SANDIA REPORTSAND#UURPrinted February 2014

Hybrid-renewable processes for biofuels production: concentrated solar pyrolysis of biomass residues

Anthe George, Manfred Geier, Daniel Dedrick

Prepared bySandia National LaboratoriesAlbuquerque, New Mexico 87185 and Livermore, California 94550

Sandia National Laboratories is a multi-program laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the U.S. Department of Energy's National Nuclear Security Administration under contract DE-AC04-94AL85000.

Approved for public release; further dissemination unlimited.

SAND2014-19004R

2

Issued by Sandia National Laboratories, operated for the United States Department of Energy by Sandia Corporation.

NOTICE: This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government, nor any agency thereof, nor any of their employees, nor any of their contractors, subcontractors, or their employees, make any warranty, express or implied, or assume any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represent that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government, any agency thereof, or any of their contractors or subcontractors. The views and opinions expressed herein do not necessarily state or reflect those of the United States Government, any agency thereof, or any of their contractors.

Printed in the United States of America. This report has been reproduced directly from the best available copy.

Available to DOE and DOE contractors fromU.S. Department of EnergyOffice of Scientific and Technical InformationP.O. Box 62Oak Ridge, TN 37831

Telephone: (865) 576-2087Facsimile: (865) 576-5728E-Mail: reports@adonis.osti.govOnline ordering: http://www.osti.gov/bridge

Available to the public fromU.S. Department of CommerceNational Technical Information Service5301 Shawnee RdAlexandria, VA 22312

Telephone: (800) 553-6847Facsimile: (703) 605-6900E-Mail: orders@ntis.govOnline order: http://www.ntis.gov/help/ordermethods.aspx#online

3

SAND200X-XXXXUnlimited Release

February 2014

Hybrid-renewable processes for biofuels production: concentrated solar pyrolysis of biomass residue

Anthe George, Manfred Geier, Daniel Dedrick08367

Sandia National LaboratoriesP.O. Box 5800

Albuquerque, New Mexico

ABSTRACT

The viability of thermochemically-derived biofuels can be greatly enhanced by reducing the process parasitic energy loads. Integrating renewable power into biofuels production is one method by which these efficiency drains can be eliminated. There are a variety of such potentially viable "hybrid-renewable" approaches; one is to integrate concentrated solar power (CSP) to power biomass-to-liquid fuels (BTL) processes. Barriers to CSP integration into BTL processes are predominantly the lack of fundamental kinetic and mass transport data to enable appropriate systems analysis and reactor design. A novel design for the reactor has been created that can allow biomass particles to be suspended in a flow gas, and be irradiated with a simulated solar flux. Pyrolysis conditions were investigated and a comparison between solar and non-solar biomass pyrolysis was conducted in terms of product distributions and pyrolysis oil quality. A novel method was developed to analyse pyrolysis products, and investigate their stability.

4

ACKNOWLEDGMENTS

ThisworkwassupportedbytheLaboratoryDirectedResearchandDevelopmentprogram

at SandiaNational Laboratories.SandiaNational Laboratories is amulti-program laboratory

managedandoperatedbySandiaCorporation,awhollyownedsubsidiaryofLockheedMartin

Corporation,fortheU.S.DepartmentofEnergy'sNationalNuclearSecurityAdministrationunder

contractDE-AC04-94AL85000.

We thank Christopher Shaddix, SNL and Rafael Kandiyoti, Imperial College London, for their

constructive comments during this work and, Trevor Morgan, at the Hawaiian institute of

technologyforhiscontributionstothemethoddevelopmentstudiesinthiswork.

5

CONTENTS

Acknowledgements 04

Contents 05

1.0 Introduction 07

2.0 Simulatedconcentratedsolarpyrolysisreactors andbio-oilproduction 09

2.1 Fluidizedandfixedbedsolarthermochemicalconversionreactorsand

conventionallyheatedbatchpyrolysisreactor 09

2.2 Materials andmethods 14

2.3 GC-FIDanalysis 15

2.4 Productdistribution 15

3.0 Methoddevelopmentforanalyzingheavypyrolysisoils andbio-oilstability

studies 17

3.1 Challengesinbio-oilcharacterization 17

3.2 Materialsandmethods 22

3.2.1 Standardsandsolvents 22

3.2.2 Targeneration 23

3.2.3 Tarrecovery 23

3.2.4 Tarstorage 23

3.2.5 Bulksamplepreparationandtaryieldcalculation 25

3.2.6 Planarchromatography(PC)fractionation 25

3.2.7 Ultimateanalysis(UA) 27

3.2.8 Gas-chromatography(GC-FID) 27

3.2.9 Sizeexclusionchromatography(SEC) 28

3.2.10Laserdesorption/ionisationmassspectrometry(LD-MS) 29

3.2.11UV-fluorescencespectroscopy(UV-F) 30

3.3 Resultsanddiscussion 31

3.3.1 Tar-yields,GC-FIDandultimateanalysis 31

3.3.2 PCfractionationoftheRecoveredDrytars 38

3.3.3 BulkanalysesofRecoveredDrytar 55

6

3.3.4 Precipitatedmaterialsanalyses 58

3.3.5 Summary 62

3.4 Conclusions 64

S3 Supportinginformation 65

S3.1 FuelProperties 66

S3.2 PlanarChromatographyImages 67

S3.3 TarYields,GCandUAFurtherDiscussion 69

S3.4 SECCalibrationandInterpretation 71

S3.5 LD-MSAdditionalInformation 73

S3.6 SynchronousUV-FInterpretation 74

S3.7 PCFractionsSEC,LD-MSandUV-FResults 77

S3.8 Figures,byPCFraction(SEC,LD-MSandUV-F) 87

S3.9 Figures,bySample(SEC,LD-MSandUV-FofPCfractions) 100

S3.10 Figures,LD-MSoftheBulkTars 106

4.0 Pyrolysisofligninresidues fromtransgenic plantmaterial 108

4.1 Transgenicligninsamples 108

4.2 Reactorconfigurationforpyrolysisoftransgenicligninplants 110

4.3 PyrolysisofArabidopsissampleswithreduceddegreeofpolymerization 111

(FCA)

4.4 PyrolysisofArabidopsissampleswithreducedamountoflignin(Qsub2) 116

5.0 References 126

7

1.0 Introduction

The viability of thermochemically-derived biofuels can be greatly enhanced by

reducing the process parasitic energy loads. Gasification processes are globally

endothermic, requiring 20-45% of the feedstocki to be consumed allothermally.

Additionally, in air-blown gasification, high CO2 and N2 dilute the product syngas.

Likewise pyrolysis processes also require external energy to progress. Integrating

renewable power into biofuels production is one method by which these energy

efficiency drains can be eliminated. There are a variety of such potentially viable

“hybrid-renewable” approaches;one is to integrate concentrated solarpower (CSP) to

power biomass-to-liquid fuels (BTL) processes, both during pyrolysis and/or

gasification operations as well as downstream. Barriers to CSP integration into BTL

processesarepredominantlythelackoffundamentalkineticandmasstransportdatato

enableappropriatesystemsanalysisandreactordesign.

There is currently little understanding of the fundamental behavior ofbiomass

particles subjected to solar flux. When considering thermochemical conversion of

carbonaceous feedstocks, many parallel as well as sequential reactions involving

neighboring particles need to be considered. In these complex reactions final product

distribution is a function of the time-temperature-pressure distribution, as well as

history,ofthereactionmixture.Pyrolysisexperimentsarethereforeverysensitivetothe

heatingmethod,flowpatterns,designof reactionzoneandconfigurationofthesample

(shape,size,composition),anddataareoftenanartifactofthemethodsusedtoproduce

them.Thepaucityofdata onsolartreatmentofbiomassisexacerbatedbythefactthat,

to-date,attemptshavenotbeenmadetodecouplethesesecondaryeffectsfromreactor

design.Inthisworkweproposedtodevelopareactorschemethatsimulatessolarinflux

onto biomass particles but minimizes these artifact effects, allowing high quality

fundamental data to be obtained. This will enable rational science-based design of

process scale-up, rather than the post-hoc approaches taken to-date. Part of this

investigation necessitated the development of methods to analyse the bio-oil in its

entirety. Current analytical techniques do not adequately cope with the heavier

components of pyrolysis oils and gasification tars, and it is thought that it is these

components which contribute to the instability of the bio-oils (one of the main

challengesassociatedwith these typesof fuels).Therefore techniquesweredeveloped

8

that could give more information on the higher molecular weight components in

thermochemicallyderived liquids, and the agingand stabilityof gasification tarswere

investigatedviathetechniquesdeveloped,inordertohighlighttheireffectivenessinthis

context.

Otherthanprocessenergy, feedstockisanotherimportantparameterinbiofuels

production.Lignocellulosic biomass typically consistsof20-30%lignin.Utilizing lignin

that is currently a waste stream from the lignocellulosic biofuels process to produce

biofuels would have an enormous impact on the economics of the biorefinery, which

only uses the cellulose portion of the biomass. Feedstock cost is one of the main

economicbarrierstotheuptakeofbiofuelstechnologies,sousingawastestreamasthe

feedstock to the CSP process would make a significant contribution to the process

viability. In this research we investigated the pyrolysis of lignin. Thermochemical

conversionisparticularlychallengingduetothehighquantitiesofcharproduced.Oneof

the great advances in plant science is the ability to manipulate the structure of the

growingplant.Ligninitselfhasbeenshowntobealteredintheplantcellwall,interms

ofbothitsdegreeofpolymerization,theamountofligninthatcanproduceaviableplant

and also the structure of the lignin. In this studywe investigated geneticallymodified

lignins to ascertain if the thermochemical conversion in these plants produces a

differentqualitybio-oilandifthisbio-oilisofasuperiorquality(orhasthepotentialto

besuperior)comparedtotheconversionofthewild-typeplant.

9

2.0 Simulatedconcentratedsolarpyrolysisreactorsandbio-oilproduction

Thermochemicalbehavior,particularlypyrolysisbehaviorishighlydependenton

reactor configuration. Three reactors were developed in this study to enable the

investigation of different aspects of solar thermochemical conversion of biomass. The

firstreactor,(Figure2.1)wasthemostcomplexsystemandwasdesignedtoallowedthe

carefulstudyofthermochemicalbehaviorbyminimizingtoasgreatanextentaspossible

masstransfer limitationswithin fuelparticles that inhibitvolatilereleaseandpromote

char formation. This is particularly important when investigating the conversion of

biomass to fuel as here char conversion should be minimized and liquids yield

maximized. The fluidizingmedium in this instancewas found to lower theachievable

temperature in the system, which whilst higher than required for typical pyrolysis

processes is lower than would be desirable when simulating higher temperature

gasification.Thereforea secondfixed bedreactorwasdevelopedwhichwouldremove

the heat-transfer from the fuel particle to the flowing gas and allow for gasification

studies to be conducted (Figure 2.2). Higher temperature gasification reactions are

somewhat less sensitive to particle size andmass transfer limitations associatedwith

lower temperature biomass pyrolysis reactions attempting to maximize oil yield and

therefore a fixed bedwas deemed to be appropriate for this typeof study. Finally to

allow for comparison between the solar case and conventional pyrolysis systems a

furnacetypepyrolysisreactorwasdesigned(Figure2.3).Thesereactorsaredescribed

below.

2.1 Fluidized and fixed bed solar thermochemical conversion reactors and

conventionallyheatedbatchpyrolysisreactor

Severalassumptions andrequirements weremadeinordertoarriveatthebest

approximation for a concentrated solar powered fuel conversion reactor. All the

reactorsinthisinvestigationwerecustomfabricatedfromquartz,whichalthoughcanbe

fragile, is used in this instance because a) it is transparent to the simulated solar

radiation,b)allowsforaninertatmosphereforthepyrolysisoils(preventingcatalytic

reactionsoccurringwhenpyrolysisproductscomeintocontactwithmetals)andc)also

10

for easy recovery of any condensed material on the reactor. An added benefit is the

possibility of allowing visual monitoring and imaging of the system whilst the

experimentisunderway.

The heat source for this system was intended to simulate concentrated solar

power. Typically argon lamps are often used for this purpose, but due to budget

considerations IR lampswere investigated as suitable alternatives. To enable an even

distributionoflighttothebiomassparticles acolumnatedlightsourcewasrequiredata

narrow wavelength range. The wavelength chosen would need to be transparent to

quartz. ResearchIncStripIR5360lampswereused.Figure2.4depicts thecolumnated

lightproducedfromasinglelampandhowaconfigurationofseverallamps canbeused

to improve the density of the radiance. In this system two lamps were used. The

tungsten emitter in the lamps has an operating temperature of up to 2205 °Cwith a

spectralenergypeakwavelengthof1.15microns(Figure2.5).Thepoweroutletofthese

heaterswas1000Wat240V.Theheaterconsistsofaspecularaluminumreflectorthat

directstheinfraredenergygeneratedbyoneceramicend-seal‘T-3style’quartzhalogen

lamp factory-installed in theheater. Thequartzhalogen lampsheatupandcooldown

instantly inresponsetopowercontrol signals.Theyreach90percentof fulloperating

temperaturewithin three secondsof a cold start.The radiantenergydissipates to ten

percentfivesecondsafterthepowersupplyisdisconnected.

Despitetheresponsivenatureofthelamp,fastpyrolysisneedstobecontrolledto

withinmilliseconds,andwehavethisrequirementhere.Thereforeashuttersystemwas

designed and controlled by LABVIEW, whereby a heat resistant shield made of

refractorymaterialwasdesigned.UsingthisshuttersystemandLABVIEWtheexposure

ofthebiomassparticlestotheheatsourcecanbecontrolledto+/- 5ms.

The operation of the fluidized bed reactor given the design described is as

follows:thefeedingtubeand lampcoolingwatersystemsareactivated,afterwhichthe

hotfluidizinggas(nitrogen,180°C)ispassedthroughthereactor.Thefluidizinggasis

heatedtobelowthepyrolysis temperatureof theparticlesas the short reactiontimes,

and high heating rates required to conduct these fundamental studies could not be

achievedwith the lamp set-up in this reactor scheme, if the experiment was starting

from“cold”.A carriergasat the topof the reactorof low flowrate (Figure2.1) is then

11

introduced into the reactor. The charge of biomass particles are introduced into the

fluidized zone via opening a simple ball valve and are carried through by the small

flowrateoftheaforementionedgas. Theexperimentisnowreadytoproceed.Bymeans

ofLABVIEW,thelampsareswitchedonandafter3secondstheshuttersareopened.The

fluidizedbiomassparticlesareexposedtotheheatsource fortheprescribedtimeafter

whichtheshuttersareautomaticallyclosed.Anysolids(charandunreactedbiomass)is

collectedinthechar-trapafterthefluidizinggasseshavebeenswitchedoff.Thevolatile

pyrolysis products are condensed in a tar-trap, and in theory after the tar trap a gas

analysercouldquantifythepermanentgasesfromtheexperiment.However,thisfacility

wasnotavailablefortheseexperimentsandgasyieldswerequantifiedbydifference. To

comparethepyrolysisproductdistributionfromthesolarheatedandnon-solarheated

caseafurnacebasedpyrolysisreactorwasdesigned(Figure2.3).

12

Batch feeder

Water cooled feeder

tube

Sintered disk to allow

volatiles escape and

prevent particle elutriation

Quartz fluidized

reaction zone

Char/solids catch-pot

Feed carrier gas

Feeder water cooling

Condensable products

outlet

Heated fluidizing gas

Heat resistant

electronically controlled

shutter in front of IR lamp

Thermocouple inlet to

reaction zone

Figure 2.1 Fluidized solar pyrolysis reactor, illustrating key operating components

13

Figure2.2 Fixedbedsolarpyrolysisreactor,showingelectronicallycontrolledshuttercarriergasin-

letandvolatile(bio-oil)outlet.

Figure2.3Columnated solarpyrolysisreactor,showingelectronicallycontrolledshuttercarrier

gasin-letandvolatile(bio-oil)outlet.

14

2.2Materials

Switchgrass (Panicum virgatum) was provided by Dr. Daniel Putnam, University of

CaliforniaatDavis.Switchgrass wasgroundbyaWileyMillthrougha2mmscreenand

separated by a vibratory sieve system (Endecotts, Ponte Vedra, FL). The switchgrass

fractions falling between 60 and 80 mesh were collected for use in this study. The

Figure2.4ColumnatedIRradiationproducedbyIRlampandmulti-lampconfigurationfor

increasingdensityofradiationtotarget.Inthisworkatwolampconfigurationwasused.

Figure 2.5 Emission range of IR lamp used in fluidized solar pyrolysis reactor. Peak

emission is approximately 1 µm which allows for transmittance through quartz and isa

representativewavelengthforsolarradiation.

15

biomasswasdried inaconvectionovenat50degreesCelsius for24hoursbeforeuse.

Analyticalgradereagents(Sigma)wereusedthroughout.

2.3GC-FIDAnalysis

For pyrolysis oil analysis, a Varian GC3800-FID with a Varian ‘FactorFour’ capillary

column (VF-5ms 30M, 0.25mM ID,DF =0.25)was used toquantitatively examine the

tars. 5 μL injection volume (split ratio 50), helium carrier gas (2 mL/min), injector

temperature325°C,columnstartingtemperatureof50°Cfor4minutes,25°C/minramp

to150°C,5°C/minrampto325°C,heldat325°Cfor2minutes.FIDdetectortemperature

was 325°C. For the GC analysis a volume of the isopropanol-tar solution was taken

before the samplewas vacuumdried. Further details of peak identification, standards

andcalibrationareprovidedinS3.

2.4Productdistribution

Switchgrass was pyrolysed at varying conditions (residence time and temperature),

table2.1. Thisdataiscomparedwithexperimentsconductedinthewire-meshreactor

described in section 4.1. Particle heating rate calculations are always challenging in

thermochemical conversion experiments. The challengewith the solar reactor system

designedhereisthattheheatingrateoftheparticlecouldnotbeasaccuratelyinferred

asinthewiremesh,asthetemperaturebeingmeasuredwasthebulktemperatureinthe

reaction zone.Nevertheless heating rates up to 500 °C per second could be achieved

with lamp power at 80% of maximum, with a heated inlet gas of 180° C with the

fluidized systemand1200 °C for the static system. For the char tar and gasproduct

distributions, char isdefinedas the solidsmaterial recoveredafter reaction, tar as the

condensableliquidmaterialrecoveredinthetrapafterreactionandthepermanentgas

yield was calculated by difference. Reactions were repeated in quadruplicate, and

measurement uncertainties were in most cases <5% but always <10% and data are

givenonadry,ashfreebasis.

16

Table2.1pyrolysisofswitchgrassusingsimulatedsolarirradiationviaanIRlamp

Reactor

type

Peak

temperature

(oC)

Holding

time(s)

Char

(wt.%)

Tar

(wt.%)

Gas

(wt.%)

Solar 400 5 24.3 48.34 27.36

Solar 400 30 22.1 47.6 30.3

Solar 550 5 17.8 50.1 32.1

Solar 550 30 17.4 50.4 32.2

Wire-mesh 400 5 19.9 52.8 27.3

Wire-mesh 400 30 19.2 52.6 28.2

Wire-mesh 550 5 18.4 53.5 28.1

Wire-mesh 550 30 18.3 53.2 28.5

In general the solar system showed greater sensitivity to parameter change than the

wire-mesh reactor. With regard tovariations in temperature, it appears for the solar

system that forbothholding times charyieldsdecreasedwith increasing temperature

and whilst both tar and gas yields increased. Pyrolysis product yields are generally

thoughttobemoreaffectedbyheatingratethantemperature,butat400°Cit is likely

that the higher solids yields are due to unreacted material, rather than a greater

productionof char.This isbourneoutby the lower charyields in the solar reactor at

longerresidencetimesatbothtemperaturetests.

Thedataaboveshowthatholdingtimedidnotmakeadifferenceinthecaseofthewire-

meshreactorforcharproductionataparticulartemperaturebutwasamoresignificant

variableinthesolarsystemat400°C.However,particularlyathighertemperatures,the

taryieldswerereducedatlongerholdingtimes Inthesolarreactor.Thewire-meshwas

designed to approximate an ideal system in terms of secondary tar reactions and

cracking. It appears for the solar system, heat transfer within the particle is more

efficientleadingtolowercharyields,however,theevolvedproductsarefurthercracked

to permanent gases. Further investigation of the heat transfer phenomena via this

radiativesystemattheselowertemperatureswouldbemostinteresting.

17

3.0 Methoddevelopmentforanalyzingheavypyrolysisoilsandbio-oilstability

3.1 Challengesinbio-oilcharacterization

Aknownissuewithpyrolysisoilandliquidproductsfromgasificationisaccurate

characterization of themolecular structure, particularly of the heavier components of

the biomass which sample instability. This is particularly challenging if these

componentsaretobeusedforbio-fuelproductionwhich,aswithall fuels,willneedto

havestableattributesthatfallwithincertainstandardvalues.Amethodwasdeveloped

tocharacterizeheaviercomponentsinbiomassbyevaluatingtheheaviesttarsproduced

in the co-gasification of biomasswith coal. A description of the study and results are

showninthissection.

Gasification has been recognised as one of the most efficient thermochemical

processes for converting solid fuels into energy [1]. Despite this high efficiency,

relatively few power stations have adopted this technology, due to economic and

downstream technical drawbacks. The environmental concerns about the levels of

carbon dioxide in the atmosphere have provoked a renewed interest [1-8]. However,

several unsolved problems related to fixed- and fluidised-bed biomass gasification

remainwhich hamper the implementation of the technology.Ofmain concern for any

applicationofthesynthesisgasisthepresenceofimpuritiessuchastar,NH3,HCl,HCN,

H2S and COS [1]. In particular, the amount of tar produced reduces the carbon

conversionandefficiencyandcausessignificantoperationalproblems duetodeposition

on exhaust lines [1, 9]. There are also difficulties with their subsequent storage and

processing[1,5-10].

Gasification tars also contain high levelsof polyaromatic hydrocarbons (PAHs)

thatare recognisedcarcinogensbytheenvironmentalprotectionagency(EPA[11]).A

numberofgascleaningsystemsarecommerciallyavailableandinroutineoperation,but

thesearecostlyandnotcurrentlyoptimised[8-10].

Atpresent,tarsfromtheco-gasificationofbiomassandcoal,aswellasfrompure

biomass,arebeingproducedonasmallbutincreasingscaleaftersynthesisgascleaning.

Recoveredgasification tars are typically re-usedasboiler-fuel. This isnotnecessarily

environmentallyfavourableorcosteffective[12,13].Moreover,theincreasedemphasis

18

on using biomass as a substitute for coal for power generation will lead to greater

amounts of tars being generated in the future. Thesewill need tobe used, stored,or

otherwisedisposed.Gasification tars frombiomass,which containhigherquantitiesof

oxygen than those from coal, will have analogous issues to biomass pyrolysis liquids

withregardtotheirstorage,upgradingand/oruse.

Biomass pyrolysis liquids are composed of a complex mixture of oxygenated

hydrocarbons(e.g.alcohols,aldehydes,carboxylicacids,esters,ethers,ketones,phenols

andsugarderivatives)with15-30wt%waterfromtheoriginalmoistureandasreaction

product [1, 14, 15]. The effects due to ageing on biomass pyrolysis liquids have been

widely studied. Reactionsoccur between oxygenated compounds, volatile components

are lostduetoevaporation leadingtoan increaseofmolecularweight,andchanges in

water content and viscosity can result in phase separation [1, 14-22]. The probable

reactionpathwaysinvolvedintheageingprocesshavebeenreportedtoconferpossible

changes tobulkproperties suchasboiling range,densityandviscosity [14].However,

limitedinformationisavailableregardingchangesonthemolecularlevel.

Ageing studies of biomass pyrolysis liquids have focused on theanalysis of the

trendsinmolecularweight,basedonsizeexclusionchromatographyorviscosity,andon

the determination of the water content [14, 15]. Less attention has been paid to the

studyof chemical composition through techniques suchasGC/GC-MS,FT-IRandNMR

[19,23,24].Theinfluenceofstoragetemperatureandsolventadditiononthestabilityof

oilsandtarshasbeenwidelyexamined,typicallybasedonviscositydatawhichcanonly

provide indicative information on other important properties [14-16, 25, 26]. Several

approaches to improve the stabilityofpyrolysisoilshavebeen reported [15,27].The

additionofapolarsolventisacommonapproachtoimprovethestabilityoftarsduring

storageandcanresultin atwentyfolddecreaseintheageingrate,reductioninviscosity

andacidity,aswellasincreasingheatingvalueandmiscibilitywithfossilfuels [15,18,

23, 26]. The upgrading of biomass pyrolysis liquids by addition of alcohols at room

temperaturehasbeenconductedat the industrialscale since1995 [28]. Similarly, the

storage of gasification and pyrolysis tars in alcohol solution at reduced temperature

(5°C)hasbeenreportedtominimiseageingreactions[14].

19

Standard analytical methods are available for the recovery [29] and

characterisation [30] of tars from gasification, which enable comparisons to bemade

betweendifferentreactorconfigurationsandfuels.Thestandardmethodshowever,are

basedongaschromatography(GC)andthegravimetricweightofthetar.Thislimitsthe

analysistomaterialsthatarevolatileatGCcolumnconditions.Ingeneral,littleisknown

about the composition of materials labelled as ‘tars’ in terms of molecular mass

distribution,averagemolecularmassorstructure,beyondinformationavailablefromGC

methods [1, 8, 12, 13, 31]. For compounds present in gasification tars the typical GC

upper limit is ~350 u for aromatic compounds and ~600 u for aliphatic compounds,

C60-C70 or a little higher (C100), using high temperature GC [32-34]. The smaller

molecular weight components typically constitute between 30-60 wt% of the entire

sample(dependingonreactionconditionsandfeedstock),sothereexistsalargegapin

understandingoftheimportantremainingportionofthetar[12].

Studies in the literature reveal that more advanced analytical techniques have

rarely been applied to gasification tars and in particular those from biomass. This is

despite the fact thatbiomass and coal gasification tars havebeen of studied since the

1990sandhavebeen indentifiedasakey issue tobeaddressed for improvingenergy

conversion [35-37]. This point has also been reiterated more recently in a

comprehensivereviewofthechemistryofthermochemicalprocessesusedtosynthesise

transportationfuelsfrombiomass[1].Muchoftheearlyworkusingadvancedanalytical

methods focused on coal tars and pitch [36, 37], petroleummaltene and asphaltenes

[38], coal [39], partial oxidized aromatic hydrocarbons [40] and products from wood

pyrolysis[41-43].Mostof thesestudies focusonthe identificationofsmall tomedium

sizedmolecules(<800u)[39-42],insomecasesusingGCbasedmethods[39,41] orin

combinationwithliquidchromatography[40,42].

Some of these early studies looked at the highermassmolecules (>1000 u) in

attempts tomore fully characterise the samples [36-38,43,44] mostlybasedon laser

desorption-MS(LD-MS).Thesestudiesshowedevidenceformolecularmassgreaterthan

2000uinawoodtarpitch[44],woodpyrolysisliquids[43],coaltars[36,37] petroleum

asphaltenes [38, 45] and pyrolysis products from waste plastics [46]. However, no

studiesusingLD-MSonbiomassgasificationtars,ortars fromco-gasificationwithcoal,

couldbefoundintheliterature.Inaddition,greatprogresshasbeenmadeinoptimizing

20

the experimental conditions for LD-MS and interpretation of the results, in particular

whenappliedtocomplexsamplessuchasthosementionedabove,asdetailedinarecent

reviewarticle[47].

Morerecentlyioncyclotronresonancemassspectrometry(ICR-MS)hasreceived

attentionforstudyingcomplexsamplessuchasoilsandtarsduetounprecedentedlevel

ofmassresolution,typicallyto0.1ppmofmass,atmassesofafewhundredmassunits.

Themethod has been applied to coal tar pitch [48], carbonblack [49], products from

pyrolysis of shale oils [50], heavy bitumen derived samples [51], and petroleum

asphaltenes [52]. The method was able to provide extremely valuable and detailed

information formolecularmass up to amaximumof 1200u (typically< 800 u) inall

these samples. However, the method is known to suffer from selective sampling

(incomplete sampling) issues, which is not fully understood [47]. In addition, all the

samplesmentionedaboveareknowntocontainhighermassmoleculesthandetectedby

ICR-MS. It cannot therefore be used to determine average mass estimates or mass

distributionsforcomplexmixturesofhydrocarbons.Thesepointsandthewiderbenefits

andlimitationsofICR-MShavebeenaddressedinarecentreviewarticle[47].

Similar problems persist when trying to fully characterise pyrolysis derived

liquidsandtarsfrombiomass,coalandpetroleumresidues,intermsofmolecularmass

distribution, averagemolecularmassor structural compositionbyotheradvancedMS

methods[12,13,47].Typically,thestandardmethodforgasificationtaranalysisisalso

usedonpyrolysisliquids[8,29].Inmanyofthestudiesmentionedabovethatlookedat

materials beyond the GC range, size exclusion chromatography (SEC) was used for

molecular weight determination, with tetrahydrofuran (THF) as eluent and solvent,

oftenwith a refractive index detector [14, 15, 27, 43]. These SEC systems have been

showntobeincapableofresolvingsmalltomediumsizedpoly-aromatichydrocarbons

intermsofmolecularsizeormass[53-55].ThisisthoughttobeduetoTHFactingasa

weaksolventforthesamplesinquestionandenablinginteractionswiththeSECcolumn,

and also the usefulness of refractive index for detection is questionable. A number of

studies have addressed these, and the wider issues, over the last decade for coal,

petroleum,biomassandbitumenderivedoils,tarsandpitches[12,13,31,53-57] anda

reviewofthesestudieshasbeenreported[47].

21

TheaforementionedinvestigationsdemonstratedthatN-methyl-2-pyrrollidinone

(NMP)wasasuperiorSECeluentandsolventfortarsthatcontainhigharomaticityand

fewerhetero-atomssuchascoalorbiomassderivedgasificationoils/tars[47].Despite

theirhighoxygencontent,pyrolysisliquidsfrombiomasscanalsobeanalysedwithNMP

astheSECeluent[58,59].

ItisimportanttonotethatusingSECwithNMPeluentdoesnotprovidecomplete

informationorquantificationofmolecularmassdistributions,although ithasproveda

valuable tool for making relative comparisons between similar samples and is better

understoodthanotherSECsystems[47,53,55].Itisimportantthatthelimitationsofthe

technique are understood and accounted for when interpreting SEC results to avoid

over-interpretationofdata.

Therefore, there remains a great need in the gasification and pyrolysis

communitiestoobtainmoreaccurateinformationonmolecularweightdistributionsand

structuralfeaturesofliquid/solidproductsbeyondthe1000umassrange[12,13,47].

Tothisend,arecentlyreportedanalyticalapproachwasassessed[31,47,56,60].This

methodologyhas been successfully applied to thedetermination of averagemolecular

mass number estimates, molecular mass distributions and detailed structural

informationonpetroleumandcoal tarpitchderivedmaltenesandasphaltenes[47,56,

60,61] aswellasbitumen(oil/tarsand)samplesandsolubilitysub-fractions[47,62].

The study at hand assesses the validity of the above mentioned analytical

approach [31,47,56, 60] for tars from the co-gasificationof a softwood (pine)anda

Polish black (sub-bituminous) coal. To study the validity of the above methods, we

attemptedto identifychanges incomposition(intermsofmolecularmassdistribution

and aromatic structural features) after ageing the tars under different storage

conditions.Findings fromthescopingstudyarereported. It is intendedthatdata from

this scoping study would first determine if using this methodology is justified, and

secondly,assessifthisnewmethodologycanspurfurtherresearchinthefuture.

Changesdue toaging in the compositionof gasification tarsareexpected tobe

lesssignificantcomparedtochangesimpartedontarsfromtheuseofcatalysts,thermal

upgradingprocesses,orseenduringtheageingofbiomasspyrolysisliquids.Therefore, if

themethodsusedherearesensitiveenoughtoidentifychangesingasificationtarsdue

22

to ageing then they would likely also be suitable for studying all the processes

mentionedabove.

3.2 Materialsandmethods

The tars used for this study were recovered from a 20 kWth internal circulating

fluidised bed gasification and combustion reactor [63, 64] using the tar protocol (TP)

method [29]. The collected tars were then subjected to different storage conditions

where temperature, ageing timeandexposure to lightwerevaried; these samplesare

defined in Section 2.4. The results from analysing these samples are presented as

follows:

1. Amassbalance,showingtaryields,GC-FIDanalysisandultimateanalysisof the

bulksamples.

2. Planar chromatography (PC) [31,56] wasused to fractionate thebulk samples,

anddataobservedfromthePCfractions(withoutfurtherchemicalanalysis)are

presented.

3. PCfractionsfromthedifferenttarswerethensubjectedtofurtheranalysisusing

SECandLD-MStodeterminemolecularmassdistributionestimates.Synchronous

fluorescence spectroscopy (UV-F) was used to compare relative extents of

conjugation.Finally,thedatawerecombinedtomakepossibleinterpretations,in

terms of chemical reactivity of the tars based on known chemistry from the

literature.Thesemethodswerebasedonthosedevelopedinpreviousstudies[31,

47,56,60].

4. LD-MSwasthenusedtoanalysethebulktars,anddatacomparedwiththePCfor

amorerobustinterpretationofthewholematerial.

5. Finally, the aged samples exhibited precipitation. The final component of this

studywas to analyse this precipitated solid tarmaterial to gain further insight

intothetarageingprocess.

3.2.1 Standardsandsolvents

Polystyrene, amongst other polymer and PAH standards, was used to calibrate

the SEC system as previously reported [53, 54]. The solvents were from VWR,

23

(chloroform, heptane and acetone as HPLC grade), Rathburn Chemicals Ltd.,

Walkerburn, Scotland, UK (NMP – peptide synthesis grade) and used without further

purification.

3.2.2 Targeneration

The tarwasgeneratedusinga20kWth internal circulating fluidisedbed (ICFB)

gasifierwhichhasbeendescribedelsewhere[63,64].Gasificationtestswereperformed

usingpelletsmade from ablendofpinewood (ovendried,~12wt%moisture) anda

black Polish sub-bituminous coal (~8 wt% moisture), at a ratio of 7:3 wt%. The

propertiesofthefuelaregiveninSupportingInformation(S)SectionS1.Thecylindrical

pelletsizewas2mmindiameterand6mminlength.Thesewerefedatarateof4.0kg

perhour.

The tarwas collectedunder steamgasification conditionsat800°C,witha flow

rateof3.75Nm3/hofsteamand3.0Nm3/hofairtothegasifier;airtofuelequivalence

ratio0.15;steamtofuelratio0.75.Thecombustorwasoperatedwithanairflowrateof

11 Nm3/h. The fluidised bed contained approximately 12 kg of inert silica sand of

particlesize0.2-0.4mm.Gasresidencetimeinthebedwas1-2secondsand3-5seconds

inthefreeboard.

3.2.3 Tarrecovery

Thetar samplewas recoveredusingthetarprotocol(TP)method[29,30].The

methodusesanimpingertrainwhereaslipstreamoftheproducergas ispassedthrough

a series of seven bottles containing isopropanol, some at -20°C and others at 40°C.

Approximately0.1Nm3 oftheproducergaswaspassedthroughtheimpingertrainover

aperiodofonehourduringstableoperation.The isopropanol tarsolutionwaspooled

from the seven impingerbottles and filteredwithin twohoursof collection to remove

particulates(i.e.charandbedmaterial). A1µmglassfibrefilterwasused;thefiltrate

wasasinglephasesolution.

3.2.4 Tarstorage

To study the effect of storage conditions on the composition of tar, the tar

solutionswerestoredunderfourdifferentconditions,wherethetime,temperatureand

exposuretolightwerevaried,asdescribedbelowforeachsample:

24

TN2-0h:Thetarsample recovered in isopropanolwas frozen in liquidnitrogenwithin2

hoursofrecoveryandstoredinthisway.Immediatelypriortotheanalysis, thesample

wasthawedinthedarkat5°Candthenvacuumdried.Thisisdeemedtobe“freshtar”.

T5C-6m:The isopropanol tar solutionwas stored for6monthsat5°C, in theabsenceof

light,priortobeingvacuumdriedandanalysed.

T20C-6m: The isopropanol tar solution was kept at room temperate (~20°C), in the

absenceoflightfor6months,priortobeingvacuumdriedandanalysed.

T20C-6m-L: The remainder of the tar solutionwas left at room temperature in the fume

cabinet and thereby exposed to indirect sunlight for 6months prior to being vacuum

driedandanalysed.

Whilst these conditions were chosen in order to be systematic in varying

temperature,timeandexposuretolight,resultsfromsampleT20C-6m showedsimilardata

to thoseobtained fromT5C-6m. Therefore, forbrevityandclarity, thepresentedwork

will focus on the samples that showed the most significant differences between one-

another and highlight the benefits of using this analytical approach: TN2-0h, T5C-6m and

T20C-6m-L.TheresultsfromT20C-6m arenot shownordiscussedexplicitlysincethefindings

aresimilartothoseforT5C-6m.

Additionally, to study the effect of time during storage, aliquots of all the tar

solutionswere taken for GC analysis after 20 hrs, 3 days,20 days and 6months, and

compared to those from the fresh tar solution immediately after it was thawed from

liquid nitrogen (TN2-0h). The results obtained after 3 and 20 days storage provided

limited information and therefore will not be discussed herein, although they are

reported in S3. The GC analysis of the fresh tar (TN2-0h) was performed before any

precipitatewasobservedto haveformed,whileforthe20hoursoldsamplesprecipitate

had already formed and was removed before performing the GC analysis. Therefore,

comparingresults forTN2-0h withthesamplesagedfor20hoursgivesacomparisonof

thetarcompositionbeforeandafterprecipitation.

As an additional check, the thawedTN2-0h sample,whichwas observed to form

precipitatewithin 14 hours of standing (at 5°C in the dark),was also analysed byGC

afterremovingtheprecipitate.Itshouldbenotedthattheexactpointintimewhenthe

25

precipitate formedwas not determined.Moreover, this 14 hour old sample was only

usedtoexaminechangesinthetar’sGCcompositionbeforeandafterprecipitationand

wasnot subjectedtofurtheranalyses.

Themain focus of this study is the application of LD-MS, SEC and UV-F to the

analysisof thetarsamples.Onlythe freshtar (TN2-0h)andthe6monthagedtarswere

analysed by all of these techniques, not the intermediate samples mentioned above

whichwere analysed byGC only. A summary of the nomenclature, storage conditions

andanalyticaltechniquesthatwereappliedtoeachsamplearelistedinTable1.

Table 1, Nomenclature of the tar samples, their storage conditions and the analytical

techniquesapplied.

Label

Storage AnalyticalTechnique

Temp

(°C)Sun-Light Age GC

Recovered

Dry

weight

UA UV-F SEC LD-MS

TN2-0h -196 No <5hrs X X X X X X

TN2-14h -196 No 14hrs* X - - - - -

T5C-20h 5 No 20hrs X X - - - -

T5C-6m 5 No 6months X X X X X X

T20C-20h-L 20 Yes 20hrs X X - - - -

T20C-6m-L 20 Yes 6months X X X X X X

*14hrsreferstothetimethesamplewasstoredinafridge(5°C)afterbeingthawedfromliquidnitrogen

3.2.5Bulksamplepreparationandtaryieldcalculation

Gravimetric analysis and tar yield calculations were performed based on the

standardmethod described in detail elsewhere [29, 30] and only briefly summarized

here,usingthefollowingstepstoprovidefourfractions:

1. GCTotal:The tar-isopropanol solutionwas directly analysed byGC to quantify

thematerialintheGCrange.

26

2. Precipitate:The tar-isopropanol solutionswere filteredusinga glass-fibre filter

of1µmtorecoveranyprecipitate.Theprecipitatewaswashedwith five lotsof

freshisopropanol(10mLeach)beforebeingdriedinafanassistedovenat105°C

untilnosolventremained,approximately2hours.This fractionwasweighedto

yieldtheprecipitateweightinthesolution

3. Recovered Dry: The filtrate after the precipitated material was removed was

vacuumdriedfor4hoursatatemperatureof75°Cand10-3bar.Thismaterialwas

weighedtodeterminethemassofthedried,recoveredmatter.Anylightmaterial

wouldbelostduetovacuumdrying.

4. GC Recovered Dry: The Recovered Dry fraction was then redissolved and

analysedbyGC.ThedifferencebetweentheGCTotalandtheGCRecoveredDry

fractionisthemassofvolatileslostduringvacuumdrying.

TheTotalTarmaybecalculatedasfollows:

TotalTar=Volatiles+Precipitate+RecoveredDry

Where:Volatiles=GCTotal– GCRecoveredDry

Thebulksampleusedinallfurtheranalysiswasthe“RecoveredDry”fraction,i.e.

thetarwithoutprecipitateorlightvolatiles.Vacuumdryingwasusedhereasthesolvent

cannot be completely removed by other means. Additionally, vacuum drying was

employed to remove themost volatile components from the samples so that a better

comparison could be made between SEC and LD-MS mass estimates. Without this

vacuumtreatmentthe samples forSECandLD-MSwouldbequitedifferentduetothe

highvacuumintheLD-MSsamplechamberleadingtothelossofmoleculeswithmassof

lessthan~200u.Itisimportanttonote,however,thatthedryingprocesscouldresultin

changes in the sample composition (due to reactions) and it is not possible to isolate

theseeffectsfromthosethatoccurredduringstoragealone.

3.2.6Planarchromatography(PC)fractionation

27

Thebulk(dried)tarsamplesdescribedabovewerefractionatedbyusingplanar

chromatography to aid in their analysis. Aluminium backed PC plates of 20 cm2 with

silica gel thickness of 250 µm (Whatman, UK) were used. The plates washed with

acetone and then chloroform before use. The tar was dosed onto the PC plate as a

solution inchloroform;multiplesampledoseswereaddedat theoriginof theplate to

increasetheamountofsample.Allsampleswerecompletelysolubleinchloroform.The

PCplatewassuccessivelydevelopedwithchloroform,acetone,andheptane,beingdried

every time before the application of a new solvent. Figure 1 gives an example of the

mobility-fractions forthetarTN2-0h,FigureS2.1shows imagesof thePCplates fromall

three tars and provides further information. Each sample was eluted with the same

solvent three times consecutively, to the same height, before moving on to the next

solvent. No attempt was made to quantify the PC fractionation due to the inherent

difficulties involved; although the apparent abundances from visual inspection are

describedinSection3.2.

Inordertooptimizethesamplefractionation,aseriesoftestswereperformedto

select the solvents and order of elution. The aim of the PC separationwas to provide

relatively few discrete mobility-fractions and to isolate some material at the origin,

whichinpreviousstudieswascomposedofthehighestmassspecies[31,56].

This material, which was immobile in all of the eluents, was denoted as PC

fractionF1.Subsequentfractionsofmaterialwithincreasingmobility(higherupthePC

plate)weretakenatregularintervalsandgroupedbynumberbasedonthesolventthey

weremobilein.FractionF2wasacetonemobilebutchloroformandheptaneimmobile;

FractionF3waschloroformmobile,butheptaneimmobile.Thematerialatthefurthest

solventfrontwaslabelledfractionF4(heptanemobile);cf.Figure1andS2.1forimages

oftheplateswiththenumberingschemeoverlaid.

ForLD-MSanalysisthedifferentmobility-fractionswerecutoutfromthePCplate

anddirectlyadheredtotheLD-MStarget.

For SEC and UV-F analysis the sampleswere recovered by removing the silica

fromtheplatesandextractingthiswithNMP.TheresultingNMPsolutionwas filtered

(1µm).

3.2.7Ultimateanalysis(UA)

28

The carbon, hydrogen, sulphur and nitrogen contents of the dried bulk tar

sampleswere determinedwith a LECO-CHNS-932microanalyzer. The oxygen content

was obtained directly using a LECO-VTF-900 furnace coupled to the microanalyzer.

Thesesamplescontainedno moistureandno ash.Duplicateanalysesshoweddeviation

fromthemeanwerelessthan+/- 0.5%oftheabsolutevalue.

3.2.8Gas-chromatography(GC-FID)

GCwasusedtostudythesmallermoleculesinthesamples.TheGCanalysisofthe

fresh tar(TN2-0h)wasperformedimmediatelyafter thawing.Thesamplewas inspected

after14hours(at5°C intheabsenceoflight)anditwasobservedthatprecipitatehad

formed. Therefore the precipitate was removed by filtration and GC analysis was

performedonthesupernatant.ComparingtheGCresultsfromTN2-0h andafter14hours

storagegivesacomparisonofthetarcompositionbeforeandafterprecipitation.Asan

additional check, the samples storedat5°C in thedarkandat20°C in thepresenceof

indirectsunlight,whichwereobservedtoformprecipitatewithin20hoursofstanding,

wasalsoanalysedbyGCafterremovingtheprecipitate(Table3).Itshouldbenotedthat

these samples were not checked between the ages of 5-20 hours due to practical

experimentalconstraints.

AVarianGC3800-FIDwithaVarian‘FactorFour’capillarycolumn(VF-5ms30M,

0.25mM ID, DF = 0.25) was used to quantitatively examine the tars. 5 μL injection

volume (split ratio 50), helium carrier gas (2 mL/min), injector temperature 325°C,

columnstartingtemperatureof50°Cfor4minutes,25°C/minrampto150°C,5°C/min

ramp to325°C,heldat325°C for2minutes.FIDdetector temperaturewas325°C.For

theGCanalysis avolumeof the isopropanol-tar solutionwas takenbefore the sample

wasvacuumdried.Furtherdetailsofpeak identification,standardsandcalibrationare

providedinS3.

3.2.9Sizeexclusionchromatography (SEC)

The operating conditions andmethodology have been reported elsewhere [53,

54].Briefly,aMixed-Dcolumn(5µmparticlesize,300mmx7.5mmi.d.)packedwith

polystyrene/ polydivinylbenzene beads, was operated at 80°C with a Knauer M100

isocraticHPLCpump. NMPwasusedaseluent(0.5mLmin-1)andsolvent.

29

Material eluting from the columnwas detected by UV-absorbance at 270, 300,

350,and370nm.Theresultsobtainedat300nmareconsideredrepresentativeofthe

maintrendsobservedatallwavelengths;onlythoseresultswillbeshownanddiscussed.

The SEC system was calibrated using standard polystyrene (PS),

polymethylmethacrylate (PMMA) andpolysaccharide (PSAC) samples, aswell as small

standardPAH,O-PAHandN-PAHcompounds[53-55].

In the conversion of elution time tomass estimate, thematerials eluting early

from the column (<15 minutes) which is excluded from column porosity cannot be

accuratelyaccountedfor.Thisisbecausethenatureofthisearlyeluting(apparentlyhigh

mass)materialremainsuncertainandisthoughttohaveahydro-dynamicvolumewhich

departs from those of the calibration materials, possibly due to three-dimensional

conformationwheresizeisnoteasilyrelatedtomolecularmass[53,55].Previouswork

onsimilarmaterialshasshownthisearlyelutingmaterial(<15minutes)tobeofhigher

averagemolecularmass(>2500u) thanthe laterelutingmaterials [55].Theseaspects

andtheapplicationofthiscalibrationaredescribedinsaidpublicationsandoutlinedin

S4.

3.2.10Laserdesorption/ionisationmassspectrometry(LD-MS)

ABrukerDaltonicsReflexIVMALDI-TOFmassspectrometerwasusedforLD-MS.

Nomatriceswereusedbecause the samples in this studybehaveas self-matrices [31,

56].Nomatrixdeflectionvoltagewasused;anitrogen laserof337nmwasemployed.

Themethodappliedinthisstudyhasalsobeendescribedindetailelsewhere[31,56].

Linear-modewasusedwithadelayedionextraction(DIE)timeof0,and600ns.

The mass range was m/z 0-300,000; Ion source 2 = 16.5 kV and Lens = 9.5 kV. The

digital gain (DG)was set to its lowest level (1x). In all cases shown, 10 spectrawere

addedusingthepulsed ionextractionmethodonthesamepoint.50spectrawerealso

addedforsomeofthesamples;thesespectrawerefoundtomatchthosewhereonly10

spectra were summed, with an increased signal-to-noise ratio. Due to time

considerations theadditionof10 spectraof eachanalysiswas consideredsatisfactory

forthepurposeofthisstudy.

30

AHIMASdetector (Bruker’s ‘high-mass detector’) operating in the linearmode

wasusedto investigatethehighermolecularmass region.Detectionofhighmass ions

canbeenhancedthroughtheuseofavariablehighmassaccelerator(HMA)voltage.

Whenthebulksampleswereanalysed,theionintensityofthesmallermassions

was reduced (to avoid overloading the HIMAS detector) by reducing the high mass

acceleratorvoltagefromthemaximumvalueof10kVto6kV.Atthesametime,thelaser

powerwasincreasedbeyondthatnecessaryforionisationofthesmallmoleculesofthe

samples.Thisistoexaminetheinfluenceoflaserpoweronthemassdistribution,which

is generally significant. A laser power of 30-50% (of the maximum available) was

typicallyfoundtobeadequatetoionisethesamples.Thebulktarsampleswereapplied

totheLD-MStargetneatfollowingtheproceduredescribedelsewhere[31,56].Mainly

theresultsobtainedwithaDIEof600nswillbereportedanddiscussed inrelationto

the‘bulk’samples,forthereasons outlinedinS5.

DuringtheexaminationofthePCmobility-fractions,theHMAvoltagewassetto

itsmaximumvalue(10kV)inallcases.Bydesorbingthesampledirectlyfromthesilica

surfaceofthePCplate,highlaserpowerscouldbeappliedwithnoresultingsignificant

changeofthemassspectra,apart fromproducingflat-toppedpeaksduetooverloading

ofthedetector.ForthePCfractions,onlyLD-MSspectraacquiredinlinear-modewithno

DIEareshown.AcleansilicasurfacefromthePCplatewasfoundtogivenoobservable

ioncurrentunderanyoftheLD-MSoperatingconditionsused[56].

AmajorobstacletotheuseofLD-MSisthevariabilityofthe datadependingon

theconditionsandtheoperator[31,56].Theapproachusedinthisstudywasfoundto

giveconsistentfindings independentoftheoperator.Fouroperatorsrecordedthedata

reported. This provides adegree of confidence in the reproducibilityof the approach.

TheresultsfromLD-MSarenotquantitative.

3.2.11UV-fluorescencespectroscopy(UV-F)

ThePerkin-ElmerLS55bluminescencespectrometerwassetwithaslitwidthof5

nm,toscanat500nmmin-1;synchronousmodefluorescencespectrawereacquiredata

constantwavelengthdifferenceof20nm.Aquartzcellwith1cmpathlengthwasused.

Theprocedure hasbeendescribedelsewhere[31,47,65].

31

Thespectrometer featuredautomatic correction for changes in source intensity

as a function of wavelength. Emission, excitation, and synchronous spectra of the

sampleswereobtainedinNMPsolutionforallofthesamples; onlysynchronousspectra

are shown. Solutionswere dilutedwith NMP to avoid self-absorption effects: dilution

wasincreaseduntilthefluorescencesignalintensitybegantobothdecreaseinintensity

andtherelativeintensitiesofthedifferentmaximainthespectraceasedtochange.The

UV-F spectra are displayed as peak normalised because this enables relative

comparisonsbetweenthesamples;theresultsarenotquantitative.

To ease the discussion of the UV-F data the results will be described with

referencetotheapproximatenumberofconjugatedaromaticringsthatwouldfluoresce

at the equivalent wavelength to the sample in question. This is based on a recently

noticed correlationbetween thepeakmaximum fromUV-F in synchronousmode, and

thenumberofconjugatedaromaticringsasdeterminedbyNMRandaveragestructural

parametercalculationsforcoalandpetroleumderivedsamples[47,60].Themaximum

intensity of fluorescence shifts steadily to longer wavelengths by about 30 nm per

additionalaromaticringinaconjugatedaromaticsystem,where1ring=270nm,2rings

=300nm,andsoon.

It shouldbenotedhowever, that someof thetar samples contain slightlymore

oxygen than the coal andpetroleumderived samples studied previously,which could

affect the UV-F results. The influence of oxygen on the UV-F spectrum of large PAH

molecules(>500u),however,isnotwellenoughunderstoodtobeabletocommenton

its effects in detail. The definitions outlined above are not a literal description of

chromophores such as those present in the tars but as ameans to discuss qualitative

differencesbetweensamplesandtheirdominantfeatures.Furtherdetailsareprovided

inS6regardingtheinterpretationoftheUV-Fresultsandtheinfluenceofoxygen.

3.3 Resultsanddiscussion

3.3.1 Tar-yields,GC-FIDandultimateanalysis

32

The tar TN2-0h and the aged samples were analyzed by means of the standard

approach (GC-FID and gravimetric yield) [30] as described in Section 2.5. TN2-0h was

analysedimmediatelyafterbeingthawedfromliquidnitrogenstorage;T5C andT20Cwere

both analysed after 20 hours and 6months of storage; the results are summarized in

Table2.TN2-0h wastheonlysamplethatwasanalysedbeforeanyprecipitationoccurred,

whereasprecipitatewasobservedforallothertarsolutionswithin20hoursofresting.

Theprecipitatewasremovedfromthesolutionsbeforetheywereanalysed.

The yields are presented in Table 2 for each tar as determined after different

periods of storage, results are normalised to grams per normal cubic meter of the

producergas(g/Nm3).ErrordeterminationisdescribedinS3.

Table2:Taryieldsafterdifferentdurationsofstorage(g/Nm3producergas)

SampleAgeTN2 T5C T20C

g/Nm3

GCTotal

0hours 5.2

20hours 4.6 4.9

6months 2.3 2.9

Precipitate

0hours 0.0

20hours 0.06 0.09

6months 0.03 0.07

Total 0.09 0.16

RecoveredDry

0hours 1.3

20hours 1.4 1.4

6months 1.8 1.9

GCRecoveredDry

0hours 0.6

20hours 0.6 0.6

6months 0.2 0.2

Volatiles 0hours 3.9

33

20hours 3.2 3.5

6months 0.5 1.0

TotalTar

0hours 5.9

20hours 5.5 5.8

6months 4.0 4.8

Relativedeviations+/- 5%forGC,15%forRecoveredDry,and30%forprecipitate.

Blankdenotesnotapplicable.

The mainfindingsfromstudyingthetaryields(Table2)werethatTN2-0h contains

thegreatestproportionofmolecules intheGCrangeandthe lowest inthegravimetric

range,withnooccurrenceofprecipitation.After6months storageT5C-6m andT20C-6m-L

showadecreaseinthequantityofGCrangemolecules(from~5g/Nm3 to~2.5g/Nm3)

andanincreaseinRecoveredDrytars(from1.3g/Nm3to1.8-1.9g/Nm3),comparedto

TN2-0h(errorsaregiven in the footnotetoTable2andS3).Precipitatewasobserved in

these samples. These results are an indication of a change inmass distributionof the

storedsamplestowardshigheraveragemolecularmasses,which is inaccordancewith

reportedstudiesofpyrolysisliquids[15,16,18,19,21,22].

ThedifferencesbetweentheGC results fromT5C-6m andT20C-6m-L wererelatively

minor after 20 hours of storage (Table 3). Both samples also exhibit the presence of

precipitate,toaslightlylargerextentinT20C-6m-L,althoughtheerrorinthedetermination

hastobeconsidered(cf.Table2andS3).Thetarsolutionswereregularlycheckedover

a 3 month period and no further precipitationwas observed; however, when the tar

solutionswerecollectedafter6monthsstorageprecipitatewasapparentinbothcases.

Todetermineiftheprecipitate(after20hoursand6months)containedanyGC

range species its dichloromethane soluble fraction was examined; no evidence of GC

rangemoleculeswasfound.Therefore,aggregation ofsmallmoleculescanberuledout

asamechanismfortheprecipitateformation.

Bycomparingthetaryieldsforthestoredtars(T5C andT20C-L)after20hoursand

6months,alarge reductionintheGCrangecomponentsisevident,by~2.0-2.5grams

34

per Nm3 forbothsamples(Table2).Ifthosemoleculeshadreactedtoformlargerones

they would be observed by an equivalent increase in the ‘Recovered Dry’ and

‘precipitate’yields.However,thegravimetricyieldsonlyincreasedby<1gramperNm3

andtheprecipitateswerelessthan200mgperNm3.Itislikelythatthediscrepancyis

duetosomeoftheGCrangecompoundsbeinglosttotheglasswareduringstorage.This

is thought to be due to attractionand adhesion ofmolecules to the glass surface, not

necessarilyduetoreactionsfollowedbylosetothesurfaces.Otherfactorsaffectingthe

quantification include evaporation of solvent and/or volatile compounds over the 6

monthstorageperiodandwaterformationduringtheageingreactions(notquantifiable

byGC).Infuturestudiesitwouldbebeneficialtodeterminethewatercontentusinge.g.

aKarl-Fischertitration[66].

The tar solutions were stored in ground glass stoppered bottles and sealed with

parafilm; however, some loses due to evaporation are still thought to have occurred

after 6months of storage and during sampling. This is thought to be the case as the

greatest changeswere in the naphthalene concentrationwhich is quitevolatile and is

unlikelytobereactiveunderthestorageconditions.

ToseeifGCcanprovideinformationontheprecipitationprocess,thethawedTN2-0h

solution was examined before any precipitation occurred, and then again after

precipitation hadoccurred (after filtration to remove the precipitate).Theprecipitate

formed within 14 hours of standing in the dark at 5°C after thawing. No significant

differenceswere observed for GC tars in TN2 before and after precipitation (Table 3).

Specifically, comparing these resultswith those from the20hoursold samples,which

hadprecipitatedmaterial,showsthatonlysmallchangesoccurredintheGCtars(Table

3). There is a slight decrease in acenaphthylene and phenanthrene, the decrease in

naphthalene is only slightly greater than the deviation. The only other statistically

significant changes are a slight increase in the amounts of benzo[a]pyrene,

indeno[1,2,3]pyreneandbenzo[g,h,i]pyrene.

Of the molecules detected in the GC range acenaphthylene and phenanthrene are

likely tobethemostreactive,but therewas littleevidence forthesehaving reacted in

the14houroldsample(Table3).Thissuggeststhedifferenceobservedinthe20hours

35

old sampleswere not related to the formationof precipitate andweremost probably

duetolossesofmaterialtotheglasswareandotherexperimentalerrors.Itwouldseem

that the precipitate is primarily formed by reactions between molecules that are not

detectedbyGC.

Theseresults andthetaryieldsarediscussed further inS3,asarethesourcesand

scalesoftheerrors.TheweightoftheprecipitatefromtarTN2 after14hourscouldnot

bedeterminedduetothesmallvolumeofsolutionfrozeninliquidnitrogen;hence,not

enoughmaterialcouldberecoveredtoweigh.

Table3,GC-FIDresultsforthetarsamplesbeforevacuumdrying,displayedasmgper

cubicmeterofproducergas.Relativeerrorswerelessthan+/- 5%.

SampleName

TN2 T5C T20C

mg/Nm3

0h 14h 20h 20h

Naphthalene 3000 3000 2600 2700

Acenaphthylene 710 700 640 670

Acenaphthene 10 10 10 10

Fluorene 50 60 40 50

Phenanthrene 460 450 380 410

Anthracene 90 90 80 80

Fluoranthene 230 240 220 240

Pyrene 240 250 220 240

Chrysene 30 30 20 30

Benzo[a]anthracene 30 40 30 30

Benzo[k]fluoranthene 30 40 30 30

36

Benzo[b]fluoranthene 10 10 10 10

Benzo[a]pyrene 40 60 40 40

Indeno[1,2,3-cd]pyrene 30 40 20 20

Dibenz[a,h]anthracene 5 10 5 5

Benzo[g,h,i]perylene 5 30 5 20

TotalEPA16 5000 5100 4400 4600

Unknowns 280 280 230 270

GCTotal 5300 5400 4600 4900

Ultimateanalysis(UA)wasusedtogain further informationabout thetars.The

resultsobtainedforsamplesTN2-0h,T5C-6mandT20C-6m-L (Table4)showanincreaseinthe

amount of oxygen in T20C-6m-L compared to T5C-6m and a corresponding decrease in

hydrogen.TN2-0h showsthehighestC/Hratio.TheamountofoxygeninTN2-0h wasgreater

thaninthecoldstoredtar,but lessthantheroomtemperaturesample.Thetrendsinthe

UAresultswererepeatable.Table5representsthesameresultsintermsofthenumber

ofatomsinamoleculewithamassof500u,itcanbeseenthatthenumberofoxygen

atomsrangesfrom0.2to1.4permolecule,whereasthereare36to38carbonatoms.

Taken together these results indicate that in TN2-0h (no precipitate) there are

significant quantities of highly aromatic PAH and oxygen containing PAH compounds.

Afterprecipitation(i.e.theagedsamples),lesscarbonisdetectedinthemoleculesthat

remained in solution andmore hydrogen, and the C/H decreased. Therefore itwould

seemthat themosthighlyaromaticcompoundshadprecipitated. InT5C-6m lessoxygen

wasfoundthaninthefreshtarwhichsuggestsmanyoftheO-PAHcompoundshavealso

precipitated.ThematerialthatremainedinsolutionforT5C-6m hasalowerC/Hratiothan

thedissolvedtarsinT20C-6m-L whichsuggeststhemoleculesarelessaromatic.InT20C-6m-L

moreoxygenwasobservedthan inTN2-0h which indicatesthissamplehasprecipitated

37

mainlyPAHsandpossibleundergoneadditional reactionswithoxygen from theairor

thesolventduringitsstorage.

The increase in nitrogen and sulphur content of dissolved tars observed with

ageingsuggests thatmoleculescontainingNorSdonotprecipitateasreadilyas those

that do not contain them; therefore N and S become concentrated in the solution.

However,theabsolutedifferencesinNandSbetweenthesamplesareminor,Table5.As

these findings are somewhat unexpected and only duplicate analyses of each tarwas

possible (due to limited amounts of sample), future studies should investigate these

aspectsmore closely to confirm these results. Inaddition, theprecipitate couldnotbe

examinedbyUA in this studybecause theamountof sample recovered from the filter

was insufficient for theanalysis. In future studies, adifferent filtrationmethodwill be

employed and larger sample volumes treated so that UA can be performed on the

precipitate.

Table4,Ultimateanalysisofthevacuumdriedtars,TN2-0h andafter6monthsstorage.

Element TN2-0h T5C-6m T20C-6m-L

Wt.% Wt.% Wt.%

C 91.5 86.6 86.9

H 5.5 11.6 7.3

N 0.3 0.6 0.6

S 0.5 0.8 0.7

O 2.2 0.5 4.5

Deviationestimatedat+/- 0.5%oftheabsolutevalue,basedontworepeats.

Table5,Numberofatomsinatarmoleculewithamassof500u,forTN2-0h andafter6

monthsstorage.

Element TN2-0h T5C-6m T20C-6m-L

38

Numberofatoms

C 38 36 36

H 28 58 37

N 0.1 0.2 0.2

S 0.08 0.13 0.11

O 0.7 0.2 1.4

C/H 1.4 0.6 1.0

Summarizing the results from the standard/commonly applied techniques (GC,

gravimetricweight andUA) for studying tars and oils fromgasification and pyrolysis,

changes to higher mass materials in the aged samples can be implied. However, the

differencesobservedcouldinprinciplealsobeduetootherreasons,suchaschangesin

polarity/solubilitywithageing.Thiscouldreducetheamountofcompoundsdetectedby

GCandproduceprecipitate.The information obtained from these techniquescan thus

only be regarded as indicative. In addition, there was no evidence for a significant

changeintheGC-FIDresultsintarTN2 beforeandafterprecipitationofmaterial,Table3.

To obtainmoredetailed information about changes inmolecularmass distribution or

structuralfeatures(conjugation)duetoageing,techniquessuchasSEC,LD-MSandUV-F

canproveuseful, aswillbedemonstratedbelow.

3.3.2 PCfractionationoftheRecoveredDrytars

Toaid the investigationof thevacuumdried tarsand thematerial they contain

beyond the range of GC the Recovered Dry material as defined in Section 2.5 was

fractionated via planar chromatography (PC). Themethodology behind this analytical

approachhas beenreportedelsewhere[31,47,56].ImagesofthePCplatesbearingthe

tarsamplesafterdevelopmentareshowninS2,FigureS2.1.Anexampleisshownbelow

inFigure1forthetarTN2-0h:

39

Figure1, PCplatebearingthevacuumdriedtarTN2-0h afterbeingelutedwith chloroform(F3)

followedby acetone (F2) and finallyheptane (F4).The imageon the left-hand sidewas taken

underwhitelight,andtheright-handsideunderUV-light(260nm).

ThePCfractionationisnotquantitative;nonetheless,itwaspossibletomakearelative

comparison between the apparent amounts of material in each mobility-fraction by

visual inspection. The apparent order of abundance of the PC mobility-fractions was

determined internally for each tar sample and is listed below (the main fluorescent

bandsareshowninbold):

TN2-0h F3c≥F4a≥F4b ≥F3b >F3a ≥F2>F1

T5C-6m F1>F2≥F4b ≥F4a>F3c(noF3b,noF3a)

T20C-6m-L F2≥F1>F3b >F3c>F4a(traceF4b,noF3a)

Where F1 = immobile, F2 = acetone mobile / chloroform and heptane immobile, F3

chloroformmobile/heptaneimmobile,F4heptanemobile;thelabelsc,bandareferto

materialwithdecreasingmobilityinthesamesolventasdenotedbythecorresponding

number– i.e.theyarenot associatedtorepeatedelutionwiththesamesolvent.

BycomparingtheapparentorderofabundanceforTN2-0h withthoseoftheaged

tars indicative information regarding changes in composition can be obtained. Only

significantdifferencesarementioned.TN2-0h containslessoftheimmobilematerial(F1)

F4bF4aF3c

F3b

F3a

F2

F1

40

relativetoitsotherfractionsthanwasthecaseforthe6monthagedsamples.Itwasalso

evidentthattherewassomefluorescentmaterialassociatedwiththeimmobilematerial

(F1)inTN2-0h thatwasnotseenfortheothersamples.

Theagedsamplesallappearedtocontainsimilaramountsofimmobilematerial,

althoughT5C-6m wastheonlytar thatappearedtocontainmorematerial in fractionF1

than inF2.Another cleardifference wasobservedbetweenT5C-6m andT20C-6m-L,where

themainfluorescentbandfromT5C-6m wasF4b(notobservedforT20C-6m-L)whileforT20C-

6m-L it was F3b (not observed for T5C-6m). In addition, the highly fluorescent material

observed inF3aofthe fresh tarwasnotseen inthe6monthagedtars.These findings

aredifficult to interpreton theirown (i.e.without thedataLD-MS, SEC andUV-Fwill

provide).However,theydoprovideclearevidenceofachangeincompositionofthetar

uponageingthatis dependentonthestorageconditions.Furtherdetailsregardingthe

planarchromatographyresultsaregiveninS2.

3.3.2.1SEC,LD-MSandUV-FanalysesofPCfractions

ToexaminetheRecoveredDrytarscomprehensivelytheirPCmobility-fractions

wereanalysedbySEC,LD-MSandUV-F.Thefindingsfromtheseanalyseswereusedto

determinethemassrangeandextentofaromaticconjugationforeachtar.Astheseare

notstandardreferencetechniquesforanalyzingtars(theyareinthevalidationphaseof

development) the completedata sets for each samplearepresented in the supporting

information(S7-S9).Forbrevity,asummaryofthekeyfindingsfromtheanalysisofthe

PC fractions is presented below with a complete account provided in S7. To aid the

comparisonoftheequivalentPCfractionfromdifferentsamplestheLD-MS,SECandUV-

F results are displayed by PC fraction in S8. To observe trends between PC fractions

fromasinglesampletheresultsarealsoshownbysample inS9.Someofthesefigures

arealsoshowninthearticletohighlightthemainfindings.

Table6 summarizes themass estimates from SECandLD-MS for thePC fractions; the

methodology for estimating average mass and mass ranges/distributions has been

described previously [47, 55, 56]. To discuss the results and draw conclusions the

following assumptions are used, but this does not however discount other possible

41

assumptions and interpretations. Further work is needed to confirm the assertions

outlinedbelow.

1. The results fromTN2-0h are considered to represent themoleculesoriginally

presentinthetar.

2. WhenUV-FsignalisobservedinaPCfractionfromTN2-0h butthesamesignal

isnotobservedintheequivalentfractionfromtheagedtars(forexamplePC

fraction F3c, Figure 2a), the molecules containing the chromophore

responsible for the fluorescence are assumed to have reacted (either with

another tarmolecule,orwith the solventoroxygen in theair).Hence these

molecules have different mobility during PC and are found in different

locationsorasprecipitate.

3. Themost probable ageing reactions are those involving tarmoleculeswith

highly conjugated aromatic systems eitherwith themselves or with oxygen

thatispresentinthesolventortheair.Thesereactionsareunlikelytocausea

significant change in the chromophore (conjugation) present in the original

tarmoleculewhenobservedbyUV-F if theyproceedas showninFigure3a.

WenotethattheadditionofacarboxylgrouptoaromaticsasshowninFigure

3b (in the presence of sunlight) can have two effects, depending on the

positiontheyhaveontheringsystem.Anaddedcarboxylicgroupcaneither

destroyconjugation,or increaseconjugation(causingaredshift) [67],cf. S6

forfurtherdetail.However,duetothelowoxygencontentsofthesetars,these

effectsareunlikelytosignificantlyinfluencetheresults.

4. Some of the UV-F results show that molecules with significantly greater

conjugationareproducedduringageing(suchasthoseintheprecipitateorF1

fromT5C-6m,discussed below).Inthesecases,itispossiblethatPAHmolecules

havereactedwithoneanotherresultinginlarger,highermassmoleculeswith

increased aromatic conjugation (the role of oxygen in this process is

42

unknown).Therearenoreportsof the formationofaromaticcarbon-carbon

bonds, or of deoxygenation as shown in Figure 3a, happening at room

temperature.Thereisalsolimitedinformationforoxygenactinginawaythat

increases conjugation in large aromatic molecules (>500 u) at room

temperature.However, the shift of fluorescence to higherwavelengthswith

ageinghastobearesultofoneoftheseinterpretations,orfromacombined

effect.

5. Fragmentation of some molecules occurs during LD-MS analysis. This is

probablyduetonon-conjugatedbonds(suchasaliphaticoroxygencontaining

bridges)betweenaromaticcores.Structuressuchasthesewereidentifiedin

recentstudiesofpetroleumsamplesasbeingthemostdifficulttoobserveby

LD-MSduetoexcessivefragmentation[47,56,61].

6. WhenSECshows evidence for largermolecules thanwitnessed fromLD-MS

this is an indication that the larger molecules fragment during ionization,

hence the mass of the parent molecule is not detected or is greatly

underestimated byLD-MS.

43

Figure2a,SynchronousUV-Fspectra(peaknormalised)ofPCfractionF3c.

Figure 2b, LD-MS spectra of PC fraction F3c at low and high laser power (left and right

respectively),noDIE,HMAvoltagewas10kV.

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

250 300 350 400 450 500 550 600

Wavelength / nm

Norm

alis

ed Inte

nsity

T-N2-0h F3c

T-5C-6m F4a + 3c

T-20C-6m-L F3c

Solvent

10 100 1000 100000

200

400

600

800

1000

1200

Ion C

ount

m/z

T-20C-6m-L LP35% T-5C-6m LP60% T-N2-0h LP65%

10 100 1000 100000

400

800

1200

1600

2000

Ion

Cou

nt

m/z

T-20C-6m-L LP45% T-5C-6m LP70% T-N2-0h LP80%

44

Figure2c,SECchromatograms(areanormalised)ofPCfractionF3cat300nm.

In addition to the assumptions above, information from the literature can also

giveinsightintopossiblemechanismsofreaction,andthisisdiscussedherebeforethe

dataareinterpretedbelow.Fromthelittlethatisknownaboutreactionmechanismsof

large PAHs, alkyl aromatics and oxygen containing aromatic molecules, it is the

molecules containing the largest conjugated chromophores that are thought to be the

mostreactive[47,60,68-70].Oxygencanreactwithmoleculessuchasthese,undermild

conditions (260-290°C [69]), and is thought to cause cross-linking through carboxylic

(Figure3a)andpossiblyetherbridgesbetweenaromaticcompounds[14,68].

It has been reported that PAHs with aliphatic hydrogen are themost reactive

during oxidative thermal treatments of oils and pitches [70, 71]; i.e. polymerization

reactionsbetweenPAHsappeartoproceedviacross-linkingwheretheactivesitesare

mainly located on the aliphatic groups (the radical chainmechanism). The amount of

oxygeninthereactionproductsincreasesbyafewweightpercent;whereasthechanges

inmassdistributionandaromaticityaremoreprofoundthanwouldbeexpecteddueto

theincorporationofoxygenalone[60,69,71].Theexactroleofoxygenremainsunclear.

Itisknownhowever,thatfurtherthermaltreatment(>440-460°C)ofthereaction

productsfromoxidativethermaltreatmentcanleadtoremovalofoxygenandresultsin

0.0E+00

1.0E-03

2.0E-03

3.0E-03

4.0E-03

5.0E-03

6.0E-03

7.0E-03

8.0E-03

9.0E-03

8 10 12 14 16 18 20 22 24 26 28

Time / minutes

Rela

tive Inte

nsity

T-N2-0h F3c

T-5C-6m F4a + 3c

T-20C-6m-L F3c

45

anincreaseinaromaticityasshowninFigure 3a,andthesizeoffusedaromaticcarbon

ringsystems[68-71].Thereisnoreportedevidencehowever,ofthishappeningatroom

temperature. It shouldbe noted that the incorporation of oxygen into alkyl aromatics

and their decomposition as described above, and shown in Figure 3a, would not

significantly alter the conjugation of the molecules and is unlikely to significantly

influence their UV-F spectrum (in synchronousmode). This is because previous NMR

andUV-Fstudiesofsamplessimilar tothosebeingexaminedhere(anthraceneoiland

their air-blown reactionproducts [31, 60,69,71]) found the aromatic coreswere the

leasteffectedduringthefirstphaseofair-blowingreactions;thereactionsoccurmainly

at aliphatic sites in themolecules that contain the largest aromatic chromophores, as

mentionedabove.Thesetypesofreactionsproceedintheabsenceoflight.

In the presence of sunlight a different mechanism can occur leading to

incorporationofoxygenintoPAHmoleculesthroughreactionswithoxygenfromtheair

[72-74].Themechanismproposedisthroughsingletoxygenformedbyenergytransfer

fromaPAHmoleculeinitstripletstate(formedviaUV-VISphoto-excitationofthePAH

groundstatefollowedbyintersystemcrossing).TheoxygencanbondtothePAHgiving

peroxideswhichareinturnphotolysetogivecarbonylsandeventuallyhydroxyls[72].

Figure 3b displays a simplified example of the photo-oxidation of phenanthrene to

phenanthrenequinone[72].Despitethedepictedapparentsimplicityitisimportantto

recall that the formation and breaking of peroxides implies the possibility of a large

rangeof(radical)reactionsincludingfurtheroxidationwithgroundstateoxygentoform

e.g.moreperoxides,radicals,carbonylsandhydroxylgroups,butalsofragmentationand

polymerisationofmolecules.

Hydroxyl groupsonPAHmolecules can react further leading topolymerization

processes occurring as described in reference [14]; this provides a comprehensive

reviewofpossiblereactionsofoxygenatedhydrocarbonsthatoccurinbiomasspyrolysis

oils. However, most of the reactions described therein are unlikely to occur in the

gasificationtarsbeingexaminedduetotheirmuch loweroxygencontents(lessthan5

wt%)thanbiomasspyrolysisoils(15-30wt%).Therefore,theinfluenceofoxygenonthe

resultsfromthesetarswillberelativelyminor. Inaddition,withoutinformationonthe

oxygencontentoftheprecipitateitisfutiletospeculateonoxygen’sroleintheageingof

46

this tar, or without NMR, FT-ICR-MS and FT-IR data. It is intended to address these

aspectsinfuturestudies.

Ingasificationandpyrolysistarsandoilsfreeradicalsarealsolikelytoplayarole

inageingreactions;however,noinformationisavailableontheiryieldinthesetarsor

theirroleduringageing.Additionally,thetarmoleculesrecoveredfromgasificationare

not thermodynamically stable, as their residence time at the high temperature in the

gasifierisinsufficientforequilibriumtobereached[1];itisthereforealsopossiblethat

part of the aging involves molecules re-arranging slowly to more energetically

favourable configurations, that they would have reached more quickly at high

temperature in the gasifier. However, detailed information on these processes is not

available forthesizesandstructuresofmoleculesbeing investigatedhere. Inaddition,

tracesofashorcharcouldremaininthetarsolutionevenafterfiltrationandifpresent

couldwellplayaroleintheageingprocesses.

Figure 3a, Simplified schematic example of the incorporation of oxygen into alkyl

aromaticsanddecompositionleadingtocross-linkingorincreasedlevelsofaromaticity

[68].

CH3

COOH

O

O

O

O

O

+ O2

+ O2

-CO2

-CO2

-CO2

-CO

47

Figure 3b, Example of the photo-oxidation of phenanthrene to phenanthrene quinone

[72].

Table6,Molecularmassestimates forthePCfractionsof theRecoveredDrytars from

SECandLD-MS.Theresultsarepresentedaspeakmaximumvaluesforthemainbands

ofsignal inorderofabundance; for theLD-MSresults therangeoverwhich ionswere

observedisalsogiven.

TN2-0h T5C-6m T20C-6m-L

SEC LD-MS SEC LD-MS SEC LD-MS

Fraction Definition Mass/u Massm/z Mass/u Massm/z Mass/u Massm/z

F4bPeakMax 150 210 140 200

Range <200-400 <200-300

F4a

PeakMax 190 260 180+ 250 160 325

2nd Peak 1100 500 1000+ 600 1000 650

Range <200-600 220-1500 250-1250

F3c

PeakMax 220 275 180+ 330 180 350

2nd Peak 900 1000+ 1000 650

Range 250-550 200-1100 220-1100

F3b

PeakMax 275 <200 460

2nd Peak 900 >2000 <200

3rd Peak >2000

Range <200-400 <200-750

F3aPeakMax 220 <200

2nd Peak 900

O

O

O2

UV-light

48

3rd Peak >2000

Range <200-800

F2

PeakMax 210 550 200 700 650 600

2nd Peak 650 600 >2000

3rd Peak >2000 >2000

Range 200-1300 220-1500 220-1500

F1

PeakMax >2000 650 >2000 725 >2000 725

2nd Peak 1000 900 1300 1100

Range 200-2000 250-2000 300-1900

+SECresultsforF3candF4afromT5C-6m arefromthecombinedF3candF4afractionsas

theycouldnotberecoveredseparately.

Combining SEC, LD-MS, and UV-F analysis: When results for equivalent PC

mobility-fractionswhere compared from the different samples it was generally found

theydidnotcontainidenticalmolecularmassesorstructures(aswasanticipated).Often

they shared common features but show significant differences in either mass

distribution or extents of conjugation (or occasionally both) due to the ageing

conditions. It is thus an important finding that it is not correct to simply assume that

equivalent PC mobility-fractions represent the same component(s) in different tar

samples.

Forexample,Figures2a-2cshowtheUV-F,LD-MSandSECspectra,respectively,

forPCfractionF3cfromthefreshandagedtars.AsignificantlydifferentUV-Fspectrum

was obtained from F3c from the fresh tar than the aged tars. Larger aromatic

chromophoreswere observed in TN2-0h than in the aged tars. Differences between the

LD-MSspectra are alsoapparent,where F3c fromTN2-0h showed the lowestm/z peak

maximum, a narrower distribution at low laser power and extensive fragmentation

when a higher laser power was used as well as some new ions at m/z >400. T5C-6m

showed a similar m/z peak maximum to TN2-0h but has a much greater tendency to

fragment during ionization and produce awiderm/z range of ions.No fragment ions

49

wereobservedfromPCfractionF3cofT20C-6m-L andagreaterabundanceofhighermass

ions(m/z>400)weredetectedherethanintheothertars,Figure2b.

SEC shows that fraction F3c from the fresh tar contains more larger-sized

moleculesthanthesamefractionfromtheagedtars(Figure2c).Thetwoagedtarsgave

almost identical SEC chromatograms; whereas their LD-MS spectra are noticeable

different. This is partly due to the lower resolution of SEC than LD-MS and may be

evidenceofthelargestmoleculesobservedbySEC(inTN2-0h)fragmentingduringLD-MS

analysis.Furtherworkisneededtobetterunderstandthedifferencesobservedbetween

theSECandLD-MSresults.

It is possible that themolecules containing the largest aromatic chromophores

present in F3c of the fresh tar react upon ageing and are no longer found in the F3c

fractionof theaged tars.Although, thetwoaged tarsappeartobeverysimilarbySEC

andUV-F,LD-MSindentifiescleardifferences.ThemoleculesinT20C-6m-L aremorestable

during ionisationwhilst those in T5C-6m are less stable. Thematerial observed in this

fraction in the aged samples contains molecules with less-conjugated chromophores

thanwerepresentinsamefractionfromTN2-0h.

It shouldbenoted that the trendsobserved forPC fractionF3cof thedifferent

tarsarenotrepresentativeofthebehaviouroftheotherfractions;eachfractionshowed

subtle anduniquedifferences.However, forbrevityonlya summaryof the findings is

presentedbelowwiththefullaccountprovidedinS7andthefigurespresentedinS8.

InspectionofthecompletesetofLD-MSresultsrevealsfurtherdifferencesinthe

tendency for fragmentation or susceptibility towards laser ionisation (i.e. strength of

laser power needed to observe satisfactory signal) for the different PC fractions, and

different tars. T20C-6m-L was least prone to fragmentation and its ions were easiest to

observe.T5C-6m showedthegreatesttendencytofragmentandwasthemostdifficultto

observe.Ingeneral,themoremobilefractionsweremostlikelytofragment.TheLD-MS

resultsalsoprovidesomeconfirmationofthemassestimatesderivedfromSEC(cf.Table

6andS7,TablesS7.1– S7.7).

Allthetarsamplesshowedevidenceofcontainingmoleculeswithmasses<200uto

>2000u.Thereweredistinctdifferenceshowever,intheirmassdistributionbeforeand

after ageingdepending also on the storage conditions. In general, therewas a shift to

50

highermassesintheagedtarscomparedtothefreshtar,andforT20C-6m-L relativetoT5C-

6m i.e.heaviermaterial inwarmerconditionsandthepresenceof light.Manyof thePC

fractionswerecomposedentirelyofmaterialsbeyondtherangeofGC,suchasfractions

F1andF2,cf.Table6,FiguresS8.5aandS8.6a.

Examining the extents of conjugation present in the tars (via UV-F) shows that

conjugated4-6aromatic ring systemsdominate inalmostall cases. It shouldbenoted

however, that fluorescence quantum yields are dramatically lower for conjugated

aromaticringsystemsthatcontainmorethan~8ringsasdiscussedinS6andelsewhere

[47,55],whichmeans,ifpresent,theyarealwaysunderestimated.Thefreshtarcontains

morearomaticmoleculeswithhighdegreeofconjugationthantheaged tarswhenthe

mobilePC fractions (F3 and F4) are compared. Correspondingly,when comparing the

leastmobilefractions(F1andF2),thereverseistypicallyfoundfortheagedtars,where

theycontainmorearomaticmoleculeswithhighdegreeofconjugationthaninthefresh

tar (there are exceptions, cf. S7). PC fraction F1 of T5C-6m contained the largest

conjugated systems of all the tar samples (equivalent to an average of 8 conjugated

aromaticrings).

When the UV-F spectra for all the PC mobility-fractions from all the tars are

consideredthefollowingobservationscanbemade:

i) The chromophores responsible for the fluorescence observed at 340-360 nm in

TN2-0h F4b move to F3b in T20C-6m-L, and possibly to F1; whereas for T5C-6m they

remaininF4b.

ii) The chromophores responsible for the fluorescence observed at 440-550 nm in

TN2-0h F3cappearstomovetoF3bandF1inT20C-6m-L;andtoF1forT5C-6m.

iii) Thechromophores identified in theprecipitate (cf. Section3.4,Figures6a-d)and

T5C-6m F1 (Figure 4a) showed maximum fluorescence intensity at >450 nm (>6

conjugated aromatic rings); these chromophores seem to originate from PC

51

fractions F4a, F3c, F3a and F2 in TN2-0h, (Figure 4b); the evidence for this is

discussedfurtherinS7andinthefinalsummary (Section3.5).

Figure 4a, Synchronous UV-F spectra (peak normalised) of PC fraction F1; for TN2-0h

therewasweaksignalduetosamplelowabundance.

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

250 300 350 400 450 500 550 600

Wavelength / nm

No

rmalis

ed I

nte

nsi

ty

T-N2-0h F1

T-5C-6m F1

T-20C-6m-L F1Solvent

52

Figure4b,SynchronousUV-Fspectra(peaknormalised)ofthePCfractionsfromTN2-0h.

AlthoughtheUV-Fresultsaredifficult tointerprettheydoshowthatchangesin

conjugationoccurduringageinginadditiontotherebeingachangeinthedistributionof

thetarsonthePCplate.However,itisnotpossibletodeterminewhetherthisissolely

due to changes in aromaticity or a result of the incorporation of oxygen into the

molecules as mentioned previously. It was often the case that the biggest differences

betweenanequivalentPCfractionfromthefreshandagedtarswereobservedbyUV-F.

Information on relative extents of conjugation is difficult to obtain by othermethods.

NMRcouldprovidemoredetailedinformationandwouldgreatlyaidthe interpretation

oftheUV-Fresults;however,thedifficultyofrecoveringtherequisitequantityofsample

fromplanarchromatographyhampersstudiesviaNMR.

In future studies, column chromatography fractionation could be performed to

obtainquantitative information;however, thiswillnotbedirectly comparable toaPC

fractionation as the mechanisms of separation differ. It would be best to do this in

parallel to a PC fractionation as it is more difficult to perform LD-MS analysis on

fractionsrecoveredfromcolumnchromatographythanfromPC[31].

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

250 300 350 400 450 500 550 600

Wavelength / nm

Norm

alis

ed I

nte

nsi

tyT-N2-0h F4b

T-N2-0h F4a

T-N2-0h F3c

T-N2-0h F3b

T-N2-0h F3a

T-N2-0h F2

T-N2-0h F1

F3c

F3a

F4b

F2

F3b

F1

Solvent

53

From a visual comparison of the PC plates bearing the fractionated tar some

initial assertions were drawn. TN2-0h had very little immobile material (F1), larger

quantitiesofF3candF2,andnoprecipitatewasobservedbeforethesamplewasdried.

Therefore, it ispossible that itwasthedarkbandofmaterialdenotedasF3cthatwas

involvedinreactionswithF4bandF3btogeneratethematerialthatislabelledF1and

precipitate in the aged samples. The interpretation of the UV-F results as described

above(andinS7)concurswiththeseassertionswhencombinedwiththemassestimates

(from SEC and LD-MS, S9), i.e. these are the PC fractions that showed the biggest

differencesafterageing.ItissuggestedthatitwasmainlyPCfractionsF4bandF3cthat

react with one another to generate F1 and the precipitated materials during ageing;

fractionsF3b,F3aandperhapsF2alsoseem tobe involved inthesereactionsbut toa

lesserextent.

Theresultsalsoimplythatdifferentprocessesoccurredwhenthetarwasstored

atroomtemperatureinthepresenceofindirectsunlightthanwhenstoredat5°Cinthe

absenceoflight.Thisisasopposedtothesamechangesoccurringinbothsamples,but

slower in the caseof the sample stored at5°C.Whenstored at5°Call of thematerial

seen asF3b inTN2-0h disappears, probablydue to reactions formingmaterial that can

mobilize to another fraction or precipitate. Meanwhile, storing the tar at room

temperatureandexposedtoindirectsunlightresultedintheabsenceoffractionF4band

achangedincomposition ofF3b;significantdifferenceswerealsoobservedbetweenthe

respective F1 and F2 fractions. These differences are probably related to the altered

oxygencontentsoftheagedtars(Table4)andtheinfluenceofsunlight,oralackofit,on

the reaction mechanism available to the tarmolecules, as discussed earlier in Section

3.2.1.

Itisinterestingherealsotospeculateifthephotochemistryofbiomassmaygive

cluestothephotochemistryofitstars.Thephotochemistryofbiomassandinparticular

ligninhasbeenwidelystudied,duetoitssusceptibilitytoabsorbbothvisibleandnear

UVlight[75].Theresultingexcitedstateshaveshowntoenablelignindepolymerisation

inthepresenceofoxygen.Afterlightabsorptionthemoleculesintheirexcitedstatecan

formradicalspecies,usuallybycleavageofthecommonβ-O-4aryletherlinkage,which

thenreadilyreactwithoxygentoformnewchromophores.Atthetemperatureatwhich

54

the gasification tarswereproduced here, the lignin in the pine is highlymodified but

maycontainsomeβ-O-4aryletherlinkagespresentintheoriginalbiomass.

No oxygenated species were identified from the GC analysis of the tars. This

impliesoxygen is concentrated in themoleculesbeyondthe rangeofGC.On theother

hand,oxygenatedhydrocarbonsareknowntoco-elutewithpurehydrocarbonsduring

GC and this may explain their absence [76]. These aspects warrant further

investigations;andarediscussedfurtherinS7.

An unexpected finding was that tar T20C-6 m-L, which has the highest oxygen

content of the three tars and a greater C/H ratio than T5C-6 m, contains aromatic

moleculeswithlowdegreeofconjugation.EvenmoresurprisingisthatT20C-6m-L wasthe

moststabletowardsLD-MSanalysis(leastpronetofragmentation,withthemosteasily

observable ions). In contrast, T5C-6m has the least oxygen (although the absolute

difference is small, Tables 4 and 5) and the lowest C/H ratio but contains the most

aromatic molecules with the highest degrees of conjugation of the three tars; it also

showedthegreatesttendencytowardsfragmentationduringLD-MSanalysis.

Theresultsdescribedaboveappeartobeevidenceofphoto-oxidationreactions

havingoccurredduringstorageofthetar.Thepresenceofindirectsunlightresultedin

partialphoto-oxidation of tarT20C-6m-L, via themechanismdescribed earlier in Section

3.2.1anddisplayedinFigure3b.Itwouldseemthattheincorporationofoxygengroups

into these predominantly aromatic molecules improves there susceptibility towards

laser desorption and ionization, and destroys (or reduces) aromatic conjugation.

WhereastarT5C-6m wasnotexposedto light, therefore itreactedmainlyviaadifferent

mechanismleadingtomoleculeswithgreatermolecularmassesandlessoxygencontent

(remaining in solution). These molecules also contain aromatic chromophores which

havehighdegreesofconjugation,buttheywerenotstabletowardslaserdesorptionand

ionisation.

Theseobservations show that theC/H ratiowhich is typicallyused toestimate

the aromaticity of tar samples does not give information regarding the extents of

aromaticconjugation.Forexample, theC/HratioshowsatrendofTN2-0h>T20C-6m-L>T5C-

6m, however, UV-F showed the extent of conjugation to be T5C-6m>TN2-0h>T20C-6m-L. In

addition, therelationshipbetweenextentof aromaticconjugationandsusceptibility to

55

LD-MSanalysisiscontrarytothatwhichwouldbeexpectedbasedonpreviousstudiesof

lignin (in preparation for publication), petroleum, bitumen and coal derived samples

[31,47, 55,56,60].Where lignin is themost difficult toanalyse followedbybitumen,

with petroleum and coal-derived samples the easiest. However, in those previous

studies all the samples were stable under the same conditions as used to age the

gasificationtarsinthepresentstudy.Additionalstudiesarerequiredtounderstandthe

relevanceoftheseobservations.

3.3.3 BulkanalysesofRecoveredDrytar

To enable a significantly more robust interpretation of LD-MS data, the PC

fractiondatamustbeconsidered intandemwithbulksampleLD-MSdata,ratherthan

eitherof these approachesalone.Thishasbeendiscussed indetail elsewhere [31,47,

56]; briefly, whilst information gained from the bulk tars is considered more

representativethanresultsfromPCfractionationalone,theLD-MSanalysisofthebulk

tarsgiveshighlyvariableresultsdependingontheconditionsused[31,47,56].This is

demonstrated in Figures S10.1 and S10.2 for T5C-6m and T20C-6m-L. Therefore combined

informationfrombothdatasetsisusedininterpretationofresults.

Figures5a-cshowthechromatogramsandspectrafromtheanalysisofthebulk

RecoveredDrytarsTN2-0h,T5C-6m andT20C-6m-L bySEC,LD-MS andUV-F,respectively.The

LD-MS spectra displayed in Figure 5bwere selected after analysing low to high laser

powersandtheinfluenceofdifferent‘delayedionisationextraction’timesontheresults,

anexampleofthesetestsandtheirresultsisgiveninS10.Thechoiceofwhichspectrato

usetorepresentthesamplewasbasedontheinformationgainedfromtheexamination

ofthePCfractions,cf.S7toS9,asdescribedpreviously[56].

SEC reveals that TN2-0h contains smaller sized molecules, on average, than the

aged tars (Figure 5a). The aged tars (T5C-6m and T20C-6m-L) show very similar SEC

chromatograms to one-another and give almost identical mass spectra from LD-MS

(Figure 5b), with T20C-6m-L possibly containing more higher-mass (m/z) ions. The

differencesobservedaresubtlebutnoticeable.Withtheinformationobtainedfromthe

analysis of the PC fractions, it can be safely concluded that there is a trend towards

higheraveragemassmaterials:TN2-0h <T5C-6m <T20C-6m-L.

56

UV-F spectra of the different bulk tars are similar (main peak ~390 nm,

equivalent to 5 aromatic rings – Figure 5c); the main difference is that TN2-0h shows

slightlymorefluorescenceatwavelengthsgreaterthan400nm(>5rings),followedby

T5C-6m withT20C-6m-L givingtheweakestsignalinthatregion.Thiscouldberelatedtothe

aromaticmoleculeswiththehighestdegreesofconjugation,originallypresentinTN2-0h,

having reacted and precipitated from the aged samples. This explains thedecrease in

fluorescence at >400 nm in those samples (this confirms the findings from the PC

fractions– S7).

WhentheUV-FspectrumfromthebulksampleiscomparedtoitsPCfractionsit

can be seen that the bulk samples resemble theirF3c and F4a PC fractions (themost

mobile,lowmassmaterials).Thisisconsistentwithpreviousobservationswheresignal

from larger aromatic chromophores is difficult to observe in the presence of less

conjugatedaromatics.Thisisthoughttoberelatedtoareductioninthequantumyields

asthearomaticsystemsbecomemoreconjugated[55,56,61,77].

The analysis of the bulk tars reveals that only limited information could be

obtained about the changes that have occurred during ageing and highlights the

importanceof fractionatingthetarspriortotheiranalysis. It is important tonotethat

fractionation of the tars by PC cannot readily be done quantitatively; therefore, it is

beneficialtoanalyzethebulksamplesaswellastheirPCfractions.

57

Figure5a,Areanormalsizeexclusionchromatogramsofthebulk tarsTN2-0h,T5C-6m,andT20C-6m-L,

detectionat300nm.

0.0E+00

1.0E-03

2.0E-03

3.0E-03

4.0E-03

5.0E-03

6.0E-03

7.0E-03

8.0E-03

9.0E-03

1.0E-02

8 10 12 14 16 18 20 22 24 26 28

Time / minutes

Re

lativ

e Inte

nsi

ty

T-N2-0h

T-5C-6m

T-20C-6m-L

10 100 1000 100000

200

400

600

800

Ion

Co

unt

m/z

T-N2-0h LP40% H8 kV T-5C-6m LP35% H9 kV T-20C-6m-L LP40% H8 kV

58

Figure5b,LD-MS spectraofbulktarsTN2-0h,T5C-6m,andT20C-6m-L withnoDIE, laserpower (LP)

andhighmassaccelerator(H)voltageareshowninthelegend.

Figure5c,PeaknormalisedsynchronousUV-FspectraofthebulktarsTN2-0h,T5C-6m,andT20C-6m-L.

3.3.4 Precipitatedmaterialsanalyses

SECandUV-Fresultsfromtheanalysisoftheprecipitatesthatformedwithin20hours,

andafter6monthsofstorageofthetarsaredisplayedinFigures6a-d.Itshouldbenotedthat

theprecipitatedmaterialswere fullysoluble inNMPat theconcentrationsused.ForTN2-0h the

precipitate was recovered after 14 hours storage in a fridge after being thawed from liquid

nitrogen – this sample is labelled TN2-14h-ppt. LD-MS was not applied to the analysis of the

precipitates as part of this scoping study; however, it would provide additional useful

informationandwillbeconsideredinfuture.

Figure6ashowstheSECchromatogramsfromtheprecipitaterecoveredafter20hours

storageofT5C (T5C-20h-ppt)andT20C (T20C-20h-ppt)and14hoursforTN2 (TN2-14h-ppt),alongsidetarT20C-

6m-L tohighlightthedifferencesbetweenthespectrafromthetarsandtheprecipitates.Evidence

ofasmallbutsteadyincreaseinthesizeofthemoleculespresentintheprecipitatecanbeseen

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

250 300 350 400 450 500 550 600 650Wavelength / nm

No

rma

lise

d I

nte

nsity

T-N2-0h

T-5C-6m

T-20C-6m-L

Solvent

59

fromTN2-14h-ppt <T5C-20h-ppt <T20C-20h-ppt.Alltheprecipitatescontainmoleculesoflargersizethan

thosepresentinthetars.

Comparingtheprecipitatefrom20hoursstoragetothatrecoveredafter6monthsshows

virtuallynochangeinsizedistributionforT5C-6m;however,forT20C-6m-L there isashifttolarger

moleculesintheprecipitaterecoveredafter6months,cf.Figures6aand6b.

UV-F reveals that all the 20 hour precipitates gave spectra almost identical to one

another,andtheprecipitatefromTN2 (14hourprecipitate)showedslightlylessfluorescenceat

wavelengthsgreaterthan470nm(>7conjugatedaromaticrings),Figure6c.Thespectrafrom

theprecipitatesweresignificantlydifferenttothosefromthebulktarsortheirPCfractions,with

theexceptionofT5C-6m PCfractionF1whichgaveanalmostidenticalspectrum(Figure4a).

ComparingtheUV-Fspectraobtainedfromtheprecipitateafter6monthsstoragetothat

after20hours shows that forT5C-6m there isnodiscernabledifference; forT20C-6m-L there is an

increaseinfluorescenceat~500nm,cf.Figures6cand6d.Theanalysisoftheprecipitateshows

thatthismaterialwasnotoriginallypresentinthefreshtarandmustthereforebetheproductof

reactions between the tar molecules (or with the solvent, or dissolved oxygen) during their

storage.Itislikelythatgreaterdifferencescouldbedetectedbetweentheprecipitatedmaterials

ifplanarchromatographyandLD-MShadalsobeenused,asfortheliquidtarsamples;however,

that was beyond the scope of the present investigation. In addition, ultimate analysis would

providevaluableinformationandwillbeusedinfuturestudies.

60

Figure6a,Areanormalsizeexclusionchromatogramsoftheprecipitateafter20hoursfrom,T5C

andT20C, andTN2 recoveredafter14hoursafterthawingfromliquidnitrogen,thebulktarT20C-6m-

Lisshownforcomparison,detectionat300nm.

Figure6b,AreanormalsizeexclusionchromatogramsoftheprecipitatefromT20C-6m-L,T5C-6m

after6monthsstorage,andthePPTfromTN2-14h (after14hoursstorage).

0.0E+00

1.0E-03

2.0E-03

3.0E-03

4.0E-03

5.0E-03

6.0E-03

7.0E-03

8.0E-03

8 10 12 14 16 18 20 22 24 26 28Time / mintues

Rela

tive I

nte

nsi

tyT-N2-14h-ppt

T-5C-20h-ppt

T-20C-20h-L-ppt

T-20C-6m-L

0.0E+00

1.0E-03

2.0E-03

3.0E-03

4.0E-03

5.0E-03

6.0E-03

8 10 12 14 16 18 20 22 24 26 28Time / minutes

Re

lative I

nte

nst

iy

T-N2-14h-ppt

T-5C-6m-ppt

T-20C-6m-L-ppt

61

Figure6c,PeaknormalisedsynchronousUV-Fspectraoftheprecipitateafter20hoursstorageof

the tar solutions T5C and T20C, and after 14 hours for TN2, alongwith the fresh tar T1N2-0h for

comparison.

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

250 300 350 400 450 500 550 600 650Wavelength / nm

No

rmalis

ed I

nte

nsity

T-N2-0h

T-N2-14h-ppt

T-5C-20h-ppt

T-20C-20h-L-ppt

Solvent

62

Figure6d,PeaknormalisedsynchronousUV-Fspectraoftheprecipitateafter6monthsstorage

ofthetarsolutionsT20C-6m-L,andT5C-6m; and14hrsprecipitatefromTN2-14h.

3.3.5Summary

When all the findings from the analyses of the tars, their PC fractions and the

precipitatedmaterials are considered together inferences canbemade regarding the changes

thatoccurredduringageingunderdifferentstorageconditions.

TheresultsstronglysuggestevidenceofagingreactionstakingplaceandthatPCfraction

F1intheagedtarsiscomposedmostlyofreactionproductsofaging.Thisisbecausetherewasa

verylowabundanceofF1inthefreshtar,solittleso,that itwasdifficulttoobtainsatisfactory

SEC and UV-F results. There is evidence that F1 of the aged tars contains higher molecular

masses and different chromophores than the fresh tar (Figures 4a and S8.6a-c); this is most

evidentforPCfractionF1fromT5C-6m.Therewasnoprecipitationnotedinthefreshtarandthe

precipitatefromtheagedsampleswasfoundtocontainmoleculesofhighermass(largersize)

andwithgreaterconjugationthanthematerialinthefreshtar(Figures5a-cand6a-d).Thisisall

evidenceofaseriesofreactionsoccurringduringageingwhichresultinanincreaseinmolecular

massandconjugationinthetars.Theprecipitatesrecoveredafter20hoursfromtheT5C-6m and

T20C-6m-L samples contained molecules of similar size distribution and extents of conjugation.

After6monthstheprecipitatefromT5C-6m showedlittlechangeby SECorUV-F,whereasT20C-6m-L

had increased in size (mass) and contained more aromatic molecules with high degree of

conjugation(Figures6a-d).

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

250 300 350 400 450 500 550 600 650Wavelength / nm

Norm

alis

ed Inte

nsi

tyT-N2-14h-ppt

T-5C-6m-ppt

T-20C-6m-L-ppt

Solvent

63

Itwasdeterminedthatthelargestaromaticchromophores(8conjugatedaromaticrings,

UV-Fpeakmax475nm) andlargestmoleculeswerefoundinPCfractionF1ofT5C-6m(Figures4a,

6b,6dandS8.6b).PCfractionF1fromT20C-6m-L containedmoleculesapproximatelythesamesize

asF1fromT5C-6m,howeverthechromophoresaremuchlessconjugated,onaverage(5aromatic

ringsversus8).TheprecipitaterecoveredfromT20C-6m-Lcontainedthenextlargestmoleculesand

sizes of conjugated chromophores (7-8 rings, 460 nm),whichwere slightly larger than those

fromtheT5C-6m precipitate.AtthesametimeT20C-6m-L contained less conjugatedchromophores

thanT5C-6m orthefreshtar.

LD-MS analysis of the PC fractions shows that fraction F1 from the fresh tar has an

average mass (peak max. m/z) of ~650 and the aged tars m/z ~725, with a shift in the

distributiontowardshighermasses in theagedtars,extendingtoat leastm/z2000(Table6).

ComparingthesemassestotheaveragenumberofaromaticringsasdeterminedbyUV-F(TN2-0h

5aromaticrings,T5C-6m 8ringsandT20C-6m-L 5rings)itcanbedeterminedthatthesemolecules

probablycontainmorethanonearomaticcore(as5rings=~250uand8rings=~400u), i.e.

more probably archipelago- and island-like structural configurations rather than continental

[61]. However, NMR and ultimate analysis data are needed to determine the amounts of

aromaticandaliphaticmaterial,ideallyforeachPCfraction,butthereweresample limitationsto

conductingNMR.

WhentheresultsforbulktarsandtheirPCfractionsareconsideredtogether,itcanbe

determinedthatTN2-0h containssignificantquantitiesofaromaticmoleculeswithhighdegreeof

conjugationandafterprecipitation(i.e.theagedsamples)theC/Hratiodecreased.Thereforeit

wouldseemthatthemosthighlyaromaticcompoundshadprecipitated.InT20C-6m-L moreoxygen

was observed than in TN2-0h which indicates mainly PAHs had precipitated and possibly

undergoneadditionalreactionswithoxygenfromtheairorthesolventduringitsstorage.InT5C-

6m lessoxygenwasfoundthaninthefreshtar.ThissuggestsmanyoftheoxygencontainingPAH

compoundshavealsoprecipitated.TheC/HratioofT5C-6m (0.62)andT20C-6m-L (0.99)impliesthat

the latter is more aromatic; however, UV-F reveals it is T5C-6m that contains molecules with

greater extents of conjugation (Figure S9.1c, S9.2c and S9.3c). T20C-6m-L shows evidence of

containing slightly highermolecularmasses and largermolecules. This sample ismuchmore

stableduringLD-MSanalysisthanT5C-6m(FigureS9.2a-bandS9.3a-b).

The results suggest that the ageing process for T20C-6m-L proceeded via a different

mechanismthanthatforT5C-6m. Indeed, it isprobablethat thepresenceof indirectsunlighton

T20C-6m-L inducesphoto-oxidationreactionswhichcannotoccurinT5C-6m storedcoldinthedark.It

islessprobablethattheresultsaresimplyduetokineticeffectswheretarT20C-6m-L hasreacheda

64

thermodynamically more stable state than T5C-6m due to enhanced kinetics from higher

temperatures. In other words, it is unlikely that the ageing / polymerization reactions are

progressingthroughthesamemechanismbuthaveprogressedtoalesserextentinT5C-6m thanin

T20C-6m-L. It is rather the case that there are a number of reaction mechanisms occurring

simultaneously with competing influences from kinetic, thermodynamic, and photochemical

effectswhensamplesarestoredexposedtolight.

T5C-6m and T20C-6m-L did not show statistically significant differences in degree of

precipitation.Itisnotpossibletoderiveconclusiveinformationregardingtheageingmechanism

fromthisstudyduetothelackofkeyinformationsuchastheoxygencontentoftheprecipitate

andideallyforthePCfractions.Inaddition,NMRstudiesarerequiredtoaidtheinterpretationof

the UV-F results and to reveal mechanistic information. However, to perform all of these

analysescomprehensivelyalargeamountoftar(>10g)wouldberequired.

However,theinvestigationsheredemonstratethatdetailedinformationcanbeobtained

bythecombineduseofSEC,LD-MSandUV-Fmethods,whichcanidentifydifferencesbetween

tars that are difficult to obtain by other approaches. UV-F and LD-MS are very sensitive

techniquesthatcanprovideevidence for thepresenceof largemolecules(>1000u)andthose

with highly-conjugated chromophores (>7 conjugated aromatic rings equivalent) that are

formed during ageing. The methods are able to identify these materials, even when they are

presentinlowabundance,ifusedinconjunctionwithplanarchromatography.Knowledgeofthe

presence of these high mass molecules and/or aromatic molecules with high degrees of

conjugation in tars and oils, even in low quantities, is essential when considering their

properties and furtheruse.These tars andoilshave applications inmanyprocesses including

combustion, reforming and upgrading but these large molecules can cause problems of char

formation/cokinginengines/boilersorpoisoningofcatalysts.

3.4 Conclusions

The tars recovered in this study using the tar protocol method rapidly exhibited

instability,withprecipitateformingwithin14hoursofstanding;thiswasindependentofstorage

at5°Corat20°C.Onlystoringthesolutioninliquidnitrogenhaltedthisprocess.

The study revealed that the molecules that contained the largest sizes of conjugated

aromatic ring systems, rather than the molecules with the greatest masses, were primarily

involvedinageingreactions,resultinginprecipitationoccurring.

Storingthetarisopropanolsolutioninthedarkatreducedtemperature(5°C)appearsto

result in a different ageing reaction mechanism to that when the sample is left at room

65

temperature and exposed to indirect sunlight. There are probably numerous mechanisms

occurring,andforthesampleexposedtoindirectsunlightthereisevidenceofadditionalphoto-

oxidationreactions.Fromthispreliminaryinvestigationthesedetailscannotbedeterminedwith

certaintyandtheinfluenceofsunlightandoxygenshouldbeinvestigatedfurther.

The differences observed between these biomass/coal mixture samples were greater

thananticipated frompreviousstudiesofcoal,petroleumandbitumenderivedmaterials [47].

However,itremainsdifficulttodefinitivelyinterprettheresults,whichhighlightsthecomplexity

ofattemptingtoelucidatethemolecularpropertiesoftarsortheirageingmechanisms.

An important finding of this scoping study is that the combined analytical approach

includingSEC,LD-MSandUV-F,whichwasoriginallydevelopedforstudyingheavysamplessuch

aspitchesandasphaltenesrather thanoilsandtars [47],canbeequallywellappliedto these

(lower molecular weight) biomass tars. In fact, the methodology could provide clearer

information for tarsandoils thanpitchesandasphaltenes,as themoleculesherehaveamore

suitable mass and size range for study via SEC, LD-MS and UV-F (mass range <5,000 u and

chromophores<10rings).However,thehigheroxygencontentofbiomasstarscomparedtothat

ofheaviersamplessuchaspitchesandasphaltenesmakesresultsinterpretationlessexact.

Future studies should include NMR, FT-ICR-MS and FT-IR analyses to aid the

interpretationoftheUV-Fresultsandtoclarifythefindingsingeneral.Ultimateanalysisofthe

bulk precipitate, and ideally also the PC fractions, would provide additional valuable

information.

Throughtheapplicationofthecombinedanalyticalapproachoutlinedhereitispossible

tobuildupadetailedunderstandingoftarsamplesandothersimilarbiomass,coalorpetroleum

derived liquids in terms of molecular mass range, average mass estimates and extents of

aromaticconjugation.For amorethoroughunderstandingof theagingprocess thesemethods

shouldbecombinedwiththosecurrentlybeingdevelopedforbio-oils.

One of the more troublesome aspects of getting useful energy from tars is their high

viscosityand their tendencytoage intoevenmoreviscousmaterial, suchthat learningdetails

aboutthechemicalroutesofageingandhowtopreventthatchemistryfromhappeningcanhave

importantimplicationsforutilizingtars.

3. S: Supportinginformation

S1;FuelProperties

66

S2;PlanarChromatographyImages

S3;TarYields,GCandUAFurtherDiscussion

S4;SECCalibrationandInterpretation

S5;LD-MSAdditionalInformation

S6;SynchronousUV-FInterpretation

S7;PCFractionsSEC,LD-MSandUV-FResults

S8;Figures,byPCFraction(SEC,LD-MSandUV-F)

S9;Figures,bySample(SEC,LD-MSandUV-FofPCfractions)

S10;Figures,LD-MSoftheBulkTars

S3.1;Fuelproperties

ThefuelpropertiesreportedinTableS1.1belowweredeterminedinpreviousstudies

[1].Approximatecontentsofcellulose,hemicelluloseandlignininatypicalpinewood

sampleis48.0,23.5and28.5wt%respectively(excludingextractives,3.9wt%)[2].

TableS1.1,Propertiesofthefuelsusedtogeneratethetarsamples

Fuel PolishcoalPine

woodchips

Origin Poland Portugal

Typebituminous

coalsoftwood

67

Proximateanalysis

FixedC,

%54.9 13.6

Volatiles,

%28.8 74.5

Ash,% 8.6 0.3

Moisture,

%7.7 11.6

Ultimateanalysis

C,%daf 79.1 51.6

H,%daf 4.5 4.9

S,%daf 0.5 0.2

N,%daf 1.3 0.9

Cl,%daf 0.4 0.07

O,%daf 14.2 42.4

Deviationsdeterminedaswithin+/-0.5%,absolute.

S3.2;PCfractionation

Imagesoftheplanarchromatographyplatesbearingthethreetarsamplesareshownin

FigureS2.1.

68

TN2-0h T5C-6m T20C-6m-L

Figure S2.1, PC plates bearing the vacuum dried tar samples after being eluted with

chloroform(F3) followed by acetone (F2) and finally heptane (F4). The image on the

left-handsidewastakenunderwhitelight,andtheright-handsideunderUV-light(260

nm).

Themainbandsofmaterialarelabelledasfollows:

F1=immobile(darkband)

F2=acetonemobile/chloroformandheptaneimmobile(darkband)

F3=chloroformmobile/heptaneimmobile(darkandfluorescentbands)

F4=heptanemobile(yellowandfluorescentbands)

Note:InFigureS2.1thesmallsilvercirclevisibleforT20C-6m-L F1,under‘whitelight’isdue

to the aluminium backing-plate showing through due to loss of the silica coating. In

addition, the lighter colour of T20C-6m-L observed under UV when compared to the other

imageswasdue to the photo being takenduring theday time,whereas the otherswere

takenatnight.Cautionisrequiredwhentryingtodrawconclusionsfromtheimagesofthe

F4b

F4a

F3c

F3b

F4b

F4a

F3c

F4a

F3c

F3b

69

PCplatesdisplayedinFigureS2.1asthephotosarenotofhighenoughresolutiontoconvey

thefullextentofthevisualinformationthatwasobservedanddescribedinthemanuscript

(whichwasbasedontheexaminationofagreaternumberofPCplates).

Chloroformwasthefirsteluentusedwhichresultedinonemainbandofdarkmaterial

atthesolventfront(F3c)andnumerousweakerbandsfurtherdowntheplate(F3band

F3abeingthemainsones),manyofwhichwerehighlyfluorescentinthegreen,yellow,

orangeandblue(thiscannot beclearlyseenfromthephotographsdisplayed inFigure

S2.1). Less emphasis was placed on the separation and identification of these mobile

components,theaimwasto isolatethehighermasscomponentsandtogenerateafew

representativebandstoaidthecharacterisationoftheparentsamples.

S3.3;Taryields,GCandUAfurtherdiscussion

Errorinthetaryields: ByexaminingthetaryieldsinTable2itisapparentthereare

errors inthemassbalancespresented.This isnoticeablewhencomparingthe20hour

GCresultstothoseafter6monthsofstorage,asdiscussedinthemaintext(Section3.1).

Thedeviationsinthedeterminationsofthe‘recovereddry’tarweightandtheweightof

the precipitate were also fairly significant (+/- ~15 and 30 % respectively). For the

recovereddrytaryield,theerrorswereduetothelimitedvolumeandlowconcentration

of thesolutionaswellas thehighvolatilecontentof thetars; theactualweightsbeing

determined were in the tens of milligrams range at best. Similarly, the quantity of

precipitaterecoveredwasinthemilligramrange.

GC-FIDexperimental – additionalinformation:Beforeeachsetofanalysesastandard

solution(containingthe16PAHsindentifiedascarcinogensbytheEnvironmentProtect

Agency, EPA-16)was used for peak identificationand area calibration. Dodecanewas

usedasaninternal quantitativestandard.Alltheresultswerenormalisedtogramsoftar

pernormalcubicmeterofexhaustproducergasfromthegasifier(g/Nm3).Tocheckthe

repeatabilityof the results triplicate sampleswere analysed from each stock solution

andthescatter foundtobe lessthan+/- 2%(relativestandarddeviation- RSD).Table

70

S3.1 presents the concentrations of the quantified compounds (EPA-16) in the tar.

Unknown peaks were included in the calculation of the total GC tar yield. This was

achievedbycomparingthecombinedareacountsoftheunknownstonaphthalene,the

calculated g/Nm3 was then halved and that amount reported. Benzene, toluene and

xylenewerenot accountedforbytheGCmethodused.

GC results: Some information can be inferred from the analysis of the relative peak

intensities, to seehow the tars changewithageing.There is a strong trendshowing a

decrease intheamountofGCrangemoleculeswithageing,TableS3.1;however, there

wasnoevidenceforanequivalentincreaseinthecorrespondinggravimetrictaryields

orprecipitate.Therefore,itseemsunlikelythattheselosesareduetoreactionsbetween

GC rangesmolecules resulting in higher mass species; it is more probably a result of

evaporationorlossestotheglasswareasdiscussedinthemanuscript.Thisisthoughtto

bethecasebecausethegreatestchangeswereinthenaphthaleneconcentrationwhichit

isunlikelytobereactiveunderthestorageconditions.

ExaminationoftheGCresultsfortarTN2-0h beforeandafterprecipitateconfirmsthatno

significant change can be detected (cf. the 0 and 14 hour samples, Table S3.1). The

biggestchangeswereaslightdecrease inacenaphthyleneandphenanthrene,although,

thesearewithinthescatteroftheresults.Theonlychangeslargerthanthescatterwere

a slight increase in the amounts of benzo(a)pyrene, indeno(1,2,3)pyrene and

benzo(g,h,i)pyrene,andasmalldecreaseintheamountofnaphthalene,asdescribedin

themanuscript.

Tarcomponentsmorevolatile thannaphthalenewerenot examined; it ispossible that

the apparent increase in the amount of benzo(a)pyrene, indeno(1,2,3)pyrene and

benzo(g,h,i)pyrenewasduetosomeofthemorevolatilecompounds(thannaphthalene)

reacting, possibly with naphthalene to form these species, and/or due to solvent

evaporation or loss of other volatile components. However, it would be surprising if

naphthalenewasreactiveundertheseconditions.

TableS3.1,GC-FIDresultsforthetarsamplesbeforevacuumdrying,displayedasmgper

cubicmeterofproducergas

71

SampleName TN2 T5C T20C

Units mg/Nm3

SampleAge 0h 14h 20d 20h 3d 20d 6m 20h 3d 20d 6m

Naphthalene 3000 3000 2800 2600 2700 2800 1000 2700 2900 2700 1500

Acenaphthylene 710 700 680 640 630 680 400 670 680 660 420

Acenaphthene 10 10 10 10 10 10 10 10 10 10 5

Fluorene 50 60 50 40 50 50 20 50 50 50 20

Phenanthrene 460 450 470 380 400 440 230 410 420 420 230

Anthracene 90 90 100 80 80 90 40 80 90 90 40

Fluoranthene 230 240 250 220 220 230 140 240 230 230 150

Pyrene 240 250 250 220 220 240 150 240 230 230 150

Chrysene 30 30 30 20 30 30 20 30 30 30 20

Benzo(a)anthracene 30 40 40 30 40 40 20 30 40 40 20

Benzo(k)fluoranthene 30 40 40 30 40 30 20 30 40 30 20

Benzo(b)fluoranthene 10 10 10 10 10 10 5 10 10 10 10

Benzo(a)pyrene 40 60 50 40 50 50 40 40 50 50 40

Indeno(1,2,3)pyrene 30 40 30 20 30 30 20 20 30 30 20

Dibenzo(a,h)anthracene 5 10 5 5 5 5 5 5 5 5 5

Benzo(g,h,i)perlyene 5 30 30 5 30 5 20 20 30 30 20

TotalEPA16 5000 5100 4800 4400 4500 4700 2100 4600 4900 4600 2700

Unknowns 280 280 270 230 250 240 170 270 250 220 180

GCTotal 5300 5400 5100 4600 4750 4900 2300 4900 5150 4800 2900

Relativeerrorswerelessthan+/- 5%

S3.4;SECcalibrationandinterpretation

72

TheSECelutiontimewasconvertedtomassusingafivepointPScalibrationperformed

onthedayofanalysis.Thisequation(Eq1)wasappliedfrom15.0– 20.7minutes.After

this timeacalibrationbasedonsmallpolyaromatichydrocarbon(PAH)standards(Eq

2)wasused,basedonprevious studies[3-5].Theequation fromthePAHstandards(Eq

2)wasappliedtothe20.7– 24.0minregion.24.0minutesrelatestoamassofabout100

u;anysignalintheregion24-25minuteswasgivenavalueof100u.25minutes is the

permeation limit for this column. Materials eluting earlier than 15.0 minutes were

assignedanestimatedmassof2500u.Thereasonsforthisarebrieflydescribedinthe

followingsection.

EquationsusedtoconvertSECelutiontimetomass:

Eq1: 15.0– 20.5minutesregion: y=9.683-0.346.x PScalibration

Eq2: 20.5– 24.0minutesregion: y=6.902-0.210.x PAHcalibration

Wherey=log10 (MM);andx=elutiontimeinminutes

12.0– 15.0minutesregion: >2,500uaveragemassLD-MScalibration.

24.0– 25.0minutesregion: Valuekeptconstantat100u

CommentonthedisparitybetweentheSECandLD-MSmassestimates: Thereisan

apparentdisparitybetweenthemassestimatesderivedfromaSECPScalibrationwith

thosefromLD-MSmeasurementswhenanalysingcomplexhydrocarbonmixtures[3-6].

73

This is thought toariseforanumberofreasonsasdescribedinarecentreviewarticle

[6].Inbrief,ithaspreviouslybeenshownthatthematerialwhichelutesintheexcluded

region(<15minutes)of thisSECsystemhasahigheraveragemassthanthematerials

eluting in the retained region (15-25 minutes), for similar materials. However, it is

difficult to accurately determine the mass of the excluded material, as discussed

elsewhere [3-6]. The PS calibration appears to greatly overestimate the mass of the

excluded material observed by SEC, an LD-MS study found average masses (m/z) of

2500 - 3500 when the excluded SEC material was examined in isolation of retained

material; whereas, the SEC PS calibration indicated masses >100,000 u. In addition,

whenthereissomeretained SECmaterialalsopresentinthesamplebeinganalysedby

LD-MS, as for these samples, it tends to dominate the LD-MS spectra evenwhen it is

presentinlowerabundance(massdiscriminationandotherfactors[3,5-9]).

S3.5;LD-MSadditionalinformation

For theLD-MSanalysis all the sampleswere investigatedusinga0 (zero) and600ns

delayedionextraction(DIE)time.Spectrawerealsorecordedinreflector-mode(results

not shown). It isprudent toexamine the samplesbyaswide a rangeof conditionsas

possible to aid interpretation. The LD-MS results presented in the manuscript were

thoseobtainedunderthefollowingconditions:

ForthePCfractions: onlyspectraobtainedinlinear-modeoperationwithaHMAvoltage

of 10 kV, no DIE time and with varying levels of laser power applied, are displayed.

Theseconditionsarebasedonpreviousstudies[3,7].Briefly,thepurposeofthe analysis

ofthePCfractionsistoobtaininformationonthemassrangeofthesample.Therefore,

theconditionsareselectedtoaidtheobservationof the fragment ionsandthehighest

masscomponents(ions).

Forthebulktars: theaimwastoobtainmassspectrathatmoreaccuratelyrepresentthe

massdistributionofthesampleinitsentirety.Therefore,onlyspectraobtainedinlinear-

mode operation with the following conditions are reported, DIE time of 600 ns and

wheretheHMAvoltagewasreducedasthelaserpowerwasincreased.Theseconditions

areusedtoaidthedetectionofthehigherm/zionsandavoidoverloadingthedetectors

withlow m/zions.

74

A few LD-MS spectra are presented that were obtained under different conditions to

thosedescribedabovetohighlightcertainpoints;inthese instancestheconditionswill

bestatedinthetext.

Theuseofthepeakofmaximumintensity(peakmax.)toestimateaveragemassvalues

fromSECandLD-MSwasconsideredsufficientforthepurposesofthisscopingstudy(i.e.

to draw relatively comparisons) based on previous experience [3, 6,7]. Details of the

methods used to determine molecular mass estimates are also provided in said

publications.

S3.6;SynchronousUV-Finterpretation

The approach used to interpret the UV-fluorescence results is based on previous

investigations.AnumberofstudieshavereportedthatUV-Fspectrashowbathochromic

shifts (to longer wavelengths, red-shift) and emit lower fluorescence intensities with

increasing sizes of conjugated aromatic ring systems [5, 8, 10, 11].More recentwork

alsosupportstheseconclusions.Aqualitativerelationshiphasbeennotedbetweenthe

wavelength of maximum fluorescence and number of conjugated aromatic rings in a

polynucleararomatic(PNA)system,asdeterminedbysynchronousmodeUV-FandNMR

spectroscopy, respectively [6]. Thiswas for a number of coal,petroleum andbitumen

derivedoils, tars,pitchesandasphaltenes,andtheirsolubilitysub-fractions[3,10,12-

14]. ThiscorrelationwasdrawnfromacomprehensivereviewofUV-Fspectroscopyand

otheranalyticaltechniquesusefulforanalysiscomplexhydrocarbonmixtures[6].

Figure S6.1 and Table S6.1 demonstrate the relationship between synchronous UV-

fluorescence spectra for a series of coal and petroleum derived samples with their

numberofconjugatedaromaticrings(determinedbyNMR)superimposedonthepeak

withmaximumintensityof fluorescence.Themaximumintensityof fluorescence shifts

steadilytolongerwavelengthsbyapproximately30nmperadditionalaromaticringina

conjugated aromatic system, based on NMR results [3, 10, 12-14]. Fluorescence

intensitiesintheUV-fluorescencespectrumofthesamples4P1and5P1wereverylow

75

andshowedabroaddistribution,rangingfrom350tomorethan650nm.Thesefindings

suggest the presence of molecules containing a wide range of chromophores, some

correspondingtopolycyclicaromaticringsystemswithmorethan10conjugatedrings–

asdeterminedby13C-nmr[6,14].

It should be noted that oxygen and other substituents on aromatic ring systems also

influence fluorescence characteristics. Limited information is available however,

regarding their influence on the fluorescence of poly aromatic hydrocarbons with

molecular masses greater than ~400 u. A summary of some relevant fundamental

aspectsofUV- fluorescencearegivenbelow,furtherdetailscanbefoundelsewhere[15].

When substituent groups are introduced to an aromatic molecule the effect on

fluorescencecharacteristics dependsonposition ontheringtheyoccupyaswellastheir

functionality.Electron-acceptingunsaturated functionalgroups,suchas -COOH, -NO2, -

C=O, NH2, or -C=S-R can influence fluorescence characteristics of PAH’s. Such groups

have low lying vacant π* orbitalswhich can become occupied by an excited electron

fromthearomaticring.Electrondensity is transferredfromtheringtothesubstituent

(intramolecularchargetransfer).Anothercomplicationis the introducedofC=O,C=S, -

NO2,orhetero-atoms(N,S).Eachofthesegroupsprocessesalonepairofelectronsinan

orbitalparalleltotheplaneofthearomaticring,whichcanbepromoted.Thiscanhave

profoundeffectonfluorescence,dramaticallyreducingfluorescenceintensity[15].

When groups containing ‘n’ electrons (n = non-bonding electrons, as in C=O) are

conjugated with a π electron group, the effect is the same as increasing aromatic

conjugation onfluorescencecharacteristics.Therefore,additionofC=OgroupstoPAHs

and alkyl-aromatics results can result in a red shift of the fluorescence (to longer

wavelengths),dependingonthepositiontheyoccupy.Aromaticcarbonylcompoundsare

stronglyinfluencedbythelow-lying(n,π*)singletexcitedstates(asn-to-π*transitionis

100xlessintensethanπ-π *);therefore,thissignificantlyreducesfluorescencequantum

yield[15].

Auxochromes,whichgenerallydonotabsorbsignificantlyinthe200-800nmregionwill

affect the fluorescence spectrumof the chromophore towhich it is attached.OH is an

auxochrome; however, it typically has a minor influence on the chromophore it is

76

attached to [15]. Ether groups typically have no influence on conjugation or

fluorescence.

DuetothecomplexnatureoffluorescencecharacteristicsinlargePAHmoleculesitwas

notpossibletoaccountfortheinfluenceofoxygenontheUV-Fresultsofthetarsbeing

studiedorinthecorrelationdescribedaboveandshowninFigureS6.1andTableS6.1.

Tosimplifythediscussionofthe UV-Fresultstheterm‘numberofconjugatedaromatic

rings’or‘numberofrings’willbeusedinreferencetothewavelengthwherethepeakof

maximum fluorescence intensity is observed, using the correlation provided in Table

S6.1. In should be noted that this is not meant as a literal description of the

chromophore, it is only used to make relative comparisons between samples. The

changesin fluorescencecharacteristicsreferredtocouldalsobeduetothe influenceof

oxygenorothersubstituents,notnecessarilychangesinaromaticconjugationalone.

Figure S6.1, Synchronous UV-F spectra (absolute intensity - mV) of sample fractions

showing correlation with conjugated aromatic rings as determined by NMR. CO is

creosote oil, AO1 anthracene oil, PPS pyridine soluble fraction of coal tar pitch, P-1 a

syntheticpitch-likematerial,3P1,4P1and5P1aresolubilityfractionsofP-1.[3,10,12-

14].

0

50

100

150

200

250

300

350

250 300 350 400 450 500 550 600 650 700

Wavelength / nm

Ab

so

lute

In

ten

sit

y /

mV

CO = 2 rings

AO1 = 2-5 rings

PPS = 7-9 ring

P-1 = 4-7 rings

3p1 = 3-12 rings

4p1 >10 rings

5p1 = >>10 rings

Normal Response >250 mV

Weak Response <250 mV

1 2 3 4 5 6 7 >8

Number of Conjugated Aromatic Rings (based on NMR results)

CO AO1

P-1

3P1

PPS

4P1

5P1

77

TableS6.1,Correlationsbetweentheaveragenumbersofringsinpolynuclear aromatic

ring systems (determined by NMR) and the wavelengths of maximum fluorescence

intensityintheUV-fluorescencespectra[6]

Wavelengthof

peakwith

maximum

intensity

Approximate

Numberof

AromaticRings

Wavelengthof

peakwith

maximum

intensity

Approximate

Numberof

Aromatic Rings

270nm 1ring 390nm 5ring

300nm 2ring 420nm 6ring

330nm 3ring 450nm 7ring

360nm 4ring ≥480nm ≥8ring

S3.7;PCfractionsSEC,LD-MSandUV-Fresults

InthissectionadetailedaccountoftheSEC,UV-FandLD-MSresultsfromtheanalysisof

the PCmobility-fractions is given. The PC fractions thatwere not observed in all the

samples are discussed first followed by those common to all the samples. This is

beneficialbecausethefractionsthatwerenot observedinallthesamplesareofgreatest

interestastheyarethematerialsthathavereactedoraretheproductsofreactions.

Fluorescentmaterial,PCfractionF3a(SISectionS9,FiguresS9.1a-c)

This material was only observed in TN2-0h. It contains some material with molecular

massesapproaching1000u,andalsocontainssignificantamountsofsmallermolecules

(~200u).UV-Fshows5-9aromaticringsequivalentchromophores,onaverage(Table

S7.1). It isnot surprising that thematerial in F3awas not found in the aged samples

consideringpreviousstudies[3,14,16].WhereithasbeenfoundthatthePAHmolecules

containing the most highly-conjugated chromophores (largest poly nuclear aromatic

groups) are the most reactive, described further in the manuscript. PC fraction F3a

contains aromaticmolecules with some of the most highly-conjugated chromophores

78

thatwereobservedforthefreshtar;thereforeitisnotsurprisingthesemoleculeshave

reacted,probably toproduce larger,highermassmoleculeswithdifferentmobilityon

thePCplateafterageingthetar(orhaveprecipitated).Hence,PCfractionF3awasnot

observedintheagedsamples.

TableS7.1,SummaryofSECandLD-MSmassestimates,andUV-FresultsforPCfraction

F3afromTN2-0h

F3a Definition Units TN2-0h

LD-MSPeakMax./m/z

Uppermassm/z

<200

800

SECPeakMax.

2nd Peaku

220

900

UV-F

PeakMax.,(Ar.

Rings+)

2nd Peak,(Ar.

Rings+)

3rd Peak,(Ar.

Rings+)

nm

(rings)

390(5)

440(7)

470(8)

+ Ar.Ringsreferstotheequivalentnumberofconjugatedaromaticringsthatwould

fluoresceatthesamewavelength,asdescribedinSISectionS6.

Fluorescentmaterial,PCfractionF4b: (FiguresS8.1a-c,SISectionS8).

TN2-0h andT5C-6m weretheonlysamplestonoticeablycontainthisPCfraction.Theresults

indicatethatitwasroughlythesamematerialinbothsamples.UV-Fshowsthespectra

from TN2-0h and T5C-6m were very similar (Table S7.2, Figure S8.1c). There was one

significant differencehowever;TN2-0h containedachromophorethatfluorescedstrongly

at360nm(~4ringseq.)thatwasnotpresentinT5C-6m.Inaddition,thefreshtarcontains

agreaterabundanceoftheleast-conjugatedchromophores(3-4rings)andfewer5ring

79

species, relatively. The ring sizes predicted from UV-F correspond with the mass

estimatesfromLD-MSandSEC.

Despitethe lowaveragemassof this fraction itwas stillpossible toobtainreasonably

strongionscounts,andrepeatablespectra,duringLD-MSanalysis.Thiswasnotthecase

forfractionF3b(discussednext).

TableS7.2,SummaryofSECandLD-MSmassestimates,andUV-FresultsforPCfraction

F4bfromTN2-0h andT5C-6m

F4b Definition Units TN2-0h T5C-6m

LD-MSPeakMax.

Uppermassm/z

210

400

200

300

SEC PeakMax. u 150 140

UV-F

PeakMax.,(Ar.

Rings+)

2nd Peak,(Ar.

Rings+)

3rd Peak,(Ar.

Rings+)

4th Peak,(Ar.

Rings+)

nm

(rings)

390(5)

355(4)

360(4)

345(3)

390(5)

355(4)

345(3)

-

+ Ar.Ringsreferstotheequivalentnumberofconjugatedaromaticringsthatwould

fluoresceatthesamewavelength,asdescribedinSISectionS6.

Fluorescentmaterial,PCfractionF3b: (FiguresS8.4a-c,SISectionS8).

PC fractionF3bwas onlyobserved for TN2-0h andT20C-6m-L. The combined information

from SEC, LD-MS and UV-F reveals that fraction F3b from the two samples is not

composed of the same materials (Table S7.3). TN2-0h contains molecules of lower

80

molecular weight (~300 u) and with less-conjugated aromatic systems (4-5 rings

equivalent),onaverage, thanT20C-6m-Lwhichhasanaveragemassof~450u fromSEC.

However,LD-MSanalysisresultedinmainlyfragmentionsbeingobservedwithfewions

detectedabovem/z450.UV-FconfirmedthatT20C-6m-L islikelytocontainmore,higher

massmoleculesthanTN2-0h asitprovidesevidenceforlarger-sizesofconjugatedsystems

(5-8 ringsequivalent inT20C-6m-L compared to4-5 rings inTN2-0h). Fraction3bwasnot

observedinT5C-6m.

UV-F revealsmarkeddifferences in theextentsof conjugationof the chromophores in

theF3bfractions.Thefreshtar(TN2-0h)mainlycontainschromophoresequivalentto4-5

fused aromatic rings,while the aged sample (T20C-6m-L) containmainly 5-8 fused rings

equivalents. ItshouldberelativelyeasytoobservethesematerialsviaLD-MS;however,

forT20C-6m-L therewaslittleevidenceofmoleculeslargeenoughtocontainthenumberof

ringssuggestedbyUV-F. It is likely that themolecules inT20C-6m-L fractionF3bcontain

bridges between 3-4 ring aromatic units in a way that maintains conjugation; this is

possibly related to the incorporation of oxygen into the samples (Table 4).Moreover,

this may also explain why the UV-F and SEC results indicate the presence of larger

molecules than could be observed by LD-MS, and why fragment ions were mainly

observed. In addition, as T20C-6m-L was exposed to indirect sunlight the sample has

probably undergone photo-oxidation reactions during ageing; this could be related to

theobservationoutlinedabove.

Of the three techniques UV-F provided the clearest evidence for their being different

moleculespresent in theF3bPC fractionofTN2-0h andT20C-6m-L (notpresent inT5C-6m).

Therewasasurprisingrangeofdiversity inextentsof conjugationandmassrange for

the two samples. It is possible that the bright green fluorescence observed from this

material,whenobservedunderUVforbothTN2-0h andT20C-6m-L,wasfromasingletypeof

chromophorethatiscommontodifferentmoleculesinthetwosamples.

Comparing fractions F3b and F4b shows that F3b appears to contain higher mass

materials,onaverage;however,thehighermassmoleculeswerenotstabletowardlaser

ionisation.BothfractionsF3bandF4bpredominantlycontain5aromaticringequivalent

81

chromophores however, F3b contains chromophores that were more conjugated, on

average.

TableS7.3,SummaryofSECandLD-MSmassestimates,andUV-FresultsforPCfraction

F3bfromTN2-0h andT20C-6m-L

F3b Definition Units TN2-0h T20C-6m-L

LD-MSPeakMax.

Uppermassm/z

<200

400

<200

750

SEC PeakMax. u 275 460

UV-F

PeakMax.,(Ar.

Rings+)

2nd Peak,(Ar.

Rings+)

3rd Peak,(Ar.

Rings+)

nm

(rings)

400(5)

370(4)

-

415(5)

445(7)

490(8)

+ Ar.Ringsreferstotheequivalentnumberofconjugatedaromaticringsthatwould

fluoresceatthesamewavelength,asdescribedinSISectionS6.

PCfractionscommontoallthesamples:

PCfractionF4a (FiguresS8.2a-c,SISectionS8)

Theanalysisof theF4aPCfractions fromthe threetar samples revealssomecommon

features (TableS7.4).TheSECresultswerealmost identical asweretheUV-Fspectra;

however, LD-MS revealed differences. TN2-0h and T5C-6m have similar ion distributions,

centred aroundm/z250;however,T5C-6m also containsa significant amountofhigher

m/zions,~650,tailingtom/z~1500.T20C-6m-L showsamainbandofionsaroundm/z

82

325 and a second band at m/z ~650; these higher mass (m/z) ions were present in

lowerabundancethanseenforT5C-6m.

ThehighestmassmaterialpredictedinTN2-0h fromSECappearstofragmentduringLD-

MS as few ionswere observed with m/z >500. Whereas, in the aged samples a new

group of higher mass molecules were detected that were more amenable to LD-MS

analysis.UV-FshowedthatTN2-0h andT5C-6m containmoleculeswithverysimilarextents

ofconjugation(5-7rings);T20C-6m-L showsaslightincreasein4ringchromophoresanda

decreasein7ringequivalents.

The combined information suggests a subtle change in the materials present in PC

fractionF4adependingonthestorageconditions.Thiscouldbeinterpretedasevidence

for changes in structure rather than just a change in mass distribution. The sample

storedatroomtemperaturehaslostthematerialthatcontainedthearomaticmolecules

withthehighestdegreesofconjugationandthehighestmassspeciesasseeninTN2-0h.It

ispossiblethatmaterialhasreactedwithoxygenand/orothertarmoleculesresultingin

itbecomingpartofanotherPCfractionorformedprecipitate.ThelackofoxygeninT5C-

6m (compared to T20C-6m-L, Table 4) could be related to it containing more low-mass

moleculesthatarenotstabletowardslaserionization.

TableS7.4,SummaryofSECandLD-MSmassestimates,andUV-FresultsforPCfraction

F4afromTN2-0h,T5C-6m andT20C-6m-L

F4a Definition Units TN2-0h T5C-6m T20C-6m-L

LD-MSPeakMax.

Uppermassm/z

260

500

250

1500

325

1250

SECPeakMax.

2nd Peaku

190

1100

180&

1000&

160

1000

UV-F

PeakMax.,(Ar.

Rings+)

2nd Peak,(Ar.

Rings+)

3rd Peak,(Ar.

nm

(rings)

390(5)

420(6)

440(7)

390(5)&

420(6)&

440(7)&

390(5)

420(6)

370(4)

83

Rings+)

+ Ar.Ringsreferstotheequivalentnumberofconjugatedaromaticringsthatwould

fluoresceatthesamewavelength,asdescribedinSISectionS6.

&FractionsF4aandF3ccombined;thesamplescouldnotberecoveredseparately.

PCfractionF3c (FiguresS8.3a-c,SISectionS8)

SECshowsallthreesamplescontainmoleculeswithasimilarsizedistributionwhichis

composedmainlyofsmallmolecules(~200u)anda lowintensityshoulder,relatingto

largermolecules (~1000u).LD-MSrevealsa trendtowardshighermasses(m/z) from

TN2-0h <T5C-6m <T20C-6m-L.

UV-F reveals a dramatic difference in the extents of conjugation between the F3c

fractions.TN2-0h containsmainly7-8ringequivalentchromophoreswhereasT5C-6m and

T20C-6m-L contain5-6ringchromophores(almostidenticalspectra).

The results reveal that thematerial present in PC fraction F3c in the aged samples is

very different when compared to TN2-0h (Table S7.5). It seems that in TN2-0h this PC

fractioncomprisesofsmall- tomedium-sizedmolecules(200-500u)thatcontainhighly-

conjugated aromatic chromophores (7-8 rings eq.). Upon storage the majority of this

material disappears, presumable due to reactions to produce larger-sized molecules

withdifferentmobilityon thePCplate, or formsprecipitate.Thematerialobserved in

thisfractionintheagedsamplesappearstocontainmoleculesofgreatermassandless

conjugatedchromophoresthanwerepresentinTN2-0h.

84

TableS7.5,SummaryofSECandLD-MSmassestimates,andUV-FresultsforPCfraction

F3cfromTN2-0h,T5C-6m andT20C-6m-L

F3c Definition Units TN2-0h T5C-6m T20C-6m-L

LD-

MS

PeakMax.

Uppermassm/z

275

550

330

1100

350

1100

SECPeakMax.

2nd Peaku

220

900

180&

1000&

180

1000

UV-F

PeakMax.,(Ar.

Rings+)

2nd Peak,(Ar.

Rings+)

3rd Peak,(Ar.

Rings+)

nm

(rings)

440(7)

460(8)

410(6)

390(5)&

420(6)&

440(7)&

390(5)

420(6)

440(7)

+ Ar.Ringsreferstotheequivalentnumberofconjugatedaromaticringsthatwould

fluoresceatthesamewavelength,asdescribedinSISectionS6.

&FractionsF4aandF3ccombined;thesamplescouldnotberecoveredseparately.

PCfractionF2 (FiguresS8.5a-c,SISectionS8)

SEC results show that PC fraction F2 from TN2-0h and T5C-6m have a similar size

distributionwithapeakmaximumat~200uanda significant shoulderat~600u. In

contrastT20C-6m-L hasapeakmax.at~600uandmorematerialelutingintheexcluded

SECregion.MaterialelutingintheexcludedSECregionhaspreviouslybeenshowntobe

ofaveragemass~2500uorgreaterforsimilarmaterial[5];cf.SI S4fordetails.

The LD-MS results show a mono distribution for all the tars. TN2-0h had the lowest

averagemass(m/z)~550; theagedsampleshadroughlythesamedistributionasone

another,with an averagem/z of 600-700 and ions tailing off tom/z ~1100. The low

massmaterial(200u)predictedbySECforTN2-0h andT5C-6m wasnotobserved;thiswas

85

probably due to it being lost in the high vacuum of the MS sample chamber. The

contradiction between the SEC and LD-MS results for T5C-6m and T20C-6m-L could be

evidenceofT20C-6m-L containmorelarger-sizedmoleculesthanT5C-6m buttheyfragment

duringLD-MS,hencetheirmassspectraappearsimilar.

UV-FshowsTN2-0h containsless-conjugatedchromophores(4-5rings)thanT5C-6m orT20C-

6m-L whichgaveidenticalspectratooneanother(5-6rings).

These results show that themolecules inPC fractionF2ofTN2-0h areof loweraverage

mass and contain less-conjugated chromophores than the aged samples (Table S7.6).

T5C-6m contains more, low mass molecules than T20C-6m-L; although, the extents of

conjugationwere almost identical.This suggests thatongoing fromTN2-0h toT5C-6m or

T20C-6m-L thereisacommonmechanismwhichresultsinchromophoresofsimilarextents

of conjugation remaining in F2 (or being formed). For T20C-6m-L there appears to be a

secondprocesswhich results inhighermolecularmassmolecules remaining inF2 (or

being produced) which were not stable towards LD-MS analysis, with no additional

change in theextentof conjugation.This couldbe related to thegreaterabundanceof

oxygen in T20C-6m-L and may be evidence for oxygen bridges between tar molecules

resulting in no change in the extent of conjugationwhich are cleaved relatively easily

during LD-MS analysis. This may be related to the possibility of photo-oxidation

reactionsoccurringduring storageof tarT20C-6m-L.However, furtherwork isneededto

understandtheseaspects.

TableS7.6,SummaryofSECandLD-MSmassestimates,andUV-FresultsforPCfraction

F2fromTN2-0h,T5C-6m andT20C-6m-L

F2 Definition Units TN2-0h T5C-6m T20C-6m-L

LD-MSPeakMax.

Uppermassm/z

550

1300

700

1500

600

1500

SECPeakMax.

2nd Peaku

210

650

200

600

650

>2000

86

UV-F

PeakMax.,(Ar.

Rings+)

2nd Peak,(Ar.

Rings+)

3rd Peak,(Ar.

Rings+)

nm

(rings)

390(5)

350(4)

445(7)

415(6)

395(5)

470(8)

415(6)

395(5)

470(8)

+ Ar.Ringsreferstotheequivalentnumberofconjugatedaromaticringsthatwould

fluoresceatthesamewavelength,asdescribedinSISectionS6.

PCfractionF1 (FiguresS8.6a-c,SISectionS8)

The SEC and LD-MS results for PC fraction F1 were generally similar from all the

samples.ThemostsignificantdifferencewasseenforTN2-0h whichgaveveryweaksignal

whenanalysedbySECandUV-F.However;thiswasmainlyduetothelowabundanceof

fraction F1 in TN2-0h. It was possible to obtain satisfactory LD-MS results from all the

samples. All the F1 samples had similar mass distributions with a peak maximum

between m/z 600-750, and ions tailing off to aroundm/z 2000,with a shift towards

highermassesintheagedtars.AllthesamplesshowedaSECchromatogramwithapeak

max.intheexcludedregion(<15minutes,>2500averagemass)withsomematerialalso

eluting in the retained region (15-25minutes).Considering the limitationsof SEC and

LD-MSderivedmassestimates(cf.SIS4andS5)theseresultsshowgoodconsistency.

UV-F revealed the largest differences between the F1 samples. TN2-0h gave a very low

intensity spectrum;however, the signal that couldbedetectedhada similarprofile as

T20C-6m-L (4-6ringeq.).Ontheotherhand,T5C-6m showedaverydifferentspectrumwith

chromophoresof6-10aromaticringsequivalent,onaverage.

The results from the immobile fraction (F1) show therewasa low abundance of this

material in the freshtarand itwascomposedofhighmassmolecules(>500uaverage

mass)withchromophoresequivalentto4-6rings.T5C-6m hadamuchgreaterabundance

of thismaterial; itwasalsostructurallyverydifferenttotheequivalent fromTN2-0h,or

T20C-6m-L,despite havingasimilarmolecularsizeandmassdistribution.Themoleculesin

87

T5C-6m containaromaticmoleculeswithsignificantlyhigherdegreesofconjugationthan

anyoftheothertarsorPCfractions(TableS7.7,FiguresS9.1c,S9.2candS9.3C).

Thedifferencesoutlinedabovearepossiblyrelatedtothedifferentoxygencontentsof

the samples, where T5C-6m had less than the fresh tar or T20C-6m-L (Table 4). The

implicationsofthesefindingsarediscussedfurtherinthemanuscript(Section3.5).

TableS7.7,SummaryofSECandLD-MSmassestimates,andUV-FresultsforPCfraction

F1fromTN2-0h,T5C-6m,andT20C-6m-L

F1 Definition Units TN2-0h T5C-6m T20C-6m-L

LD-

MS

PeakMax.

Uppermassm/z

650

2000

725

2000

725

1900

SECPeakMax.

2nd Peaku

>2500*

1000*

>2500

1300

>2500

1100

UV-F

PeakMax.,(Ar.

Rings+)

2nd Peak,(Ar.

Rings+)

3rd Peak,(Ar.

Rings+)

nm

(rings)

400*(5)

470*(8)

-

475(8)

410(5-6)

535(10)

400(5)

410(5-6)

460(7)

+ Ar.Ringsreferstotheequivalentnumberofconjugatedaromaticringsthatwould

fluoresceatthesamewavelength,asdescribedinSISectionS6.

*Veryweaksignal.

3.8;Figures,byPCfraction(SEC,LD-MSandUV-F)

PCfractionF4b

88

Figure S8.1a, LD-MS spectra of PC fraction F4b at low and high laser power (top and

bottomrespectively),noDIE,HMAvoltagewas10kV.

10 100 1000 100000

100

200

300

400

500

Ion C

ount

m/z

T-5C-6m LP50% T-N2-0h LP70%

10 100 1000 100000

200

400

600

800

Ion

Cou

nt

m/z

T-5C-6m LP60% T-N2-0h LP80%

89

FigureS8.1b,SECchromatograms(areanormalised)ofPCfractionF4bat300nm.

FigureS8.1c,SynchronousUV-Fspectra(peaknormalised)ofPCfractionF4b.

PCfractionF4a

0.0E+00

2.0E-03

4.0E-03

6.0E-03

8.0E-03

1.0E-02

8 10 12 14 16 18 20 22 24 26 28

Time / minutes

Rela

tive Inte

nsity

T-N2-0h F4b

T-5C-6m F4b

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

250 300 350 400 450 500 550 600

Wavelength / nm

Norm

alis

ed Inte

nsity T-N2-0h F4b

T-5C-6m F4b

Solvent

90

Figure S8.2a, LD-MS spectra of PC fraction F4a at low and high laser power (top and

bottomandrightrespectively),noDIE,HMAvoltage10kV.

10 100 1000 100000

100

200

300

400

500

Ion C

ou

nt

m/z

T-20C-6m-L LP45% T-5C-6m LP30% T-N2-0h LP60%

10 100 1000 100000

200

400

600

800

1000

1200

1400

1600

Ion

Co

un

t

m/z

T-20C-6m-L LP65% T-5C-6m LP70% T-N2-0h LP70%

91

FigureS8.2b, SECchromatograms(areanormalised)ofPCfractionF4aat300nm.

FigureS8.2c,SynchronousUV-Fspectra(peaknormalised)ofPCfractionF4a.

PCfractionF3c

0.0E+00

2.0E-03

4.0E-03

6.0E-03

8.0E-03

1.0E-02

8 10 12 14 16 18 20 22 24 26 28

Time / minutes

Rela

tive

Inte

nsi

ty

T-N2-0h F4a

T-5C-6m F4a + 3c

T-20C-6m-L F4a

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

250 300 350 400 450 500 550 600

Wavelength / nm

No

rmla

ise

d In

ten

sity

T-N2-0h F4a

T-5C-6m F4a + 3c

T-20C-6m-L F4a

Solvent

92

Figure S8.3a, LD-MS spectra of PC fraction F3c at low and high laser power (top and

bottomrespectively),noDIE,HMAvoltagewas10kV.

10 100 1000 100000

200

400

600

800

1000

1200

Ion C

ount

m/z

T-20C-6m-L LP35% T-5C-6m LP60% T-N2-0h LP65%

10 100 1000 100000

400

800

1200

1600

2000

Ion C

ount

m/z

T-20C-6m-L LP45% T-5C-6m LP70% T-N2-0h LP80%

93

FigureS8.3b,SECchromatograms(areanormalised)ofPCfractionF3cat300nm.

FigureS8.3c,SynchronousUV-Fspectra(peaknormalised)ofPCfractionF3c.

PCfraction F3b

0.0E+00

1.0E-03

2.0E-03

3.0E-03

4.0E-03

5.0E-03

6.0E-03

7.0E-03

8.0E-03

9.0E-03

8 10 12 14 16 18 20 22 24 26 28

Time / minutes

Rela

tive Inte

nsi

ty

T-N2-0h F3c

T-5C-6m F4a + 3c

T-20C-6m-L F3c

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

250 300 350 400 450 500 550 600

Wavelength / nm

Norm

alis

ed Inte

nsity

T-N2-0h F3c

T-5C-6m F4a + 3c

T-20C-6m-L F3c

Solvent

94

Figure S8.4a, LD-MS spectra of PC fraction F3b at low and high laser power (top and

bottomrespectively),noDIE,HMAvoltagewas10kV.

10 100 1000 100000

100

200

300

400Io

n C

ou

nt

m/z

T-20C-6m-L LP40% T-N2-0h LP50%

10 100 1000 100000

200

400

600

800

1000

Ion C

ount

m/z

T-20C-6m-L LP50% T-N2-0h LP60%

95

FigureS8.4b,SECchromatograms(areanormalised)ofPCfractionF3bat300nm.

FigureS8.4c,SynchronousUV-Fspectra(peaknormalised)ofPCfractionF3b.

PCfraction F3a(notshownasonlypresentinTN2-0h,seeFiguresS9.1a-c)

0.0E+00

1.0E-03

2.0E-03

3.0E-03

4.0E-03

5.0E-03

6.0E-03

8 10 12 14 16 18 20 22 24 26 28

Time / minutes

Rela

tive I

nte

nsi

ty

T-N2-0h F3b

T-20C-6m-L F3b

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

250 300 350 400 450 500 550 600

Wavelength / nm

Norm

alis

ed Inte

nsi

ty

T-N2-0h F3b

T-20C-6m F3b

Solvent

96

PCfraction F2

Figure S8.5a, LD-MS spectra of PC fraction F2 at low and high laser power (top and

bottomrespectively),noDIE,HMAvoltagewas10kV.

10 100 1000 100000

100

200

300

400

500

600

700

Ion C

ou

nt

m/z

T-20C-6m-L LP30% T-5C-6m LP20% T-N2-0h LP40%

10 100 1000 100000

400

800

1200

1600

Ion C

oun

t

m/z

T-20C-6m-L LP35% T-5C-6m LP35% T-N2-0h LP70%

97

FigureS8.5b,SECchromatograms(areanormalised)ofPCfractionF2at300nm.

FigureS8.5c,SynchronousUV-Fspectra(peaknormalised)ofPCfractionF2.

0.0E+00

1.0E-03

2.0E-03

3.0E-03

4.0E-03

5.0E-03

6.0E-03

8 10 12 14 16 18 20 22 24 26 28

Time / minutes

Rela

tive Inte

nsity

T-N2-0h F2

T-5C-6m F2

T-20C-6m-L F2

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

250 300 350 400 450 500 550 600

Wavelength / nm

No

rma

lise

d I

nte

nsi

ty

T-N2-0h F2

T-5C-6m F2

T-20C-6m-L F2

Solvent

98

PCfraction F1

FigureS8.6a,LD-MSspectraofPCfractionF1atlowandhighlaserpower(leftandright

respectively),noDIE,HMAvoltagewas10kV.

10 100 1000 100000

100

200

300

400

500

600

700

Ion

Co

un

t

m/z

T-20C-6m-L LP15% T-5C-6m LP10% T-N2-0h LP30%

10 100 1000 100000

400

800

1200

1600

Ion

Co

un

t

m/z

T-20C-6m-L LP35% T-5C-6m LP25% T-N2-0h LP40%

99

FigureS8.6b,SECchromatograms(areanormalised)ofPCfractionF1at300nm;forTN2-

0h therewasweaksignalduetolowabundance.

FigureS8.6c,SynchronousUV-Fspectra(peaknormalised)ofPCfractionF1; forTN2-0h

therewasweaksignalduetosamplelowabundance

0.0E+00

5.0E-04

1.0E-03

1.5E-03

2.0E-03

2.5E-03

3.0E-03

3.5E-03

4.0E-03

4.5E-03

5.0E-03

8 10 12 14 16 18 20 22 24 26 28

Time / minutes

Rela

tive

In

ten

sity

T-N2-0h F1

T-5C-6m F1

T-20C-6m-L F1

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

250 300 350 400 450 500 550 600

Wavelength / nm

No

rmalis

ed I

nte

nsi

ty

T-N2-0h F1

T-5C-6m F1

T-20C-6m-L F1Solvent

100

S3.9;Figures,bysample(SEC,LD-MSandUV-FofthePCfractions)

TarSampleTN2-0h

Figure S9.1a, SEC chromatograms (area normalised) of PC fractions from TN2-0h, at

300nm.

0.0E+00

2.0E-03

4.0E-03

6.0E-03

8.0E-03

1.0E-02

8 10 12 14 16 18 20 22 24 26 28Time / minutes

Rela

tive Inte

nsi

ty

T-N2-0h F4b

T-N2-0h F4a

T-N2-0h F3c

T-N2-0h F3b

T-N2-0h F3a

T-N2-0h F2

T-N2-0h F1

F4b

F1

F2

F3b

F3a

F3c

F4a

101

Figure S9.1b, LD-MS spectra of the PC fractions from TN2-0h when low and high laser

powerwasused(leftandrightrespectively),noDIE,HMAvoltagewas10kV.

10 100 1000 100000

200

400

600

800

1000

1200

1400

Ion C

ount

m/z

F4b LP70% F4a LP60% F3c LP70%

10 100 1000 100000

400

800

1200

1600

2000

Ion C

ount

m/z

F4b LP80% F4a LP70% F3c LP80%

10 100 1000 100000

200

400

600

800

1000

1200

1400

Ion

Co

un

t

m/z

F3b LP50% F3a LP70% F2 LP40% F1 LP30%

10 100 1000 100000

400

800

1200

1600

2000

2400

Ion

Cou

nt

m/z

F3b LP60% F3a LP80% F2 LP70% F1 LP40%

102

FigureS9.1c,SynchronousUV-Fspectra(peaknormalised)ofthePCfractionsfromTN2-

0h.

TarSampleT5C-6m

Figure S9.2a, SEC chromatograms (area normalised) of PC fractions from T5C-6m, at

300nm.

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

250 300 350 400 450 500 550 600 650

Wavelength / nm

Norm

alis

ed I

nte

nsi

tyT-N2-0h F4b

T-N2-0h F4a

T-N2-0h F3c

T-N2-0h F3b

T-N2-0h F3a

T-N2-0h F2

T-N2-0h F1

F3c

F3a

F4b

F2

F3b

F1

Solvent

0.0E+00

1.0E-03

2.0E-03

3.0E-03

4.0E-03

5.0E-03

6.0E-03

7.0E-03

8.0E-03

9.0E-03

8 10 12 14 16 18 20 22 24 26 28

Time / minutes

Rela

tive I

nte

nsity

T-5C-6m F4b

T-5C-6m F4a + 3c

T-5C-6m F2a

T-5C-6m F1

F4b

F1

F2

F4a + 3c

103

Figure S9.2b, LD-MS spectra of the PC fractions from T5C-6m when low and high laser

powerwasused(leftandrightrespectively),noDIE,HMAvoltagewas10kV.

FigureS9.2c,SynchronousUV-Fspectra(peaknormalised)ofthePCfractionsfromT5C-

6m.

10 100 1000 10000

0

100

200

300

400

500

Ion

Co

un

t

m/z

F4b LP50% F4a LP40% F3c LP40%

F3b LP45% F2 LP15% F1 LP10%

10 100 1000 100000

400

800

1200

1600

2000

2400

Ion C

ount

m/z

F4b LP60% F4a LP70% F3c LP65% F3b LP60% F2 LP40% F1 LP40%

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

250 300 350 400 450 500 550 600 650

Wavelength / nm

No

rmalis

ed I

nte

nsi

ty

T-5C-6m F4b

T-5C-6m F4a + 3c

T-5C-6m F2

T-5C-6m F1

F1F2F4a + 3c

F4b

Solvent

104

TarSampleT20C-6m-L

Figure S9.3a, SEC chromatograms (area normalised) of PC fractions from T20C-6m-L, at

300nm.

0.0E+00

1.0E-03

2.0E-03

3.0E-03

4.0E-03

5.0E-03

6.0E-03

7.0E-03

8.0E-03

9.0E-03

8 10 12 14 16 18 20 22 24 26 28

Time / minutes

Rela

tive I

nte

nsity

T-20C-6m-L F4a

T-20C-6m-L F3c

T-20C-6m-L F3b

T-20C-6m-L F2

T-20C-6m-L F1

F4a

F1

F2F3b

F3c

105

FigureS9.3b, LD-MSspectraof thePCfractions fromT20C-6m-L whenlowandhigh laser

powerwasused (left and right respectively),noDIE,HMAvoltagewas10kV.Note, it

wasverydifficulttoobtainaspectrumfromF3.

FigureS9.3c,SynchronousUV-Fspectra(peaknormalised)ofthePCfractionsfromT20C-

6m-L.

10 100 1000 100000

100

200

300

400

500

600

700

800

Ion C

ount

m/z

F4a LP45% F3c LP30% F3b LP40% F2 LP20% F1 LP10%

10 100 1000 100000

400

800

1200

1600

2000

2400

2800

Ion

Co

un

t

m/z

F4a LP60% F3c LP50% F3b LP50% F2 LP40% F1 LP45%

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

250 300 350 400 450 500 550 600 650

Wavelength / nm

No

rmalis

ed I

nte

nsi

ty

T-20C-6m-L F4a

T-20C-6m-L F3c

T-20C-6m-L F3b

T-20C-6m-L F2

T-20C-6m-L F1

F3b

F4a

F1F3c F2

Solvent

106

S3.10;Figures,LD-MSofthebulktars

ExamplesoftheinfluenceofthelaserpowerstrengthanduseofdifferentDIEtimes,on

the LD-MS spectra obtained from the bulk tar T5C-6m are shown in Figures S10.1 –

S10.2.

FigureS10.1,LD-MSspectraofthebulkT5C-6m whenzeroand600nsDIEtimewereused

(leftandrightrespectively);laserpowerstrengthandHMAvoltage(kV)aregiveninthe

legend.

10 100 1000 100000

400

800

1200

1600

Ion C

ount

m/z

LP55% H6.5 LP45% H7.5 LP40% H7.5 LP35% H9.0

10 100 1000 100000

400

800

1200

1600Io

n C

oun

t

m/z

LP70% H6.0

LP50% H7.5 LP40% H10.0 LP30% H10.0

107

Figure S10.2, LD-MS spectra of the bulk tar T20C-6m-L when zero DIE time was used,

showingthe influenceof laserpower strengthandHMAvoltage(kV)onthespectrum;

laserpowerandHMAaregiveninthelegend.

10 100 1000 100000

200

400

600

800

1000

1200Io

n C

ount

m/z

LP70% H6.5 LP45% H7.5 LP40% H8.0 LP35% H10.0 LP30% H10.0 LP20% H10.0

108

4.0Pyrolysisofbiomassresidues fromtransgenicplantmaterial

Inmodels of future bio-refineries it is deemed that useof geneticallymodified

plantsthatareeasiertodegradebyenzymeswillbebeneficial.Theresiduesfromthese

biorefinerieswillneedtobeutilizedandthermochemicalconversion isconsideredone

of the most technically advanced methods by which to achieve valorization of these

residues. Recently significant advances have been made in modifying biomass

composition in planta. This includes manipulating the type, amount and degree of

polymerizationof lignin,whichcausesmanyof thepyrolysisproblemsassociatedwith

thethermochemicalconversionofbiomass. Thepossibilityexiststoremovesomeorall

of these R&D challenges by conducting pyrolysis experiments on the genetically

modifiedplants and understanding how thesemodificationsaffect pyrolysisbehavior.

Theoretically,forexample,reducingthelignincontentinplantsshouldreducethefinal

charcontentandquality,andenhancetheyieldsofthelightercomponentsinthebio-oil.

Investigations need to be carried out to ascertain the effects of altering the biomass

compositioninthelignin.

The Joint BioEnergy Institute (JBEI) has developed plants which contain less

lignin and lignin of a smaller degree of polymerization aswell as lignin of a different

composition. We propose to conduct pyrolysis on these plants and ascertain how

modificationsintheplantaffectfundamentalpyrolysisbehavior.Inthispartofthestudy

weaimed tounderstandthedifferencesintheproportionofchar,tarandgasproduced

by the genetically modified plants versus the wild type, and also to understand

differences in the composition of the bio-oil, and char reactivity as a function of the

geneticmodification.

4.1Transgeniclignin samples

Twosamplessetshavethusfarbeeninvestigated,fromplantsengineeredatJBEI.

The first set used an alternative strategy to reduce lignin recalcitrance. A dominant

approachwasdevelopedthatdivertsprecursors fromtheligninpathwayandenhances

productionof C6C1 aromaticsthatareknownasnon-conventionallignin monomersafter

exporttotheapoplast.Comparedwithregular C6C3 monolignols, theseC6C1 monomers

have reduced polymerization properties as they lack propanoid side-chain and its

conjugated double bond, disabling them from undergoing condensation at their b-

109

position. We also demonstrated that C6C1 monomersaccumulate as end-groups in the

ligninofFCA transgenics,resultinginreducedligninDPandcellwallslessrecalcitrantto

enzymatichydrolysis. Thereduceddegreeofpolymerizationisshownforthecellulolytic

enzymaticlignin(CEL)ofArabidopsisplantsinFigure4.1aandbbelow. Duetothelack

of calibrations standards for lignin and difficulty in understanding precise lignin

structurepolydispersityvaluesarealwaysrelativeratherthanabsolute.Neverthelessa

reduceddegreeofpolymerizationwasshowninthesamplesbelow.

Figure 4.Polydispersity of cellulolytic enzymatic lignin (CEL) in thewild type (solid

black line) FCA transgenics (grey line) as analysed by size exclusion chromatography

usingUVabsorbancedetection(Figure4a)andUVfluorescencedetection(b).

110

Thesecondsetof samples,socalledQsubsamples, weredevelopedtohaveareduced

lignincontent,ratherthanareduceddegreeofpolymerization. Twonewsampleswere

createdtohave11%and7%lignincomparedto19%lignininthewildtype.

4.2 Reactorconfigurationforpyrolysisoftransgenicligninplants

Transgenic plantswere available invery small quantities (100’smg), therefore itwas

deemed themost appropriate reactor toanalyses these sampleswas theatmospheric

pressurewire-mess,availableatthelaboratoriesofourcollaboratorsatImperialCollege

London. The reactor is shown in Figure 4.2. The design and principle of wire-mesh

reactorsisdescribedindetailelsewhere.Thebasicdesignconceptoftheseinstruments

is straightforward. Milligram quantities of sample particles are placed between two

layers of folded wire-mesh. This assembly is weighed and stretched between two

electrodes. Finewire thermocouples are attached. A controlled current is then passed

throughthemesh,whichalsoservesasa resistanceheater.After the samplehasbeen

exposed to a pre-programmed time-temperature profile, the weight change of the

assemblyisdetermined.Dependingonthepurposeoftheexperiment,volatilesand/or

tarsmayberecoveredandcharacterised.Thisreactorconfigurationallowsexperiments

tobecarriedoutusingwiderangesofheatingrates(1– 20,000°Cs-1),temperatures(to

2,000°C)andpressures(to160bar).Thisreactorconfigurationprovidesthecurrently

mostoptimizedmethodbywhichtopyrolysesmallquantitiesof transgenicplantsand

theirlignintounderstandsmallchangesinpyrolysisbehaviorwhich mightoccur.

111

4.3 PyrolysisofArabidopsissampleswithreduceddegreeofpolymerization(FCA)

Thepyrolysis oftwosamplesFCAandthecorrespondingwildtype(WT)wascarriedout

intheatmosphericwiremeshreactoratapeaktemperatureof400oC,5sofholdingtime

andaheatingrateof1000oC/s.Theseoperatingconditionswereselectedwiththeaim

ofobservingcleardifferencesbetweenthesamplesduetoincompletepyrolysis.Before

being used, the two sampleswere crushed and sieved between 106 and150µm and

Figure4.1 Theatmosphericpressurewire-meshreactorwithtartrap,maximumheatingrate10,000°Cs-1,batch.Legend:[1]CopperCurrentCarrier;[2]LiveElectrode;[3]BrassClamping Bar; [4] Sample Holder Support Plate; [5] Mica Strip; [6] Wire-mesh SampleHolder; [7] Electrode; [8] Stainless Steel Tubes; [9]Mica Layer; [10] Brass Pillars; [11]SinteredPyrexGlassDisk;[12]BasePlate;[13]PyrexBell;[14]O-ringSeal;[15]Off-takeColumn;[16]O-ring;[17]CarrierGasEntryPort;[18]ConnectionforVacuumPump.

112

driedovernight. ThenomenclatureFCA3andWT3isgivenasthreeseparatesamples

wereused.

Table2showstheproductyieldsobtainedfromthepyrolysisofbothsamples.Duplicate

experimentswerecarriedoutforeachdatapoint.FCA3yieldedslightlylargeramountof

charthanWT3.FCA3hasanincreasedlevelofcondensation(C-Cbonds)whichexplains

the increasedcharyield.On theotherhand,FCA3gaveriseto lesstaryieldthanWT3.

Tarproportion inwithinthetotalvolatile fractionalsodiminished inthecaseofFCA3

comparedtothatinthecaseofWT3.

Table 4.1. Experimental conditions and product yields from Wire Mesh Reactor

experiments of biomasswith reduced degreeof polymerization and its corresponding

wildtype.Heatingrate– 1,000oC/s,Pressure– 1bar,Atmosphere- He

Sample Peak Temperature (oC)Holding time

(s)

Char

(wt.%)

Tar

(wt.%)

Gas

(wt.%)

WT3 400 5 27.8±0.6 37.4±0.2 34.9±0.8

FCA3 400 5 27.1±0.4 34.9±0.1 38.1±0.5

Tars frompyrolysisofWT3andFCA3wereanalysedbygaschromatography(GC)and

sizeexclusionchromatography(SEC).

GCanalyseswereperformedinaPerkinElmer“Clarus500”chromatographwith flame

ionizationdetector(FID).TheGCwasequippedwithanon-polarcapillarycolumnHT5

(25 mx0.32mmand0.1μmfilmthickness).Aflowrateof10mL/minofHewasusedas

carriergaswithasplitratioof5.4:1.Theinitialoventemperaturewas40oC,whichwas

held for 1.0 min. It was then programmed to rise to 380 oC at 10 oC/min with an

isothermheld for10min.Theprogrammedtemperature injectorwasrampedfrom80

oC(held for0.5min)to380oCat100oC/min.TheflamewasmaintainedwithH2 flow

rateof45mL/minandairflowrateof450mL/min.Calibrationswithasetofn-alkanes

(C8toC30)andastandardPAHwereusedtoevaluatethepercentageofelutionof the

material.

Figure 1 shows the chromatograms obtained for the analyses of tars from FCA3 and

WT3. As can be seen, no significant differences are observed between the two tar

samples.

113

Figure4.2. GC chromatograms of tars obtained from thepyrolysis of FCA3 andWT3.

WMRat400oC,5sand1000oC/s.

In addition, Table 4.2 summarizes the data calculated from the n-alkanes and PAH

calibrations. Approximately 80% of the area under the evolved peaks was identified

based on the n-alkanes and PAH calibrations. However, the area identified only

corresponds to around 1 % of the area expected based on the calibration. In other

words,99%oftheinjectedsamplesdidnotevolvethroughtheGCcolumn(materialwith

boilingpointhigherthan550 oC). Thisisbeinginvestigatedfuthre.

Table 4.2.Results from GC analyses of tars obtained from the pyrolysis of FCA3 and

WT3.WMRat400oC,5sand1000oC/s

CompoundBoiling

point(oC)

Tarfrom

WT3(wt.%)

Tarfrom

FCA3(wt.%)

Acenaphthylene 280 0.12 0.14

C29 441 0.63 0.47

Benzo(A)Pyrene 496 0.30 0.24

Indeno(1,2,3,CD)Pyrene&

Dibenz(A,H)Anthracene524&536 0.09 0.00

2.5 7.5 12.5 17.5 22.5 27.5 32.5 37.5 42.5

Time (min)

Tar from FCA3

Tar from WT3

114

Benzo(G,H,I)Perylene 550 0.07 0.00

Percentageidentifiedfromtheareaobtained 74.65 84.58

A 300mm long, 7.5mm i.d. polystyrene/polydivinylbenzene-packedMixed-D column

with5μmparticleswasusedforSECanalyses.Thecolumnwasoperatedat80°Canda

flowrateof0.5mL/min.NMPwasusedasthemobilephase.Detectionwascarriedout

usingaKnauerSmartlinediodearrayUV-absorbancedetector.As NMPisopaqueat254

nm, detection of standard compounds and sampleswas performed at 300nm,where

NMPispartiallytransparent.

A calibration of the Mixed-D column was carried out using two sets of standards,

polystyrene (PS) and polyaromatic hydrocarbons (PAH). The PS-based calibration is

appliedtothe11- 20mintimerange,whilethePAH-basedcalibrationisusedin20-24

minregion,resultinginthefollowingcalibrationequations:

11-20minregion:log10[MM]=10.6320– 0.4038[elutiontime(min)]

20-24minregion:log10[MM]=6.8495– 0.2095[elutiontime(min)]

Figure4.3 showsmolecularweightdistributionbysizeexclusionchromatographyofthe

tarsfromthepyrolysisofFCA3andWT3.Inbothcases,bimodaldistributionsofsignal

areobserved.Theearlyelutingpeakcorrespondstomaterialofmolecularsizeunableto

penetratetheporosityofthecolumnpacking,andisreferredtoas“excluded”fromthe

columnporosity.Theexclusionlimitofthecolumn,definedaccordingtothebehaviour

of polystyrene standards, is about 200,000 u (although molecular conformation is

consideredtobethefactorthatcausesmoleculestobecomeexcludedfromthecolumn

porosity rather than molecular weight). The second eluting peak corresponds to the

materialabletopenetratetheporosityofthecolumnpacking.

Thetwotarsshowsimilarmolecularweightdistributions,withnosignificantdifferences

between them. Both chromatograms show a lift-off in the baseline around 16.7 min,

whichcorrespondstoapolystyrenemassofabout7,740u.Thisisconsideredtheupper

limit in the molecular weight distribution of these samples. Both tars present a

maximumpeakat19.7min,which corresponds toapolystyrenemassof about475u.

115

Tar fromWT3seems tohaveamolecular weightdistribution slightly shifted to lower

valuesthanthetarfromFCA3.However,thisdifferenceisnotsignificant.

Figure4.3. SECchromatogramsof tarsobtainedfromthepyrolysisofFCA3andWT3.

WMRat400oC,5sand1000oC/s.

APerkin-ElmerLS50luminescencespectrometerwasusedforUV-Fanalyses.Thedevice

wassetwithaslitwidthof5nm, toscanat240nmmin-1;synchronousspectrawere

acquired at a constant wavelength difference of 20 nm. A quartz cell with 1 cm path

lengthwasused.Thespectrometerfeaturedautomaticcorrectionforchangesinsource

intensityasafunctionofwavelength.Synchronousspectraofthesampleswereobtained

inNMP.SolutionsweredilutedwithNMP toavoidself-absorptioneffects:dilutionwas

increased until the fluorescence signal intensity began to decrease. All shown spectra

have been peak-normalized to account for the different fluorescence yields of the

differentsamplesandhighlighttheshiftsinthepeakmaximum.

Figure4.4 showstheUV-FspectraofthetarsfromthepyrolysisofFCA3andWT3.The

two tars exhibit similar fluorescence. Themost intense fluorescence for both samples

was centered around 315 nm. Additionally, both tars presented a second peak at the

longerwavelengthof350nm,thetar fromWT3showingaslightlymoreintensesignal.

0.0000

0.0005

0.0010

0.0015

0.0020

0.0025

0.0030

0.0035

0.0040

8 13 18 23 28

No

rmal

ised

Are

a

Time (min)

Tar from FCA3

Tar from WT3

116

Figure4.4. UV-FspectraoftarsobtainedfromthepyrolysisofFCA3andWT3.WMRat

400oC,5sand1000oC/s.

ThemainconclusionsobtainedfromthepyrolysisofWT3andFCA3thusfarare:

Charyield fromFCA3washigher than that fromWT3.On the contrary,FCA3gave

risetolesstaryieldthanWT3.

BasedonGC,SECandUV-Fanalyses,therearenosignificantdifferencesbetweenthe

tarsobtainedfromFCA3andWT3.

The change in the degree of polymerization of the samples seems to affect the

pyrolysis product yields but not the composition of the tars produced. More

experimentsareneededtoprovethisstatement.

Sincepyrolysisofthebiomasssamplesisincompleteundertheoperatingconditions

studied,clearerdifferenceswereexpected.Asfurtherconditionsneedtobeevaluate.

4.4 PyrolysisofArabidopsissampleswithreducedamountoflignin(Qsub2)

The pyrolysis of two samples from Set 3 (WT2 and Qsub2) was carried out in the

atmosphericwiremeshreactoratdifferentpeaktemperaturesandholdingtimes,andat

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

250 350 450 550 650 750

Sig

nal (N

orm

.)

Wavelength (nm)

Tar from WT3

Tar from FCA3

117

aheatingrateof1000oC/s.Beforebeingused,thetwosampleswerecrushedandsieved

between106and150µmanddriedovernight.

Table 4.3 shows the product yields obtained from the pyrolysis of both samples.

Duplicateexperimentswerecarriedoutforeachdatapoint.Qsub2yieldedslightlylarger

amountofcharthanthecorrespondingwildtypeWT2at400oCand5sofholdingtime.

Ontheotherhand,Qsub2gaverisetolessamountoftarfractionthanWT2.Therefore,

tar proportion in within the total volatile fraction also diminished in the case of the

modifiedbiomassQsub2compared to that in the caseofWT2.On theotherhand, the

increase in temperature from400 to 550 oC gave rise to a slight higher yield of total

volatiles.Nevertheless, change in tar yieldwas negligible. Additionally, product yields

werenotaffectedbytheincreaseinholdingtimefrom5to30s.

Table 4.3. Experimental conditions and product yields from Wire Mesh Reactor

experimentsofbiomasswithlesslignincontentanditscorrespondingwildtype.Heating

rate– 1,000oC/s,Pressure– 1bar,Atmosphere– He

Sample Peak Temperature (oC)Holding time

(s)

Char

(wt.%)

Tar

(wt.%)

Gas

(wt.%)

WT2 (S3) 400 5 18.3±0.5 41.7±0.4 40.1±0.1

Qsub2 (S3)

400 5 18.8±1.2 39.6±0.1 41.6±1.3

Qsub2 (S3)

550 5 17.1±0.4 39.9±0.2 43.1±0.6

Qsub2* (S3)

550 30 16.6 40.0 43.4

*furtherreplicatesrequired

TarsfrompyrolysisofWT2andQsub2wereanalysedbyGC,SECandUV-F.

GCanalyses

GCanalyseswereperformedinaPerkinElmer“Clarus 500”chromatographwith flame

ionizationdetector(FID).TheGCwasequippedwithanon-polarcapillarycolumnHT5

(25 mx0.32mmand0.1μmfilmthickness).Aflowrateof10mL/minofHewasusedas

carriergaswithasplitratioof5.4:1.Theinitialoventemperaturewas40oC,whichwas

held for 1.0 min. It was then programmed to rise to 380 oC at 10 oC/min with an

isothermheld for10min.Theprogrammedtemperature injectorwasrampedfrom80

oC(held for0.5min)to380oCat100oC/min.TheflamewasmaintainedwithH2 flow

118

rateof45mL/minandairflowrateof450mL/min.Calibrationswithasetofn-alkanes

(C8toC30)andastandardPAHwereusedtoevaluatethepercentageofelutionof the

material.

Figure 4.5 shows the chromatograms obtained for theanalysesof tars fromWT2 and

Qsub2. As can be seen, no remarkable results were obtained from these analyses.

AlmostnopeakwasdetectedforthetarfromQsub2.Somepeakswereobservedinthe

caseoftar fromWT2,whichshowsthepresenceofsomecompoundelutingwithinthe

temperature range relative to GC analyses. In general, it can be said thatmost of the

componentsinthetarsfromWT2andQsub2presentaMWdistributionthatexceedthe

operatingwindowof theGC(materialwithboilingpointlowerthan570oC).Thus, the

area identified corresponds to less than0.1%of theareaexpectedbasedon thePAH

calibration.

Figure4.5. GCchromatogramsof tarsobtainedfromthepyrolysisofWT2andQsub2.

WMRat400oC,5sand1000oC/s.

Table2 shows the resultsobtained from theGCanalysesofWT2andQsub2biomass.

Resultsaregivenastheratioofthemassobtainedindifferentboilingpointrangeswith

theinitialmassofbiomasschargedontheWMR.Resultsoftars fromSeries1biomass

areincludedforcomparison.

500000

520000

540000

560000

580000

600000

620000

640000

2.5 7.5 12.5 17.5 22.5 27.5 32.5 37.5 42.5

Inte

nsi

ty (

a.u

.)

Time (min)

Tar from WT2

Tar from Qsub2

119

Table4.4.ResultsfromGCanalysesoftars obtainedfromthepyrolysisofFCA3andWT3(Series1)andWT2andQsub2(Series2)

Mass in the BP range/Initial Mass of Biomass (µg/µg)

Boiling point range (oC)

Tar from WT3 (S1)

Tar from FCA3 (S1)

Tar from WT2 (S3)

Tar from Qsub2 (S3)

Tar from Qsub2 @550oC, 5s

Tar from Qsub2 @550oC, 30s

147 - 266 0.027 0.041 0.037 0.005 0.046 0.027

267 - 339 0.080 0.090 0.034 0.010 0.011 0.005

340 - 399 0 0.002 0.020 0.005 0.003 0

400 - 449 0.202 0.152 0.017 0.002 0.033 0.001

450 - 491 0.197 0.078 0.002 0.002 0.016 0

592 - 572 0.050 0 0.004 0.001 0 0.001

573 - 587 0 0 0 0 0 0.021

Total amount analysed (wt.%)

1.53 1.11 0.28 0.07 0.36 0.15

120

Ascanbeseen,theamountanalysed byGCrespecttotheinitialmassofbiomasschargeon

theWMRwasslightlylargerinthecaseofthesamplesfromSeries1.Theamountevolved

inthecaseofQsub2at400oCand5swasnegligible.RegardingtarsfromSeries1,mostof

thedetectedcompoundsevolvedintherangeofequivalentboilingpointof400-490oC.In

the caseof tar fromWT2 (Series2), the small amountofdetectedmaterial eluted in the

rangebetween150and400oC.InthecaseoftarfromQsub2(Series2)at550oCand5s,

the small amountofdetectedmaterial eluted in the rangesbetween150and340 oCand

400 and 491 oC. Regarding the tar from Qsub2 (Series 2) at 550 oC and 30s, the small

amount of detectedmaterial eluted in the ranges between 150 and 270 oC and 575 and

590 oC.

SECanalyses

A300mmlong,7.5mmi.d.polystyrene/polydivinylbenzene-packedMixed-Dcolumnwith

5μmparticleswasusedforSECanalyses.Thecolumnwasoperatedat80°Candaflowrate

of 0.5 mL/min. NMP was used as the mobile phase. Detection was carried out using a

Knauer Smartline diode array UV-absorbance detector. As NMP is opaque at 254 nm,

detectionof standardcompoundsandsampleswasperformedat300nm,whereNMP is

partiallytransparent.

A calibration of the Mixed-D column was carried out using two sets of standards,

polystyrene (PS) and polyaromatic hydrocarbons (PAH). The PS-based calibration is

appliedtothe11- 20mintimerange,whilethePAH-basedcalibrationisusedin20-24min

region,resultinginthefollowingcalibrationequations:

11-20minregion:log10[MM]=10.6320– 0.4038[elutiontime(min)]

20-24minregion:log10[MM]=6.8495– 0.2095[elutiontime(min)]

Figures4.6and4.7 showmolecularweightdistributionbysizeexclusionchromatography

ofthetarsfromthepyrolysisofWT2andQsub2.Inallcases,bimodaldistributionsofsignal

areobserved.Theearlyelutingpeakcorrespondstomaterialofmolecular sizeunable to

penetrate the porosity of the column packing, and is referred to as “excluded” from the

columnporosity.Theexclusionlimitofthecolumn,definedaccordingtothebehaviourof

polystyrenestandards,isabout200,000u(althoughmolecularconformationisconsidered

121

tobethefactorthatcausesmoleculestobecomeexcludedfromthecolumnporosityrather

than molecular weight). The second eluting peak corresponds to the material able to

penetratetheporosityofthecolumnpacking.

Figure4.6 showsmolecularweightdistributionby sizeexclusion chromatographyof the

tars from thepyrolysisofWT2and Qsub2at 400 oCand5 s.The two tars show similar

molecularweightdistributions.Bothchromatogramsshowalift-offinthebaselinearound

16.7 min,whichcorrespondstoapolystyrenemassofabout7,740 u.Thisisconsideredthe

upper limit in the molecular weight distribution of these samples. Both tars present a

maximumpeakaround19.65min,whichcorrespondstoapolystyrenemassofabout475

u.Tar fromQsub2seems tohavea slightlynarrowermolecularweightdistribution than

thetarfromWT2,whichisshiftedtolowerMWvalues.

Figure 4.6. SEC chromatograms of tars obtained from the pyrolysis ofWT2 and Qsub2.

WMRat400oC,5sand1000oC/s.

0.0000

0.0010

0.0020

0.0030

0.0040

0.0050

0.0060

8 13 18 23 28

No

rmal

ised

Are

a (a

.u.)

Time (min)

Tar from WT2

Tar from Qsub2

122

Figure4.7 showsmolecularweightdistributionbysizeexclusionchromatographyof the

tars from the pyrolysis of Qsub2 at different final temperatures and holding times. The

three tars show similar molecular weight distributions. Thus, the three chromatograms

showalift-offinthebaselinearound16.7 min,whichcorrespondstoapolystyrenemassof

about 7,740 u.This is considered the upper limit in themolecularweight distribution of

these samples. Tar from pyrolysis at 550 oC and 5 s presents a maximum peak around

19.78 min, which corresponds to a polystyrene mass of about 440 u. Although the

maximumpeakwasslightlydisplacedtowardslongerretentiontimethanthatoftarfrom

pyrolysis at400 oCand5 s, tar from550 oCand 5sexhibits abroaderMWdistribution.

Thus,thechromatogramcoversbothlighterandheavierMW.Tarfrompyrolysisat550oC

and 30 s presents a MW distribution clearly displaced towards lighter values. It has a

maximumpeakaround19.97min,whichcorrespondstoapolystyrenemassofabout370

u.Therefore,theincreaseininholdingtimeseemstogiverisetolightermolecularweight

distribution.

0.0000

0.0010

0.0020

0.0030

0.0040

0.0050

0.0060

8 13 18 23 28

No

rmal

ised

Are

a (a

.u.)

Time (min)

Tar from Qsub2 @ 400oC,5s

Tar from Qsub2 @ 550oC,5s

Tar from Qsub2 @ 550oC,30s

123

Figure4.7. SECchromatogramsof tarsobtainedfromthepyrolysisofQsub2atdifferent

temperaturesandholdingtimes.

UV-Fspectra

APerkin-ElmerLS50 luminescence spectrometerwasusedforUV-Fanalyses.Thedevice

was setwith a slit width of 5 nm, to scan at 240 nmmin-1; synchronous spectra were

acquiredataconstantwavelengthdifferenceof20nm.Aquartzcellwith1cmpathlength

wasused.Thespectrometerfeaturedautomaticcorrectionforchangesinsourceintensity

as a functionofwavelength. Synchronous spectra of the sampleswere obtained inNMP.

SolutionsweredilutedwithNMP toavoid self-absorptioneffects:dilutionwas increased

until the fluorescence signal intensity began to decrease. All shown spectra have been

peak-normalized toaccount for thedifferent fluorescenceyieldsof thedifferent samples

andhighlighttheshiftsinthe peakmaximum.

Figure4.8 showstheUV-FspectraofthetarsfromthepyrolysisofWT2andQsub2at400

oC and 5 s. The two tars exhibit fluorescence in the same region of the spectra. Similar

profiles are observed; however, Qsub2 present a spectrum slightly displaced towards

shorter wavelengths compared to that from WT2. This result points to the presence of

smaller polynuclear aromatic groups in the case of Qsub2. Thus, the most intense

fluorescence was centred around 314 nm forWT2 and 307nm for Qsub2. Additionally,

both tars presented a second peak at longer wavelength. WT2 showed a more intense

signalatthislongestwavelength,withapeakat360nm.Thesecondpeakwasaround355

nminthecaseofQsub2.The less intense florescenceat shorterwavelengthsuggests the

presence of smaller conjugated aromatic systems in the tar obtained from the modified

biomassQsub2,with reduced amount of lignin. These results are consistentwith earlier

findingsfromSEC,bothofwhichpointtowardhigherconcentrationsoflargerMWgroups

beingproducedinthetarobtainedfromthewildtypebiomass.

124

Figure4.8 UV-FspectraoftarsobtainedfromthepyrolysisofWT2andQsub2.WMRat400oC,5s

and1000oC/s.

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

250 350 450 550 650 750

Sig

nal

(No

rm.)

Wavelength (nm)

Tar from WT2

Tar from Qsub2

125

Figure 4.9. UV-F spectra of tars obtained from the pyrolysis Qsub2 at different temperatures and

holding times.

Figure4.9 showstheUV-FspectraofthetarsfromthepyrolysisofQsub2atdifferentpeak

temperatures (400 and 550 oC) and holding times (5 and 30 s). The three tars exhibit

similarfluorescenceprofiles.Thus,themostintensefluorescencewascentredaround307

nmforallthecases.Additionally,thethreetarspresentedasecondpeakaround354nm.

Conclusions

ThemainconclusionsobtainedfromthepyrolysisofWT2andQsub2inSeries3are:

Tar yield from Qsub2 at 400 oC and 5 s was smaller than that from WT2. On the

contrary,Qsub2gaverisetomorecharyieldthanWT2.

GC did not gave rise to significant results, since less than 0.3 wt.% of the injected

materialseemstohaveelutedfromthecolumn. Thisneedstoberevisited.

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

250 350 450 550 650 750

Sig

nal

(No

rm.)

Wavelength (nm)

Tar from Qsub2 @400oC,5s

Tar from Qsub2 @550oC,5s

Tar from Qsub2 @550oC,30s

126

SECandUV-F results showed slightdifferencesbetween the tarsobtained fromWT2

andQsub2.FindingsfromSECandUV-Fpointtowardhigherconcentrationsofsmaller

aromaticgroupsinthetarobtainedfromthemodifiedbiomassQsub2.

Increasing final temperature and holding time had a slight influence on the products

yields.

SECresultsshoweddifferencesintheMWdistributionofthetarsobtained.

127

5.0 References

1. Huber,G.W.,S.Iborra,andA.Corma,SynthesisofTransportationFuelsfromBiomass:

Chemistry, Catalysts, and Engineering. Chem. Rev. (Washington, DC, U. S.), 2006.

106(9):p.4044-4098.

2. Koppatz, S., et al.,H2richproductgasby steamgasificationofbiomasswith in situ

CO2 absorption in a dual fluidized bed system of 8 MW fuel input. Fuel Process.

Technol.,2009.90(7–8):p.914-921.

3. Pan,Y.G.,etal.,Fluidized-bedco-gasificationofresidualbiomass/poor coalblendsfor

fuelgasproduction. Fuel,2000.79(11):p.1317-1326.

4. Vélez, J.F., et al.,Co-gasificationofColombiancoal andbiomass in fluidizedbed:An

experimentalstudy. Fuel,2009.88(3):p.424-430.

5. Pinto,F.,etal.,Effectofcatalystsinthequalityofsyngasandby-productsobtainedby

co-gasificationofcoalandwastes.1.Tarsandnitrogencompoundsabatement. Fuel,

2007.86(14):p.2052-2063.

6. Brage,C.,etal.,Tarevolutionprofilesobtainedfromgasificationofbiomassandcoal.

BiomassBioenergy,2000.18(1):p.87-91.

7. AbuEl-Rub,Z.,E.A.Bramer,andG.Brem,ReviewofCatalystsforTarEliminationin

BiomassGasificationProcesses. Industrial&EngineeringChemistryResearch,2004.

43(22):p.6911-6919.

8. Rabou,L.P.L.M., etal.,TarinBiomassProducerGas,theEnergyresearchCentreofThe

Netherlands (ECN) Experience: An Enduring Challenge. Energy Fuels, 2009.

23(Copyright (C)2012AmericanChemical Society (ACS).AllRightsReserved.):p.

6189-6198.

9. Pfeifer,C.andH.Hofbauer,Developmentofcatalytictardecompositiondownstream

fromadualfluidizedbedbiomasssteamgasifier. PowderTechnology,2008.180(1–

2):p.9-16.

10. Wolfesberger-Schwabl,U.,I.Aigner,andH.Hofbauer,MechanismofTarGeneration

during Fluidized Bed Gasification and Low Temperature Pyrolysis. Industrial &

EngineeringChemistryResearch,2012.51(40):p.13001-13007.

11. EPA. http://www.epa.gov/region9/waste/solid/projects/pdf/ColumbiaFinal-

reportJune20.pdf.2012[cited2012Sept].

128

12. Herod,A.A.,K.D.Bartle,andR.Kandiyoti,CharacterizationofHeavyHydrocarbonsby

Chromatographic and Mass Spectrometric Methods:  An Overview. Energy & Fuels,

2007.21(4):p.2176-2203.

13. Kandiyoti, R., Herod, A.A., Bartle, K.D.,Solid Fuels andHeavy Hydrocarbon Liquids:

ThermalCharacterisationandAnalysis2006:Elsevier.

14. Diebold, J.P., A Review of the Chemical and Physical Mechanisms of the Storage

Stability of Fast Pyrolysis Bio-Oils., 2000, National Renewable Energy Laboratory:

Golden,Colorado.

15. Oasmaa,A.andS.Czernik,FuelOilQualityofBiomassPyrolysisOilsStateoftheArtfor

theEndUsers. Energy&Fuels,1999.13(4):p.914-921.

16. Boucher,M.E.,etal.,Bio-oilsobtainedbyvacuumpyrolysisofsoftwoodbarkasaliquid

fuel for gas turbines. Part II: Stability and ageing of bio-oil and its blends with

methanol and a pyrolytic aqueous phase. BiomassBioenergy, 2000.19(5): p. 351-

361.

17. Bridgwater,A.V.andG.V.C.Peacocke,Fastpyrolysisprocessesforbiomass. Renewable

andSustainableEnergyReviews,2000.4(1):p.1-73.

18. Diebold,J.P.andS.Czernik,AdditivesToLowerandStabilizetheViscosityofPyrolysis

OilsduringStorage. Energy&Fuels,1997.11(5):p.1081-1091.

19. Fahmi,R.,etal.,Theeffectofligninandinorganicspeciesinbiomassonpyrolysisoil

yields,qualityandstability. Fuel,2008.87(7):p.1230-1240.

20. Oasmaa,A.andE.Kuoppala,FastPyrolysisofForestryResidue.3.StorageStabilityof

LiquidFuel. Energy&Fuels,2003.17(4):p.1075-1084.

21. Oasmaa,A.,etal.,FastPyrolysisofForestryResidueandPine.4.Improvementofthe

ProductQualitybySolventAddition. Energy&Fuels,2004.18(5):p.1578-1583.

22. Oasmaa,A.L.,E.;KoponenP.;Levander,J.andTapolaE.().,Physicalcharacterisation

ofbiomass-basedpyrolysisliquids– Applicationofstandardfueloilanalyses., inVTT

Publications306.1997,TechnicalResearchCentreofFinland.:Espoo,Finland.

23. Doshi, V.A.,H.B. Vuthaluru, andT. Bastow, Investigations into the control of odour

and viscosity of biomass oil derived from pyrolysis of sewage sludge. Fuel Process.

Technol.,2005.86(8):p.885-897.

24. Dufour,A.,etal.,InSituAnalysisofBiomassPyrolysisbyHighTemperatureRheology

inRelationswith1HNMR. Energy&Fuels,2012.26(10):p.6432-6441.

129

25. Stamatov,V.,D.Honnery,andJ.Soria,Combustionpropertiesofslowpyrolysisbio-oil

produced from indigenous Australian species. Renewable Energy, 2006.31(13): p.

2108-2121.

26. Yu,F.,etal.,Physicalandchemicalpropertiesofbio-oilsfrommicrowavepyrolysisof

corn stover. Applied Biochemistry and Biotechnology, 2007. 137-140(1): p. 957-

970.

27. Qi, Z.J., C.; Tiejun, W. and Ying, X., Review of biomass pyrolysis oil properties and

upgradingresearch. EnergyConversionandManagement.,2007.48:p.87-92.

28. Majerski,P.A.,Methodofupgradingbiomasspyrolysisliquidsforuseasfuelsandasa

sourceofchemicalsbyreactionwithalcohols.,E.p.office,Editor1996.

29. Neeft, J.P.A.,Rational for setupof ImpingerTrain, technicalbackground report, Jan.

2005,CENBT/TF143.,inEuropeanStandard2005:TheNetherlands.

30. EuropeanStandard,Biomassgasification.Tarandparticlesinproductgases.Sampling

andanalysis.DDCEN/TS15439:2006.,2006.

31. Morgan, T.J., et al., Characterization of Molecular Mass Ranges of Two Coal Tar

DistillateFractions(CreosoteandAnthraceneOils)andAromaticStandardsbyLD-MS,

GC-MS,Probe-MSandSize-exclusionChromatography. Energy&Fuels,2008.22(5):

p.3275-3292.

32. DelRio,J.C.,R.P.Philp,andJ.Allen,Natureandgeochemistryofhighmolecularweight

hydrocarbons (above C40) in oils and solid bitumens. Organic Geochemistry, 1992.

18(4):p.541-553.

33. Hsieh, M. and R. Paul Philp, Ubiquitous occurrence of high molecular weight

hydrocarbonsincrudeoils. OrganicGeochemistry,2001.32(8):p.955-966.

34. Heath,D.,etal.,Quantificationof theC30+ fractionofNorthseagascondensatesby

high temperature capillary gas chromatography. Analytical Proceedings including

AnalyticalCommunications,1995.32(12):p.485-487.

35. Milne, T.A., Evans, R.J., Abatzoglou, N., Biomass gasification tars: their nature,

formation, destruction, and tolerance limits in energy conversion devices., in

Proceedingsof3rdbiomassconferenceoftheAmerica’s.1997:Quebec,Canada.p.729.

36. Parker,J.E.,etal.,Identificationoflargemolecularmassmaterialinhightemperature

coal tars and pitches by laser desorptionmass spectroscopy. Fuel, 1993.72(10): p.

1381-1391.

130

37. Johnson, B.R., et al., A test of the MALDI-MS calibration of SEC with N-methyl-2-

pyrrolidinoneforcoalderivedmaterials. Fuel,1998.77(14):p.1527-1531.

38. Artor,L.,Su,Y.,Yoshihisa,H.,Murata,S.,Nomura,M.,StructuralCharacterizationof

asphaltene fraction fromvacuumresidueofArabian crudemixtureby spectroscopic

andpyrolyticmethods. SekiyuGakkaishi.,1999.42:p.206.

39. Janitschke,W., et al., Investigationsof coalsby on-line coupled laser desorption/gas

chromatography/mass spectrometry(LD/GC/MS). JournalofAnalyticalandApplied

Pyrolysis,2000.56(1):p.99-111.

40. Letzel,T.,etal.,Phenyl-ModifiedReversed-PhaseLiquidChromatographyCoupled to

AtmosphericPressureChemicalIonizationMassSpectrometry: AUniversalMethodfor

theAnalysisofPartiallyOxidizedAromaticHydrocarbons. AnalyticalChemistry,2001.

73(7):p.1634-1645.

41. Marin,N.,etal.,Copyrolysisofwoodbiomassandsyntheticpolymersmixtures.PartII:

characterisation of the liquid phases. Journal of Analytical and Applied Pyrolysis,

2002.65(1):p.41-55.

42. Sharypov,V.I., et al.,Co-pyrolysisofwoodbiomassand syntheticpolymersmixtures.

Part III: Characterisation of heavy products. Journal of Analytical and Applied

Pyrolysis,2003.67(2):p.325-340.

43. Bayerbach, R., et al., Characterization of the water-insoluble fraction from fast

pyrolysisliquids(pyrolyticlignin):PartIII.MolarmasscharacteristicsbySEC,MALDI-

TOF-MS,LDI-TOF-MS,andPy-FIMS. JournalofAnalyticalandAppliedPyrolysis,2006.

77(2):p.95-101.

44. Lazaro,M.J.,etal.,Fractionationofawoodtarpitchbyplanarchromatographyforthe

characterisation of large molecular mass materials. Journal of Chromatography A,

1999.840(1):p.107-115.

45. Tanaka, R., et al., Analysis of the Molecular Weight Distribution of Petroleum

AsphaltenesUsingLaserDesorption-MassSpectrometry. Energy&Fuels,2004.18(5):

p.1405-1413.

46. Purevsuren, B., Davaajav, Y., Karaca, F., Morgan, T.J., George, A., Herod, A.A. and

Kandiyoti, R. , Pyrolysis of waste polypropylene and characterization of the tar.

EuropeanJournalofMassSpectrometry,2009.15:p.23-33.

47. Herod, A.A., et al., Analytical Methods for Characterizing High-Mass Complex

PolydisperseHydrocarbonMixtures:AnOverview. Chem.Rev.(Washington,DC,U.S.),

131

2012. 112(Copyright (C) 2012 American Chemical Society (ACS). All Rights

Reserved.):p.3892-3923.

48. Karaca, F., et al., Molecular mass ranges of coal tar pitch fractions by mass

spectrometry and size-exclusion chromatography. Rapid Communications in Mass

Spectrometry,2009.23(13):p.2087-2098.

49. Podgorski, D.C., et al., Characterization of Pyrogenic Black Carbon by Desorption

Atmospheric Pressure Photoionization Fourier Transform Ion Cyclotron Resonance

MassSpectrometry. AnalyticalChemistry,2011.84(3):p.1281-1287.

50. Jin, J.M., S. Kim, and J.E. Birdwell, Molecular Characterization and Comparison of

ShaleOilsGeneratedbyDifferentPyrolysisMethods. Energy&Fuels,2011.26(2):p.

1054-1062.

51. Smith, D.F., et al., Characterization of Acidic Species in Athabasca Bitumen and

Bitumen Heavy Vacuum Gas Oil by Negative-Ion ESI FT−ICR MS with and without

Acid−Ion Exchange Resin Prefractionation. Energy & Fuels, 2008. 22(4): p. 2372-

2378.

52. McKenna,A.M.,etal., IdentificationofVanadylPorphyrins inaHeavyCrudeOiland

Raw Asphaltene by Atmospheric Pressure Photoionization Fourier Transform Ion

Cyclotron Resonance (FT-ICR) Mass Spectrometry. Energy & Fuels, 2009. 23(4): p.

2122-2128.

53. Karaca, F., et al., The Calibration of Size Exclusion Chromatography Columns: 

MolecularMassDistributions ofHeavyHydrocarbon Liquids. Energy& Fuels, 2004.

18(3):p.778-788.

54. Berrueco, C., et al., Calibration of Size-Exclusion Chromatography Columns with 1-

Methyl-2-pyrrolidinone (NMP)/Chloroform Mixtures as Eluent: Applications to

Petroleum-DerivedSamples. Energy&Fuels,2008.22(5):p.3265-3274.

55. Morgan, T.J., et al., Isolation of Size Exclusion Chromatography Elution-Fractions of

Coal and Petroleum-Derived Samples and Analysis by Laser Desorption Mass

Spectrometry. Energy&Fuels,2009.23(12):p.6003-6014.

56. Morgan,T.J.,etal.,Estimatingmolecularmassesofpetroleum-derivedfractions:High

mass(>2000u)materialsinmaltenesandasphaltenesfromMayacrudeoil. Journalof

ChromatographyA,2010.1217(24):p.3804-3818.

57. Islas,C.A.,etal.,Matchingaveragemassesofpitchfractionsofnarrowpolydispersity,

derived from matrix-assisted laser desorption ionisation time-of-flight mass

132

spectrometry,withthepolystyrenecalibrationofSEC. JournalofSeparationScience,

2003.26(15-16):p.1422-1428.

58. Pindoria, R.V., et al., A two-stage fixed-bed reactor for direct hydrotreatment of

volatiles fromthehydropyrolysisofbiomass:effectofcatalyst temperature, pressure

and catalyst ageing time on product characteristics. Fuel, 1998. 77(15): p. 1715-

1726.

59. Somrang, Y., Effect of Operating Conditions on Product Distributions and Bio-oil

Ageing in Biomass Pyrolysis., in Chemical Engineering.2011, Imperial College:

London,UK.

60. Morgan,T.J.,Molecularmassandstructuralcharacterizationofheavy

hydrocarbonmaterials,inChemicalEngineering.2008,ImperialCollege:London.

61. Morgan,T.J., et al.,CharacterizationofMayaCrudeOilMaltenesandAsphaltenes in

TermsofStructuralParametersCalculatedfromNuclearMagneticResonance(NMR)

Spectroscopy and Laser Desorption−Mass Spectroscopy (LD−MS). Energy & Fuels,

2010.24(7):p.3977-3989.

62. P.Álvarez,C.B.,S.Venditti,A.George,T.J.Morgan,M.Millan,A.A.Herod,R.Kandiyoti.

Syncrude Characterization inACS Fuel Div, Nationalmeeting. 2009. Salt Lake City,

USA:ACS.

63. S. Kalisz, F.M., J.Martinec, J. Ungeheuer, L. Andersen,D. Baxter,Air and air-steam

biomass gasification in interconnected, internally circulating fluidized beds, in 17th

EuropeanBiomassConference&Exhibition2009:Hamburg,Germany.p.980.

64. Miccio,F.,etal.,Combustionofliquidbio-fuelsinaninternalcirculatingfluidizedbed.

ChemicalEngineeringJournal,2008.143(1–3):p.172-179.

65. Li,C.-Z.,etal.,UV-FluorescenceSpectroscopyofCoalPyrolysisTars. Energy&Fuels,

1994.8(5):p.1039-1048.

66. Mohan,D.,C.U.Pittman,Jr.,andP.H.Steele,PyrolysisofWood/BiomassforBio-oil:A

Critical Review. Energy Fuels, 2006. 20(Copyright (C) 2012 American Chemical

Society(ACS).AllRightsReserved.):p.848-889.

67. Guilbault., G.G., Practical Fluorescence, Second Edition.1990: Marcel Dekker, New

York,USA.

133

68. Metzinger, T. and K.J. Hüttinger, Investigations on the cross-linking of binder pitch

matrix of carbon bodies with molecular oxygen — Part I. Chemistry of reactions

betweenpitchandoxygen. Carbon,1997.35(7):p.885-892.

69. A� lvarez, P., et al., Characterization and Pyrolysis Behavior of Novel Anthracene Oil

Derivatives. Energy&Fuels,2008.22(6):p.4077-4086.

70. Fernández, A.L., et al., Air-blowing of anthracene oil for carbon precursors. Carbon,

2000.38(9):p.1315-1322.

71. Álvarez,P.,etal.,Aninsightintothepolymerizationofanthraceneoiltoproducepitch

usingnuclearmagneticresonance. Fuel,2012(0).

72. Lee,R.F.,Photo-oxidationandPhoto-toxicityofCrudeandRefinedOils. SpillScience

&amp;TechnologyBulletin,2003.8(2):p.157-162.

73. Thominette, F. and J. Verdu, Photooxidative behavior of crude oils relative to sea

pollution. Part I. Comparative study of various crude oils and model systems. Mar.

Chem., 1984.15(Copyright (C) 2012 American Chemical Society (ACS). All Rights

Reserved.):p.91-104.

74. Thominette, F. and J. Verdu, Photooxidative behavior of crude oils relative to sea

pollution. Part II. Photo-induced phase separation. Mar. Chem., 1984.15(Copyright

(C)2012AmericanChemicalSociety(ACS).AllRightsReserved.):p.105-115.

75. Heitner., C., Lignin and Lignans: Advances in Chemistry. Chapter 16. The

photochemistryoflignin2010:CRCPress.

76. Omais,B.,etal.,CharacterizationofOxygenatedSpeciesinCoalLiquefactionProducts:

AnOverview. Energy&Fuels,2010.24(11):p.5807-5816.

77. Strausz, O.P., I. Safarik, and E.M. Lown,Cause of Asphaltene Fluorescence Intensity

Variation with Molecular Weight and Its Ramifications for Laser Ionization Mass

Spectrometry. Energy&Fuels,2009.23(3):p.1555-1562.

References:

1. Aigner, I., C. Pfeifer, and H. Hofbauer, Co-gasification of coal and wood in a dual

fluidizedbedgasifier. Fuel,2011.90(7):p.2404-2412.

2. Vassilev, S.V., et al.,Anoverviewof theorganicand inorganicphase compositionof

biomass. Fuel, 2012.94(Copyright (C) 2012American Chemical Society (ACS). All

RightsReserved.):p.1-33.

134

3. Morgan, T.J., et al., Characterization of Molecular Mass Ranges of Two Coal Tar

DistillateFractions(CreosoteandAnthraceneOils)andAromaticStandardsbyLD-MS,

GC-MS,Probe-MSandSize-exclusionChromatography. Energy&Fuels,2008.22(5):

p.3275-3292.

4. Karaca, F., et al., The Calibration of Size Exclusion Chromatography Columns: 

MolecularMassDistributions ofHeavyHydrocarbon Liquids. Energy& Fuels, 2004.

18(3):p.778-788.

5. Morgan, T.J., et al., Isolation of Size Exclusion Chromatography Elution-Fractions of

Coal and Petroleum-Derived Samples and Analysis by Laser Desorption Mass

Spectrometry. Energy&Fuels,2009.23(12):p.6003-6014.

6. Herod, A.A., et al., Analytical Methods for Characterizing High-Mass Complex

PolydisperseHydrocarbonMixtures:AnOverview. Chem.Rev.(Washington,DC,U.S.),

2012. 112(Copyright (C) 2012 American Chemical Society (ACS). All Rights

Reserved.):p.3892-3923.

7. Morgan,T.J.,etal.,Estimatingmolecularmassesofpetroleum-derivedfractions:High

mass(>2000u)materialsinmaltenesandasphaltenesfromMayacrudeoil. Journalof

ChromatographyA,2010.1217(24):p.3804-3818.

8. Strausz, O.P., I. Safarik, and E.M. Lown,Cause of Asphaltene Fluorescence Intensity

Variation with Molecular Weight and Its Ramifications for Laser Ionization Mass

Spectrometry. Energy&Fuels,2009.23(3):p.1555-1562.

9. Strausz,O.P.,etal.,ACritiqueofAsphalteneFluorescenceDecayandDepolarization-

Based Claims aboutMolecularWeight andMolecular Architecture. Energy& Fuels,

2008.22(2):p.1156-1166.

10. Morgan,T.J., et al.,CharacterizationofMayaCrudeOilMaltenesandAsphaltenes in

TermsofStructuralParametersCalculatedfromNuclearMagneticResonance(NMR)

Spectroscopy and Laser Desorption−Mass Spectroscopy (LD−MS). Energy & Fuels,

2010.24(7):p.3977-3989.

11. Zander,M.andM.W.Haenel,Regularitiesinthefluorescencespectraofcoal-tarpitch

fractions. Fuel,1990.69(9):p.1206-1207.

12. P.Álvarez,C.B.,S.Venditti,A.George,T.J.Morgan,M.Millan,A.A.Herod,R.Kandiyoti.

Syncrude Characterization inACS Fuel Div, Nationalmeeting. 2009. Salt Lake City,

USA:ACS.

135

13. Morgan, T.J., et al., Optimization of 1H and 13C NMR Methods for Structural

Characterization of Acetone and Pyridine Soluble/Insoluble Fractions of a Coal Tar

Pitch. Energy&Fuels,2008.22(3):p.1824-1835.

14. Morgan,T.J.,Molecularmassandstructuralcharacterizationofheavy

hydrocarbonmaterials,inChemicalEngineering.2008,ImperialCollege:London.

15. Guilbault., G.G., Practical Fluorescence, Second Edition.1990: Marcel Dekker, New

York,USA.

16. A� lvarez, P., et al., Characterization and Pyrolysis Behavior of Novel Anthracene Oil

Derivatives. Energy&Fuels,2008.22(6):p.4077-4086.

136

137

138