2and10‐3M,tbvaluesat50°Caregreaterwithrespecttotheoneobtainedfor

F600‐900raw(21,24and23sforF600‐900raw,[Emim][Gly]10‐3and10‐2Mrespectively),

eveniftheadsorptioncapacitiesarepracticallyequivalent:obviously,eqvaluealsodepends

on the sorbent capture contribution for times greater than tb. The analysis of data

highlights that the only significant difference in dynamic performances can be observed at

30°CforwhichthisparameterisgreaterforF600‐900raw(sloweradsorptionkinetic)with

respect to the impregnated materials (which display similar values discrepancies are

practicallynegligibleathighertemperatures.Thispatterncouldbeexplainedconsideringthat

at lower temperaturedifferences in adsorption capacities of the impregnated sampleswith

respecttotheparentonearegreater,thusthehigherthenumberofactivesitestobeoccupied

(forF600‐900raw)theslowerthekinetics.Asthetemperaturerises,eqdifferencesreduce

andintraparticlediffusivityishigher,determiningsimilarCO2capturekinetics.Moreover,itis

hererecalledthattheporesizedistributionofthesubstrateisnotappreciablyinfluencedby

the impregnationprocess(cf.Section5.1)andtheonlyobservedeffect isareduction inthe

totalmicroporevolume(Vmesoispracticallyconstant):thisparameter(Vmicro)itselfisnotable

to explain kinetic differences between the raw and impregnated materials. Vice versa the

distributionofbothILsinthesorbentmicropores(widthoftheILfilm,positioningatthepore

mouth,completeorpartialporefillingetc.),whichcanbealsoinfluencedbythetemperature

andthecaptureprocessitself, islikelyafurtherfactoraffectingthesorptionkineticsforthe

impregnatedsamples.

 

63

5.2.3 CO2 adsorption tests onto N.RGC30 raw and functionalized with

[Hmim][BF4]/[Emim][Gly]

The thermodynamic behaviour of N.RGC30 raw and impregnated with

[Hmim][BF4]/[Emim][Gly] ILs in the CO2 capture process is illustrated in Figure 5.18 for

adsorptiontemperaturesof(a)30,(b)50and(c)80°C.

Peq, bar

0.0 0.1 0.2 0.3 0.4

eq

, mm

ol g

-1

0.0

0.2

0.4

0.6

0.8

1.0

N.RGC30

N.RGC30 [Hmim][BF4] 10-3

M

N.RGC30 [Hmim][BF4] 10-2

M

N.RGC30 [Emim][Gly] 10-3

M

N.RGC30 [Emim][Gly] 10-2

M

(a)

Peq, bar

0.0 0.1 0.2 0.3 0.4

eq

, mm

ol g

-1

0.0

0.1

0.2

0.3

0.4

0.5

0.6

N.RGC30

N.RGC30 [Hmim][BF4] 10-3

M

N.RGC30 [Hmim][BF4] 10-2

M

N.RGC30 [Emim][Gly] 10-3

M

N.RGC30 [Emim][Gly] 10-2

M

(b)

 

64

Peq, bar

0.0 0.1 0.2 0.3 0.4

eq

, mm

ol g

-1

0.0

0.1

0.2

0.3

0.4N.RGC30

N.RGC30 [Hmim][BF4] 10-3

M

N.RGC30 [Hmim][BF4] 10-2

M

N.RGC30 [Emim][Gly] 10-3

M

N.RGC30 [Emim][Gly] 10-2

M

(c)

Figure5.18CO2adsorptionisothermsonN.RGC30rawandimpregnatedwith[Hmim][BF4]/[Emim][Gly]

ILsatC°=5.6×10‐3and2.2×10‐2Mandata)30,(b)50and(c)80°C

Theanalysisoftheadsorptionisothermsrevealsthecomplexeffectofthetemperatureon

CO2 adsorption performances of the investigated adsorbents and the main aspects can be

summarizedasfollows:

For an adsorption temperature of 30°C the eq ranking is N.RGC30 raw>N.RGC30

[Emim][Gly] 10‐3 M>N.RGC30 [Emim][Gly] 10‐2 M≈N.RGC30 [Hmim][BF4] 10‐3 M≈

N.RGC30[Hmim][BF4]10‐2M.

At 50°C N.RGC30 raw and impregnated with [Emim][Gly] under more diluted

impregnationconditionsexhibitthehighesteqvaluesandtheirisothermspractically

overlap up to Peq=0.15 bar, while for higher CO2 concentrations the active phase

determines an improvement of the adsorption capacity with respect to the parent

sorbent. The other functionalized materials show generally similar adsorption

performancesexceptN.RGC30impregnatedwith[Hmim][BF4]atC°=2.2×10‐2Mwhich

ischaracterizedbyworsecapturecapacitiesforPeq>0.15bar.

Finallyat80°C, [Emim][Gly] ionic liquid iseffective inamelioratingN.RGC30capture

capacities for all the investigated Peq‐range under more diluted impregnation

conditions,whileeqisslightlygreaterforN.RGC30[Emim][Gly]10‐2Mwithrespectto

therawactivatedcarbononlyforCO2equilibriumpartialpressuresgreaterthanthose

of a typical‐flue gas.Moreover, N.RGC30 [Hmim][BF4] 10‐2M is again theworst CO2

 

65

adsorbent, while the sample impregnated with the same IL at C°=5.6×10‐3 M is

equivalenttotherawmaterial.

Inordertoshedlightontheintertwiningamongtheimpregnationconditions,thesorbent

properties and their CO2 capture capacities, Table 5.9 reports a comparison of the main

parametersderivedfromTGAandporosimetricanalysesforN.RGC30rawandimpregnated

with [Hmim][BF4]/[Emim][Gly] ILs and their adsorption capacities obtained under typical

flue‐gasconditionsandforoperatingtemperaturesof30,50and80°C.

Table5.9Comparisonamong %valuesandthemainparametersderivedfromporosimetricandTGA

analysesforN.RGC30rawandimpregnatedwith[Hmim][BF4]and[Emim][Gly]ILs

Sample

%[mmolg‐1] (V0‐raw‐V0‐impr)/V0‐raw

%mmolIL‐adsg‐1AC

(TGA)T=30°C T=50°C T=80°C

N.RGC30 0.499 0.276 0.139

N.RGC30[Hmim][BF4]10‐3M

0.391 0.246 0.141 6 0.013

N.RGC30[Hmim][BF4]10‐2M

0.386 0.240 0.122 8 0.051

N.RGC30[Emim][Gly]10‐3M

0.426 0.275 0.163 2 0.027

N.RGC30[Emim][Gly]10‐2M

0.381 0.238 0.137 8 0.057

Experimentaldatahighlightthatat30°Cporeblockingistherulingfactorindetermininga

reductionofthe impregnatedmaterialsadsorptionperformanceswithrespecttotheparent

one. In particular, the comparable micropore volume reduction observed for N.RGC30

[Hmim][BF4] 10‐2M and 10‐3M andN.RGC30 [Emim][Gly] 10‐2Mdetermines similarω %

values forthesesamples.Conversely, the lowermicroporevolumereductiondeterminedby

[Emim][Gly]at10‐3M(2%)producesalessmarkedω %decreasewithrespecttotheparent

materialalsoforthepossibleaccessofCO2moleculestotheILactivesites,whichseemstobe

hindered in the case of higher IL loading. At 50°C the contribution of [Emim][Gly] active

phaseinCO2capturebalancestheporecloggingeffectonlyundermoredilutedimpregnation

conditionsmakingω %values similar for the functionalizedand rawactivated carbon; for

the other impregnated adsorbents the considerations expressed for an adsorption

temperature of 30°C still hold at 50°C, even if differences in capture performances with

respecttotheparentmaterialreduceforthelowercontributionofphysisorptionofthelatter

 

66

as the temperature rises. Finally, at 80°C the pore blocking is negligible for N.RGC30

[Hmim][BF4] 10‐3M andN.RGC30 [Emim][Gly] 10‐2M determining equivalentω % values

with respect to the parent material, while this effect is still important for N.RGC30

[Hmim][BF4]10‐2M.InthecaseofN.RGC30functionalizedwith[Emim][Gly]atC°=5.6×10‐3M,

theactivephaseisabletodetermineanincreaseinCO2adsorptioncapacitywhencompared

to the raw substrate (ω % is approximately 16% greater) thanks to stronger chemical

interactionsbetweenacidicCO2moleculesandbasicaminegroupsandtheeasieraccessibility

inthesorbentporesdeterminedbythelowILloading.

TheconsiderationsconcerningthecorrelationsbetweensorbentpropertiesandtheirCO2

capturecapacitiesunder typical flue‐gasconditionscanbegenerallyextended to thewhole

Peq‐range,evenifat50and80°CandforCO2equilibriumpartialpressuresgreaterthan0.15

bar, theprocessdrivingforceexertsa furtherrole.Forexample,at50°Cthecontributionin

CO2captureexertedby[Emim][Gly]undermoredilutedconditionsisprevailingwithrespect

toporeblockingforPeq>0.15bar(producinganincreaseinadsorptioncapacitieswithrespect

toN.RGC30raw),while theybalance for lowerprocessdriving force;similarargumentsare

alsovalidforN.RGC30[Emim][Gly]10‐2Mtestedatanadsorptiontemperatureof80°C.

The comparison of the dynamic performances in CO2 capture process for both raw and

ILs‐functionalizedN.RGC30 sorbents under typical flue‐gas conditions (CO2 15%by vol.) is

reportedinFigure5.19foroperatingtemperaturesof(a)30,(b)50and(c)80°C.Table5.10

liststhevaluesofbreakpointtimetbandofthecharacteristicparameterderivedfromthe

kineticprofiles.

Resultsevidence,onceagain,thatathighertemperaturestheadsorptionkineticsisfaster

foralltheinvestigatedsorbentswithamonotonicdecreaseofcharacteristicbreakpointand

saturation times. For example, N.RGC30 [Hmim][BF4] 10‐3 M experiences a saturation 3.5‐

timesfasterat80°Cwithrespectto30°C(t*=165and47sat30and80°C),whileinthecaseof

N.RGC30 [Emim][Gly] 10‐3 M, t* is 1.6‐times lower for the same temperature increase.

Moreovertherankingof tbvaluesobtainedfor thedifferent investigatedsamples is ingood

agreement with the one derived in terms of CO2 adsorption capacities (cf. Tables 5.9 and

5.10). The comparison ofdata reveals thatmain kinetic differences can be identified at

30°C forwhich thisparameters ishigher for rawN.RGC30with respect to the impregnated

onesindicatingslowercapturekineticsforthissample.Moreover,N.RGC30[Emim][Gly]10‐3

Mdisplaysaslightlyhighervalueofwithrespecttotheotherfunctionalizedmaterialsand

thiscanbeascribedtoitslowerdifferenceinω %whencomparedtorawmaterial(cf.Table

 

67

5.9).Athighertemperaturestheintrinsicincreaseinintraparticlediffusivityandthereduced

differences in adsorption capacity contribute in determining similar capture kinetics for all

samples.Finally, theway inwhich the IL isdispersed in the substratemicropores (without

affectingremarkablythesubstratePSD,cf.Section5.1)shouldalsocontributeindetermining

different adsorptionkinetics for the impregnated sampleswith respect to the rawmaterial

especiallyat30°C.

t, s0 20 40 60 80 100 120

QC

O2ou

t (t)/

QC

O2in

, -

0.0

0.2

0.4

0.6

0.8

1.0

N.RGC30

N.RGC30 [Hmim][BF4] 10-3

M

N.RGC30 [Hmim][BF4] 10-2

M

N.RGC30 [Emim][Gly] 10-3

M

N.RGC30 [Emim][Gly] 10-2

M

(a)

 

t, s0 20 40 60 80

QC

O2ou

t (t)/

QC

O2in

, -

0.0

0.2

0.4

0.6

0.8

1.0

N.RGC30

N.RGC30 [Hmim][BF4] 10-3

M

N.RGC30 [Hmim][BF4] 10-2

M

N.RGC30 [Emim][Gly] 10-3

M

N.RGC30 [Emim][Gly] 10-2

M

(b)

 

 

68

t, s0 10 20 30 40

QC

O2ou

t (t)/

QC

O2in

, -

0.0

0.2

0.4

0.6

0.8

1.0

N.RGC30

N.RGC30 [Hmim][BF4] 10-3

M

N.RGC30 [Hmim][BF4] 10-2

M

N.RGC30 [Emim][Gly] 10-3

M

N.RGC30 [Emim][Gly] 10-2

M

(c)

 Figure5.19BreakthroughcurvesofCO2onN.RGC30rawandimpregnatedwith

[Hmim][BF4]/[Emim][Gly]fora15%byvol.CO2gaseousstreamat(a)30,(b)50and(c)80°C

Table5.10Comparisonbetweenandtbvaluesobtainedfromkineticadsorptiontestsat30,50and80°CforN.RGC30rawand[Hmim][BF4]/[Emim][Gly]‐impregnated; =15%byvol.

Sample

[s]

tb[s]

T=30°C T=50°C T=80°C T=30°C T=50°C T=80°C

N.RGC30 15 8 5 32 18 8

N.RGC30[Hmim][BF4]10‐3M 11 8 6 25 15 8

N.RGC30[Hmim][BF4]10‐2M 11 8 5 25 15 7

N.RGC30[Emim][Gly]10‐3M 13 8 5 27 17 10

N.RGC30[Emim][Gly]10‐2M 11 8 6 25 14 7

 

69

5.2.4 IntertwiningamongsolidspropertiesandCO2captureperformances

Table5.11schematicallysummarizes themainresultsobtained in thiswork in termsof

bothsorbentsCO2captureperformancesundertypicalflue‐gasconditions(CO215%byvol.)

andmainparametersderivedfromTGAandporosimetricanalyses.

Asageneralcomment, itcanbeobservedthatforbothactivatedcarbonsporesblocking

induced by the presence of the ILs is the prevailing effect at 30°C which determines a

worseningofCO2 captureperformancesof each functionalizedmaterialwith respect to the

parentone,duetothemaincontributionofphysisorptionforthelatteratlowertemperatures.

Moreover, the impregnation of F600‐900 and N.RGC30 with [Hmim][BF4] IL (a physical

solvent for CO2) appears to be not suitable (at least under the investigated experimental

conditions) for flue‐gas treatment, because the IL contribution to CO2 capture does not

counterbalancethereductionofthesubstrateadsorptionperformancesinducedbythepores

obstruction,alsoathighertemperatures.Ontheotherhand,theuseofmore‐chemicallyCO2

affine [Emim][Gly] IL can be potentially apt to ameliorate the parent carbon sorptive

performancestowardsCO2at80°C,evenifatthecurrentstageoftheresearchtheobtained

adsorptioncapacitiesarestilltoolowforatechnicalapplicabilityofthesematerialsinlarge

scaleCO2captureprocesses.

The main aspects derived from a deeper analysis of the effects of the impregnation

conditions for each IL on the activated carbons capture performances are discussed in the

following.

For [Hmim][BF4], at 30°C both the impregnation conditions determine a similar

percentagereductionoftherawmaterialsω %values(nearly23%)despiteaslightlyhigher

pore volume reduction for F600‐900 at C°=2.2×10‐2 M. Moreover, for both F600‐900 and

N.RGC30 functionalized with [Hmim][BF4] under more diluted impregnation conditions

differencesinadsorptioncapacitywithrespecttorawsubstratesreduceasthetemperature

risesandtheporescloggingeffectbecomesnegligibleat80°C.TheimpregnationofF600‐900

with[Hmim][BF4]atC°=2.2×10‐2Mdeterminesapracticallyconstantpercentagereductionof

ω %valueswithrespecttotherawmaterialasthetemperatureincreases,whileinthecase

of N.RGC30 impregnatedwith the same IL and adopting the same initial concentration the

poreblockingeffectontheadsorptionperformancesislessrelevantat80°C.Inthiscontext,

differences between F600‐900 and N.RGC30 impregnated at a higher [Hmim][BF4]

concentration could be related to the combined effects of a larger contribution of wider

microporesforN.RGC30andtemperatureonamorefavourableILdispersioninthesorbent

 

70

pores.ThismeansthatforN.RGC30,asthetemperatureincreases,abetterdistributionofthe

ILonthesubstrateporessurfacemakestheaccessofCO2moleculestotheadsorbentactive

siteseasierthaninthecaseofF600‐900.

For both F600‐900 and N.RGC30 impregnated with [Emim][Gly] at C°=5.6×10‐3 M

comparableamountsofILadsorbedproducesimilareffectsontheadsorptioncapacitywhen

the temperature increases. In particular, each functionalized material displays similar

adsorptioncapacitywithrespecttotherawsubstrateat50°C,whileat80°Ctheactivephase

is able to ameliorate the parent activated carbons capture performances thanks to a good

compromisebetweenILloadingandporeaccessibility(i.e.alowporevolumereduction).In

the case of higher [Emim][Gly] loading, N.RGC30 functionalized material displays similar

adsorptioncapacitywithrespecttotherawoneat80°C,whileF600‐900[Emim][Gly]10‐2M

isequivalent to therawmaterialat50°Candslightlyworseat80°C.Thispatterncannotbe

directlyjustifiedonthebasisofthedifferenttexturalpropertiesoftheinvestigatedactivated

carbons,butshouldbesomehowrelatedtoaverycomplexeffectof thetemperatureonthe

distribution of this IL into the sorbents pores (hardly explicable with the information

currentlyavailable).

As an additional remark, it should be considered that it is a really hard task trying to

discriminatethecontributiontoCO2capturedeterminedbyeachILwithrespect to theone

exertedbytheparentmaterial:abetterunderstandingoftheeffectofthetemperatureonthe

ILdistributiononthesubstrateporeswillbehelpfulinthisdirection.

Tothebestofourknowledge,similarfindingsconcerningtheeffectsoftheimpregnation

conditionsonCO2sorptivepropertiesofmicroporousactivatedcarbons functionalizedwith

ILshavenotyetbeenreportedintheliterature.Nevertheless,itisworthyofmentioningthe

workofPlazaetal.(2007)inwhichanactivatedcarbon(NoritCGPSuper)impregnatedwith

differentamine‐based compoundswasemployed forCO2 capture.Authorsobserved that at

room temperature the activephases arenot able to exert their peculiar captureproperties

becauseofaprevailingporeblockingeffect;athighertemperature,thisphenomenonisstill

prevailing but to a less extent because of the reduced contribution of physisorption of the

parentmaterial.Finally,theyevidencedanincreaseoftherawmaterialcaptureperformances

onlyfortheactivatedcarbonimpregnatedwithdiethylentriamine(withthehighestnitrogen

content)andtestedattemperaturesgreaterthan60°C.

 

71

Table5.11Comparisonamong %valuesandthemainparameters

derivedfromporosimetricandTGAanalysesforthesorbentsinvestigatedinthiswork

Sample

%[mmolg‐1] V0

[cm3g‐1]Vn

[cm3g‐1](V0‐raw‐V0‐impr)/V0‐raw

%mmolIL‐adsg‐1AC

(TGA)T=30°C T=50°C T=80°C

F600‐900 0.575 0.315 0.137 0.41 0.32

F600‐900[Hmim][BF4]10‐3M 0.437 0.264 0.139 0.39 n.a.† 4.9 0.015

F600‐900[Hmim][BF4]10‐2M 0.443 0.230 0.101 0.36 n.a.† 12.2 0.052

F600‐900[Emim][Gly]10‐3M 0.502 0.324 0.157 0.39 n.a.† 4.9 0.028

F600‐900[Emim][Gly]10‐2M 0.502 0.317 0.125 0.36 n.a.† 12.2 0.057

N.RGC30 0.499 0.276 0.139 0.50 0.32

N.RGC30[Hmim][BF4]10‐3M 0.391 0.246 0.141 0.47 n.a.† 6 0.013

N.RGC30[Hmim][BF4]10‐2M 0.386 0.240 0.122 0.46 n.a.† 8 0.051

N.RGC30[Emim][Gly]10‐3M 0.426 0.275 0.163 0.49 n.a.† 2 0.027

N.RGC30[Emim][Gly]10‐2M 0.381 0.238 0.137 0.46 n.a.† 8 0.057

†notavailable

 

72

Table 5.12 reports a comparison of the adsorbents dynamic performances at different

temperatures(fora15%CO2gasstream)intermsofanalysistogetherwiththesolidsmain

microstructuralparameters.Itisonceagainrecalledthatthedifferentsorbentdosesusedfor

adsorptiontests(15and13gforF600‐900andN.RGC30sorbentsseries,respectively)donot

allowtodefinekineticsdifferencesbetweenN.RGC30andF600‐900materialsonthebasisof

breakpoint times. Results underline thatmain differences can be observed at 30°C, where

generally higher values obtained for F600‐900 functionalized solids suggest slower

kinetics for these sorbents, while differences reduce at higher temperatures to become

practicallynegligibleat80°C.Onceagain,theoccurrenceofwidermicroporesandthegreater

contribution of mesopores for N.RGC30 adsorbents should determine faster CO2 diffusion

rates with respect to F600‐900 impregnated materials. Moreover, the generally lower

adsorptioncapacityobservedforN.RGC30impregnatedmaterialsat30°C(cf.Table5.11)also

contributes in determining faster kinetics. At higher temperatures, intraparticle diffusivity

increases (see also Section5.4.2) anddifferences in adsorption capacity reducemaking the

sorbentskineticallyequivalentinthecaptureprocessat80°C.

Table5.12Comparisonbetweenvaluesobtainedfromkineticadsorptiontestsat30,50and80°Candthemaintexturalparametersderivedforthesorbentsinvestigatedinthiswork; =15%byvol.

Sample

[s]

Microstructuralparameters

T=30°C T=50°C T=80°C V0[cm3g‐1]

Vn[cm3g‐1]

Vmeso[cm3g‐1]

F600‐900 22 10 5 0.41 0.32 0.17

F600‐900[Hmim][BF4]10‐3M 16 9 6 0.39 n.a.† 0.16

F600‐900[Hmim][BF4]10‐2M 17 9 5 0.36 n.a.† 0.16

F600‐900[Emim][Gly]10‐3M 17 10 6 0.39 n.a.† 0.16

F600‐900[Emim][Gly]10‐2M 17 10 6 0.36 n.a.† 0.16

N.RGC30 15 8 5 0.50 0.32 0.65

N.RGC30[Hmim][BF4]10‐3M 11 8 6 0.47 n.a.† 0.65

N.RGC30[Hmim][BF4]10‐2M 11 8 5 0.46 n.a.† 0.65

N.RGC30[Emim][Gly]10‐3M 13 8 5 0.49 n.a.† 0.65

N.RGC30[Emim][Gly]10‐2M 11 8 6 0.46 n.a.† 0.65

†notavailable

 

73

5.3 Adsorption/desorption cycles and regeneration experiments for

F600‐900raw

InthisSectionpreliminaryadsorption/desorptionandregenerationexperimentscarried

outonthesorbentwhichdisplayedthehighestadsorptioncapacityamongalltheinvestigated

experimentalconditions,namelyrawF600‐900testedat30°C(cf.Table5.11),areshown.

It is worthy to mention that exploratory CO2 adsorption/desorption tests over 3

consecutive cycles at 30°C have been performed on [Hmim][BF4] and [Emim][Gly]

functionalizedsorbents(notshownhereforthesakeofbrevity)andresultsshowedthatthe

processisreversibleforthesematerials.

InFigure5.20CO2equilibriumadsorptioncapacityofrawF600‐900isreportedover10

consecutiveadsorption/desorptioncyclesat30°Candfora15%CO2gasstream.

Cycle

0 1 2 3 4 5 6 7 8 9 10 11

eq

, mm

ol g

-1

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Figure5.20EquilibriumadsorptioncapacityofrawF600‐900obtainedover10adsorption/desorption

cyclesat30°C; =15%byvol.

Resultsindicatethatthevalueofeqispracticallyconstantuponthenumberofcyclesand

consequently, F600‐900 can be completely regenerated (reversible adsorption). This

behaviourhasbeencommonlyobservedintheliteratureforactivatedcarbonsandisascribed

to the establishment of weak interactions between CO2molecules and the sorbent surface

active sites (physisorption) (Choi et al., 2009; Sayari et al., 2011). In comparison, sorbents

suchasthosecalciumoxide‐basedsufferfromarapiddegradationofCO2capturecapability

 

74

during multiple carbonation/calcination cycles caused by pore blocking and adsorbent

sintering, thus requiring a continuous make‐up of fresh sorbent (Abanades and Alvarez,

2003).

Figure5.21reportsthetime‐dependentCO2outletconcentrationprofilesobtainedduring

theregenerationofF600‐900activatedcarbon(previouslysaturatedwitha15%CO2gaseous

stream at 30°C) carried out with desorption temperatures of (a) 60 and (b) 100°C and

adoptingthreedifferentN2purgeflowratelevels(6.95×10‐3Ls‐1,1.11×10‐2Ls‐1,1.39×10‐2L

s‐1,evaluatedatT=20°CandP=1bar).

t, s

0 20 40 60 80 100 120 140 160

CC

O2ou

t (t),

% v

ol.

0

10

20

30

40QN2

des=1.39x10-2 L s

-1

QN2

des=1.11x10-2 L s

-1

QN2

des=6.95x10-3 L s

-1

(a)

 

 

75

t, s

0 20 40 60 80

CC

O2ou

t (t),

% v

ol.

0

10

20

30

40

50

60QN2

des=1.39x10-2 L s

-1

QN2

des=1.11x10-2 L s

-1

QN2

des=6.95x10-3 L s

-1

(b)

Figure5.21CO2concentrationprofilesobtainedfromregenerationexperimentsofrawF600‐900asafunctionofN2purgeflowrateat(a)60and(b)100°C(adsorptionstep: =15%byvol.,T=30°C)

Fromaqualitativepointofview,experimentalresultsevidencethatforbothregeneration

temperaturesalltheconcentrationprofilesreachamaximuminCO2outletconcentrationfor

relatively low desorption times, indicating that most of the adsorbed pollutant is quickly

removed from the solid. The regeneration curve becomes narrower as the N2 flow rate

increases, thus testifying a faster desorption process associated to an increase in the gas

velocitythroughthepacked‐bed(higherstrippingrate).However,theconcentrationprofiles

show quite long tails indicating that residual CO2 desorption takes place slowly when the

driving force decreases. Finally, for each investigated purge flow rate, an increment in the

desorptiontemperaturedeterminesapositiveeffectontheregenerationkinetic,whichcanbe

ascribed to a decrease of CO2 adsorption capacity (thermodynamic factor) coupled to an

increaseinthepollutantintraparticlediffusivity(kineticfactor).

Themainquantitativeparametersobtainedfrompost‐processingofregenerationprofiles

at different N2 purge flow rates ( des2NQ ) and desorption temperatures ( des

C60T and desC100T ),

according to the procedure described in Section 4.4.2, are listed in Table 5.13. It is here

recalled thati

2COC ,i

2COV and i2NV represent themean CO2 concentration in the desorbing

flow,thetotalCO2volumedesorbedandthepurgegasvolumefedtothecolumnuptotimeti,

respectively(computedforrecoverypercentagesi=50,70,80and90%).

 

76

Table5.13MainparametersobtainedfromregenerationexperimentsofF600‐900rawactivatedcarbon

des2NQ

[Ls‐1]

eq

[mmolg‐1]

des

[mmolg‐1]

t50

[s]

t70

[s]

t80

[s]

t90

[s]

502NV

[L]

702NV

[L]

802NV

[L]

902NV

[L]

50

2COC

[%vol.]

70

2COC

[%vol.]

80

2COC

[%vol.]

90

2COC

[%vol.]

desC60T

6.95x10‐3 0.559 0.569 33 50 62 80 0.229 0.348 0.431 0.556 29.5 28.6 27.0 24.5

1.11x10‐2 0.565 0.561 25 37 46 61 0.278 0.411 0.511 0.677 25.4 24.4 22.8 19.91.39x10‐2 0.566 0.554 17 25 32 44 0.236 0.348 0.445 0.612 30.4 28.6 26.8 23.1

desC100T

6.95x10‐3 0.567 0.557 15 22 28 37 0.104 0.153 0.195 0.257 47.0 46.4 43.7 401.11x10‐2 0.562 0.559 9 13 16 21 0.100 0.144 0.178 0.233 49.4 48.2 46.2 42.31.39x10‐2 0.560 0.555 7 11 13 17 0.097 0.153 0.181 0.236 47.7 47.0 45.7 42.2

 

77

It isworthy toobserve that thecomparisonbetweeneqanddesvalues,obtained from

the integration of the adsorption and desorption kinetic profiles respectively, allows the

verificationoftheCO2massbalanceforeachtest.Resultsconfirmthepositiveeffectsonthe

desorption kinetics induced by an increase of both des2NQ and desorption temperature. As a

matterof fact, ata fixeddesorption temperature, the timerequired toobtainadefinedCO2

recovery percentagemonotonically decreaseswith the N2 purge flow rate: for example, at

bothtemperatures,t50andt90approximatelydoublewhen des2NQ decreasesfrom1.39×10‐2Ls‐1

to 6.95×10‐3 L s‐1. Moreover, a similar trend for ti is observed when the desorption

temperatureincreases:for des2NQ =1.39×10‐2Ls‐1,t50andt90at des

C100T arenearly0.4‐timesthe

correspondingvaluesdeterminedat desC60T .Theresultsobtainedintermsof

i

2COC revealeven

more interesting features. For each des2NQ and for each temperature, when the desired CO2

recoverylevelisincreasedthemeanCO2concentrationinthedesorbingstreamdecreases.In

fact, as the recovery percentage increases the time required for desorption is higher, the

desorptionratedecreaseswithtimeduetoaloweringinthedrivingforceand,consequently,

agreaterpurgevolumeisrequiredtoremoveresidualadsorbedCO2,determiningadilution

effect.Differently,foreachtemperatureandforeachfixedCO2recoverylevel,i

2COC doesnot

substantiallyvarywhencomparingthevaluesobtainedat1.39×10‐2Ls‐1and6.95×10‐3Ls‐1.

In fact, from eq. (4.5) (Section 4.4.2), at fixed regeneration leveli

2COV is constant but i2NV

depends on both desorption time and purge flow rate ( ides

2N tQ , in eq. (4.5)): this product is

comparable at the highest and lowest purge flow rate investigated determining equivalent

concentrationlevels.FromdatareportedinTable5.13,itcanbealsohighlightedthatfor desC60T

arecoverylevelof90%adoptingthelowestN2flowrateallowstoobtainadesorbinggaswith

i

2COC ≈25%, while the mean CO2 concentration only slightly increases for a 50% solid

regeneration(i

2COC ≈30%),but thiswouldeventuallyproduceasignificantreduction in the

sorbentutilizationtime,ifasubsequentadsorptioncyclehastobeperformed.Aregeneration

temperature of 100°C appears to be the best operating condition to obtain higheri

2COC

values: amore concentrated gas is obtained by recovering 90% of the adsorbed CO2 with

respectto desC60T atthesameregenerationlevelandfor des

2NQ =6.95×10‐3Ls‐1andwithahalved

desorption time (t90=80 and 37 s for desC60T and

desC100T , respectively). Finally, regeneration

 

78

levelsof70%and80%,obtainedat100°Candfor des2NQ =1.11×10‐2Ls‐1,canbe identifiedas

theoptimalsolutionsamongthoseinvestigated,asacompromisebetweentheamountofCO2

recovered, its concentration in the desorbed gas (46‐48% by vol.) and time required to

performthatrecovery.

 

79

5.4 Adsorptionthermodynamicsandkineticsmodelling

In this Section, modelling analysis (using equations described in Chapter 3) of both

adsorptionisothermsandbreakthroughcurves,experimentallydeterminedforallthesolids

investigatedinthiswork,ispresented.

5.4.1 Thermodynamicaspects

Tables 5.14 and 5.15 list the main parameters obtained from the application of

Henry/Langmuir/Freundlich/Dubinin‐Radushkevich models to the adsorption isotherms

obtainedat30,50and80°CforF600‐900andN.RGC30bothrawandILs‐functionalized.

It ishighlighted that: i) for theLangmuirmodelasimultaneous fittingof theadsorption

isotherms at different temperatures was performed by imposing a unique max value

(temperature‐independent) according to the recommendation provided inRuthven (1984);

ii) for Dubinin‐Radushkevich model, the characteristic energy E was obtained by

simultaneously fitting equilibrium data at different temperatures according to eq. (3.5) (cf.

Chapter3),andforeachsolidthemicroporevolumedeterminedfromN2porosimetricdataat

‐196°C (cf. Chapter 5) was imposed as a constant parameter. Moreover for DRmodel, the

molar volume of the liquid adsorbate and the pseudo‐vapor pressure were computed

accordingtotheexpressionsreportedinDo(1998)andSahaetal.(2011).

Asageneralconsideration,itcanbeobservedthattheHenrymodelprovidesthepoorest

qualityofdata fitting forall sorbentsas testifiedby the lowestvaluesof thedetermination

coefficient R2 above all at 30°C; at 80°C the model provides a better fitting of adsorption

isothermsduetoamore lineartrendofadsorptiondata(cf.alsoFigure5.22).Similarly, the

DRmodelisnotadequatetointerpretsatisfactorilyadsorptionisothermsaboveallathigher

temperaturesandthisispossiblyrelatedtotheabsenceofadsorptiondataathighpressures

for a proper fitting of the characteristic curve (eq. (3.5)). Freundlich and Langmuirmodels

determine the best fitting of equilibrium data for all the investigated materials and

experimentalconditionsastestifiedbythehighestR2values(practicallyunitaryforboth).In

particular, for each sorbent KL and KF values decrease with temperature, due to the

exothermicnatureof the adsorptionprocess (Ruthven, 1984;Do, 1998). It is interesting to

observethatinallthecases,thevaluesoftheFreundlichheterogeneityparameter1/ndonot

differ toomuch fromunity,whichclearly indicates that thesorbent surfacesarepractically

energetically homogeneous in the CO2 capture process, also in the case of functionalized

sorbents(videinfra)(Do,1998).

 

80

Finally, in thecaseofFreundlichmodel it ispossible toobserveageneralbetteragreement

betweentherankingderived intermsof theaffinityparameter(KF)andthoseobservedfor

the adsorption capacities of the investigated sorbentswith respect to a comparison on the

basisoftheLangmuirconstantKL.Forinstance,whencomparingthevaluesofthisparameter

fortherawactivatedcarbons,weobtainKF=2.45and2.03at30°Cand0.80and0.81at80°C

for F600‐900 and N.RGC30 respectively, which agrees with the general trend of the

adsorption isotherms with temperature described in Section 5.2.1 (higher CO2 capture

capacitiesforF600‐900at30°Cwhileequivalentadsorptionperformancesat80°C).

As an example, Figure 5.22 reports the comparison between experimental adsorption

isothermsandthermodynamicmodelspredictionsforF600‐900rawat(a)30,(b)50and(c)

80°C.

Peq, bar

0.0 0.1 0.2 0.3 0.4

eqm

mol

g-1

0.0

0.2

0.4

0.6

0.8

1.0

exp.HenryLangmuirFreundlichDubinin-Radushkevich

(a)

 

81

Peq, bar

0.0 0.1 0.2 0.3 0.4

eqm

mol

g-1

0.0

0.1

0.2

0.3

0.4

0.5

0.6

exp.HenryLangmuirFreundlichDubinin-Radushkevich

(b)

Peq, bar

0.0 0.1 0.2 0.3 0.4

eqm

mol

g-1

0.00

0.05

0.10

0.15

0.20

0.25

0.30

exp.HenryLangmuirFreundlichDubinin-Radushkevich

(c)

Figure5.22Comparisonbetweenexperimentaladsorptionisotherms(symbols)andHenry,Langmuir,FreundlichandDubinin‐Radushkevichmodels(lines)forF600‐900rawat(a)30,(b)50and(c)80°C

 

82

Table5.14MainparametersofHenry,Langmuir,FreundlichandDubinin‐RadushkevichmodelsforCO2adsorptionontoF600‐900rawandimpregnatedwith[Hmim][BF4]/[Emim][Gly]ILsat30,50and80°C

F600‐900 F600‐900[Hmim][BF4]10‐3M

F600‐900[Hmim][BF4]10‐2M

F600‐900[Emim][Gly]10‐3M

F600‐900[Emim][Gly]10‐2M

Henry

KH[mmolg‐1bar‐1]

T=30°C 3.18 2.38 2.35 2.95 2.87T=50°C 1.76 1.63 1.38 1.91 1.89T=80°C 0.88 0.92 0.66 1.01 0.83

R2T=30°C 0.922 0.969 0.970 0.966 0.951T=50°C 0.965 0.989 0.982 0.977 0.983T=80°C 0.990 0.994 0.994 0.999 0.996

Langmuir

KL[bar‐1]

T=30°C 2.07 1.31 1.8 1.28 1.49T=50°C 0.91 0.79 0.87 0.75 0.84T=80°C 0.40 0.41 0.37 0.34 0.32

max[mmolg‐1]

2.51 2.54 2.01 3.19 2.81

R2T=30°C 0.998 0.998 0.999 0.998 0.999T=50°C 0.997 0.999 0.999 0.999 0.999T=80°C 0.998 0.997 0.999 0.999 0.998

Freundlich

KF[mmolg‐1bar‐1/n]

T=30°C 2.45 1.95 1.85 2.45 2.32T=50°C 1.46 1.46 1.20 1.62 1.65T=80°C 0.80 0.88 0.61 1.00 0.78

1/nT=30°C 0.76 0.82 0.78 0.83 0.81T=50°C 0.83 0.90 0.88 0.86 0.88T=80°C 0.91 0.95 0.93 0.99 0.95

R2T=30°C 0.999 0.999 0.998 0.996 0.999T=50°C 0.999 0.999 0.999 0.999 0.999T=80°C 0.999 0.997 0.999 0.997 0.998

DubininRadushkevich

max[mmolg‐1]

T=30°C 8.85 8.42 7.77 8.42 7.77T=50°C 8.42 8.01 7.39 8.01 7.39T=80°C 7.81 7.43 6.86 7.43 6.86

E[kJmol‐1]

9.36 9.05 9.09 9.36 9.41

R2T=30°C 0.978 0.976 0.972 0.987 0.990T=50°C 0.975 0.967 0.977 0.975 0.980T=80°C 0.975 0.686 0.898 0.895 0.945

 

83

Table5.15MainparametersofHenry,Langmuir,FreundlichandDubinin‐RadushkevichmodelsforCO2adsorptionontoN.RGC30rawandimpregnatedwith[Hmim][BF4]/[Emim][Gly]ILsat30,50and80°C

N.RGC30N.RGC30

[Hmim][BF4]10‐3MN.RGC30

[Hmim][BF4]10‐2MN.RGC30

[Emim][Gly]10‐3MN.RGC30

[Emim][Gly]10‐2M

Henry

KH[mmolg‐1bar‐1]

T=30°C 2.64 2.20 2.12 2.48 2.18T=50°C 1.68 1.56 1.48 1.78 1.57T=80°C 0.88 0.89 0.75 1.05 0.95

R2T=30°C 0.921 0.964 0.944 0.967 0.970T=50°C 0.981 0.995 0.991 0.996 0.998T=80°C 0.990 0.997 0.992 0.998 0.996

Langmuir

KL[bar‐1]

T=30°C 2.04 1.20 1.60 1.06 0.89T=50°C 1.04 0.74 0.94 0.67 0.57T=80°C 0.48 0.39 0.43 0.37 0.32

max[mmolg‐1]

2.10 2.52 1.98 3.14 3.15

R2T=30°C 0.999 0.999 0.999 0.999 0.997T=50°C 0.999 0.999 0.999 0.997 0.995T=80°C 0.998 0.999 0.999 0.999 0.992

Freundlich

KF[mmolg‐1bar‐1/n]

T=30°C 2.03 1.82 1.68 2.07 1.83T=50°C 1.45 1.44 1.33 1.70 1.53T=80°C 0.81 0.84 0.68 1.01 1.03

1/nT=30°C 0.76 0.83 0.79 0.83 0.84T=50°C 0.87 0.93 0.91 0.95 0.98T=80°C 0.92 0.95 0.91 0.96 0.98

R2T=30°C 0.998 0.999 0.999 0.999 0.998T=50°C 0.999 0.999 0.999 0.999 0.998T=80°C 0.999 0.999 0.999 0.999 0.992

DubininRadushkevich

max[mmolg‐1]

T=30°C 10.78 10.15 9.93 10.58 9.93T=50°C 10.27 9.65 9.45 10.06 9.45T=80°C 9.53 8.96 8.76 9.34 8.76

E[kJmol‐1] 8.84 8.67 8.64 8.80 8.67

R2T=30°C 0.931 0.917 0.910 0.925 0.918T=50°C 0.928 0.942 0.933 0.955 0.955T=80°C 0.730 0.501 0.616 0.508 0.534

 

84

The energetic aspects of the adsorption phenomena for the investigated systems were

evaluatedbycomputingtheisostericheatofadsorptionqstasafunctionofthespecificloading

bymeans of theClausius‐Clapeyron equation applied to experimental adsorptiondata at

differenttemperatures(cf.eq.(3.8)inSection3.1).Theqstvs.trendsarereportedinFigure

5.23forboth(a)F600‐900basedand(b)N.RGC30materials.

, mmol g-1

0.00 0.05 0.10 0.15 0.20 0.25 0.30

q st

kJ

mol

-1

0

5

10

15

20

25

30

35

F600-900

F600-900 [Hmim][BF4] 10-3 M

F600-900 [Hmim][BF4] 10-2 M

F600-900 [Emim][Gly] 10-3 M

F600-900 [Emim][Gly] 10-2 M

(a)

, mmol g-1

0.00 0.05 0.10 0.15 0.20 0.25 0.30

qstk

J m

ol-1

0

5

10

15

20

25

30

35

N.RGC30

N.RGC30 [Hmim][BF4] 10-3 M

N.RGC30 [Hmim][BF4] 10-2 M

N.RGC30 [Emim][Gly] 10-3 M

N.RGC30 [Emim][Gly] 10-2 M

(b)

Figure5.23Isostericheatofadsorptionasafunctionofthespecificloadingfor(a)F600‐900and(b)

N.RGC30adsorbentsbothrawandfunctionalizedwith[Hmim][BF4]/[Emim][Gly]ILs

 

85

Asitcanbeobserved,theisostericheatofadsorptionispracticallyconstantwithloading

foralltheinvestigatedadsorbents,indicatingthatthesurfacesareenergeticallyhomogeneous

towardsCO2capture(Chakrabortyetal.,2006), inagreementwiththeobservationsderived

from thermodynamic modelling analysis. For impregnated materials, the low amounts of

activephasechargeddonotappreciablymodifytheoverallsurfaceenergetichomogeneityof

theparentsubstratethusdeterminingaconstantqstvaluewith loading. Inthiscontext, it is

hardatthecurrentstageoftheresearchtounderstandtheinterplayofthecontributionsof

the substrate active sites and the ones belonging to the ionic liquid to CO2 capture, and

consequently to explain the trends observed for the functionalized materials in terms of

interactionenergies.On theotherhand, themeanvaluesof the isostericheatofadsorption

obtainedfortherawactivatedcarbonsare28.8and25.5kJmol‐1forF600‐900andN.RGC30

respectively:thisconfirmsthatthepresenceofanarrowermicroporesizedistributionwitha

prevailingcontributionofverysmallporediameters(<10Å)observedforF600‐900produces

stronger interactionswithCO2moleculeswith respect toN.RGC30andconsequentlyhigher

adsorptionperformancesaboveallatlowertemperatures(cf.Section5.2.1).

 

86

5.4.2 Kineticaspects

The main kinetic parameters determined from the modelling analysis of breakthrough

curves (according to the theoretical frameworks described in Section 3.2) obtained at

differentoperatingtemperaturesandfora15%CO2gasstreamarereported inTables5.16

and5.17forF600‐900andN.RGC30sorbentsseries,respectively.Itishererecalledthatmass

and momentum balance equations were numerically solved with Aspen AdsimTM software

adoptingalineardrivingforceapproximation(LDF)forthemasstransferrate;moreoverthe

onlyfittingparameterofthemodelwasthemicroporediffusivityDmicro,whereastheexternal

and macropore (Knudsen) mass transfer coefficients were calculated from gas/solids

properties (cf. Section 3.2). Figure 5.24 reports, as an example, the comparison of the

experimental and theoretical breakthrough curves obtained for raw F600‐900 for a typical

flue‐gascompositionandat(a)30,(b)50and(c)80°C.

Asageneralconsideration,itishighlightedthattheFreundlichthermodynamicmodelwas

adopted as equilibrium isotherm in the rate of adsorption equation (3.13) (cf. Section 3.2)

becauseitsuppliedslightlybetternumericalsolutionswithrespecttotheLangmuirisotherm.

Noteworthy,thecomputedfixed‐bedPécletnumberwashigherthan100inallcases,thusit

was possible to consider a plug‐flow for all the investigated systems (Inglezakis and

Poulopoulos, 2006); in addition, the pressure drops across the fixed bed, calculated from

Ergun’sequation(3.14),werepracticallynegligible(orderofmagnitude10‐3bar).

Thegeneralfeaturesinferablefromacomparisonamongfilm,macroporeandmicropore

diffusionresistancesaresummarizedinthefollowing.

Microporediffusionmechanismrepresentstherate‐determiningstepoftheadsorption

process in almost all the investigated systems. As the temperature increases,

differences betweenmicropore andmacropore diffusion resistances tend to reduce,

andinanycasetheCO2transportthroughtheexternalfluidfilmisveryfast(negligible

film‐diffusion resistance). Dmicro is generally 2 or 3 orders of magnitude lower than

Dmacroat30°C,andwhiletheformerhasastrongdependenceontemperaturethelatter

onlyslightlyincreasesfrom30to80°C.

When comparing each raw activated carbon with the corresponding functionalized

materials, it can be observed that micropore diffusion resistance is higher for the

formeranddifferencesdiminishwithtemperaturetobecomepracticallynegligibleat

80°C.Thisbehaviourshouldbeonceagainimputedtothealreadydescribedtrendsin

adsorptioncapacityandmicroporediffusivity togetherwith thepossible influenceof

 

87

the temperature on the IL distribution inside the sorbent pores in the case of

impregnated adsorbents (cf. also Sections 5.2.2 and 5.2.3). Finally, for N.RGC30

sorbents,thediffusionresistancesaregenerally lowerthanthoseobservedforF600‐

900 materials and differences reduce with temperature: this testifies again the

important role exerted by both the occurrence ofwidermicropores and the greater

contributionofmesoporesforN.RGC30adsorbentswithrespecttoF600‐900ones in

determining faster kinetics above all at lower temperatures. The observations here

providedconfirmtheresultsdiscussedintermsofanalysisinthepreviousSections.

 

88

Table5.16MainkineticparametersderivedfrommathematicalmodellingofbreakthroughcurvesforCO2adsorptionontoF600‐900rawandimpregnatedwith[Hmim][BF4]/[Emim][Gly]ILsat30,50and80°C; =15%byvol.

F600‐900

F600‐900[Hmim][BF4]10‐3M

F600‐900[Hmim][BF4]10‐2M

F600‐900[Emim][Gly]10‐3M

F600‐900[Emim][Gly]10‐2M

Filmdiffusion

[s]

T=30°C 1.37x10‐3 1.37x10‐3 1.37x10‐3 1.37x10‐3 1.37x10‐3

T=50°C 1.23x10‐3 1.23x10‐3 1.23x10‐3 1.23x10‐3 1.23x10‐3

T=80°C 1.08x10‐3 1.08x10‐3 1.08x10‐3 1.08x10‐3 1.08x10‐3

Macroporediffusion

[s]

T=30°C 0.15 0.16 0.16 0.16 0.16

T=50°C 0.15 0.15 0.15 0.15 0.15

T=80°C 0.14 0.14 0.15 0.14 0.15

Dmacro

[m2s‐1]

T=30°C 1.13x10‐7 1.13x10‐7 1.13x10‐7 1.13x10‐7 1.13x10‐7

T=50°C 1.17x10‐7 1.17x10‐7 1.17x10‐7 1.17x10‐7 1.17x10‐7

T=80°C 1.22x10‐7 1.22x10‐7 1.22x10‐7 1.22x10‐7 1.22x10‐7

Microporediffusion

 

[s]

T=30°C 3.82 1.10 1.21 1.22 1.72

T=50°C 1.08 0.24 0.60 1.20 0.65

T=80°C 0.34 0.22 0.31 0.05 0.40

Dmicro

[m2s‐1] 

T=30°C 1.60x10‐10 8.52x10‐10 7.32x10‐10 6.19x10‐10 4.24x10‐10

T=50°C 1.15x10‐9 6.26x10‐9 2.75x10‐9 9.76x10‐10 1.79x10‐9

T=80°C 7.13x10‐9 1.15x10‐8 1.12x10‐8 5.26x10‐8 7.06x10‐9

 

89

Table5.17MainkineticparametersderivedfrommathematicalmodellingofbreakthroughcurvesforCO2adsorptionontoN.RGC30rawandimpregnatedwith[Hmim][BF4]/[Emim][Gly]ILsat30,50and80°C; =15%byvol.

N.RGC30 N.RGC30[Hmim][BF4]10‐3M

N.RGC30[Hmim][BF4]10‐2M

N.RGC30[Emim][Gly]10‐3M

N.RGC30[Emim][Gly]10‐2M

Filmdiffusion

[s]

T=30°C 1.60x10‐3 1.60x10‐3 1.60x10‐3 1.60x10‐3 1.60x10‐3

T=50°C 1.44x10‐3 1.44x10‐3 1.44x10‐3 1.44x10‐3 1.44x10‐3

T=80°C 1.26x10‐3 1.26x10‐3 1.26x10‐3 1.26x10‐3 1.26x10‐3

Macroporediffusion

[s]

T=30°C 0.09 0.10 0.10 0.09 0.10

T=50°C 0.09 0.10 0.10 0.09 0.10

T=80°C 0.09 0.10 0.10 0.10 0.10

Dmacro

[m2s‐1]

T=30°C 2.10x10‐7 2.10x10‐7 2.10x10‐7 2.10x10‐7 2.10x10‐7

T=50°C 2.11x10‐7 2.11x10‐7 2.11x10‐7 2.11x10‐7 2.11x10‐7

T=80°C 2.12x10‐7 2.12x10‐7 2.12x10‐7 2.12x10‐7 2.12x10‐7

Microporediffusion

 

[s]

T=30°C 1.66 0.54 0.51 0.63 0.61

T=50°C 0.30 0.15 0.15 0.14 0.18

T=80°C 0.15 0.15 0.08 0.07 0.08

Dmicro

[m2s‐1] 

T=30°C 5.89x10‐10 9.34x10‐10 9.58x10‐10 6.95x10‐10 8.33x10‐10

T=50°C 2.41x10‐9 5.89x10‐9 5.57x10‐9 5.23x10‐9 4.46x10‐9

T=80°C 9.57x10‐9 2.15x10‐8 2.02x10‐8 1.75x10‐8 2.03x10‐8

 

90

t, s0 50 100 150 200 250 300

QC

O2ou

t (t)/

QC

O2in

, -

0.0

0.2

0.4

0.6

0.8

1.0

exp.LDF model

(a)

t, s0 20 40 60 80 100 120 140 160

QC

O2ou

t (t)/

QC

O2in

, -

0.0

0.2

0.4

0.6

0.8

1.0

exp.LDF model

(b)

 

91

t, s0 10 20 30 40 50 60 70 80

QC

O2ou

t (t)/

QC

O2in

, -

0.0

0.2

0.4

0.6

0.8

1.0

exp.LDF model

(c)

Figure5.24Comparisonbetweenexperimental(symbols)andtheoretical(lines)breakthroughcurves

obtainedforF600‐900rawat(a)30,(b)50and(c)80°C;CinletCO2=15%byvol.

 

92

CHAPTER6

CONCLUSIONSANDFUTUREPERSPECTIVES

Theneedofdevelopinghighperformanceandcost‐effectivepost‐combustionpurification

systems for CO2 capture has recently stimulated the research of new adsorbentsmaterials

with tailored microstructural and chemical properties. Contextually, the use of supported

ionic liquid phasematerials (SILP) is a very attractive but limitedly explored investigation

area. In this PhD project the effect of confining ionic liquids (ILs) into activated carbons

characterizedbydifferentporosimetricstructuresontheirCO2captureperformancesunder

typicalflue‐gasconditionshasbeeninvestigated.Theobtainedresultshaveprovidedadeeper

understandinginthisfieldwithrespecttotheavailablescientificliterature.

CO2 adsorption tests havebeen carried inmodel flue‐gas streamsonto two commercial

activated carbons, namely Filtrasorb 400 and Nuchar RGC30, both raw and functionalized

with either 1‐hexyl‐3‐methylimidazolium tetrafluoroborate [Hmim][BF4] or 1‐ethyl‐3‐

methylimidazolium glycine [Emim][Gly] ILs adopting different impregnation conditions

(5.6×10‐3and2.2×10‐2M).

Resultsobtainedforrawactivatedcarbonshaveconfirmed,inagreementwithpreviously

literature findings, that the presence of a narrower micropore size distribution with a

prevailingcontributionofverysmallporediameters(<10Å)observedforFiltrasorb400isa

keyfactorindetermininghigherCO2capturecapacitiesaboveallatlowtemperature(30°C);

ontheotherhand,thesorbentsareequivalentinthepurificationprocessat80°Cbecauseof

thereducedcontributionofphysisorptionathighertemperatures,asexpectedinexothermal

processes.Theseexperimentalevidenceshavebeenalsocorroboratedbythehighervalueof

the isosteric heat derived for Filtrasorb400 solid, testifying stronger interactionswithCO2

moleculeswithrespecttoNucharRGC30activatedcarbon.

Thermodynamic adsorption results onto [Hmim][BF4]‐functionalized sorbents suggest

that the impregnation of micro and micro‐mesoporous activated carbons with this IL (a

physical solvent for CO2) is not suitable for CO2 removal from flue‐gas (at least under the

investigatedimpregnationconditions),becausetheactivephasecontributioninthepollutant

captureisnotabletobalancethereductionofthesubstrateadsorptionperformancesinduced

by the pores obstruction, also at higher temperatures. Conversely, the functionalization of

both activated carbonswith amore CO2 chemically‐affine IL (i.e. [Emim][Gly]) undermore

diluted impregnation condition can ameliorate the parent carbons CO2 adsorption

 

93

performances at 80°C and for a typical 15% CO2 flue‐gas stream (while pores blocking is

dominant at 30°C) thanks to a good compromisebetween IL loading andpore accessibility

(low pore volume reduction). Nevertheless, the improvement in CO2 adsorption capacity

determined by [Emim][Gly] is still not adequate for the potential application of these

materialsinlarge‐scaleflue‐gaspurificationsystems.

Dynamic adsorption results on the investigated sorbents highlighted the important role

playedbybothagreatercontributionofmesoporesandthepresenceofwidermicroporesfor

Nuchar RGC30‐based materials in determining faster capture kinetics with respect to

Filtrasorb400sorbents, inparticularat low temperature.As the temperature increases the

reduced differences in adsorption capacity coupled with the increase in intraparticle

diffusivitymakethetwoclassesofsorbentskineticallyequivalentinthepurificationprocess.

Furthermore, modelling analysis of breakthrough curves allowed identifying micropore

diffusionastherate‐determiningstepofCO2adsorption,foralmostalltheanalysedsystems.

Preliminary regeneration studies on raw Filtrasorb 400, which displayed the highest

capture performances among all the adsorbents investigated, showed a complete

regenerability of this sorbent under multiple adsorption/desorption cycles. Moreover,

desorptionexperimentscarriedoutonFiltrasorb400(afterasolidsaturationwitha15%CO2

gasstreamat30°C)atdifferenttemperatures(60°Cand100°C)andN2flowrates(6.95×10‐3

L s‐1, 1.11×10‐2 L s‐1, 1.39×10‐2 L s‐1) evidenced that regeneration levels of 70 and 80%

obtained at 100°C and adopting a 1.11×10‐2 L s‐1 purge flow rate can be considered as the

optimal solutions among those investigated, being a fair compromise between CO2

concentrationinthedesorbingflow(46‐48%)andtimenecessarytoperformtherecovery.

As a general consideration, the results obtained in thiswork encourage future research

efforts in the field of porous activated carbons functionalizationwith amine‐based ILs as a

potentialroutetoimprovetheirCO2captureperformances,inparticularathightemperatures

atwhichtheparentmaterialcontributionisquitelowtoallowacost‐effectivetreatmentofa

real flue‐gas. In this context, the choice of activated carbons characterized by largermean

pore diameters (i.e.mainlymesoporous) could be apt to favour a higher dispersion of the

ionic liquid over the substrate surface avoiding the undesired pore clogging effect. The

synergistic collaborationwith research groups belonging to different fields of investigation

(ChemicalEngineering, InorganicandOrganicChemistry)couldbean importantstrategy to

fosterthedevelopmentofhighlyCO2‐affineSILPmaterialsandacceleratetheirapplicabilityin

post‐combustion systems, aiming at the following goals: i) synthesis of amine‐based ILs

characterized by reduced molecular sizes to minimize potential pore blocking; ii)

 

94

development of new functionalization protocols, e.g. covalent tethering of the IL to the

support surface which could be an interesting option for both minimizing the amount of

active phase used, with consequent economic benefits, and avoiding the continuous

redistributionoftheliquidphaseinsidethesorbentporesinducedbytemperaturevariations;

iii)adsorptiontestsinmulticomponentsystemsincludingthepresenceofNOx,SO2andwater

vapour; iv) dedicated tests using different reactor configurations (fluidized‐bed, circulating

fluidized beds etc.); v) regeneration studies aimed atmaximizing CO2 concentration in the

desorbing flow (for subsequent storage of the pollutant) while minimizing energy

requirements.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

95

Bibliography

Abanades J.C. and Alvarez D., Conversion limits in the reaction of CO2with lime. Energy&Fuels17(2003)308‐315.

Abanades J.C., Rubin E.S. and Anthony E.J., Sorbent cost and performance in CO2 capturesystems.Industrial&EngineeringChemistryResearch43(2004)3462‐3466.

AnJ.andRosiN.,TuningMOFCO2adsorptionpropertiesviacationexchange.JournaloftheAmericanChemicalSociety132(2010)5578‐5579.

ArouaM.K.,DaudW.M.A.W.,YinC.Y.andAdinataD.,Adsorptioncapacitiesofcarbondioxide,oxygen,nitrogenandmethaneoncarbonmolecularbasketderived frompolyethyleneimineimpregnation on microporous palm shell activated carbon. Separation and PurificationTechnology62(2008)609‐613.

BalsamoM.,DiNataleF.,ErtoA.,LanciaA.,MontagnaroF.andSantoroL.,Arsenateremovalfromsyntheticwastewaterbyadsorptionontoflyash.Desalination263(2010)58–63.

BalsamoM.,DiNataleF.,ErtoA.,LanciaA.,MontagnaroF.andSantoroL.,Gasificationofcoalcombustionashforitsreuseasadsorbent.Fuel106(2013)147‐151.

Bara J.E., Gabriel C.J., Hatakeyama E.S., Carlisle T.K., Lessmann S., Noble R.D. and Gin D.L.,Improving CO2 selectivity in polymerized room‐temperature ionic liquid gas separationmembranes through incorporation of polar substituents. Journal ofMembrane Science 321(2008)3‐7.

BatesE.D.,MaytonR.D.,NtaiI.andDavisJr. J.H.,CO2capturebyaTask‐SpecificIonicLiquid.JournaloftheAmericanChemicalSociety124(2002)926‐927.

BelmabkhoutY.andSayariA.,AdsorptionofCO2fromdrygasesonMCM‐41silicaatambienttemperature and high pressure. 2:Adsorption of CO2/N2, CO2/CH4 and CO2/H2 binarymixtures.ChemicalEngineeringScience64(2009)3729‐3735.

BerlierK.andFrèreM.,AdsorptionofCO2onactivatedcarbon:simultaneousdeterminationofintegralheatandisothermofadsorption. JournalofChemical&EngineeringData41(1996)1144‐1148.

BlameyJ.,AnthonyE.J.,WangJ.andFennelP.S.,ThecalciumloopingcycleforlargescaleCO2capture.ProgressinEnergyandCombustionScience36(2010)260‐279.

BoschettiA.,MontagnaroF.,RienzoC.andSantoroL.,Apreliminaryinvestigationontheuseoforganicionicliquidsasgreensolventsforacylationandoxidationreations.JournalofCleanerProduction15(2007)1797‐1805.

BourbigouH.O.,Magna L. andMorvanD., Ionic liquids and catalysis: Recent progress fromknowledgetoapplications.AppliedCatalysisA:General373(2010)1‐56.

 

96

Brúder P. and Svendsen H.F., Capacity and kinetics of solvents for post‐combustion CO2capture.EnergyProcedia23(2012)45‐54.

Chakraborty A., Saha B.B., Koyama S. and Ng K.C., On the thermodynamicmodeling of theisosteric heat of adsorption and comparisonwith experiments. Applied Physics Letters 89(2006)171901.

Chang F.Y., Chao K.J., Cheng H.H. and Tan C.S., Adsorption of CO2 onto amine‐graftedmesoporoussilicas.SeparationandPurificationTechnology70(2009)87‐95.

ChewT.L.,AhmadA.L.andBathiaS.,Orderedmesoporoussilica (OMS)asanadsorbentandmembraneforseparationofcarbondioxide(CO2).AdvancesinColloidandInterfaceScience153(2010)43‐57.

ChoiS.,DreseJ.H.andJonesC.W.,Adsorbentmaterialsforcarbondioxidecapturefromlargeanthropogenicpointsources.ChemSusChem2(2009)796‐854.

CoppolaA.,MontagnaroF.,SalatinoP.andScalaF.,AttritionoflimestoneduringfluidizedbedcalciumloopingcyclesforCO2capture.CombustionScienceandTechnology184(2012a)929‐941.

CoppolaA.MontagnaroF.,SalatinoP.andScalaF.,Fluidizedbedcalciumlooping:TheeffectofSO2 on sorbent attrition and CO2 capture capacity. Chemical Engineering Journal207‐208(2012b)445‐449.

CrosthwaiteJ.M.,MuldoonM.J.,DixonJ.N.K.,AndersonJ.LandBrenneckeJ.F.,Phasetransitionanddecompositiontemperatures,heatcapacitiesandviscositiesofpyridiniumionic liquids.JournalofChemicalThermodynamics37(2005)559‐568.

Delgado J.A., Uguina M.A., Sotelo J.L. and Ruíz B., Fixed‐bed adsorption of carbon dioxide‐helium, nitrogen‐helium and carbon dioxide‐nitrogen mixtures onto silicalite pellets.SeparationandPurificationTechnology49(2006)91‐100.

DevadasM.,RamalingamB.,RameshK.,TasrifM.andSharratP.,CO2capturebyadsorptiononmesoporous MCM‐68 solid sorbent materials. 21st International Symposium on ChemicalReactionEngineering,Philadelphia,USA(2010).

Ding Y. and Alpay E., Equilibria and kinetics of CO2 adsorption on hydrotalcite adsorbent.ChemicalEngineeringScience55(2000)3461‐3474.

DoD.D.,Adsorptionanalysis:equilibriaandkinetics.SeriesonChemicalEngineeringsvol.2,ImperialCollegePress(1998).

DoD.D.andWangK.,Anewmodelforthedescriptionofadsorptionkineticsinheterogeneousactivatedcarbon.Carbon36(1998)1539‐1554.

Dubinin M.M. and Radushkevich L.V., Equation of the characteristic curve of activatedcharcoal. Proceedings of the Academy of Sciences, Physical Chemistry Section, U.S.S.R55(1947)331‐333.

 

97

EddaoudiM.,KimJ.,RosiN.,VodakD.,WachterJ.,O’KeefeM.andYaghiO.,Systematicdesignof pore size and functionality in isoreticularMOFs an their application inmethane storage.Science295(2002)469‐472.

FigueroaJ.D.,FoutT.,PlasynskiS.,McIlvriedH.andSrivastavaR.D.,AdvancesinCO2capturetechnology‐The U.S. Department of Energy's Carbon Sequestration Program. InternationalJournalofGreenhouseGasControl2(2008)9‐20.

GarcíaS.,GilM.V.,MartinC.F.,PisJ.J.,RubieraF.andPevidaC.,Breakthroughadsorptionstudyof a commercial activated carbon for pre‐combustion CO2 capture. Chemical EngineeringJournal171(2011)549‐556.

GomesV.G.andYeeK.W.K.,Pressureswingadsorptionforcarbondioxidesequestrationfromexhaustgases.SeparationandPurificationTechnology28(2002)161‐171.

GrayM.L,SoongY.,ChampagneK.J.,PennlineH.,BaltrusJ.P.,StevensJr.R.W.,KhatriR.,ChuangS.S.C.andFilburnT.,Improvedimmobilizedcarbondioxidecapturesorbents.FuelProcessingTechnology86(2005)1449‐1455.

Hasib‐ur‐RahmanM., SiajM. and Larachi F., Ionic liquids forCO2 capture‐Development andprogress.ChemicalEngineeringandProcessing49(2010)313‐322.

HerzogH.,MeldonJ.andHattonA.,Advancedpost‐combustionCO2capture.ReportpreparedfortheCleanAirTaskForce(2009).

HughesR.W.,LuD.Y.,AnthonyE.J.andMacchiA.,Design,processsimulationandconstructionofanatmosphericdual fluidizedbedcombustionsystem for insituCO2captureusinghigh‐temperaturesorbents.FuelProcessingTechnology86(2005)1523‐1531.

Inglezakis V.J. and Poulopoulos S.G., Adsorption, Ion Exchange and Catalysis. Design ofoperationsandenvironmentalapplication.Elsevier(2006).

International Energy Agency (IEA), Energy Technology Perspectives(2010).http://www.iea.org/techno/etp/etp10/English.pdf.

JangH.T.,ParkY.K.,KoY.S.,LeeJ.Y.andMargandanB.,HighlysiliceousMCM‐48fromricehuskashforCO2adsorption.InternationalJournalofGreenhouseGasControl3(2009)545‐549.

KannicheM., Gros‐Bonnivard R., Jaud P., Valle‐Marcos J., Amann J.M. and Bouallou C., Pre‐combustion, post‐combustion and oxy‐combustion in thermal power plant for CO2 capture.AppliedThermalEngineering30(2010)53‐62.

KasaharaS.,IshigamiTandMatsuyamaH.,Aminoacidionicliquid‐basedfacilitatedtransportmembranesforCO2separation.ChemicalCommunications48(2012)6903‐6905.

Khalil S.H., Aroua M.K. and Daud W.M.A.W., Study on the improvement of the capacity ofamine‐impregnated commercial activated carbon beds for CO2 adsorbing. ChemicalEngineeringJournal183(2012)15‐20.

 

98

KimY.S.,ChoiW.Y.,JangJ.H.,YooK.P.andLeeC.S.,Solubilitymeasurementandpredictionofcarbondioxideinionicliquids.FluidPhaseEquilibria228‐229(2005)439‐445.

KittelJ.,IdemR.,GelowitzD.,TontiwachwuthikulP.,ParrainG.andBonneauA.,CorrosioninMEAunitsforCO2capture:pilotplantstudies.EnergyProcedia1(2009)791‐797.

KoldingH.,FehrmannR. andRiisagerA.,CO2Capture technologies:Current statusandnewdirectionsusing supported ionic liquidphase (SILP)absorbers. ScienceChinaChemistry55(2012)1648‐1656.

KrumdieckS.,WallaceJ.andCurnowO.,Compact, lowenergyCO2managementusingaminesolutioninapackedbubblecolumn.ChemicalEngineeringJournal135(2008)3‐9.

Krutyeva M., Grinberg F., Furtado F., Galvosas P., Kärger J., Silvestre‐Albero A., Sepulveda‐Escribano A., Silvestre‐Albero J. and Rodríguez‐Reinoso F., Characterization of carbonmaterialswiththehelpofNMRmethods.MicroporousandMesoporousMaterials120(2009)91‐97.

Lee Z.H., Lee T.K., Bhatia S. and Mohamed A.R., Post‐combustion carbon dioxide capture:Evolution towardsutilizationofnanomaterials.RenewableandSustainableEnergyReviews16(2012)2599‐2609.

Lemus J., Palomar J., Gilarranz M.A. and Rodriguez J.J., Characterization of supported ionicliquidphase (SILP)materials prepared fromdifferent supports.Adsorption17 (2011) 561‐571.

Leofanti G., Padovan M., Tozzola G. and Venturelli B., Surface area and pore texture ofcatalysts.CatalysisToday41(1998)207‐219.

Li F. and Fan L.S., Clean coal conversion processes–progress and challenges. Energy &EnvironmentalScience1(2008)248‐267.

Lu C., Bai H., Wu B., Su F. and Hwang J.F., Comparative study of CO2 capture by carbonnanotubes,activatedcarbons,andzeolites.Energy&Fuels22(2008)3050‐3056.

MarshH.andRodrίguez‐ReinosoF.,ActivatedCarbon.ElsevierScience&TechnologyBooks(2006).

McCabeW.L.,SmithJ.C.andHarriotP.,UnitOperationsofChemicalEngineering.McGraw‐Hill5thedition(1993).

Metz B., Davidson O., de Coninck H., LoosM. andMeyer L., International Panel on ClimateChange(IPCC)Carbondioxidecaptureandstorage.CambridgeUniversityPress(2005).

MontagnaroF.,SalatinoP.andScalaF.,The influenceof temperatureon limestonesulfationand attrition under fluidized bed combustion conditions. Experimental Thermal and FluidScience34(2010)352‐358.

 

99

Morlay C. and Joly J.P., Contribution to the textural characterization of Filtrasorb 400 andother commercial activatedcarbons commonlyused forwater treatment. JournalofPorousMaterials17(2010)535‐543.

MuhammadN.,ManZ.B.,BustamM.A.,MutalibM.I.A,WilfredC.D.andRafiqS.,Synthesisandthermophysical properties of low viscosity amino acid‐based ionic liquids. Journal ofChemical&EngineeringData56(2011)3157‐3162.

NaB.K.,KooK.K.,EumH.M.,LeeH.andSongH.K.,CO2recoveryfromfluegasbyPSAprocessusingactivatedcarbon.KoreanJournalofChemicalEngineering18(2001)220‐227.

Neimark A.V., Lin Y., Ravikovitch P.I and Thommes M., Quenched solid density functionaltheoryandporesizeanalysisofmicro‐mesoporouscarbons.Carbon47(2009)1617‐1628.

PatrickJ.W.,PorosityinCarbons:CharacterizationandApplications.HalstedPress(1995).

PerryR.H. andGreenD.W.,Perry'sChemicalEngineers'Handbook.McGrawHill, 7th edition(1997).

Plaza M.G., Pevida C., Arenillas A., Rubiera F. and Pis J.J., CO2 capture by adsorption withnitrogenenrichedcarbons.Fuel86(2007)2204‐2212.

Plaza M.G., Pevida C., Arias B., Fermoso J., CasaL M.D., Martin C.F., Rubiera F. and Pis J.J.,Developmentoflow‐costbiomass‐basedadsorbentsforpostcombustionCO2capture.Fuel88(2009)2442‐2447.

Plaza M.G., Pevida C., Pis J.J. and Rubiera F., Evaluation of the cyclic capacity of low‐costadsorbentforpost‐combustionCO2capture.EnergyProcedia4(2011)1228‐1234.

Polanyi M., Theories of the adsorption of gases. A general survey and some additionalremarks.TransactionoftheFaradaySociety28(1932)316‐333.

Rege S.U. and Yang R.T., A novel FTIR method for studying mixed gas adsorption at lowconcentrations:H2OandCO2onNaXzeolitesandγ‐alumina.ChemicalEngineeringScience56(2001)3781‐3796.

Roquerol J., Llewellyn P. and Rouquerol F., Is the BET equation applicable tomicroporousadsorbents?.StudiesinSurfaceScienceandCatalysis160(2007)49‐56.

RuthvenD.M.,Principlesofadsorptionandadsorptionprocesses.JohnWiley&Sons(1984).

Saha B.B., Jribi S., Koyama S. and El‐Sharkawy I.I., Carbon dioxide adsorption isotherms onactivatedcarbons.JournalofChemical&EngineeringData56(2011)1974‐1981.

SahaD.,BaoZ.,JiaF.andDengS.,AdsorptionofCO2,CH4,N2OandN2onMOF‐5,MOF‐177,andzeolite5A.EnvironmentalScience&Technology44(2010)1820‐1826.

Samanta A., Zhao A., Shimizu G.K.H., Sarkar P. and Gupta R., Post‐combustion CO2 captureusingsolidsorbents:Areview.Industrial&EngineeringChemistryResearch51(2012)1438‐1463.

 

100

Sayari A., Belmabkhout Y. and Serna‐Guerrero R., Flue gas treatment via CO2 adsorption.ChemicalEngineeringJournal171(2011)760‐774.

Serna‐Guerrero R. and Sayari A., Modeling adsorption of CO2 on amine‐functionalizedmesoporous silica. 2: Kinetics and breakthrough curves. Chemical Engineering Journal 161(2010)182‐190.

Shafeeyan M.S., Daud W.M.A.W., Houshmand A. and Shamiri A., A review on surfacemodification of activated carbon for carbon dioxide adsorption. Journal of Analytical andAppliedPyrolisis89(2010)143‐151.

ShenC.,GrandeC.A.,LiP.,YuJ.andRodriguesA.E.,AdsorptionequilibriaandkineticsofCO2andN2onactivatedcarbonbeads.ChemicalEngineeringJournal160(2010)398‐407.

ShenC.,YuJ.,LiP.,GrandeC.A.andRodriguesA.E.,CaptureofCO2fromfluegasbyvacuumpressureswingadsorptionusingactivatedcarbonbeads.Adsorption17(2011)179‐188.

Shim H.L., Udayakumar S., Yu J.I and Park D.W., Synthesis of cyclic carbonate from allylglycidyletherandcarbondioxideusingionicliquid‐functionalizedamorphoussilica.CatalysisToday148(2009)350‐354.

Silvestre‐Albero J., Silvestre‐Albero A., Rodrίguez‐Reinoso F. and Thommes M., Physicalcharacterizationofactivatedcarbonswithnarrowmicroporositybynitrogen(77.4K),carbondioxide(273K)andargon(87.3K)adsorption incombinationwith immersioncalorimetry.Carbon50(2012)3128‐3133.

Silvestre‐Albero J.,WhabyA., Sepúlveda‐EscribanoA.,Martίnez‐EscandellM.,KanekoK.andRodrίguez‐ReinosoF.,UltrahighCO2adsorptioncapacityoncarbonmolecularsievesatroomtemperature.ChemicalCommunications47(2011)6840‐6842.

Sjostorm S., KrutkaH., Starns T. and Campbell T., Pilot test results of post‐combustion CO2captureusingsolidsorbents.EnergyProcedia4(2011)1584‐1592.

Soriano A.N., Doma Jr. B.T. and Li M.H., Carbon dioxide solubility in some ionic liquids atmoderate pressures. Journal of the Taiwan Institute of Chemical Engineers 40(2009) 387‐393.

Strube R. and Manfrida G., CO2 capture in coal‐fired power plants‐Impact on plantperformance.InternationalJournalofGreenhouseGasControl5(2011)710‐726.

TanakaS.,KidaK.,FujimotoH.,MakinoTandMiyakeY.,Self‐assembling imidazolium‐basedionicliquidsinrigidnanoporesinducesanomalousCO2adsorptionatlowpressure.Langmuir27(2011)7991‐7995.

TangJ.,TangH.,SunW.,RadoszM.andShenY.,Poly(ionicliquid)sasnewmaterialsforCO2absorption.JournalofPolymerScience:PartA:PolymerChemistry43(2005)5477‐5489.

TliliN., GrévillotG. andVallièresC., Carbondioxide capture and recoverybymeansofTSAand/orVSA.InternationalJournalofGreenhouseGasControl3(2009)519‐527.

 

101

Udayakumar S., Raman V., Shim H.L. and Park D.W., Cycloaddition of carbon dioxide forcommercially‐imperative cyclic carbonates using ionic liquid‐functionalized porousamorphoussilica.AppliedCatalysisA:General368(2009)97‐104.

UdayakumarS.,ShimH.L.,RamanV.andParkD.W.,Thecompleteoptimizationofionicliquid‐functionalizedporousamorphoussilicaunderone‐potsynthesisconditions.MicroporousandMesoporousMaterials129(2010)149‐155.

Wahby A., Silvestre‐Albero J., Sepúlveda‐Escribano A. and Rodrίguez‐Reinoso F., CO2adsorptiononcarbonmolecular sieves.MicroporousandMesoporousMaterials164(2012)280‐287.

Whaby A., Ramos‐Fernández J.M., Martίnez‐Escandell M., Sepúlveda‐Escribano A., SilvestreAlbero J. andRodrίguez‐ReinosoF.,High‐surface‐area carbonmolecular sieves for selectiveCO2adsorption.ChemSusChem3(2010)974‐981.

XueZ.,ZhangZ.,HanJ.,ChenY.andMuT.,Carbondioxidecapturebyadualaminoionicliquidwith amino‐functionalized imidazolium cation and taurine anion. International Journal ofGreenhouseGasControl5(2011)628–633.

Yang H., Xu Z., Fan M., Gupta R., Slimane R.B. and Bland A.E., Progress in carbon dioxideseparationandcapture:Areview.JournalofEnvironmentalSciences20(2008)14‐27.

YongZ.,MataV.G.andRodriguesA.E.,Adsorptionofcarbondioxideonchemicallymodifiedhighsurfaceareacarbon‐basedadsorbentsathightemperature.Adsorption7(2001)41‐50.

Zhang J., Zhang S., Dong K., Zhang Y., Shen Y. and Lv X., Supported absorption of CO2 bytetrabutylphosphoniumaminoacid ionic liquids.ChemistryAEuropean Journal12 (2006b)4021‐4026.

Zhang J., Sun J.,ZhangX.,ZhaoY.andZhangS.,TherecentdevelopmentofCO2 fixationandconversionbyionicliquid.GreenhouseGasesScienceandTechnology1(2011)142‐159.

ZhangS.,ChenY.,LiF.,LuX.,DaiW.AndMoriR.,FixationandconversionofCO2usingionicliquids.CatalysisToday115(2006a)61‐69.