THESISFORTHEDEGREEOFDOCTOROFPHILOSOPHY

HydrometallurgicalRecoveryofRareEarthElementsfromFluorescentLampWasteFractions

CRISTIANTUNSU

DepartmentofChemistryandChemicalEngineeringCHALMERSUNIVERSITYOFTECHNOLOGY

Gothenburg,Sweden,2016

HydrometallurgicalRecoveryofRareEarthElementsfromFluorescentLampWasteFractionsCRISTIANTUNSUISBN978-91-7597-434-7

©CRISTIANTUNSU,2016

DoktorsavhandlingarvidChalmerstekniskahögskolaNySerieNr.4115ISSN0346-718X

NuclearChemistryandIndustrialMaterialsRecyclingDepartmentofChemistryandChemicalEngineeringChalmersUniversityofTechnologySE-41296,GothenburgSwedenTelephone:+46(0)31-7721000

Cover:ScanningElectronMicroscopypictureofoneofthefluorescentlampwastephosphorsfractionsinvestigated(5000xmagnification)

Printedby:ChalmersReproserviceGothenburg,Sweden,2016

HydrometallurgicalRecoveryofRareEarthElementsfromFluorescentLampWasteFractions

CRISTIANTUNSUNuclearChemistryandIndustrialMaterialsRecyclingDepartmentofChemistryandChemicalEngineering

ChalmersUniversityofTechnology

ABSTRACT

Recoveryandreuseofmaterialsisimportantforacirculareconomy.Inrecentyearstherecoveryofcritical metals from end-of-life products has received increased attention. Various streams, e.g.permanentmagnets,nickelmetalhydridebatteriesandfluorescentlampsareconsideredtargetsforthe recovery of rare earth elements (REEs). The last can be a source of up to six different REEs:europiumandyttrium(primarily),aswellaslanthanum,cerium,terbiumandgadolinium(secondarily).

Becausefluorescentlampsusemercurytogeneratelight,adecontaminationstepneedstobecarriedoutpriortoprocessingdiscardedproductsforREEsrecovery.Thisisoftencarriedoutusingthermaltreatment(upto800°C)butthismethodhassomedrawbacks,e.g.energyconsumptionandthefactthatitisnotbestsuitedforwastestreamscontaininghighamountsofmoisture.

Hydrometallurgical methods for the decontamination of fluorescent lamp waste fractions andsubsequentrecoveryoftheREEscontainedarepresentedinthisstudy.Aselectiveleachingprocessfollowedbyseparationofmetalsusingsolventextractionwasdeveloped.Mercurywasleachedinafirststageusingiodineinpotassiumiodidesolutions.Furtherprocessingofthemercuryinsolutionwasinvestigatedusingvarioustechniques,e.g.ionexchange,reductionandsolventextraction.Inasecondleachingstep,impuritymetals,e.g.calcium,barium,etc.,wereselectivelyleachedfromtheREEswithnitricacidsolutionbymakinguseoftheirfastdissolutionkinetics.Furtherleaching,carriedoutwithmoreconcentratedacidicsolutionsforlongertime,ledtothedissolutionoftheREEs.Partialleachingselectivity between yttrium + europium and the other four REEs was achieved by controlling theleachingtime,acidconcentrationandtemperature.

AgroupseparationoftheREEionsinsolutionwascarriedoutusingsolventextractionwithCyanex923,acommercialmixoftrialkylphosphineoxides.Testingoftheprocessatlaboratorypilotscaleinmixer-settlersshowedpromisingresults,leadingtoafinalproductconsistingofayttrium/europium-rich solution. Over 99% of the REEs present in lamp leachates were extracted and stripped,respectively,inamixer-settlersystemcomprisedofthreeextractionstagesandfourstrippingstages.The metals were then further separated using Cyanex 572, a novel phosphorus-based chelatingextractantaimedattheseparationofindividualREEs.SelectiveseparationofyttriumandeuropiumwasachievedbycontrollingtheequilibriumpH(pHeq)duringextraction.YttriumwasextractedatpHeq=0andeuropiumatpHeq=1.Rareearthoxideswerepreparedviaoxalicacidprecipitationandthermaltreatmentoftheobtainedoxalatesat800°C.AmixedREEoxide(99.96%REEs,with94.61%yttrium,5.09% europium and 0.26% others) was synthetized from the strip product after extraction withCyanex 923. Yttrium oxide (99.82%) and europium oxide (91.6%) were synthetized from the stripproductsafterextractionwithCyanex572.

KEYWORDS: fluorescent lamp waste, rare earth elements, recycling, waste treatment, leaching,solventextraction,Cyanex923,Cyanex572.

LISTOFPUBLICATIONS

Thisthesisisbasedontheworkcontainedinthefollowingpublications,referredtobyRomannumeralsinthetext:

PublicationI

Tunsu,C.,Ekberg,C.,Retegan,T.,2014.Characterizationandleachingofrealfluorescentlampwastefortherecoveryofrareearthmetalsandmercury.Hydrometallurgy144-145,91-98.

Contribution:mainauthor,allexperimentalwork,processingandinterpretationofdata.

PublicationII

Tunsu,C.,Ekberg,C.,Foreman,M.,Retegan,T.,2014.StudiesontheSolventExtractionofRareEarthMetalsfromFluorescentLampWasteUsingCyanex923.SolventExtractionandIonExchange32,650-668.

Contribution:mainauthor,allexperimentalwork,processingandinterpretationofdata.

PublicationIII

Tunsu, C., Ekberg, C., Foreman, M., Retegan, T., 2015. Investigations regarding the wetdecontamination of fluorescent lamp waste using iodine in potassium iodide solutions. WasteManagement36,289-296.

Contribution:mainauthor,allexperimentalwork,processingandinterpretationofdata.

PublicationIV

Tunsu,C.,Petranikova,M.,Ekberg,C.,Retegan,T.,2016.Ahydrometallurgicalprocessfortherecoveryofrareearthelementsfromfluorescentlampwastefractions.SeparationandPurificationTechnology161,172–186

Contribution:mainauthor,majorityoftheexperimentalwork,processingandinterpretationofdata.

PublicationV

Tunsu,C.,Lapp,J.B.,Ekberg,C.,Retegan,T.SelectiveseparationofyttriumandeuropiumusingCyanex572forapplicationsinfluorescentlampwasteprocessing.Hydrometallurgy,decisionpending.

Contribution:mainauthor,experimentalwork,processingandinterpretationofdata.

Otherpublicationsthatwerenotincludedinthisthesis

Tunsu,C.,Petranikova,M.,Gergorić,M.,Ekberg,C.,Retegan,T.,2015.Reclaimingrareearthelementsfromend-of-lifeproducts:Areviewoftheperspectivesforurbanminingusinghydrometallurgicalunitoperations.Hydrometallurgy156,239-258.

Contribution:mainauthor.

Tunsu,C.,Retegan,T.,2016.Chapter6:HydrometallurgicalprocessesfortherecoveryofmetalsfromWEEE,in:Chagnes,A.,Cote,G.,Ekberg,C.,Nilsson,M,Retegan,T.(Eds.),WEEERecycling:Research,Development,andPolicies.Elsevier,139-175.

Contribution:mainauthor.

TABLEOFCONTENTS

ABBREVIATIONSANDDEFINITIONS.................................................................................................11.INTRODUCTION...........................................................................................................................3

1.1.Goalanddrivingforces.................................................................................................................42.BACKGROUNDANDTHEORY........................................................................................................7

2.1.Rareearthelements.....................................................................................................................72.2.Mercury........................................................................................................................................82.3.Fluorescentlamps........................................................................................................................92.4.Industrialprocessingoffluorescentlamps.................................................................................112.5.Hydrometallurgicalunitoperations...........................................................................................12

2.5.1.Leaching...............................................................................................................................132.5.2.Solventextraction...............................................................................................................13

2.6.HydrometallurgicalrecoveryofREEsfromfluorescentlamps...................................................163.EXPERIMENTAL..........................................................................................................................21

3.1.Overviewoftheexperimentalwork...........................................................................................213.2.Generalinformationaboutthefluorescentlampwastefractionsused....................................223.3.Chemicalsandtechniques..........................................................................................................233.4.Characterizationofwastefractions...........................................................................................243.5.Hydrometallurgicalremovalofmercury....................................................................................243.6.Leachingofmetals......................................................................................................................243.7.SolventextractionofREEswithCyanex923..............................................................................253.8.Testingoftheprocessinlaboratorypilotscale..........................................................................253.9.FurtherseparationofREEswithCyanex572.............................................................................263.10.PreparationofREEoxides........................................................................................................26

4.RESULTSANDDISCUSSIONS.......................................................................................................274.1.Characterizationofwastefractions...........................................................................................274.2.Hydrometallurgicalremovalofmercury....................................................................................294.3.Leachingofmetals......................................................................................................................334.4.SolventextractionofREEswithCyanex923..............................................................................394.5.Testingoftheprocessinlaboratorypilotscale..........................................................................43

4.5.1.Processingofthermally-treatedfractions...........................................................................434.5.2.Processingofmercury-contaminatedfractions..................................................................46

4.6.FurtherseparationofREEswithCyanex572.............................................................................484.7.PreparationofREEoxides..........................................................................................................544.8.Proposedprocessflowsheet......................................................................................................56

5.SUMMARYANDCONCLUSIONS.................................................................................................597.FUTUREWORK..........................................................................................................................618.ACKNOWLEDGMENTS................................................................................................................63REFERENCES..................................................................................................................................65APPENDIX1.ANALYTICALANDINTRUMENTALTECHNIQUES.........................................................71APPENDIX2.MIXER-SETLERS.........................................................................................................73

1

ABBREVIATIONSANDDEFINITIONS

Thefollowingabbreviationsanddefinitionsareusedthroughoutthisthesis:

°C DegreesCelsius> Morethan< Lessthan%wt. Percentageweighttotal%vol. Volumetricpercentage%E PercentageextractedαA/B SeparationfactorbetweenAandB∆H0 Enthalpychangeatstandardstate∆S0 Entropychangeatstandardstateapprox. Approximatelyaq aqueousBAM BaMgAl10O17:Eu2+phosphorCAT (Ce,Tb)MgAl11O19phosphorCFL CompactfluorescentlampCRT CathoderaytubeCyMe4BTBP 6,6’-bis(5,5,8,8-tetramethyl-5,6,7,8-tetrahydro-benzo-[1,2,4-]triazin-3-yl)-

[2,20]bipyridineD DistributionratioD2EHPA Bis-2-ethylhexylphosphoricacide.g. Forexample;suchasg Gram(s)h Hour(s)HDD HarddiskdriveHEHEHP 2-ethylhexylphosphonicacidmono-2-ethylhexylesterI2/KI iodineinpotassiumiodide(solution)ICP-OES InductivelyCoupledPlasma-OpticalEmissionSpectroscopyICP-MS InductivelyCoupledPlasma-MassSpectrometryK DegreesKelvinkg Kilogram(s)L Litre(s)Lanthanides The15elements57Lato71Lu,inclusiveLanthanoids The14elements58Ceto71Lu,inclusive(thelanthanideswithout57La)LAP LaPO4:Ce3+,Tb3+phosphorLED LightemittingdiodeM Molarconcentration(mol/L)MPa Megapascalmin Minute(s)mL millilitre(s)

2

NiMH NickelmetalhydrideO:A Organic:aqueous(phaseratio)org OrganicpHeq EquilibriumpHPhosphor Compoundcapableofluminescenceppb PartsperbillionREEs Rareearthelements(thelanthanides,39Yand21Sc)rpm Revolutionsperminutes SolidSEM/EDS ScanningElectronMicroscopy/EnergyDispersiveSpectroscopyS/L Solidtoliquid(ratio)TBP Tri-n-ButylPhosphateTPH TetraPropyleneHydrogenatedTRPO TrialkylPhosphineOxides;Cyanex923 UN UnitedNationsUS UnitedStatesUV Ultravioletvpm Vibrationsperminutevs. Versusw/v Weight/VolumeXRD X-RayDiffractionYOX Y2O3:Eu3+phosphor

3

1.INTRODUCTION

Recyclingandre-useofmaterialsareimportantpartsincirculareconomies.Theincreasingdemandofrawmaterials requiredby industries isachallengethatneeds tobeaddressed.Thisdemandoftenleads to excessivemining and exploitation of natural deposits, aswell as generation of secondarystreams during resource processing. These considerations have led to an increased interest inrecycling,especiallyforscarcematerialssuchastherareearthelements(REEs).

TheREEsarepresentlyregardedasbeingsomeofthemostcriticalmaterials(EuropeanCommission,2011;UNEnvironmentProgrammeandUNUniversity,2009;USDepartmentofEnergy,2011).TheEuropean Commission declared REEs to have the highest supply risk among non-energy, non-agricultural rawmaterials twice, in 2010 and2014 (EuropeanCommission, 2010; 2014). There areseveral reasons for this, themost important ones being related to geological distribution of REEs,difficultandcomplexseparationofindividualelements,highdemandandpoliticalfactors.

ThefactthatREE-containingmineralsrarelyoccur inconcentratedformsmakesexploitationofthisgroupofelementsdifficult (GuptaandKrishnamurthy,2005).Theirgeologicaloccurrence, togetherwitheithertheuraniumorthoriumdecaychains,makestheprocessingofREEschallenginginmanycountries due to radiotoxicity (Kilbourn, 1994). Moreover, REEs have similar chemical properties,whichtranslates intoseparationandpurificationdifficulties.ProductionofREEs isoftenassociatedwith environmental issues due to the numerous hydrometallurgical separation stages needed toachievethehighdegreesofpurityrequiredincertainapplications.Thisoftenleadstohighamountsofsecondarywaste.

TheREEshaveawidevarietyofapplications,rangingfromsimpleones,e.g.polishingagents,tomoreadvancedones,e.g.laserrepeatersforhighspeedandhighdistancedatatransfer(USGS,2002).Someof these applications are characterized by high specificity; the REEs needed having no efficientsubstitutes with similar properties. Most importantly, REEs are essential in future sustainabletechnologies.Neodymium,dysprosium,europium,yttriumandterbium,thefivemostcriticalREEs(USDepartmentofEnergy,2011),areusedingreenenergyapplicationse.g.:

• neodymiumanddysprosium inpermanentmagnets,whichareused inwindturbines,harddiskdrives(HDDs),speakers,headphonesandothers;

• europium, yttrium and terbium, together with lanthanum, cerium and gadolinium, inphosphorsinlow-energyfluorescentlamps;

• neodymium,yttrium,cerium,lanthanumandpraseodymiuminnickelmetalhydride(NiMH)batteriesforelectrictransportation.

Becauseoftheir largevarietyofapplications,REEsare inhighdemand.Arecentconcernwastheirlimitedavailabilityon themarket. The closingofMountainPassmine in theUS in theearly2000sallowedChinatogainover90%oftheREEmarketshare,althoughthecountrypossesseslessthanhalfof the global deposits (Xie et al., 2014). This has led to significant price fluctuations over thepastdecade,aswellasexportquotas.

Alltheseaforementionedfactorshavefocusedattentiontowardsabetterprocessingofvariousend-of-life products and industrial waste streams for the recovery of the contained REEs. Permanent

4

magnetsinHDDs,NiMHbatteriesandfluorescentlampsarethemosttargetedproductsduetotheirabundanceonthemarketandrelativelyhighREEcontent(Tunsuetal.,2015).

UptosixREEsarefoundinthecompositionofthephosphorspowderofmodernfluorescentlamps:cerium,europium,gadolinium,lanthanum,terbium,andyttrium.Mercuryisanessentialcomponentin theseappliances,being required forphotonemission (Changetal., 2009). Itspresence requiresspecialcollectionandtreatmentofend-of-lifefluorescentlamps.Moreover,mercuryinterfereswiththefurtherprocessing,requiringremovalfromthestreampriortotherecoveryofthecontainedREEs.Hydrometallurgicalmethodse.g.leaching,solventextraction,ionexchangeandprecipitationaremostsuited for the recovery of REEs and have been studied in recent years for potential processing offluorescentlampwaste(Tunsuetal.,2015).Despitetheresearch,large-scaleapplicationstorecoverREEsareuncommon.In2011theglobalaveragefunctionalrecyclingratesofREEsandmercuryfromend-of-life products were reported to be less than 1% and 1-10%, respectively (UN EnvironmentProgramme,2011).Thisisbecausesomeoftheeffortsinthisfieldwereonlycarriedoutatbenchscale,sometimesusingcleanlamps,carefullytreatedand/ordecontaminatedmaterial,orpuresamplesthatimitatethecompositionofphosphorspowder.Manyoftheproposedprocesseswerenottestedusinghigher amounts of real, chemically complex and mercury-contaminated waste. Description of fullprocessesthatusecontaminatedfluorescent lampwastefractionsas inputfeedandseparatesolidREEcompoundsasfinalproductsarescarce.Forthisreason,improvingand/ordevelopingindustrialprocesses that remove mercury from phosphors powder fractions and individually recover thecontainedREEsisofgreatimportance.

1.1.Goalanddrivingforces

ThegoalofthisresearchisthedevelopmentofarobustprocessforrecoveryofREEscontainedinthephosphorspowderoffluorescentlampwastefractions.Emphasishasbeenplacedon:

• developingawet-basedmercurydecontaminationprocessasanalternativetothetraditional,energyintensive,thermaltreatmenttoremovethemercuryinthewaste.Thermaltreatmentcanbeanissueforwastefractionsobtainedviawet-sievingprocesses(describedinsection2.4)(Binnemansetal.,2013)e.g.someofthewastesamplesinvestigatedinthisthesis;

• separationandrecoveryofREEsfromthewasteusingahydrometallurgicalapproachbasedonselective leaching, solventextractionandprecipitation.Preparationof solidREEsoxideswasdesired;

• scaling-upthedevelopedprocessuptolaboratorypilotscale,basedonleachingreactorsandcounter-currentmixer-settlercontactors.

Theseweredeemedimportantbecause,atthebeginningofthepresentdecade,largescaletreatmentofrealfluorescentlampwastewasscarcelydescribedintheliterature,especiallycompleteprocessesthataddressallstepse.g.decontamination,selectiveleaching,individualREEsseparationandoxideproduction.

The development of effective recycling processes cannot entirely replace mining, especiallyconsideringtheincreasingdemandforrawmaterials.However,processingofend-of-lifeproductse.g.fluorescentlampscanprovideadditionalsourcesofREEsinadditiontoknownnaturaldepositsand

5

can increasetheglobalavailabilityoftheseelements.Thedrivingforcescanbegrouped intothreecategories,asshowninTable1:materialsupply,environmentalbenefitsandeconomicissues(Tunsuetal.,2015).

Table1–SomeofthedrivingforcesbehindthedevelopmentofprocessesforthedecontaminationandrecoveryofREEsfromfluorescentlampwaste.AdaptedfromTunsuetal.,2015.

Sector Benefits

Environmental

Limitthespreadingofradioactiveisotopese.g.uraniumandthoriumduringREEsoreminingandprocessing.Preventthespreadingoftoxicchemicalse.g.mercuryintheenvironment.Reducelandfillareasbothwhenitcomestominingtailingsanddisposalofdiscardedproductsorfractionsfromdiscardedproducts.Cleanlandscapes.Reduceexploitationbyreducingmining.Conservenaturaldeposits.

Materialsupply

Materialscanberecycledbackintoproductioncycles(forthesameordifferentapplications).Increasetheavailabilityofcertaincompounds(alsoduetothefactthattheconcentrationsofREEsarehigherindiscardedproductscomparedtosomeores).

Economic Reducethecostsofrawmaterialsbycreatingadditionalsupply.

6

7

2.BACKGROUNDANDTHEORY

2.1.Rareearthelements

TheREEs group, also referred toas rareearthmetals or rareearths comprises the14 lanthanoids(ceriumtolutetium),lanthanum,yttriumandscandium(GreenwoodandEarnshaw,1998).Thisgroupisoftendividedintotwogroups;lightREEsandheavyREEs.Theformer,alsoreferredtoastheceriumgroup,comprisesthefirstsevenlanthanides,uptoeuropium(GreenwoodandEarnshaw,1998;Tyler,2004).Therestofthelanthanidesandyttrium,whichhasanionicradiusbetweenthoseofdysprosiumandholmium,formtheheavyREEsgroup,alsoreferredtoastheyttriumgroup.Althoughnotthatwelldefined,anintermediategroupcontainingREEsofintermediateatomicmassandionicradius,knownasmediumREEs,issometimesmentionedintheliterature(Tyler,2004).

Althoughthenameimpliesotherwise,REEsarerelativelyabundantintheEarth’scrust,withlightREEse.g.lanthanum,ceriumandneodymiumbeingmorecommonthanlead.ThuliumandlutetiumarethescarcestREEsbuteventheseareonehundredtimesmoreabundantthangold(USGS,2002).Themainissueistheirgeologicaldistribution.Rareearthsareencounteredtogetherwithothermembersofthegroup and seldom form concentrated ore bodies. Due to the fact that they have similar chemicalproperties, their individual separation is challenging. Obtaining high purity REEs requires complexchemicalprocessinge.g.alargenumberofsolventextractionseparationandstrippingstages.

The slowdecrease in ionic radii along the lanthanide series, known as the lanthanide contraction,allowsindividualelementseparation(McLennan,1994).Thisdecreaseleadstoincreasedstrengthofcation-anion, ion-dipole and ion-induced dipole interactions, meaning that heavier REEs will formstrongercomplexeswithharddonorextractantmolecules(Nash,1993).Thetetradeffect(Peppardetal.,1969)wasnoticedinvariousextractionsystemswhenplottingthelogarithmofcertainphysicalchemical parameters vs. the corresponding atomic number. Four breakpoints, explained by thestabilizationenergy related to the inter-electron repulsionenergyof the4f electrons,werenoted;lanthanum–cerium–praseodymium–neodymium, promethium–samarium–europium–gadolinium,gadolinium–terbium–dysprosium–holmium, and erbium–thulium–ytterbium–lutetium. Thebreakpointinthemiddleoftheseriesisduetothestabilizedhalf-filled4f7electronconfigurationofgadolinium.

As mentioned in the introduction section, REEs are essential in a wide variety of technologicalapplications,somecharacterizedbyhighspecificity(Table2).

8

Table2–SomeapplicationsofREEs(USDepartmentofEnergy,2011;USEPA,2012;USGS,2002).

Element ApplicationsScandium Metalalloysfortheaerospaceindustry,nuclearapplications,lighting.

YttriumPhosphors,metalalloys,lasers,temperaturesensors,ceramics,yttriumirongarnet(transmitterandtransducerofacousticenergy),yttriumaluminiumgarnet(simulateddiamond).

Lanthanum Carbonarclamps.automotivecatalysts,catalystsforpetroleumrefining,glasses,phosphors,batteries,lasers,X-rayfilms.

Cerium Lighterflints,polishingpowders,componentanddecolourizeringlasses,carbon-arclighting,catalystinpetroleumrefining,metallurgicalandnuclearapplications.

Praseodymium Carbonarclights,glassandenamelcolorants,magnets(improvedcorrosionresistance),searchlights,airportsignallenses,photographicfilters.

Neodymium Magnets,lasers,glasscolorant(weldergoggles),fluid-frackingcatalysts,colorantforenamels.

Promethium Betaradiationsource,fluid-frackingcatalyst,portableX-raysources.

Samarium Magnets,reactorcontrolrods,guidanceandcontrolsystems,electricmotors,carbonarclamps,lasers,infraredabsorbingglass.

Europium Phosphors,lasers,glassadditive,targetingandweaponssystems,communicationdevices.

GadoliniumPhosphors,glassadditives,magneticresonanceimagingcontrastagent,gadoliniumyttriumgarnets(microwaveapplications),alloys,magnetocaloricmaterials,burnableneutronpoisoninnuclearfuel.

Terbium Phosphors,magnets,guidanceandcontrolsystems,targetingandweaponssystems,electricmotors.

Dysprosium Magnets,lasers,nuclearsector,alloys.Holmium Industrialmagnets,nuclearapplications.Erbium Lasers,glasscolorant,nuclearandmetallurgicaluses.

ThuliumRadiationsourceinportableX-rayequipment,potentiallyusefulasanenergysource(171Tm),ferrites(ceramicmagneticmaterials)usedinmicrowaveequipment,dopingfibrelasers,highpowermagnets.

Ytterbium PortableX-raydevices(radiationsource),lasers,fibre-opticcables,improvingthegrainrefinement,strengthandothermechanicalpropertiesofstainlesssteel.

Lutetium Catalystsincracking,alkylation,hydrogenation,andpolymerization,X-rayphosphors.

2.2.Mercury

Mercuryisatoxicelementforlivingorganismsandtheenvironment(CliftonII,2007).However,dueto itsphysicalandchemicalproperties,mercury isusedinseveralapplications,acommonexamplebeingfluorescentlamps(Changetal.,2009).Oneofthepropertiesofmercuryistherapidincreaseinvapour pressure and concentration in airwith an increase in temperature (Table 3) (OSHA, 1987).Becauseofthis,oneofthemethodsusedtorecovermercuryfromcontaminatedmaterialsisthermaltreatment.Analternativeisawetprocess,inwhichmercuryisleachedfromthematerialusingsuitableleachingagents.Thespeciationofmercurye.g.elementalmercuryand/orionicspecies(Hg+;Hg2+),aswell as other factors e.g. strong adsorptionofmercury in thematerial, presenceofmoisture etc.,strongly influence the typeofsuitabledecontaminationprocessesand, in thecaseof leaching, thechoiceofleachingagent.Forfluorescentlamps,thisaspectisfurtheraddressedinsection2.4.

9

Table3–Vapourpressure-saturationconcentrationofmercuryatvarioustemperatures(OSHA,1987).

Temperature(°C) VapourPressure(Torr) Mercuryconcentrationinair(µg/m3)0 0.000185 2.1810 0.000490 5.8820 0.001201 13.224 0.001691 18.328 0.002359 25.230 0.002777 29.532 0.003261 34.436 0.004471 46.640 0.006079 62.6

2.3.Fluorescentlamps

OneoftheapplicationsofREEsistheproductionofphosphors–substancescapableofluminescence(Cuifetal.,2005).Phosphorsareused inproductssuchascathoderay tubes (CRTs), lightemittingdiode(LED)lamps,plasmadisplaypanels,fieldemissiondisplays(largeformatandbillboarddisplays),fluorescentlamps(generalservicefluorescentlamps,compactfluorescentlamps(CFLs),coldcathodefluorescent lamps),andelectronicproducts thatuseLEDor fluorescent lampbacklight illumination(flatscreentelevisions,computerscreens,etc.).

In recent decades, fluorescent lamps have experienced increased usage due to the advantagescompared to incandescent bulbs, political and marketing campaigns, and energy-relatedenvironmental considerations. Fluorescent lamps consume about 75% less energy to produce thesame light output as an incandescent bulb and have up to ten times longer life expectancy (USDepartmentofEnergy,2010).

AschematicrepresentationofatypicalfluorescentlampandtheprincipleofoperationarepresentedinFigure1.Aglasstube,whichcanhavedifferentshapesandsizes,isfilledwithaninertgase.g.argon.Theinsideoftheglassiscoatedwithathinlayerofphosphorspowder.Mercuryisavitalcomponentinallfluorescentlampsanditisaddedinelementalformduringmanufacturing.Invisibleultraviolet(UV)photonsaregeneratedwhenmercuryatomsinteractwithelectronsemittedbythecathode.ThephosphorsabsorbtheUVenergy,re-emittingitasvisiblelight.

Figure1–Schematicrepresentationandoperationprincipleofatypicalfluorescentlamp.

10

Modern fluorescent lamps use variousmixes of red, green and blue REE-containing phosphors toproducethedesiredlightoutput(Table4)(Cuifetal.,2005;Rondaetal.,1998;Songetal.,2013;USDepartmentofEnergy,2009).Theselampsareknownastri-phosphorsfluorescentlampsortri-bandlamps. Europiumandyttriumaccount for thevastmajorityof the contentofREEs in lamps,beingpresenttogetherintheredphosphor.Europiumisalsousedinbluephosphors.Allgreenphosphorscontainterbium,alongsideotherREEse.g.cerium,gadoliniumandlanthanum.

Table4–SomeoftheREE-basedphosphorsusedinfluorescentlamps(Cuifetal.,2005;Rondaetal.,1998;Songetal.,2013;USDepartmentofEnergy,2009).

Phosphortype Possiblecompounds Typicalamountintri-bandlampsRed Y2O3:Eu3+(YOX)* 55%(containsmostlyyttrium;4–10%europium)

Blue BaMgAl10O17:Eu2+(BAM)

(Sr,Ca,Ba)5(PO4)3Cl:Eu2+ 10%(contains<5%europium)

Green

CeMgAl10O17:Tb3+LaPO4:Ce3+,Tb3+(LAP)(Ce,Tb)MgAl11O19(CAT)(Ce,Gd,Tb)MgB5O10

35%(containsapprox.10%terbium)

*Thecolonsymbolspecifiesdopedcompounds,inthiscaseyttriumoxidedopedwitheuropium.

Typically,thephosphorspowderaccountsfor2%wt.ofatypical40Wfluorescentlamp,whichisintherangeof4-6g(Raposoetal.,2003).Rareearths-freephosphorsaresometimesused,especiallyincheaperlamps,andareregardedasimpuritiesfromarecyclingpointofview(DupontandBinnemans,2015).Halophosphorsfallintothiscategorye.g.(Sr,Ca)10(PO4)6(Cl,F)2:Sb3+,Mn2+.

Thetypeoflamp,themanufacturer,andtheyearofmanufactureinfluencetheamountsofmercuryinfluorescentlamps.Theaveragemercurycontentinfluorescentlampsdecreasedfrom48.2mgin1985to8.3mgin2001(NEMA,2005).AtthebeginningofthiscenturytheEuropeanUnionsetalimitof5mgmercury/CFL(TheEuropeanParlimentandtheCounciloftheEuropeanUnion,2003).Despitethis,lampscontainingmercuryabovethislimitarestillfoundonthemarket.Mercurycontentsintherangeof1.6-27mg/lampwerequantifiedin15differenttypesofCFLs,with40%oftheinvestigatedsamples failing to pass the aforementioned requirement (dos Santos et al., 2010). Unfortunately,mercuryisrequiredinfluorescentlightinganditsreplacementwithasubstitutewithsimilarpropertiesisnotenergy-feasible.Suchalampwouldconsumethreetimesmoreenergytogenerateasimilarlightoutputasamercury-basedone(NEMA,2005).

Duringlampoperationmercuryisadsorbedmainlyinthephosphorpowderandinsmalleramountsintheglass,endcapsandelectrodes(Dangetal.,1999;Doughtyetal.,1995;Rey-RaapandGallardo,2012;Thaleretal.,1995).Theconcentrationofmercuryinthephosphorspowderofend-of-lifelampscaninsomecasesbe40-foldhighercomparedtothepowderofnewlampsofthesametype(Raposoetal.,2003).Ithasbeenreportedthat85%ofthemercuryinCFLsbindstothephosphorpowderandabout13%totheglass(Rey-RaapandGallardo,2012).Elementalmercury(Hg0),Hg+andHg2+speciesweremeasuredinfluorescentlampwaste,withthefirsttwospeciesfoundtobepredominantoverdivalentmercury(Raposoetal.,2003).

11

2.4.Industrialprocessingoffluorescentlamps

Thepresenceofmercuryclassifiesend-of-lifefluorescentlampsashazardouswaste.Specialcollectionandtreatmenttoreducethemercurycontentofthewasteunderadefinedthresholdisrequiredunderthelegislationofmanycountries.

An initial processing step to separate various macro-fractions e.g. glass, aluminium end caps,phosphors,plastics,electronicsetc.,andtoaddressmercurycontaminationistypicallycarriedoutbylampprocessers(Binnemansetal.,2013).Linearfluorescentlampscanbeprocessedusingtheend-cutmethod,inwhichtheendsoftheglasstubesaremechanicallycutandthephosphorspowderisblownout.Theclearadvantageistherecoveryofrelativelycleanphosphorsfractions,butthismethodisnotabletohandlealargenumberoflampsinagiventimeanditisnotsuitedforlampswithmorecomplexconstructione.g.CFLs.

Anothercommonpractice,whichovercomesthesedisadvantages,iscrushingandshredding,followedbyseparationofmacro-fractionsusingtechniquessuchasdry/wetsievingorEddycurrentseparation.Inthiscase,separationofcleanphosphorsisimpossibletoachieveduetoveryfineglassparticlesbeinggenerated during crushing. Glass can amount up to 50%wt. of the collected phosphors fractions(Binnemansetal.,2013).Othershreddedlampcomponentse.g.plastics,electrodefragmentsetc.canalsocontaminatethecollectedphosphors.

Thermaltreatment(distillation),chemicalleachingandstabilizationarethethreeoptionsforhandlingmercuryinfluorescentlamps(Duraoetal.,2008).Distillationmakesuseoftherapidvolatilizationofmercury with increasing temperature (Table 3) and decomposition of mercury compounds toelementalmercuryatelevatedtemperatures.Hightemperatures(800°C)areneededforcompletedecontamination,however,duetothestrongadsorptionofmercuryintheglassmatrix(Raposoetal.,2003).Theuseofreducingagents,e.g.sodiumborohydrideandcitricacid,canlowerthetemperaturesneededformercurydesorptionbyupto30%(Duraoetal.,2008)but,despitethis,distillationisanenergy-intensive process. Moreover, special equipment is needed for this, e.g. perfectly sealeddistillation units that operate under negative pressure in order to prevent the escape ofmercuryvapours (Changetal.,2009).Themethod isnotbestapplicable forwasteobtainedviawet-sievingprocessesduetothehighamountsofmoisturepresentinsuchstreams(Binnemansetal.,2013).Analternativeishydrometallurgicalprocessing,whichinvolvesthechemicalleachingofmercuryfromthewaste, followed by its further recovery from solution using selective adsorbents, electrochemicalmethodsorsolventextraction.Strongoxidizingsolutions,e.g.hypochloriteandnitricacid,areefficientforleachingmercuryfromthephosphorspowder(Sobraletal.,2004).

ChemicalstabilizationcanbeusedtocontainelementalmercurybyoxidationtoyieldmorestableandlessvolatileHg+andHg2+species.Althoughmercuryisnotcompletelyremovedfromthestream,themethodallowsforsaferstoringandprocessingoffractions.Removalofmercuryfromthephosphorspowder fractions is indicated for subsequent recoveryof REEs, however, to prevent the spreadofmercuryinvariousprocesssteps,e.g.leaching,solventextraction,striping,refiningetc.Stabilizationviawetprocessescanaffectthefurthertreatmentoffractions.Distillationofmercurybecomesmorecomplicatedbecauseofthemoisturepresentinthematerial.Additionalenergyinputwillberequiredto evaporate water. Also, higher temperatures are needed for the desorption of the mercurycompounds inhigheroxidationstates that formduringchemicalstabilization (Raposoetal.,2003).Becauseof thesedrawbacks, finding selectivehydrometallurgicalprocesses formercury removal is

12

important as an alternative to the thermal treatment of wet, stabilized, fluorescent lamp wastephosphorsfractions.

2.5.Hydrometallurgicalunitoperations

When it comes to recoveryand separationof individualREEs,hydrometallurgicalprocessing is themethod of choice. Low grade and chemically complex streams can effectively be processed usinghydrometallurgy,leadingtohighpurityproducts.Applicationsofpyrometallurgyintheprocessingofend-of-lifeproductscontainingREEstypicallyleadstotheaccumulationoftheREEsintheslagphasesinlowconcentration,makingtheireconomicrecoverymoredifficult(Binnemansetal.,2013).

Hydrometallurgicalrecoveryofmetalsfromsolidstreamsrequirestwomainsteps:

• Leaching,inwhichthesolublefractioncontainedinasolidphaseisremovedasasolution.Thisstep dissolves the metals of interest and, depending on conditions, other undesiredconstituentspresentinthematerial;

• Separationofthemetalsofinterestfromeachotherand/orfromundesiredelementspresentinsolutionusinge.g.solventextraction,ionexchangeand/orprecipitation.

Additionalprocessingsteps,e.g.pre-concentrationofmetalionsinsolution,refiningandpurificationof separated feeds, complement these two main steps. A simplified representation of a typicalhydrometallurgicalprocessfortheseparationofREEsispresentedinFigure2(Tunsuetal.,2015).

Figure2–Asimplifiedflowsheetbasedonpre-treatment,leaching,solventextraction,ionexchangeandprecipitationfortheseparationofREEsfromsolidstreams(Tunsuetal.,2015).

13

2.5.1.Leaching

Water,acid,baseorsaltsolutionsareusedforleaching,dependingonthesolubilityandthechemicalforminwhichthemetalsofinterestarepresentinthestream.Theefficiencyoftheprocessisaffectedby leaching agent concentration, temperature, contact time, stirring, solid-to-liquid ratio and pre-treatmentmethods,e.g.crushingandgrinding.Thedistributionofsolublecompoundsinthematerial,possible adsorption in the matrix, entrapment of soluble compounds in less soluble or insolubleparticles, and slow reactions between the soluble constituents and the leaching agent affect theleachingkinetics(Rao,2006).Fromaneconomicalpointofview,fastkineticsaredesired,butslowerkineticscansometimesbebeneficial,e.g.forselectiveleaching.

Thechoiceofleachingagentishighlyimportantasitinfluencesnotonlytheleachingprocessbutalsothefurtherrecoveryofmetalsfromsolution.Inasolventextractionprocessthenatureoftheaqueousphase (the leachate)andthepresenceofcertain ions in thesystemcan limit thechoiceoforganicsolventsandaffecttheseparationefficiency.

2.5.2.Solventextraction

Ina solventextractionprocess thecompoundsof interestarepartitionedbetween two immiscibleliquidphases;anaqueousandanorganicphase(CoxandRydberg,2004).AschematicrepresentationofasolventextractionprocessforREEsispresentedinFigure3.

Figure 3 – Schematic representation of a solvent extraction process for REEs. Themetals, initiallypresentas soluble ions inanaqueousphase,arepartitionedbetween theaforementionedaqueousphase and an organic phase containing one ormore specific extractants dissolved in an adequatediluent.

Thedistributionratio(D),definedastheratiobetweenthetotalconcentrationofanelementintheorganicphaseandthetotalconcentrationofthesameelementintheaqueousphase,characterizestheextractionprocess(Equation1).

14

𝐷" =[𝐴]𝑜𝑟𝑔[𝐴]𝑎𝑞

(1)

The distribution ratio can be used to calculate the percentage extracted (Equation 2); and theseparation factor,whichcharacterizes thedegreetowhichtwosolutes,AandB, canbeseparated(Equation3).

%𝐸 = 100∗𝐷𝑉𝑎𝑞𝑉𝑜𝑟𝑔

+𝐷 (2),

whereVorgandVaqarethevolumesoforganicandaqueousphases,respectively.

𝛼"/5 =6768

(3)

Theextractedmetalionsaresubsequentlyrecoveredbycontactingtheorganicphaseafterextractionwithacleanaqueoussolution,calledastrippingagent.Theelementsofinterestwillpartitionbetweenthesetwophases;aprocesscalledstrippingorback-extraction.Scrubbing(removalofunwantedco-extractedspecies) iscarriedout inasimilarmanner.AcommonpracticetorecovertheREEions instripproductsisprecipitationwithoxalicacid,followedbythermaltreatmentoftheobtainedoxalatestoproduceoxides.

Fivemainextractionsystemshavebeendescribed(Cox,2004):

• extractionofalmostpurelycovalentinorganiccompoundsbynon-solvatingsolvents.• extraction of neutral coordinatively saturated complexes formed betweenmetal ions and

lipophilicorganicacids(chelateextraction).• extractionusingsolvatingextractants,whichreplacethewaterofhydrationinthecomplexes

formedbetweenthemetalionsandthecounter-ionspresentintheaqueousphase(solvateextraction).

• ionpairextraction(anion/cationexchange).• extractionofmetalcomplexesthatdonotfitintoanyofthepreviouscategories(e.g.clathrate

compounds).

FortheextractionofREEs,chelatingandsolvatingextractantsplayanimportantrole.

Thechoiceofdiluenthasanimportantroleintheextractionprocess,affectingthedistributionratiosand separation factors between metal ions extracted with different, or the same extractionmechanism.Theextractionofacomplexisafunctionoftwosteps(LöfströmEngdahletal.,2010):

• the collapsingof the cavity containing the complex in the aqueousphase, followedby thetransferfromtheaqueousphase;

• breaking the bonds between the water molecules that solvate the complex, followed bycreationofacavityintheorganicphaseanddissolvingofthecomplex.

15

Thediluentinfluencesnotonlytheattractiveenergiesbetweentheextractedspeciesandtheorganicphase,butalsotheenergyrequiredforcavityformation.

2.5.2.1.Cyanex923extractant

Cyanex923isacommerciallyavailablesolvating-typeextractant.Ithasanaveragemolecularweightof 348 g/mol and it is amixture of four trialkylphosphine oxides (TRPO) (Table 5). (Dziwinski andSzymanowski,1998).

Table5–ChemicalcompositionofCyanex923(DziwinskiandSzymanowski,1998).

Compound Generalformula AmountTrioctylphosphineoxide R3PO* 14%Dioctylmonohexylphosphineoxide R2R'PO** 42%Dihexylmonooctylphosphineoxide RR'2PO 31%Trihexylphosphineoxide R3'PO 8%

*Rdenotesn-octyl:CH3–(CH2)7–**R'denotesn-hexyl:CH3–(CH2)5–

TheprinciplefortheextractionofametalionwithasolvatingextractantisdescribedbyEquation4(Cox,2004).

Mz+(aq)+zX-(aq)+bB(org)↔MXzBb(org) (4)

The metals in the aqueous phase (Mz+) form complexes with the counter-ions present (X-). Theremainingcoordinationsitesareoccupiedbywater.Duetotheirwatercontent,thesecomplexeshaveverylowdistributioncoefficients.Inthepresenceofasolvatingextractant,thewatermoleculesarereplacedbytheorganicextractantmolecules.Morelipophiliccomplexesareformed,enhancingtheirextractionintotheorganicphase.

ThreeTRPOmoleculeswerefoundtobindtolanthanidesandyttriuminnitricacidmedia(Reddyetal.,1998)(Equation5).ChloridecomplexesofmercuryarealsoextractedbyCyanex923(Meeraetal.,2001)(Equation6).

M3+(aq)+3NO3-(aq)+3TRPO(org)↔M(NO3)3TRPO3(org) (5)

Hg2+(aq)+2Cl-(aq)+3TRPO(org)↔HgCl2TRPO3(org) (6)

16

2.5.2.2.Cyanex572extractant

Cyanex572isanovelphosphorus-basedchelatingextractanttailoredforindividualseparationofREEs.Itsexactcompositionhasnotbeendisclosedbutit ismentionedtobeamixtureofphosphinicandphosphonicacids(Cytec,2014).TheseparationfactorsbetweenadjacentelementsaresimilartothoseofphosphonicacidscommonlyusedinREEseparationse.g.EHEHPA,butextractionoccursathigherpH and stripping requires less concentrated acids. The extraction mechanism, as stated in theextractantsheet,ispresentedinEquation7.

REE3+(aq)+3HA(org)↔REEA3(org)+3H+(aq) (7)

Usually,themechanismismorecomplex,withthedimerformoftheextractantbeinginvolvedintheextraction(Equation8;Xieetal.,2014).

REE3+(aq)+3(HA)2(org)↔REE(HA2)3(org)+3H+(aq) (8),

where(HA)2isthedimerformoftheextractant.

Sinceprotons are involved in the equilibrium reaction, thedistributionofmetal ions between theaqueousandtheorganicphasewillbepH-dependent.Equations7and8showthatextractionofREEsisfavouredbyincreasesinpHintheaqueousphase.Theoppositeprocess(stripping)isfavouredbydecreases inpH,thusan increasedacidityoftheaqueousphase.SeparationofREEs isachievedbycontrollingtheequilibriumpH(pHeq)duringextraction.

2.6.HydrometallurgicalrecoveryofREEsfromfluorescentlamps

AnumberofscientificstudiesdirectedtowardstherecoveryofREEsfromfluorescentlampshavebeenpublished and are presented below. Some of these address certain steps of hydrometallurgicalprocessinge.g. leachingor testingof extractants,whileothers focuson combined leaching and/orsolventextractiontorecoverREEs.Whilesomestudiesuserealphosphorsfractionscollectedfromend-of-life lamps, part of the experimental work was also performed using pure commercialphosphors,carefullycleaned lampsthatarebrokenundercontrolledconditions,andevenartificialmixtures intended to simulatephosphorspowder.Most studiesdonotusemercury-contaminatedmaterialasfeed.

Dueto thechemicalstructureof theREEphosphorsused in fluorescent lighting (Table4),efficientleachingrequirestheuseofacidicsolutionstobringtheREEsintosolution.ThechemicalprocessesforthiscanbewrittenaccordingtoEquations9-11.

17

Redphosphor:Y2O3:Eu3++6H+→2Y3++Eu3++3H2O (9)

Bluephosphor:BaMgAl10O17∶Eu2++34H+→Ba2++Mg2++10Al3++Eu2++17H2O (10)

Greenphosphor:CeMgAl10O17∶Tb3++34H+→Ce2++Mg2++10Al3++Tb3++17H2O (11)

Acidswillalsoleachotherimpuritymetalsinthestream,notablycalcium(Equation12).

(Sr,Ca)10(PO4)6(Cl,F)2:Sb3+,Mn2++18H+→10(Sr,Ca)2++6H3PO4+2(Cl,F)-+Sb3++Mn2+ (12)

Separation of REEs, either as a group or individually, is carried out using solvent extraction.PrecipitationwithoxalicacidisalsousedtorecoverREEs,eitherdirectlyfromleachatesorfromthestrippingproductsof solventextraction (Equation13).The latterapproach leads tohigherproductpurityduetoco-precipitationofimpuritymetalswiththeREEswhentheleachateisdirectlysubjectedtooxalicacidtreatment.

2REE3+(aq)+3C2O42-(aq)→REE2(C2O4)3(s) (13)

Efficient leachingofallREEs in lampphosphors isanenergy-andreagent-intensiveprocess,ashasbeenreportedbyZhangetal.(2013).Notably,thesolubilityoftheLAP,CATandBAMphosphors(Table4)islowinacidsatambientconditions.Aconversionstep,similartotheoneusedforleachingofREEsfrommonaziteore,canbeemployed.Alkalifusionoftheselow-solubleREEphosphorscanbecarriedoutwithsodiumhydroxideat800°C.Thiswill leadtomoresolubleeuropium,ceriumandterbiumoxides,accordingtoEquations14and15.

Ce0.67Tb0.33MgAl11O19+NaOH+O2→NaAlO2+MgO+CeO2+Tb4O7+H2O (14)

BaMgAl10O17:Eu2++NaOH+O2+CO2→NaAlO2+BaCO3+MgO+Eu2O3+H2O (15)

Amulti-stepleachingprocessforfluorescentlampwastehasbeendescribedbyOttoandWojtalewicz-Kasprzak (2011). Dissolution of the halophosphors can be donewith hydrochloric acid at ambienttemperatures (<30 °C). Hydrochloric acid at 60-90 °C is then used to dissolve easily soluble REEphosphorssuchasYOX.Sulphuricacidat120-230°CisusedtoleachtheREEphosphates.Alkalifusionwith hydroxides, followed by acid leaching is finally carried out to dissolve the remaining REEcompounds.

Sulphuricacidwassuggestedtoreducetheamountsofcalcium,bariumandleadthatleachtogetherwiththeREEsfromfluorescentlampwaste(DeMichelisetal.,2011).Thiswillforminsolublecalcium,bariumandleadsulphatesbutithasthedisadvantageofco-precipitationofREEs.Directprecipitationofyttriumwascarriedoutfromthe leachateusingoxalicacid.Yttriumoxalateswithpurities intherange90-95%wereobtained.Reducingtheamountofimpuritymetals(notablyzincandcalcium)insulphuric acid based leachates of lamp phosphors and mixtures of lamp-CRT phosphors was

18

investigatedwithsodiumsulphide,afterapHadjustmentstepwithsodiumhydroxideintherange2-3(Innocenzietal.,2013).SignificantlossesofyttriumwerereportedduringthepHadjustmentstep(78%forlampphosphorsleachateand30%forthemixedphosphorsleachate)duetotheformationofamorphouscompoundsofcalcium,siliconandyttrium.Yttriumlosseswerealsoreportedforthesodiumsulphidetreatmentstep,withamaximumof20%yttriumbeingco-precipitatedinthelampleachate.

SolventextractionwasappliedforseparationofpurecommercialYOX,BAMandCATphosphorsinanaqueousphaseconsistingofpotassiumsodiumtartrateandsodiumcarbonate(Meietal.,2009).Theextractant2-thenoyltrifluoroacetonedissolvedinheptaneextractedBAMatalkalinepH.Thisisduetothechelatingadsorptionof2-thenoyltrifluoroacetonewiththeEu2+ ions,whichmakestheparticleshydrophobic. CAT is extracted with chloroform containing added 1-pentanol, due to homo-aggregation.

AcomparisonbetweentheextractionofREEs in lampphosphorsusingtri-n-butylphosphate (TBP)complexes with nitric acid and water at atmospheric pressure and under supercritical conditions(carbondioxide)hasbeencarriedout(Shimizuetal.,2005).Testingatatmosphericpressure ledto37.4% extraction efficiency for yttrium, 36.8% for europium and <3% for lanthanum, cerium andterbium.Under supercritical conditions (15MPa,333K) theextractionefficiencies for yttriumandeuropiumwere>99%whilethoseforlanthanum,ceriumandterbiumwere<7%.

FurtherinvestigationsontheextractionofoxidesofREEsusedinfluorescentlampsusingsupercriticalcarbondioxidemodifiedwithTBP-nitricacidadductswerecarriedout(Baeketal.,2016).Differenttypes of adductswere prepared and characterized, and the phase-equilibrium behaviour for eachadductinsupercriticalcarbondioxidewasstudied.Thewater,acidandTBPcontentsoftheadductsisvery important for theextractionofREEs.More acidic adducts lead to excess acid andwater thatcondensates,loweringtheextractionefficiencyofREEs.BestadductsarepreparedfromdehydratednitricacidandTBP,ensuringminimumcontentofexcesswaterandonlyenoughacidtofacilitatetheformationofextractableREEnitratecomplexes.HighestextractionefficienciesofselectedREEswereachievedusingaTBP(HNO3)1.7(H2O)0.6adductat34.5MPaand338K:>99%foryttriumandeuropium,92.1%forterbiumandonly0.12%forcerium.

Separationofeuropium,terbiumandyttriumfromnitricacidmediawith2-ethylhexylphosphonicacidmono-2-ethylhexyl ester (HEHEHP) in kerosene for applications in fluorescent lampprocessing hasbeen studied (Nakamura et al., 2007). Simulations in a counter-current mixer-settler cascade (15extractionstagestoseparateyttriumfromeuropiumandterbium,followedby10extractionstagestoseparateeuropiumfromterbium)gaverecoveriesof97.8%foryttrium,58.1%forterbiumand52.8%foreuropium.Thepuritiesinthiscasewere98.1%,85.7%and100%,respectively.

Conversionofeuropiumandyttriuminsulphuricacidleachatesoflampphosphorstothiocyanates,followedbyextractionwithtrimethyl-benzylammoniumchloridehasbeendescribed(Rabah,2008).ThetwoREEswereextractedwith>96%efficiency,followedbystrippingwith1Mnitricacid,selectiveseparationofthenitratesinethanol,andhydrogenthermalreductionat850°Cand1575°Ctoobtainmetalliceuropiumandyttrium,respectively.TheseparationfactorbetweenthetwoREEswas9.4.

In recent years, ionic liquid extraction has received increased attention. These compounds areregardedasalternativestoconventionalsolventsduetonegligiblevapourpressure,flameresistanceandtheabsenceintheircompositionofelementssuchasphosphorous(CevascoandChiappe,2014;Yangetal.,2013).

19

SelectiveandquantitativerecoveryofREEsfromnitricandsulphuricacidleachatesoflampphosphorswas achieved at pH 3 using N,N-dioctyldiglycol amic acid in the ionic liquid 1-butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide (Yang et al., 2013). The system showeddrawbacksduringstrippingwithsulphuricacid,with1.4%oftheionic liquidbeinglost.Adeclineinextractionefficiencyafterfiveextractioncycleswasalsonoted.

Applicationsoftheionicliquidbetainiumbis(trifluoromethylsulfonyl)imideonsimulatedfluorescentlampphosphorspowdershowedthatitispossibletoseparateYOXfromBAM,LAPandhalophosphors(DupontandBinnemans,2015).YOXisdissolvedintotheionicliquid,fromwhereitcanberecoveredbyprecipitationwithoxalicacid.

Bifunctional ionic liquid extractants prepared from Aliquat 336, bis-2-ethylhexyl phosphoric acid(HDEHP)andHEHEHPwereappliedfortherecoveryofREEsfromsimulatedtri-bandphosphorpowder(Yangetal.,2012).TheirefficiencywasbetterthanHDEHPandTBPbutlowerCyanex923.

20

21

3.EXPERIMENTAL

3.1.Overviewoftheexperimentalwork

TheexperimentalworkcarriedoutispresentedinFigure4andisbasedonthefivepublicationsstatedinthePublicationlistatthebeginningofthethesis.

Figure4–Asimplifiedoverviewoftheexperimentalworkpresentedinthisthesis.

Theinvestigationsweregroupedintosevenblocks:

• Characterizationofthewastefractions(PublicationsIandIV).• Hydrometallurgicalremovalofmercury(PublicationsI,IIIandIV).• BenchscaleleachingofREEsandothermetalsinthesamples(PublicationsI,IIIandIV).

22

• GroupbatchextractionofREEswithCyanex923(PublicationsIIandIV).• Testingoftheleachingandseparationprocessesatlaboratorypilotscale(PublicationIV).• FurtherseparationofindividualREEswithCyanex572(PublicationV).• PreparationofREEoxidesusingoxalic acidprecipitation followedby thermal treatmentof

obtainedoxalates(PublicationsIVandV).

Detailsabouttheexperimentalparametersaregivenineachrespectivepublicationandineachfigureandtablecaptioninthisthesis.Ashortsummaryoftheworkdoneineachblockispresentedinthefollowingsubsections.

3.2.Generalinformationaboutthefluorescentlampwastefractionsused

Variousrealfluorescentlampwastefractionsthatcontainphosphorswereusedintheexperiments.DetailsregardingthetypeandoriginoftheinvestigatedfractionsaregiveninPublicationsIandIVandaresummarizedinTable6.

Table6–Typesoffluorescentlampwastefractionsinvestigated.

Sampleorigin(typeofprocess)

Additionaldetailsabouttheprocessgeneratingthesample

SamplecharacteristicsPublicationno.investigatingthesample*

Samplenameusedinthepublication

Samplenameusedinthisthesis

Wetprocess

Crushingoflargebatchesofvarioustypesoflampsundercontrolledconditions,followedby(partial)separationofmacro-fractionse.g.metals,glassetc.Mercuryisstabilizedusinghypochloriteoxidation.

Wetsludgethatismercury-contaminated.Containsglassasmajorimpurity,fibresandothernon-solublefractions.Variousfractionsofthistype,withdifferentREEs,glassandmoisturecontentwereinvestigated.

I - IV,V

II - IV

III - IV

IV III III

Wetprocess

Water-basedprocessfortreatmentofend-of-lifefluorescentlamps.Theresultingphosphorsfractionwasthermallytreatedtoremovemercury.

Dry,finepowderthatdoesnotcontainmercury.Containsapprox.50%wt.glassandothernon-solublefractions.

IV II II

Dryprocess

Nodetailsprovidedbythesupplier.

Mixofphosphorsfromnewlampsandproductionwaste.Dry,finepowderthatismercury-contaminated.Containsapprox.40%wt.glassandothernon-solublefractions.

IV I I

*accordingtothepublicationlistpresentedatthebeginningofthethesis.

23

Emphasiswasplacedonstudyingwastesamplesoriginatingfromwetprocesses,inwhichmercurywasstabilizedusingoxidativereagentsandisstillpresentinthestreams.Thisisbecauseoftheprocessingchallengesassociatedwiththistypeofwaste,aspreviouslymentionedinsection2.4.Dryingofthematerialwascarriedoutatambientconditions inorder todetermine themoisturecontentand tofacilitate parts of the investigations. Homogenization of the samples was performed by thoroughmixing.SomeoftheinvestigatedsamplesarepresentedinFigure5.

Figure5–Samplesusedinthepresentstudy.AdditionaldetailsforeachsampleareprovidedinTable6.

3.3.Chemicalsandtechniques

The inorganic reagentsusedwereof analytical gradepurityorhigher.All inorganic solutionswereprepared using pure water (MilliQ, Millipore, >18 MΩ/cm). Pure water was also used in someexperimentsasleachingagent,strippingagentandtowashresiduesafterleaching.Theinvestigationswerecarriedoutatambientconditions(usuallyapprox.21°C),withtheexceptionoftheexperimentsthataddressedtheinfluenceoftemperature.Allorganicsolvents(extractantsanddiluents)wereusedassuppliedbythemanufacturer,withnoadditionalpurification.Dilutionsofthe inorganicaliquotscollectedforanalysisweremadeusing0.5-1Mnitricacidsolution,preparedfromconcentratedstocksolution(65%,Suprapur,Merck)andpurewater.Whereneeded,solutionswerefilteredusingeitherpolypropylenesyringefilters(0.45μm,VWR)andsyringes,orglassfibrefilters.

Phasemixingforsolventextractionbatchexperimentsusingvolumes<10mLwasperformedinglassvialssecuredwithlids.Thevialswereplacedinaclosedthermostaticsampleholderconnectedtoacooling/heating water bath and fixed to a shaking machine (IKA Vibrax VXR Basic) set at >1500vibrationsperminute(vpm).Forlargervolumes,phasemixingwasperformedinseparationfunnelsusingmanualshaking.Kineticexperimentswereinitiallycarriedouttoestimatethephasecontacttimerequired for equilibrium. Generally, after phase mixing, the samples were centrifuged to ensurecompletephaseseparation.

Most investigationsweremadeintriplicatetests inordertoaccountforuncertaintiesandmaterialinhomogeneity.

24

InductivelyCoupledPlasma-OpticalEmissionSpectroscopy(ICP-OES)(iCAP6500,ThermoFischer)andInductivelyCoupledPlasma-MassSpectrometry(ICP-MS)(iCAPQ,ThermoFischer)wereusedtoobtainqualitativeandquantitativeelementalinformation.

3.4.Characterizationofwastefractions

ScanningElectronMicroscopy/EnergyDispersiveSpectroscopy (SEM/EDS) (FEIQuanta200F/OxfordInca300EDSSystem)wasusedtomagnifythesamplesinvestigated,toobtainqualitativedataandtodetermine the occurrence and distribution of the components. X-Ray Diffraction (XRD) (SiemensDiffraktometer D5000) was used to obtain structural information about the major components.AdditionalinformationaboutthesetechniquesisprovidedinAppendix1.

To determine the solublemetals content in thewaste, aswell as in the residues collected duringleaching,dissolutionofthematerialinaquaregia(3:1mixtureofconcentratedhydrochloricandnitricacid)atelevatedtemperaturewascarriedout, followedbyanalysisoftheobtainedsolutionsusingICP-OES.Aquaregiadissolutionwasdeterminedtobeamongthebestmethodsforelementalanalysisof fluorescent lamp shredderwaste (Hobohm et al., 2016). Thismethod showed highest leachingvalues for REEs among different reagent mixtures, including nitric/perchloric/hydrofluoric acidmixture,variousmixturesofnitricandhydrochloricacids,sulphuricandhydrochloricacids,nitricacidandhydrogenperoxide,nitricacidonly,andhydrochloricacidonly.TheonlyREEforwhichdissolutionwasfavouredbyanotherleachingsystemwasterbium,whichwasleachedinslightlyhigheramounts(+14%)byasulphuric:hydrochloricacidmixture(5:1.5volumetricratio).However,atthesametime,this latter system led to poor dissolution of elements such asmercury, lead, iron, aluminium andothers.Theformationofinsolublesulphatesaccountsforsomeofthesepoorleachingresults.

3.5.Hydrometallurgicalremovalofmercury

Water, ammoniumchloride, nitric, hydrochloric and acetic acid (Publication I), aswell as iodine inpotassiumiodide(I2/KI)solutions(PublicationsIIIandIV)havebeeninvestigatedaspossibleleachingagentsformercury.Theinfluenceofparameterssuchasleachingtime,leachingagentconcentration,solid to liquid (S/L) ratioand temperatureon the leachingefficiencyofmercuryhasbeen studied,notablyforthenitricacidandI2/KIsystems.Sincethelattershowedselectivityformercury,possiblerecoveryroutesformercuryinI2/KIlampwasteleachateswereinvestigated:reductionwithsodiumhydrosulphite,ionexchangewithDowex1X8resin,andsolventextractionwithCyanex923,bis(2,4,4-trimethyl-pentyl) monothiophosphinic acid (Cyanex 302) and 6,6’-bis(5,5,8,8-tetramethyl-5,6,7,8-tetrahydro-benzo-[1,2,4-]triazin-3-yl)-[2,20]bipyridine(CyMe4BTBP).

3.6.Leachingofmetals

Theleachingbehaviourofmetalsinthewastewasstudiedusingvarioussolutions:water,ammoniumchloride,nitric,hydrochloricandaceticacid(PublicationsIandIV),andI2/KI(PublicationIII).Variousparameterswereinvestigated,theemphasisbeingplacedonthenitricacidsysteme.g.leachingtime(upto96h),leachingagentconcentration(0.5-4M),S/Lratio(10-30%w/v),temperature(20°C,60°C)andultrasound.ThefocuswasthesixREEspresentinthesamplesbutattentionwasalsopaidto

25

leachingofimpuritymetals,withthegoalofdevelopingaselectiveleachingprocess(PublicationIV).After leachingofmercurywithI2/KIsolution,theresultingresiduewaswashedandfurther leachedwith1Mnitricacid solutionatambient conditionsusing short contact times (0-60min).A10minleachingtimewasdeemedsufficienttoremovealargeamountofimpuritymetals,notablycalcium.Subsequentleaching,carriedoutwith2Mnitricacidatambientconditions,wasusedforupto30htostudythedissolutionofREEs.Mostoftheyttriumandeuropiuminthewastewasdissolvedwithin24h,alongsidetracesofcerium,lanthanum,gadoliniumandterbium.ThebulkoftheselastfourREEswasleftinthefinalresidue,togetherwithotherlowsoluble/insolublecomponentse.g.aluminaandglass.

3.7.SolventextractionofREEswithCyanex923

Batchextractionofmetal ions innitricacid-basedleachatesoffluorescent lampwastewithCyanex923dissolvedinSolvent70,acommercialkerosenewithlowaromaticcontent,wasextensivelystudiedinPublicationII.Theinvestigationstargetedtheinfluenceofvariousparametersontheextractione.g.time(0-20min), ligandconcentration intheorganicphase(0.24-2.35M),acidconcentration intheaqueousphase(0.005-4.1M)andtemperature(23-60°C).Strippingofthemetalionsextractedwascarried out with nitric (0.1-4M), hydrochloric (0.1-4M), acetic (25% vol.) and oxalic acid (0.5M)solutions.

3.8.Testingoftheprocessinlaboratorypilotscale

The leachingandsolventextractionprocesswithCyanex923was further tested in laboratorypilotscaleusingbatchesof0.5kgwaste,asdescribedinPublicationIV.Mercury-contaminatedfractionsandfractionsthatwerethermallydecontaminatedweretested.Leachingwastestedin5Lleachingreactorswithpropellerstirringsetat400revolutionsperminute(rpm).Mercuryinthecontaminatedwastewasleachedwith0.025/0.075MI2/KIsolutionfor4.5h,usingaS/Lratioof25%w/v.Leachingofimpuritymetals,andthebulkofRREs,withnitricacidwasstudiedasaonestepprocess(22h),aswellasatwo-stepselectiveleachingprocess(10min,followedby22h),usingS/Lratiosof10%w/v.ExtractionofREEionsintheobtainedleachateswasdoneinacounter-currentmixersettler-systemcomprisedof threeextractionchambersand fourstrippingchambers (120mLeach;polyvinylidenedifluorideconstruction)(Figure6).

Additional information aboutmixer-settlers is provided in Appendix 2. The organic:aqueous (O:A)phaseratiousedforbothextractionandstrippingwas2:1.Theflowratesfortheaqueousandorganicfeedswere2.5and5mL/min,respectively,andthemixingspeedinthemixingchamberswas700rpm.Extractionwasstudiedwith35%and50%vol.Cyanex923inSolvent70.Strippingwasstudiedusing2Mand4Mhydrochloricacidsolutions.

26

Figure6–LaboratorypilotscalesetupfortheseparationofREEsfromfluorescentlampwastefractions.Thedashedlinesrepresenttheflowoforganicstreams.

3.9.FurtherseparationofREEswithCyanex572

Furtherseparationof theREEs in thestrippingproductscollectedafter thesolventextractionwithCyanex 923 inmixer settlers was carried out with Cyanex 572 (Publication V). The focus was theseparationofyttriumandeuropium,thebulkREEsinsolution.Theeffectsofvariousparameterswerestudied;phasecontacttime(0–30min),equilibriumpH(pHeq=-0.35–1.15),ligandconcentration(0.75-2 M), diluent (long chain alcohols, aliphatic and aromatic diluents) and temperature (25-65 °C).Strippingwasinvestigatedwithwaterandhydrochloricandnitricacidsolutions(0.1-6M).Separationofyttriumwascarriedoutonalargerscaleintensuccessivebatchextractionstageswith1MCyanex572inSolvent70atpHeq=0usingaO:Aphaseratioof1.EuropiumwassubsequentlyextractedinonestageatpHeq=1.Strippingoftheextractedmetalionsfromtheirrespectiveorganicphaseswascarriedoutinonestagewith3MhydrochloricacidusingaO:Aphaseratioof1.

3.10.PreparationofREEoxides

TheREEsinthevariousstripproductsaftersolventextractionwithCyanex923andCyanex572wereprecipitatedbyadditionofoxalicacid(PublicationsIVandV).Theprecipitateswerewashedwithpurewateranddriedatambienttemperature.Theoxalateswereheatedfortwohoursat800°Cinanoven,inairatmosphere.ThestructureoftheformedcompoundswasdeterminedusingXRD.Thepurityofthecompoundswasdeterminedbydissolutionsintoconcentratednitricacid(65%,Suprapur,Merck),followedbyanalysisoftheobtainedsolutionswithICP-OESandICP-MS.

27

4.RESULTSANDDISCUSSIONS

4.1.Characterizationofwastefractions

All samples contained a significant amount of glass and other non-soluble fractions that weregenerated during lamp crushing. Even for some of the fractions rich in REEs the content of suchimpuritiesreached40-50%drywt.Glassaccountedforthevastmajorityofthe impurities.Metallic(aluminiumendcaps,electroderemains)andplasticpartsandevenpackagingresidues(smallcartonpieces)wereobserved in coarser fractions like theone studied inPublication I. Theglassparticlesvariedgreatlyinsize,rangingfromover1000µm(incoarsefractions)tolessthan25µm.SievingofthemoreREE-concentratedfractionsthroughmeshesofdifferentsizeshowedthat thebulkof thematerial (up to 83 % dry wt.) consists of particles <25 µm. The phosphors particles are alsoconcentrated in this fraction, making isolation of the fluorescent powder inefficient using simplefractionationtechniques.

SEM/EDSimagingshowedthatthephosphorsparticlesvaryinsize.Theycanbesmallerthan2µmandhaveatendencytoformlargeraggregatesconsistingofmultiplesmallparticles(Figure7).

Figure7–SEMmagnificationofthreesamples(5000x),aspresentedinPublicationIV.Fromlefttoright: samples I, II and III, respectively (referenced in Table 6 in this thesis). Particle numbers areindicatedasfollows:1–calciumhalophosphate,2–LAP,3–YOX,4–CAT,and5–glass.

Foroneofthefractionsthathadbeenthermallytreatedtoremovemercury(sampleII),thepresenceofsuchaggregateswaslowercomparedtonon-heatedsamples.Also,thethermaltreatmentresultedin cracking of some phosphor particles. Not surprisingly, this was shown to affect the leachingbehaviour, leading to a slightly increased leaching efficiency of metals due to larger surface areacontactwiththesolution.Yttriumwasfoundtoco-existwitheuropiumindistinctparticlesasYOXbuteuropiumwas also found independent of yttrium.Halophosphor, LAP andCATparticleswere alsodetected.Mercurywasfoundspreadthroughoutthematerial,beingpresentinthephosphorsandonglass particles. This confirms its absorption onto lamp components during usage of the lamp, aphenomenondiscussedinseveralpublications(Dangetal.,1999;Doughtyetal.,1995;Raposoetal.,2003;Thaleretal.,1995).

ThesolublemetalcontentinthevariousfractionsinvestigatedispresentedinTable7.

28

Table7-Solublecontentforseveralmetalsinthedriedwastefractions,determinedusingaquaregiadigestionatelevatedtemperature,followedbymeasurementsoftheobtainedsolutionsusingICP-OES.CompileddatafromPublicationsIandIV.Theerrorrepresentsthestandarddeviationofthreesamplereplicates;nd=notdetected.

Element UnitSamplenumber(accordingtoTable6)andmetalcontentI II III IV* V**

REEs

Eug/kg

10.5±0.3 10.8±0.3 10.9±0.3 0.4±0.03 4.7±0.2Y 164.7±3.3 162.3±3.9 174.7±2.1 6.8±0.3 94.8±4.6Ce

mg/kg

700±28 678±44 1060±61 382±23 2015±145Gd 766±10 978±35 2043±68 244±17 1937±81La 773±46 609±68 412±64 339±25 1400±215Tb 427±16 416±22 661±32 158±10 904±44

Others

Al

g/kg

7.5±0.1 7.7±0.3 9.3±0.2 1.1±0.1 14.3±0.7Ba 4.7±0.2 4.5±0.1 7.1±0.6 2.3±0.2 10.5±0.4Ca 135.0±3.9 87.8±0.7 93.1±1.8 70.0±1.8 88.3±5.4Fe 4.5±0.1 7.2±0.1 6.0±0.3 3.3±0.4 5.2±0.4Mg 2.8 1.1 2.5 2.1±0.1 2.0±0.1Mn 4.4±0.2 3.3±0.1 3.5±0.1 0.8±0.1 3.1±0.2Na 1.7 2.3 2.4±0.1 Sb 1.6 1.2 1.1 Sr 1.8±0.1 3.1±0.1 4.8 B

mg/kg

466±10 1531±19 2462±71 Cd 253±9 40 207±5 40±2 274±16Cr 56±2 36±2 57±1 25±21 30±2Cu 70±3 220±4 359±9 513±61 548±38Hg 112±2 nd 227±5 724±70 367±26K 490±18 619±20 600±17 508±3 946±43Mo nd 61±1 37±1 148±7 263±19Ni 39±1 133±3 236±4 159±168 181±16Pb 137±5 310±7 548±13 344±91 743±41Si 445±17 268±8 134±6 Sn 11±1 43±1 203±3 83±20 227±18Ti 472±13 468±4 489±6 W 136±19 435±17 441±8 666±33 905±69Zn 409±15 193±12 1299±41 1234±114 1690±117Zr 7±1 20±1 20±1

*Coarsefractioncontaininglargeamountsofglassandotherimpurities.**FractionderivedfromsampleIV,afterpartialseparationofimpurities(sieving).

Yttriumwas thedominantREE in all samples, followedby europium. In theREE-rich fractions e.g.samplesI-III,thesetwoelementsaccountedforapprox.98%ofallREEsinthestreams.TheotherfourREEs,cerium, lanthanum,gadoliniumandterbium,werealsodetected.Amongthese,onlyterbiumandpossiblygadoliniumareregardedashavingpotentialrecoveryimportance.Thisisbecauseceriumandlanthanumaretypicallyobtainedinsufficientquantitiesduringtheprocessingoforescontainingheavy REEs and are not regarded as being critical (USEPA, 2012). The cost, energy demand andenvironmentalburdensofalkaliconversiontofacilitatetheeffectivedissolutionofcerium,lanthanum,gadoliniumandterbium,coupledwiththeirseparationandpurificationcosts,directedattentioninthis

29

studytowardstherecoveryofyttriumandeuropiumalone.Nonetheless,theleachingandextractionbehavioursoftheseotherfourREEswerealsostudiedalongsideyttriumandeuropiumandarealsoreported.

TheXRDanalysisconfirmedthepresenceofyttriumandeuropiumasyttriumeuropiumoxide.TheseobservationswereinagreementwithalreadypublisheddataonthestatusofREEscontainedinspentfluorescentlampwaste(Belardietal.,2014).SimilarXRDpatternswerenotedbetweenthedifferentsamples, suggesting that the initial treatment, either thermal or wet, does not alter the crystalstructureofthemajorcomponents.

Otherimpuritymetalsweredetectedinthesamples.Theseoriginatefromthephosphors(Table4),remainingelectrodematerial,endcaps,electronicboard fragmentsgeneratedduringcrushingetc.Calcium is the impuritymetal that is present in the largest amounts, being amajor componentofhalophosphors.Othernotableimpuritymetalswerealuminium,bariumandiron.

4.2.Hydrometallurgicalremovalofmercury

Leachingofmercurywasstudiedwithwater,ammoniumchloride,nitric,hydrochloricandaceticacid(PublicationI)andI2/KIsolutions(PublicationsIIIandIV).Watergavethepoorestleachingefficiency,with0.3±0.1%ofthemercurybeingleachedduring168hfromsampleIV(Figure8).

Figure8–Comparisonsbetweenthe leachingefficiencies formercury fromsample IVusingvariousleachingagents, asdescribed inPublication III. Leaching conditions forwater, ammoniumchloride,acetic,hydrochloricandnitricacidsolutions:20±1°C,168h,S/Lratioof10%w/v,magneticstirring200±50rpm.LeachingconditionsforI2/KIsolutions:21°C,2hleachingtime,S/Lratioof10%w/v,magneticstirring300rpm.Theerrorrepresentsthestandarddeviationofatriplicatetest.

30

Undersimilarconditions,2.1±0.4%ofthemercurywasleachedwith25%vol.aceticacid,and15.8±5.2%with1Mammoniumchloride.Thisindicatesthepresenceofmercuryascompoundswithlowwatersolubilitye.g.elementalmercury,oxides,mercury(I)chlorideetc.and/orstrongabsorptioninthe glass and phosphorus powder. Nitric and hydrochloric acid solutions led to better leachingefficiencies,mostnotablyforthelatter(Figure8).

Increasedacidconcentrationand increasedtemperature/ultrasound improvedtheefficiencyof theprocess,asdescribedinPublicationI.ThemainissueisthattheseacidicsolutionswillalsoleachtheREEs,notablyyttriumandeuropium,asfurtherdiscussedinsection4.3.Therefore,aselectiveleachingagent formercury is desired in order to prevent the contamination of subsequent REE-containingstreams.Iodineinpotassiumiodidewasfoundtohaveverygoodselectivityformercury,withalmostnoREEsbeingleachedalongside.Leachingofcalciumandantimonywasalsounder1%.Noleachingofaluminium, cadmium, chromium, iron, lead, tin or zincwas noticed.Only low amounts of barium,copper,magnesiumandnickel(eachunder2%oftheirinitialcontent)weredissolved.

LeachingofmercurywithI2/KIoccursviatwomainchemicalprocesses(Foust,1993):

• oxidationofmercuryanditscompoundstodivalentmercury(Hg2+);• reactionofthepreviouslyoxidizedspecieswiththeiodideions(I-)toformHgI42-ions.

Theverystrongcomplexesformedbythedivalentmercuryionswithiodideions(β4=5.6x1029)andthehighsolubilityoftheHgI42-ion(Cleveretal.,1985)leadstotheefficientmobilizationofmercuryfrom thematrix. Leaching of elementalmercury,mercuric sulphide,mercuric oxide and adsorbedmercuryoccursaccordingtoEquations16-19(Mattigodetal.,2009).

Hg+I2+2I-→HgI42- (16)

HgS+4I2+3H2O→HgI42-+SO42-+6H++4I- (17)

HgO+4I-+2H+→HgI42-+H2O (18)

Hg-X2(s)+2K++4I-→2KX+HgI42- (19)

TheleachingefficiencyofmercurywasfoundtoincreasewithincreasedI2/KIconcentrations(Figure8)anda0.025/0.05Msolutionwasenoughtoachieveremovalofmorethan90%of themercury.Completedecontaminationwasnotachieved,possiblyduetoverystrongadsorptionofmercuryintheglasscullet(Jangetal.,2005).TheleachingbehaviourofmercuryovertimeispresentedinFigure9,suggestingaminimumleachingtimeoftwohoursforthistypeofwasteunderthetestedconditions.

31

Figure9–LeachingbehaviourofmercuryfromsampleIVusing0.05/0.1MI2/KIsolution,accordingtoPublicationIV(21°C,upto4hleachingtime,S/Lratioof10%w/v,300rpmmagneticstirring).Theuncertaintybarsrepresentthestandarddeviationofatriplicatetest.

IncreasesoftheS/Lratioledtoslightdecreasesinleachingefficiency(Figure10).

Figure10– Leachedpercentagesofmercury fromsample IVusing0.025/0.05M I2/KI solutionanddifferentS/Lratios,accordingtoPublicationIV(21°C,2hleachingtime,300rpmmagneticstirring).Theuncertaintybarsrepresentthestandarddeviationofatriplicatetest.

Duetotheoxidizingnatureofthesolution,thiol-basedadsorbentsarenotsuitedtotherecoveryofmercuryintheleachate,duetooxidationtodisulphide(Equation20).

2-SH→-S-S-+2H+ (20)

32

AnionicexchangeusingDowex1X8resin (chloride form) led tocompleteremovalofmercury fromsolution. The iodine dissolved in potassium iodide, present as a tri-iodide species (Equation 21)(Mattigodetal.,2009),wasremovedalongside(Figure11).

I2+I-=I3- (21)

Figure11–Kinetics forremovalofmercuryand iodinefromI2/KIbased leachatesusingDowex1X8resin(chlorideform),aspresentedinPublicationIII.Theinitialmercurycontentinsolutionwas24.5mg/L.25mLofleachatewastreatedwith2.5gDowex1X8resinforupto4hat21°C,undermagneticstirring(500rpm).Theuncertaintybarsrepresentthestandarddeviationofatriplicatetest.

Thispreventsthereuseoftheresultingsolutionforfurtherleachingofmercury.Theadvantageisthattheobtainedstreamcanbedisposedofasnon-hazardousindustrialwastewaterorcanbeusedasasolventforthepreparationofnewI2/KIleachant.

Analternativeistheadditionofareducingagente.g.sodiumhydrosulphitetoprecipitatethemercuryinthe leachate.Completereductionofmercurywasachievedbyaddingsodiumhydrosulphitetoaleachatecontaining149mg/LmercuryusingaS/Lratioof2.5%.Approximately99.5%ofthemercurywas reducedwhenusing a S/L ratio of 0.5%. Similar to theDowex 1X8 treatment, thismakes theresultingsolutionimpossibletoreuseforfurthermercuryleachingduetoreductionofoxidizingiodinespecies.Itdoeshowevermaketherecoveryofmetallicmercuryeasiertoachieve,e.g.isolationoftheresidue,followedbythermaltreatment.

AthirdalternativetorecovermercuryfromI2/KIleachatesissolventextraction.ExtractantssuchasCyanex302andCyanex923havepreviouslybeenstudiedformercuryrecoveryfromchloridemedia(FrancisandReddy,2003;Meeraetal.,2001).

Mercurywasextractedfromiodineleachateswithgreatefficiencyinonestage(D≥500)withbothCyanex302andCyanex923inSolvent70,respectively.WhenusingCyMe4BTBPinnitrobenzene,anaveragedistributionratioof42.1±11.1wascalculatedbasedonthreeinvestigatedreplicates.Oneoftheadvantagesofsolventextractionisthefactthattheresultingsolutionretainssomeofitsoxidizingcharacteristics,makingitsreuseforfurthermercuryleachingpossible.

33

Leachingofmercuryfromothertypesofcontaminatedsamples(IandIII)withI2/KIsolutionwasalsotestedinPublicationIVandthefindingsareinagreementwiththeaforementioneddata(Table8).

Table8–Leachedpercentagesformercuryfromsamples IandIIIafter2hwith0.025/0.05MI2/KIsolutionat21°C,accordingtoPublicationIV.

Sample Hgcontent(mg/kgdrywaste) S/Lratio(%w/v) Hgleached(%)

I 112±220 93.440 91.9

III 227±520 93.140 90.5

A clear advantage for this mercury leaching process is the use of simple, commercially available,chemical reagents. Thermal treatment of fluorescent lamp waste fractions has large energyrequirements, given the fact that temperaturesashighas800 °Careneeded toachievecompletedecontamination.Forwet samples theenergy inputwillbehigher.Thehydrometallurgicalprocessdeveloped here can be an alternative to thermal treatment and has a considerably lower energyrequirement. Because the residue obtained after the I2/KI leaching step ismeant to be processedhydrometallurgicallytorecoverthecontainedREEs,therewillbeanoverlapbetweenthespaceandequipment requirements for these twosteps.Decontamination in thiscasecanbecarriedout inaleachingtank,thelargestamountoftheenergyinputgoingintoagitationandthepumpingoffeeds.Furthertreatmentoftheresiduee.g.REEsleachingcanbecarriedoutusingthesameequipment,afterremoval of the I2/KI leachate. The initial investment in equipment, especially as this is the sameequipment needed in the subsequent treatment of the residue for REEs recovery, will be lowercomparedtodedicatedequipmentforthermaldesorption.

ThetreatmentofthecontaminatedI2/KIleachatefollowsasimpleapproachand,dependingontheprocesschosen,doesnotrequirecomplicatedequipmente.g.acolumn(ifionexchangeisused)orareduction tank (ifNa2S2O4 treatment ispreferred).Both iodineandmercuryare removed fromtheleachate,allowingfortheresultingwastewatertobereusedfurtherintheprocess,e.g.forthewashingofresiduesorpreparingnewleachingagent,thusequilibratingthewaterbalance.

Sincedecontaminationisnot100%efficientusingI2/KIleaching,themethodissuitableiftheamountsofmercury left intheresidueafterdecontaminationarebelowtherequirementssetby legislation.Thisisveryimportantwhentakingsafetyandenvironmentalaspectsintoconsideration.

4.3.Leachingofmetals

Water,ammoniumchloride,acetic,nitricandhydrochloricacidsolutionswerescreenedaspotentialleachingagentsfortheREEs inthewaste(PublicationI).Onlytheacidicsolutions ledtonoticeableleachingofREEs.

Littleover50%oftheeuropiumand75%oftheyttriumwas leachedfromsampleIVwith25%vol.aceticacid(168h,ambienttemperature,S/Lratioof10%w/v)(Figure12).

34

Figure12–LeachingofREEsfromsampleIVusing25%vol.aceticacidsolution(20±1°C,168h,S/Lratio of 10% w/v, 200 ± 50 rpm magnetic stirring). The uncertainty bars represent the standarddeviationofatriplicatetest.

SmallamountsoftheotherREEs(2-10%)wereleachedalongside.Nitricandhydrochloricacidprovedtobeveryeffectiveatleachingyttriumandeuropiumundersimilarconditions(Figure13).

Figure 13 – Leached percentages of REEs using nitric and hydrochloric acid solutions of variousconcentrations(20±1°C,168h,S/Lratioof10%w/v,200±50rpmmagneticstirring).Theuncertaintybarsrepresentthestandarddeviationofatriplicatetest.

Acidconcentrationwasfoundtoplayamore importantrole inthe leachingofcerium,gadolinium,lanthanumand terbium.While leaching efficiencies >95%were reached for yttrium andeuropiumevenwithdilutedsolutions,theeffectivedissolutionofthegreenphosphorswasfoundtoincreasewithacidconcentration.ThisisduetothesignificantlyincreasedsolubilityofYOXphosphorcomparedtoLAPandCATphosphors.

35

Nitricacidwaschosenforfurtherstudy.TheleachingbehaviourofREEsfromsampleIVwithnitricacidsolutionsofvariousconcentrationsispresentedinFigure14.

Figure 14 – Leaching behaviour of REEs from sample IV using nitric acid solutions of variousconcentrations (20 ± 1°C, S/L ratio of 10% w/v, 100 rpmmagnetic stirring). The uncertainty barsrepresentthestandarddeviationofatriplicatetest.

36

Leachingofcerium,gadoliniumandterbiumoccurredslowlyovertime,withoutreachingequilibriumundertheinvestigatedconditions.Similaritiesintheleachingpatternswerenoticedduetothefactthattheseelementsareusuallyencounteredtogetheringreenphosphors.Yttriumandeuropiumareleachedmuchfaster,withequilibriumbeingreachedafterabout24hThesedifferencesinleachingbehaviourcanbeexploitedtoachievepartialselectivitybetweenyttrium+europiumandtheotherfourREEs,facilitatingtheirsubsequentseparation,e.g.lessseparationstagesarerequiredforsolventextraction. An increased leaching efficiency for cerium, gadolinium, lanthanum and terbium wasobservedwhenincreasingtemperatureandwhenusingultrasound(Figure15).

Figure15–LeachedpercentagesofREEsfromsampleIVusing1Mnitricacidsolutionundervariousconditions:20±1°Cand60±2°Candultrasoundassisteddigestionat60±2°C,respectively(6hleachingtimeusingaS/Lratioof10%w/vandmanualshaking,followedby24hrestingat20±1°C).Theuncertaintybarsrepresentthestandarddeviationofatriplicatetest.

Therecoveryoftheseelementsfromtheresidueremainingafterleachingofyttriumandeuropiumcanthusbecarriedoutinasubsequentleachingstepusingmoreconcentratedsolutionsatelevatedtemperature.Alternatively, alkali fusion followedby acid dissolution canbeused, as suggestedbyZhangetal.(2013)(Equations14and15).

SinceaconsumptionofprotonsoccursduringtheleachingofREEs(Equation9-11),acidconcentrationisanimportantparameterinthisprocess.Thepresenceofcalcium-basedhalophosphorsinthewastefractionsposesproblemsduring leachingby increasingacidconsumptionand limitingtheS/Lratio.Leaching of calcium with acidic solutions occurs significantly faster than REEs, leading to fastconsumptionofprotons(Figure16).

37

Figure16–LeachingbehaviourofseveralmetalspresentinlargeramountsinsamplesI-IIIwith1Mnitricacidat21°CandS/Lratioof10%w/v.Fromlefttoright:samplesI,IIandIII,respectively.

According to Equation 12, 18molesH+ are consumedduring leaching of onemole halophosphate(DupontandBinnemans,2015).Leachingofcalciumandotherimpuritymetalsinthehalophosphorsalongside REEs also leads to increasedmetal concentration in the leachate and, depending on pHchanges,possibleprecipitation issuesduringsolventextraction.Leachingwithsulphuricacidavoidscalcium,bariumandleaddissolutionduetotheformationof insolublesulphates.Nonetheless,thiswasshowntoleadtoco-precipitationofREEs,andthereforelossesofpotentialrecyclablematerial(DeMichelisetal.,2011;Innocenzietal.,2013).

Thedifferenceinleachingbehaviourofcalciumandotherimpuritiescanbeusedtoselectivelyleachtheseelements.ThiswilllowertheimpuritymetalconcentrationsintheREEsleachates,asdiscussedinPublicationIV.Forthemoreconcentratedwastefractionse.g.samplesI-III,a10-15minleachingtimewith1MnitricacidatambienttemperatureandS/Lratioof10%w/vwasdeemedsufficienttoleachcalciumwith80-90%efficiency(Figure16).Thelossesofyttriumandeuropiumwerebelow2-3%,withtheexceptionofsampleII,whichwasthermallypre-treatedtoremovemercury.Analmost2-foldincreaseintheleachingofyttriumandeuropiumwasnotedforthissample.Thiswasattributedtothebreakingoftheparticlesofphosphoraggregatesandcrackingofparticlesuponheating,whichinturnleadstofasterdissolutionduetolargersurfaceareacontactwiththeleachingagent.

OthersolubleimpuritymetalsthatleachedinsufficientamountsfromsamplesI-IIIduringthefirst15min were barium (up to 70% efficiency), antimony (up to 77% efficiency), strontium (up to 92%efficiency),cadmium(upto88%efficiency),silicon(upto80%efficiency),tin(upto72%efficiency),lead (up to 78% efficiency), copper (up to 67% efficiency) and zinc (up to 88% efficiency). Thedissolution behaviour of most of these elements was similar to that of calcium, manganese ormagnesium(fasterleachingwithinthefirst10-15min,followedbyslowerdissolutionafterthispoint).Aluminiumwasfoundtohavelowerdissolutionefficiencies,partiallyduetothechemicalstabilityofaluminausedasabarrierlevelinfluorescentlamps(oneofthesourcesofaluminium).

TheREEsleftintheresidueafterleachingofthebulkcalciumfromsamplesI-III(Figure17)werefoundtohaveasimilarleachingbehaviourtothatobservedduringleachingofsampleIV(Figure14).Again,leaching of yttrium and europium occurred almost concomitantly over approx. one day, and wassignificantly faster than for the other four REEs. The same slight increase in leaching efficiency ofmetalsfromsampleII(thermallypre-treated)wasalsoobserved.

38

Figure17–LeachingbehaviourofREEsfromsamplesI-IIIafterpriorleachingofimpuritymetals.Firstleachingstep:1Mnitricacidfor10minat21°CandS/Lratioof10%w/v.SecondleachingstepforREEs(presentedhere):2Mnitricacidat21°CandS/Lratioof10%w/v.Fromlefttoright:samplesI,IIandIII,respectively.

TwoapproacheswerecomparedfortheleachingofREE-concentratedwastefractions,withthegoalofdissolvingasmuchoftheyttriumandeuropiumandaslittleoftheotherfourREEs:

1. Leachingofthewasteinasinglestepwithnitricacidsolution.Thiswilldissolvemostmetals,leadingtoacomplexsolutionrichincalciumandyttrium.Themainadvantagesarethatonlyone leaching step is performed and one liquid stream is obtained (reduced volumes ofsecondary wastes). The disadvantages are a higher metal load in solution, possibleprecipitationissuesduringsolventextractionandlowergradefinalproduct.

2. Selectiveleachingofcalcium,bariumandotherimpuritymetals,followedbyfurtherleachingof the residue to dissolve the REEs. The advantages for this are less impurities in solution(highergradefinalproduct)andlessissueswiththesolventextractionstep.ThedisadvantagesareminorlossesofREEsduringthefirstleachingstepandtheneedtoprocesstwosolutionstreams(secondarywastes).

Table9showsacomparisonbetweentheconcentrationsofmetalsintheREE-richsolutionsobtainedusingthesetwoapproaches.Aclearreductionofimpuritymetalsinsolutionwasnotedinthecaseofatwo-stepleachingprocess.

Amulti-stepleachingapproachalsohastheadvantageofreducingtheunwantedanionicspeciesintheREE-richsolution.Inthiscasethelargestamountofanionicspecieswillendupinthecalcium-richleachate.Thehalophosphatephosphorsmakeahighcontributiontothis,withsixphosphateionsandtwo chloride and fluoride ions, respectively, being released per dissolved halophosphatemolecule(Equation12).SincethebulkoftheREEsareinoxideform(YOX,BAM,CAT),thepresenceofunwantedanionicspeciesintheREE-richleachatewillbelow.TheLAPphosphorisasourceofphosphateanions,buttheamountpresentinthewasteisrelativelylow,asisitssolubilityunderthetestedconditions.

39

Table9–ComparisonsbetweentheconcentrationsofsomemetalsintheREE-richsolutionsobtainedby(1)asinglestepleachingofREEswith3Mnitricacidfor24husingaS/Lratioof10%w/v,and(2)atwo-stepleaching,asfollows:i)partialleachingofimpuritieswith1Mnitricacidfor10min,usingaS/Lratioof10%w/v,followedbyii)leachingofREEsintheremainingresiduewith2Mnitricacidfor24husingaS/Lratioof10%w/v.

Approach SampleElementconcentrationintheREE-richsolution(g/L)andchangesinconcentrationusingtwo-stepleaching(%)Al Ba Ca Fe Mg Mn Eu Y

(1)I 0.19 0.43 12.91 0.34 0.22 0.42 0.96 15.25II 0.23 0.39 8.31 0.17 0.06 0.26 1.01 15.49III 0.33 1.69 9.22 0.32 0.15 0.30 1.02 17.07

(2)

I 0.09(-53%)

0.06(-86%)

1.53(-88%)

0.03(-91%)

0.04(-82%)

0.05(-88%)

1.37(+43%)

20.88(+37%)

II 0.13(-44%)

0.13(-67%)

0.39(-95%)

0.06(-65%)

0.05(-17%)

0.01(-96%)

1.40(+39%)

22.29(+44%)

III 0.11(-67%)

0.85(-50%)

0.56(-94%)

0.10(-69%)

0.07(-53%)

0.03(-90%)

1.34(+31%)

21.83(+28%)

4.4.SolventextractionofREEswithCyanex923

TheextractionofREEswithCyanex923wasfoundtoreachequilibriumwithinonemin(Figure18).

Figure18–KineticsoftheextractionprocessforREEs, ironandmercuryusing0.7MCyanex923inSolvent70,accordingtoPublicationII.Theaqueousphasewasobtainedbyleaching20gsampleIVwith200mL2Mnitricacidsolutionfor48hat60±2°C(finalnitricacidconcentrationwas1.7M).Theextractionwascarriedoutat23±1°Cand1750vpm,usingaphaseratioof1.Theuncertaintybarsrepresentthestandarddeviationofatriplicatetest.

Thedistributionratiosoflanthanidesgenerallyincreasedwithdecreasesinionicradius.Gadoliniumshowed distribution ratios lower than those of europium, despite having smaller ionic radius(GreenwoodandEarnshaw,1998).Similarbehaviourhasbeenobservedinothersolventextractionsystemsand this is explainedby thehalf-filled electron shell of gadolinium(III) (4f7) (Nilssonet al.,

40

2006).Ironandmercuryionswerealsoextractedbutthekineticsweresignificantlyslower(Figure18).Equilibriumwasattainedafter10-15minforiron,whileextractionofmercuryoccurredcontinuouslyforthedurationoftheinvestigatedtime.AshortcontacttimebetweenthetwophasescanbeusedinordertoachievepartialseparationoftheREEsfromthesetwoundesiredelements.Chlorideionsinsolution are responsible for the extraction ofmercury (Equation 6).No extraction ofmercurywasobservedfromnitricacidsolutioncontainingonlypureREEs,mercuryandironnitrates.Forthisreason,a decontamination step prior to the solvent extraction of REEs is advised in order to prevent thespreadingofmercuryinsubsequentprocessstreams.

TheacidityoftheaqueousphaseplaysanimportantroleintheextractionofREEions(Figure19).

Figure19–ExtractionofREEsfromfluorescentlampleachatesofvariousacidityusing0.7MCyanex923inSolvent70,accordingtoPublicationII.TheaqueousphaseswereobtainedbyleachingsampleIVwith0.5M,1M,2Mand4Mnitricacidsolutionsfor168hat20±1°C,usingaS/Lratioof10%w/v(finalnitricacidconcentrationsintheleachateswere0.16M,0.67M,1.7Mand3.69M,respectively).Theextractionwascarriedoutat23±1°C,usingaphaseratioof1.Initialnitricacidconcentrationsintheleachatesareplottedonthex-axis,nottheequilibriumvalues.Theuncertaintybarsrepresentthestandarddeviationofatriplicatetest.

Adecrease inextractionefficiencywasobservedwithan increase inacidity,duetocompetitionofprotonswith the REE ions for the extractantmolecules. The data in Figure 19 refers to the initialconcentrations of nitric acid in the aqueous phases investigated, not to the actual equilibriumconcentrations,whichareexpectedtobelower.ThisisbecauseCyanex923iscapableofextractingnitricacid,accordingtoEquation22.

xH+(aq)+xNO3-(aq)+yTRPO(org)↔yTRPO·xHNO3(org) (22)

41

ItwasstatedthatprobableCyanex923-nitricacidcomplexescanbeofthetype2:1,1:1and1:2(Ansarietal.,2004).Figure19ismeanttobearepresentationofhowextractionisinfluencedbytheacidityintheaqueousphaseduringthefirststageofanextractionprocess.

ThedistributionratiosofnitricacidincreasewithincreasesinCyanex923concentrationintheorganicphase(Figure20;equilibriumdata).

Figure20–ExtractionofnitricacidfromarealfluorescentlampwasteleachateatvariousCyanex923concentrations,accordingtoPublicationII.Theaqueousphasewasobtainedbyleaching20gsampleIVwith200mL2Mnitricacidsolutionfor48h,at60±2°C(finalnitricacidconcentrationwas1.7M).The organic phase consisted of 0.24M, 0.7M, 1.18M, and 1.65MCyanex 923 in Solvent 70 andundilutedCyanex923(2.35M),respectively.Theextractionwascarriedoutfor10minat23±1°Cand1750vpm,usingaphaseratioof1.

Uponextraction,thepHoftheaqueousphasewillincrease.Thiscanleadtohydrolysisofmetalsionsandformationofinsolubleprecipitates,whichareundesiredinanysolventextractionprocess.Forthisreason,thepresenceoflargeamountsofcalciuminlampwasteleachatescanhindertheextractionprocessifthepHincreaseishighenough.Useofatwo-stepleachingprocessforthewasteisindicatedtoavoiddissolvingthebulkofthecalciumtogetherwiththeREEs.

TheextractionofREEswasfoundtodecreasewithanincreaseintemperature(Figure21).

42

Figure21–TheinfluenceoftemperatureontheextractionprocessforREEs,accordingtoPublicationII.Theaqueousphasewasobtainedbyleaching20gsampleIVwith200mL2Mnitricacidsolutionfor48hat60±2°C(finalnitricacidconcentrationwas1.7M).Theextractionwascarriedoutwith0.7MCyanex923inSolvent70for10minat23±1°C,40±1°Cand60±1°C,respectively,usingaphaseratioof1and1750vpm.Theuncertaintybarsrepresentthestandarddeviationofatriplicatetest.

Itwas concluded that the extractionof REEs is drivenby changes in enthalpy. For all six REEs theenthalpyandentropyofextractionwerefoundtobenegative(Table10).

Table10–EnthalpyandentropyofcomplexationintheextractionofREEsfromarealfluorescentlampwasteleachateofsampleIVusing0.7MCyanex923inSolvent70,accordingtoPublicationII.

Element ∆Go(23°C;kJ/mol) ∆Ho(kJ/mol) ∆So(J/mol·K)Ce -17.96±0.03 -18.3±0.4 -1±1Eu -21.28±0.04 -22.1±0.5 -3±2Gd -21.06±0.03 -22.7±0.5 -5±2La -15.36±0.06 -14±0.4 5±1Tb -21.98±0.16 -23.3±1 -5±3Y -20.99±0.03 -23.1±0.5 -7±2

Hydrochloricandnitricacidsolutions,respectively,werefoundtoeffectivelystriptheextractedmetals(Figure 22). The former was capable of selectively stripping the REEs from co-extracted iron andmercury.Nitricacidcanbeusedforthescrubbingofco-extractedspecies,afterstrippingofREEswithhydrochloricacid.AcidextractedbyCyanex923canbescrubbedbywashingtheorganicphasewithwater.ThiswasobservedbyapHdropinpurewaterto<1uponwashing,asnotedinPublicationIV.

43

Figure 22 – Stripping of extractedmetals using hydrochloric and nitric acid solutions, according toPublicationII.Anorganicphasecontaining0.7MCyanex923inSolvent70wasloadedwithmetalsbymanuallyshakingitwithanitricacid-basedleachateofsampleIVfor10minat21°C,usingaphaseratioof1.Strippingwasperformedat22±1°Cfor5min,usingaphaseratioof1.Theuncertaintybarsrepresentthestandarddeviationofatriplicatetest.

4.5.Testingoftheprocessinlaboratorypilotscale

Based on the aforementioned findings, the leaching and separation processes were tested inlaboratory pilot scale using a leaching reactorwith propeller stirring and a counter-currentmixer-settlersystem,asdescribedinPublicationIV.

4.5.1.Processingofthermally-treatedfractions

Leaching of metals from the mercury-free sample II was investigated using the two approachesdiscussedinsection4.3.Thiswastodetermineifthepresenceoflargeamountsofcalciumalongsidethe REEswould hinder the solvent extraction separation process. Themetal concentrations in thesolutionsobtainedarepresentedinTable11.

44

Table11–Concentrationsofmetals in the solutionsobtainedafter leachingof sample IIusing i)asingle-stepfor22husing3Mnitricacidsolutionandii)two-stepleaching,firstwith1Msolutionfor10min,followedbyfurthertreatmentoftheresiduewith2Msolutionfor22h.Allleachingexperimentswerecarriedoutinambientconditions.

Elements

Leachatetypeandtheconcentrationsofseveralmetalsinsolution(mg/L)Single-stepleaching(2Macid;22h;S/Lratio10%w/v)

Two-stepleaching;Ca-richleachate(1Macid;10min;S/Lratio10%w/v)

Two-stepleaching;REE-richleachate(Resultingresidue+2Macid;22h;S/Lratio10%w/v)

REEs

Ce 12.9 3.4 21.6Eu 966.6 70.7 1196Gd 29.0 4.1 55.2La 8.0 6.0 4.4Tb 10.6 4.4 14.1Y 15860.0 591.6 18020.0

Impurities

Al 213.8 137.7 102.0B 117.9 75.5 81.5Ba 278.1 314.6 78.3Ca 8143.0 7576.0 205.5Cd 3.8 4.5 0.2Cr 0.8 0.5 0.3Cu 18.6 8.8 11.3Fe 101.5 48.6 26.6K 66.8 53.0 20.4Mg 79.7 49.7 45.7Mn 208.6 257.0 4.5Mo 5.2 2.3 3.8Na 209.8 191.5 36.8Ni 7.3 2.3 6.6Pb 27.8 26.3 6.8Sb 131.9 100.3 38.0Si 122.0 164.9 60.3Sn 3.2 2.5 1.6Sr 250.0 213.9 100.3Ti 1.4 0.5 1.3W 21.3 5.9 17.5Zn 10.0 7.4 4.8Zr 1.5 0.8 3.5

The twoREE-rich leachates,denotedascalcium-rich leachateandcalcium-depleted leachate,werepumped into the counter-current mixer settler system described in Figure 6. The extraction andstrippingbehavioursofyttriumandeuropiumarepresentedinFigure23.

45

Figure 23 – Extraction/stripping behaviour of yttrium (left) and europium (middle) from differentleachates of sample II in a mixer-settler system comprised of three extraction and four strippingchambers,respectively.ThepHchangeoftheaqueousphaseduringextractionisshownontheright.Theextractionwascarriedoutwith50%vol.Cyanex923inSolvent70.Strippingwasdonewith2Mand4Mhydrochloricacid,respectively.AllO:Aphaseratioswere2:1.Theerrorrepresentsthestandarddeviationofuptosixsamplings,uptoonehourapart,duringoperationtime.

The separation trial showed good stability over time, with only minor variations in metalconcentrationsbetweenthecollectedsamples.Similaritiesintheextractionandstrippingbehavioursof REEs between the two types of leachates were observed, the minor differences being due toincreasedcontentofREEsinthecalcium-depletedleachate.MostoftheREEs(>92%)wereextractedin the first stage, followed by recovery of >99% after two stages. For the calcium-rich leachate,precipitationwasobservedtobemostobviousaftersystemshutdownovernight.PrecipitationwasstrongestinchamberE3andwasconnectedtopHchangesduringextractionwithCyanex923(Figure23).Sincethiswasnotanissueforthecalcium-depletedleachate,itisadvisedtocarryoutleachingofREEsintwo-steps,especiallyifthematerialcontainshighamountsofhalophosphatepowder.Thiswillalso lead to lower amounts of co-extracted species in solution. Iron, molybdenum, tin, titanium,tungsten and zirconium were co-extracted almost entirely alongside the REEs but they were notstrippedbyhydrochloricacid.Extractionofboron,leadandsiliconwasalmostabsentinthefirststagesbutwasnoticeableinthethirdextractionchamber.Usingtwoseparationstagesinthiscasewillleadtohigherproductpurity.

Stripping with 2M hydrochloric acid was not completely effective at recovering all the extractedmetals.Thiswasachievedbyincreasingtheacidconcentrationto4M.ThemetalcontentinthefinalstripproductsispresentedinTable12.

46

Table 12 – Concentration of impuritymetals in the final strip product derived from sample II. Theextraction and stripping were carried out with 50% vol. Cyanex 923 and 2 M hydrochloric acid,respectively.AllO:Aphaseratioswere2:1.TheinitialconcentrationsintheaqueousfeedareshowninTable6.Theuncertaintybarsrepresentthestandarddeviationofuptosixsamplings,uptoonehourapart,duringoperationtime;nd=notdetected.

Feed Metalsandtheirconcentrationinthefinalstripproduct(mg/L)Al B Ba Ca Fe Mn Pb Sb Si Sr Ce La Gd Tb

Ca-richleachate

3.1±0.1

20.9±1.0

1.4±0.1

118.8±8.5 0.1 4.1±

0.36.2±0.7 0.6 18.0±

0.6 0.9 9.8±0.5

6.6±0.1

9.3±0.5

2.8±0.1

Ca-depletedleachate

1.4±0.3

10.6±1.7

0.7±0.1

5.9±1.6 nd nd 1±

0.2 0.2 9.7±0.9

0.4±0.1

19.4±1.1

4.4±1.5

24.2±0.5

6.2±0.4

4.5.2.Processingofmercury-contaminatedfractions

Theseparationprocesswasalsotestedforsample III.Becausethissamplecontains227±5mg/kgmercury, a decontamination stepwas carried outwith 0.025/0.075M I2/KI solution. Themercuryleachingefficiencywas>96%after4.5h(Figure24).Equilibriumwasachievedfastercomparedtothebenchscaletrials(within45min),aconsequenceofsignificantlybettermixingandcontactbetweenthephases,e.g.propellerstirringvs.magneticstirring.

Figure24–Leachingofmercuryfrom500gsampleIIIwith0.025/0.075MI2/KIsolutionandaS/Lratioof25%w/v.

Inasubsequentleachingstepsomeoftheimpuritymetalsweredissolvedwith1Mnitricacidfor10mininordertoavoidanyprecipitationissuesduringsolventextraction.Asolutioncontaining8.6g/Lcalciumand1.2g/Lbariumasmajorimpuritieswasobtained.Thissolutionalsocontained0.05and0.7g/Leuropiumandyttrium,respectively.Inafinalleachingstep,thebulkoftheREEswereleachedwith2Mnitricacid.ExtractionofREEsfromthissolutionwascarriedoutusing50%vol.Cyanex923butalsowith35%vol.Cyanex923.Thiswasbasedonthepreviousobservationthattwostagesshouldbesufficient to achieve separation of these two REEs when using 50% vol. extractant. Stripping was

47

carriedoutwith4Mhydrochloricacid.TheextractionandstrippingbehaviourofyttriumandeuropiumispresentedinFigure25.

Figure 25 – Extraction/stripping behaviour of yttrium (left) and europium (middle) froma calcium-depleted leachate of sample III in a mixer-settler system comprised of three extraction and fourstrippingchambers,respectively.ThepHchangeoftheaqueousphaseduringextractionisshownontheright.Theextractionwascarriedoutwith35%and50%vol.Cyanex923inSolvent70,respectively.Strippingwasperformedwith4Mhydrochloricacid.AllO:Aphaseratioswere2:1.Theuncertaintybarsrepresentthestandarddeviationofuptothreesamplings,uptoonehourapart,duringoperationtime.

TheconcentrationofmetalsintheaqueousfeedsandinthestrippingproductsisshowninTable13.Similarities between the co-extraction of impurity elements in this leachate and the leachates ofsample IIwereobserved.The35%vol.Cyanex923wasmore thansufficient to completelyextractyttriumandeuropium.Usinglessconcentratedorganicphasesalsoleadstohigherstripproductpurity.Loweraciduptakeduringextractionwillalsooccur,thereforeresultinginaneasierregenerationoftheextractant.PhaseseparationwasobservedtobebetteratlowerCyanex923concentrations.

48

Table13–Concentrationsofmetals in theREE-rich leachatesobtainedafter leaching sample III inlaboratorypilotplantscale,aswellasinthestrippingproductsafterextractionwith35%and50%vol.Cyanex923inSolvent70;nd=notdetected.

Elements

Solutiontypeandtheconcentrationsofseveralmetalsinsolution(mg/L)REE-richleachate

Stripproduct(35%vol.Cyanex923)

Stripproduct(50%vol.Cyanex923)

REEs

Ce 11.5 15.2±1.8 15.0±0.2Eu 1253.0 1268.5±27.5 1100.3±13.5Gd 28.4 29.4±1.0 27.5±0.0La 2.2 2.7±0.4 2.5±0.1Tb 9.2 8.6±0.2 7.8±0.1Y 20080.0 21605.0±1005.0 19676.7±81.8

Impurities

Al 61.5 1.3±0.0 3.1±0.3B 186.3 8.8±1.3 16.1±1.0Ba 387.8 4.3±0.4 10.2±1.7Ca 200.6 3.6±0.3 5.8±0.2Cd 0.6 nd ndCr 0.3 nd ndCu 14.3 nd ndFe 61.8 nd ndK 13.3 nd ndMn 10.3 nd 0.4±0.0Mo 0.6 nd ndNa 9.8 nd 1.3±0.0Ni 6.5 nd 0.3±0.1Pb 9.8 1.5±0.0 2.0±0.4Sb 21.5 nd 0.4±0.0Sn 0.8 nd ndSr 300.9 3.9±0.3 6.8±0.3Ti 1.2 nd ndW 5.2 nd ndZn 8.5 nd ndZr 0.4 nd nd

4.6.FurtherseparationofREEswithCyanex572

Theobjectiveoftheseexperiments,describedindetailinPublicationV,wastoindividuallyseparateyttriumandeuropium,thepredominantREEsinthestrippingproductsobtainedafterextractionwithCyanex923.Theaqueousphasesusedoriginatedfromthemixer-settlertestsdiscussedintheprevioussection.

Underthetestedconditions,theextractionandstrippingkineticsofREEswithCyanex572werefoundtobeslowercomparedtoCyanex923(Figure26).

49

Figure 26 – Extraction (left) and striping (right) behaviour of europium and yttrium over 30 min.Extractionwascarriedoutwith1MCyanex572inSolvent70at21±1°C,usingaphaseratioof1.pHin=0.7;pHeq=-0.1.Strippingwascarriedoutwith6Mhydrochloricacidat21±1°Cfroma1MCyanex572solutioninSolvent70thatwaspreviouslycontactedwithyttrium/europiumfeed.Theuncertaintybarsrepresentthestandarddeviationofatriplicatetest.

TheCyanex572systemreachedequilibriumwithin12.5min,significantlylongerthanthe<1minforCyanex923.Strippingequilibriumwasachievedwithin5min,withover>99.9%oftheeuropiumandyttriumintheorganicphasebeingback-extractedwitha6Mhydrochloricacidsolution.

EquilibriumpHhassignificantimportance,asCyanex572isanacidicextractant(Equations7and8).TheextractionbehaviourofyttriumandeuropiumatvariouspHeqispresentedinFigure27.

Figure27–ExtractionbehaviourofyttriumandeuropiumatvariouspHeq.ExtractionwascarriedoutfromaqueousphasesofvariouspHvalueswith1MCyanex572inSolvent70at21±1°C,usingaphaseratioof1.Initialconcentrationsofyttriumandeuropiumintheaqueousphasewere18g/Land1g/L,respectively.Phasecontacttimewas15min.Theuncertaintybarsrepresentthestandarddeviationofatriplicatetest.

50

YttriumcanbeseparatedfromeuropiumataroundpHeq=0.Thiswillrequireseveralseparationstagesasapprox.22%thetotalyttriumwasextractedinonestage.Atthesametime,<0.5%oftheeuropiumwasco-extracted.TheseamountsincreaseaspHeqincreases.Aseparationfactorof61betweenyttriumandeuropiumwasattainedatpHeq=0.Sincephasedisengagementdidnotposeanyproblems,being<15seconds,andnothirdphaseformationoccurred.Phasemodifierswerenotused.

IncreasesinligandconcentrationintheorganicphaseledtoexpectedincreasesindistributionratiosofbothREEs.

ForEquation7,theextractionconstantKexcanbewrittenaccordingtoEquation23.

𝐾:; =[𝑅𝐸𝐸𝐴3(𝑜𝑟𝑔)]∙[𝐻

+(𝑎𝑞)]

3

[𝑅𝐸𝐸3+(𝑎𝑞)]∙[𝐻𝐴(𝑜𝑟𝑔)]3 (23)

ThedistributionratioofaREEiswrittenaccordingtoEquation24.

𝐷 =[𝑅𝐸𝐸𝐴3(𝑜𝑟𝑔)]

[𝑅𝐸𝐸3+(𝑎𝑞)] (24)

ByinsertingEquation24intoEquation23,Equation25isobtained:

𝑙𝑜𝑔𝐷 = 𝑙𝑜𝑔𝐾:; + 3 ∙ 𝑙𝑜𝑔[DE(FG)][D"(HIJ)]

(25)

Thedecimal logarithmof thedistributionratiosofREEsvs. theratiobetweentheconcentrationofprotonsandextractantatequilibriumwereplotted(Figure28).

51

Figure 28 – Dependency of the extraction behaviour of europium and yttrium on Cyanex 572concentration.Equilibriumvalueswereplotted.ExtractionwascarriedoutwithCyanex572inSolvent70(0.75-2M)at21±1°C,usinganaqueousphasewithpHin=0.8.Contacttimewas15min,andthephaseratiousedwas1.Theuncertaintybarsrepresentthestandarddeviationofatriplicatetest.

Theslopeofthislineardependencyshouldbeequalto3(thenumberofextractantmoleculesinvolvedintheextractionofoneREEion),accordingtoEquation7.Yttriumshowedagoodcorrelationwiththistheoretical model. The slope for europium was higher, 3.66 ± 0.09. For slope analysis it isrecommendedthattheextractantisinlargerexcesscomparedtotheamountofmetalionsextracted.Thisistoensurethatthefreeextractantconcentrationsatequilibriumdonotvarysignificantlywithloading.Forthedatapresentedhere,thecontentofREEsintheaqueousphasecanbeapproximatedto0.2M.Evenatdistributionratios<1,asisthecasehere,thisrequirementisnotentirelysatisfiedforallpointsintheslope.Thiscanbeareasonfortheobserveddeviationsfromthetheoreticalmodel.Hein(2014)reportedaslopeof3.85±0.06fortheextractionofeuropiumwithCyanex572attracelevels (approx.0.65mM) fromnitricacidmedia.Quinnetal. (2015) reported thatdeviations fromlinearityandaslopeof3arelikelyduetothechangeinpropertiesoftheorganicphase,especiallyathighextractantconcentratione.g.formationofhydrogen-bondedaggregates.

Theinfluenceofvarioustypesofdiluentsontheextractionofyttriumwasinvestigated(Table14).

Table14–DiluentstestedintheCyanex572solventextractionstudy.

Diluentcategory Commercialname/compound Characteristics

Aliphatic

Solvent70 0.5%wt.aromaticcontentTetraPropyleneHydrogenated(TPH)

Similartododecane.UsedinthePUREXprocessinLaHague.

IsoparL <0.01%wt.aromaticcontentEscaid120 0.4%wt.aromaticcontent

Aromatic

Toluene Nitrobenzene Tertbutylbenzene Solvesso150 >99%wt.aromaticcontent

Alcohols 1-octanol Longchainaliphaticalcohol

52

Bestperformancewasobtainedusingaliphaticdiluents.VerysmalldifferenceswerenotedbetweenaliphaticdiluentssuchasTPH,IsoparLandEscaid120(Figure29).

Figure29–Distributionratiosofyttriumwith1MCyanex572invariousdiluents.Extractionwascarriedoutat21±1°C,fromanaqueousphasewithpHin=0.8,usingaphaseratioof1.Phasecontacttimewas30min.Theuncertaintybarsrepresentthestandarddeviationofatriplicatetest.

Asignificantdropindistributionratios(30-45%)wasseenwhensubstitutingIsoparLwitharomaticdiluents.Whenusing1-octanolthedistributionratiodroppedevenmore.Inthissystemathirdphaseformationoccurredattheorganic-aqueousinterface.

By increasing the temperature of the system, an increase in the distribution ratios of yttrium andeuropiumwasnoted(Figure30).

Figure30–Dependencyoftheextractionofeuropiumandyttriumontemperaturesintherange25-65°C(±1°C).Extractionwascarriedoutwith1MCyanex572inSolvent70.Contacttimewas30min,andthephaseratiowas1.Theuncertaintybarsrepresentthestandarddeviationofatriplicatetest.

53

Theincreaseindistributionratioswithincreasingtemperaturewasmoresignificantforeuropium.Thisledtoadecreaseinseparationfactors,fromαY/Eu=65toαY/Eu=46,whenthetemperatureincreasedfrom25to65°C.Toachievebetterseparationandhigherproductpurityitisthereforeadvisedtocarryouttheextractionatlowertemperatures.

StrippingoftheextractedREEscanbecarriedoutwithmineralacidse.g.hydrochloricandnitricacidsolutions(Figure31).

Figure31–Strippingbehaviourofeuropiumandyttriuminonestagewithhydrochloricandnitricacidsolutionsofvariousconcentrations.Theorganicphasewas1MCyanex572 inSolvent70 thatwaspreviouslycontactedwithanaqueoussolutioncontaining1g/Leuropiumand18g/Lyttrium(pHin=0.8;15min,withmanualshakingat21±1°Cusingaphaseratioof1).Contacttimeduringstrippingwas15min,at21±1°C,usingaphaseratioof1.Theuncertaintybarsrepresentthestandarddeviationofatriplicatetest.

OneofthemainadvantagesofCyanex572overphosphonicacidsisthefactthatstrippingrequireslessconcentratedacids(Cytec,2014).Hydrochloricacidsolutions≥3MstrippedalltheREEsextractedinonestagewhenusingO:Aratiosof1.Dilutedacids(0.1M)canbepotentiallyusedtoscrubpartoftheco-extractedeuropium(42-47%efficiency)inordertoachievehigherpurityofpotentialyttriumproducts.

Usingtheseresults,theseparationofyttriumfromeuropiumwasfurthercarriedoutfrom1Lfeedintensuccessivebatchextractionstageswith1MCyanex572(pHeqintherange0-0.1).Morethan97%oftheyttriumwasextractedduringthefirstsevenstages,with>99.9%beingrecoveredafterthetenthstage. This separation more closely resembles an analytical procedure rather than a continuouscounter-currentprocess,aswouldberequiredindustrially.Furtherstrippingwascarriedoutwith3Mhydrochloric acid to obtain a yttrium-concentrate meant for further processing, e.g. oxalic acidprecipitation(furtherdescribedinthenextsection).

Theaqueousphaseleftaftertheextractionofyttriumcontainedthebulkofeuropiumandtracesofgadolinium, terbium, cerium, lanthanum and yttrium. Extraction of europium was subsequently

54

carriedoutatpHeq=1,inonestage,accordingtoFigure32.Theextractedeuropiumwasstrippedwith3Mhydrochloricacid.

Figure32–ExtractiondependencyofeuropiumandtraceREEsonpHeq,afterextractionofbulkyttrium.Extractionwascarriedoutwith1MCyanex572at21±1°C,fromaqueousphasesofvariouspH,usinga phase ratio of 1. Phase contact time was 15 min. The uncertainty bars represent the standarddeviationofatriplicatetest.

4.7.PreparationofREEoxides

RareearthoxalatescanbeprecipitatedfromsolutionusingoxalicacidinaccordancewithEquation13.TheoxalatescanbeconvertedtooxidesviathermaltreatmentinaccordancewithEquation26.

2REE2(C2O4)3(s)+3O2(g)→2REE2O3(s)+12CO2(g) (26)

Rare earthoxideswereprepared from the yttrium/europium-rich strippingproduct obtained aftersolvent extraction with Cyanex 923 (Publication IV); and the yttrium stripping product and theeuropiumstrippingproductobtainedaftersolventextractionwithCyanex572(PublicationV).Fortheformer, a mixed yttrium-europium oxide was obtained and the structure was confirmed by XRDmeasurements.TheREEscontentwas99.96%REEs:94.61%yttrium,5.09%europiumand0.26%theotherfourREEs.Onefifthofthe0.04%impuritymetals(mostlysodium,silicon,boron,calciumandbarium)originatedfromtheoxalicacidusedforprecipitation.

Scrubbingof theCyanex572 loadedwithyttriumwith0.1Mhydrochloricacidpriortostrippingofyttriumwith3Macidledtoanincreaseinpurityofyttriumoxidefrom99.77%(withoutscrubbing)to99.82%.Themajorimpurityinbothcaseswaseuropiumoxide(Table15).

55

Table15–Metalcontentofthepreparedyttriumandeuropiumoxides,respectively(%).TheassaywasperformedbydissolvingtheoxidesinSuprapurnitricacidsolution,followedbymeasurementsoftheobtainedsolutionsusingICP-MSandICP-OES.

Metal

Compound/preparationmethodandmetalcontent(%)

Y2O3;noscrubbingwith0.1MHClpriortostripping

Y2O3;scrubbingwith0.1MHClpriortostripping

Eu2O3;directprecipitationfromtheaqueousphaseafterextractionofY

Eu2O3;furtherextractionwithCyanex572afterextractionofY

Al <0.01 <0.01 nd ndB <0.01 <0.01 0.09 0.02Ba <0.01 <0.01 0.13 ndCa <0.01 <0.01 0.10 ndCe nd nd 1.87 0.38Eu 0.13 0.08 82.81 91.60Fe <0.01 <0.01 0.01 ndGd 0.01 0.01 6.24 7.16K 0.01 nd 0.41 ndLa nd nd 0.37 0.06Na 0.03 0.02 6.15 0.04Si <0.01 <0.01 0.25 ndTb 0.02 0.02 0.41 0.46Y 99.77 99.82 1.08 0.29Zn <0.01 <0.01 0.07 nd

Foreuropiumoxide,theobtainedpuritywaslower(91.6%)duetothepresenceoftheotherfourREEsinsolution(notablygadolinium,whichwaspresentinaslightlyhigheramountinthefeedusedinthisparticularexperiment(approx.90mg/L)).MinimizationoftheseunwantedREEscanbeachievedbyhaving better control over selectivity during leaching. By optimizing the leaching conditions withrespect to maximum dissolution of yttrium/europium and minimum dissolution ofgadolinium/terbium/cerium/lanthanum, higher purities of europium oxides can be obtained usingCyanex572.Asdiscussedinsections4.1and4.3,thermaltreatmenttoremovemercurywasshowntoleadtoincreasedleachingefficiencyofthesefourREEs.

Itisworthmentioningthatifprecipitationofeuropiumiscarriedoutfromtheraffinatedirectlyaftersolvent extraction of yttrium, the obtained purity of europium oxidewill be lower due to the co-precipitationofsodium(Table15).HighamountsofsodiumfromthesodiumhydroxideusedforpHadjustments were present in solution. In this case, further extraction of europium at higher pH,followedbyitsstripping,willleadtomuchhigherproductpurity.

56

4.8.Proposedprocessflowsheet

Ageneralflowsheettoprocessfluorescentlampwastefractionslikethoseinvestigatedinthepresentstudy,basedontheaforementionedobservations,ispresentedinFigure33.

Figure 33 – Proposed flowsheet for the recovery of REEs from fluorescent lamp waste fractionscontainingphosphors.Theprocesscomprisesremovalofmercury(inpurplecolour),selectiveleachingofimpuritymetals(ingreen)andREEs(indarkblue),groupsolventextractionofREEswithCyanex923(inblack),furtherseparationofyttriumandeuropiumwithCyanex572(indarkred),andpreparationofREEoxides(inlightblue).

57

Themaintargetsareyttriumandeuropium.Allhydrometallurgicalstepsintheprocessarecarriedoutatambientconditions.Adecontaminationstepisfirstcarriedouttoremovemercuryandpreventitsspreadinsubsequentprocessstreams.Thiscanbedonethermally(thetraditionalroute)orusingthehydrometallurgicalprocessbasedontheI2/KIsolutionspresentedinthisthesis.Afterawashingsteptoremoveresidualiodineandsolublemercury,thebulkmetalsinthewasteareleachedselectivelywithnitricacidbytakingadvantageoftheirdifferentleachingkinetics.Calcium,togetherwithsomeoftheotherimpuritymetals,isdissolvedafter10–15minutes.A1MsolutionandaS/Lratioof10%w/vareneededforsamplescontainingintherangeof10%wt.calcium.Theresultingresidueisleachedwith2Mnitricacidfor18horlonger,usingaS/Lratioof10%w/v,todissolveyttriumandeuropium,followedbyagroupseparationoftheREEsinthisleachateusingsolventextractionwithcommercialsolvents.A35%vol.Cyanex923solution inkerosene issufficienttocompletelyextracttheREEs inthreestagesatO:Aratiosof2:1.Strippingwith4Mhydrochloricacidwillleadtoyttrium/europium-richsolutions.Fromheretherearetwopossiblepaths;eitherprecipitateamixedREEsoxalatewithoxalicacidorprocessthissolutionwithCyanex572toseparateyttriumfromeuropiumatdifferentpHeqvalues.TheseparatedstreamsarethentreatedwithoxalicacidtoprecipitateREEoxalates,whicharesubsequentlythermallytreatedat800°Ctoobtainoxides.

58

59

5.SUMMARYANDCONCLUSIONS

ThegoalofthisstudywastodevelopahydrometallurgicalprocesstorecoverREEsfromfluorescentlampwastephosphorsfractionsatambienttemperature.Mercury-contaminatedrealwastesamplesoriginatingfromlampprocessingcompanieswereinvestigated.Thewastefractionsstudiedhadhighchemical complexity,withmany impuritymetals present alongside cerium, europium, gadolinium,lanthanum,terbiumandyttrium,thesixREEscommonlyfoundinsuchstreams.Onaverage,17-18%drywt.REEswerequantifiedinthemostconcentratedfractions,alongside>100mg/kgmercury.

Becausethermaltreatmentofthewaste(acommonmethodusedtodaytoremovemercury)isenergyintensiveandnotalwayssuitableforwetsamples,analternativehydrometallurgicaldecontaminationprocesswasproposed.LeachingofmercurywithI2/KIsolutionsatambienttemperatureswasfoundtobeselectiveoverREEs.Mercuryremovalefficienciesof>90%wereachievedusing0.025/0.05MI2/KI,andincreasesinefficiencieswerenotedwithincreasesinleachantconcentration.IonexchangewithDowex1X8resin,reductionwithsodiumhydrosulphiteandsolventextractionwithCyanex302,Cyanex923andCyMe4BTBPwereallefficientatrecoveringmercuryfromsolution.Therecoveryratesofmercuryfromsolutionwereoften>99%whenusingthesemethods.

Impuritymetals,notablycalcium,whichisthemostprevalentimpuritymetal,wereselectivelyleachedpriortotheleachingofREEsbytakingadvantageoftheirfastdissolutionkinetics.A10-15mincontacttimewith1Mnitricacidwas found tobe sufficient to leach thebulkof calcium (80-90% leachingefficiency).Thisallowsforhigherfinalproductpurityandminimizessomeoftheissuesthatcanoccurinsubsequentprocesssteps,e.g.precipitationduringsolventextraction.

Mineralacids,e.g.nitricandhydrochloricacid,weremoreeffectiveatdissolvingREEscomparedtoorganic acids, e.g. acetic acid. Partial leaching selectivity between the sixREEs in the sampleswasachievedbycontrollingtheleachingtimeandtemperature,andalsomakinguseofultrasound-assisteddigestion.Leachingofyttriumandeuropiumdidnotposeaproblematambientconditions.Leachingefficiencies>95%werenotedforthesetwoREEsover24h,evenwithdilutedacids.ThisisduetotherelativelyeasydissolutionoftheYOXphosphor(yttrium-europiumoxide)comparedtoLAP(cerium,terbium-doped lanthanum phosphate) and CAT (cerium, terbium-doped magnesium aluminate)phosphors. After dissolution of the YOX phosphor, further leaching of the remaining REEs can becarriedoutusing concentratedacids and temperature/ultrasound,orby chemically converting theundissolvedphosphorstomoresolubleoxides(alkalifusion).

Two commercial extractants, Cyanex 923 and Cyanex 572,were used to separate the REE ions insolution.TheformerwasusedforagroupseparationoftheREEsfromtheotherimpuritymetalsinlampleachates,whilethelatterwasusedtoindividuallyseparateyttriumfromeuropium.Leaching,followedbyseparationofREEswithCyanex923,wassuccessfullytestedinlaboratorypilotscaleusingleaching reactors and counter-current mixer-settler systems. Selective leaching of metals waspreferredoverleachingofallmetalsinasinglestep.Thelatterledtolowerstrippingproductpurityand to precipitation during extraction due to pH changes and the presence of large amounts ofimpuritymetals in solution. Over 99% of the REEs present in lamp leachates were extracted andstripped, respectively, in a mixer-settler system comprising three extraction stages with 35% vol.Cyanex923andfourstrippingstageswith4MhydrochloricacidatO:Aratios2:1.

60

ThestrippingproductafterextractionwithCyanex923wasfurtherprocessedwithCyanex572,anovelreagentaimedatindividualREEsseparations.YttriumwasselectivelyseparatedfromeuropiumatpHeq=0usingtensuccessivebatchextractionstages,followedbyrecoveryofeuropiumatpHeq=1inonestage.RareearthoxideswerepreparedbyprecipitatingtheREEsinthestripproductswithoxalicacid,followed by thermal treatment of the oxalates at 800 °C. Amixed REEs oxide (99.96% REEs,with94.61%yttrium,5.09%europiumand0.26%others),aswellasyttriumoxide(99.82%)andeuropiumoxide(91.6%)weresynthetized.

61

7.FUTUREWORK

Additional studies are required to improve certain steps in the process. Particular areas requiringfurtherdevelopmentaredescribedbelow.

1. Recovery of mercury from I2/KI leachates using ion exchange and solvent extraction. Theexperimentspresentedinthisstudywereaproofofconceptandmoreinsightisneeded.RegenerationoftheDowexresinrequiresaddressing,toavoidexcessiveconsumptionofionexchanger.Also,thesolvent extraction recovery of mercury requires more study to gain knowledge of the extractionsystems.Alternativesofremovingthemercuryfromtheleachateswithoutalteringtheoxidizingnatureof the solution (the iodine speciation)wouldbebeneficial, toensure the reuseof this solution forfurtherleaching.

2.FurtheroptimizationoftheleachingprocessofmetalswithregardtotheS/Lratio.Thisisneededinorder tominimize the volumesof secondarywastegeneratede.g. the calcium-rich solutionor theraffinateaftersolventextractionwithCyanex923.

3.FurtheroptimizationofthesolventextractionandstrippingprocesswithregardtotheO:Aratioandacidconsumption.Similartotheabove,thiswouldreducetheamountsofsecondarywastegenerated.Theuseof lessconcentratedacidduringstrippingwouldalsorequire lesssodiumhydroxideforpHadjustmentduringsolventextractionwithCyanex572.AciduptakebyCyanex923anditsscrubbingwithwateralsoneedsadditionalinvestigation.

4. Better control of selectivity during leaching.Minimizing the amounts of undesired constituentsduring leachingwould translates into higher product purity. Less gadolinium and terbium leachedalongsideyttriumandeuropiumwouldleadtohigherpurityofeuropiumoxide.

5.ExaminetheuseofCyanex572toextractREEsfromtheyttrium/europium-richleachatewithoutfirst carryingout a groupextractionof REEswithCyanex 923. Thiswould significantly simplify theseparation process, minimize chemical consumption and lower the amounts of secondary wastegenerated.

62

63

8.ACKNOWLEDGMENTS

Formas, Energimyndigheten and Vinnova, Sweden, are acknowledged for financial support. Thepartnersinvolvedatvariousstagesinthisresearchareacknowledgedforsupplyingguidanceandwastesamples:NordicRecyclingAB(amemberoftheHansAnderssonGroupAB),CITRecyclingDevelopmentAB,MRTSystemInternationalAB,MEABMetallextraktionABandIVLSvenskaMiljöinstitutet.

Specialthanksgotomysupervisors,TeodoraReteganandChristianEkberg,andalltheotherpeoplewhohaveguidedmeonthepathIcurrentlywalk,helpedmebewhereIamtodayandblessedmewiththeiradvice,knowledge,friendshipandlove.

64

65

REFERENCES

Ansari,S.A.,Murali,M.S.,Pathak,P.N.,Manchanda,V.K.,2004.SeparationofironfromcobaltinnitratemediumusingCyanex-923asextractant.JournalofRadioanalyticalandNuclearChemistry262,469-472.

Baek,D.L.,Fox,R.V.,Case,M.E.,Sinclair,L.K.,Schmidt,A.B.,McIlwain,P.R.,Mincher,B.J.,Wai,C.M.,2016. ExtractionofRare EarthOxidesUsing Supercritical CarbonDioxideModifiedwith Tri-n-ButylPhosphate–NitricAcidAdducts.IndEngChemRes55,7154-7163.

Belardi,G.,Ippolito,N.,Piga,L.,Serracino,M.,2014.Investigationonthestatusofrareearthelementscontainedinthepowderofspentfluorescentlamps.ThermochimicaActa591,22-30.

Binnemans, K., Jones, P.T., Blanpain, B., VanGerven, T., Yang, Y.X.,Walton,A., Buchert,M., 2013.Recyclingofrareearths:acriticalreview.JournalofCleanerProduction51,1-22.

Cevasco,G.,Chiappe,C.,2014.Areionicliquidsapropersolutiontocurrentenvironmentalchallenges?GreenChemistry16,2375-2385.

Chang,T.C.,You,S.J.,Yu,B.S.,Chen,C.M.,Chiu,Y.C.,2009.Treatinghigh-mercury-containinglampsusingfull-scalethermaldesorptiontechnology.JournalofHazardousMaterials162,967-972.

Clever,H.L.,Johnson,S.A.,Derrick,M.E.,1985.TheSolubilityofMercuryandSomeSparinglySolubleMercurySaltsinWaterandAqueousElectrolyteSolutions.JournalofPhysicalandChemicalReferenceData14(3),631-680.

CliftonII,J.C.,2007.MercuryExposureandPublicHealth.PediatricClinicsofNorthAmerica54,237-269.

Cox,M.,2004.Solventextractionequilibria,in:Rydberg,J.,Cox,M.,Musikas,C.,Choppin,G.R.(Eds.),Solventextractionprinciplesandpractice,secondedition,revisedandexpanded.MarcelDekker,Inc.,NewYork,109-201.

Cox,M.,Rydberg, J.,2004. Introduction toSolventExtraction, in:Rydberg, J.,Cox,M.,Musikas,C.,Choppin,G.R.(Eds.),Solventextractionprinciplesandpractice,secondedition,revisedandexpanded.MarcelDekker,Inc.,NewYork,1-25.

Cuif,J.P.,Rohart,E.,Macaudiere,P.,Bauregard,C.,Suda,E.,Pacaud,B.,Imanaka,N.,Masui,T.,Tamura,S.,2005.Applications,in:Adachi,G.,Imanaka,N.,Kang,Z.C.(Eds.),BinaryRareEarthOxides.SpringerNetherlands,215-255.

Cytec,2014.Cyanex572SolventExtractionReagent.Availableat:https://www.cytec.com/sites/default/files/files/CYTEC_CYANEX_572_FINAL.pdf[AccessedMay2016].

Dang,T.A.,Frisk,T.A.,Grossman,M.W.,Peters,C.H.,1999.Identificationofmercuryreactionsitesinfluorescentlamps.JournaloftheElectrochemicalSociety146,3896-3902.

DeMichelis, I., Ferella, F.,Varelli, E.F.,Vegliò, F., 2011.Treatmentofexhaust fluorescent lamps torecoveryttrium:Experimentalandprocessanalyses.WasteManagement31,2559-2568.

66

dosSantos,É.J.,Herrmann,A.B.,Vieira,F.,Sato,C.S.,Corrêa,Q.B.,Maranhão,T.A.,Tormen,L.,Curtius,A.J.,2010.DeterminationofHgandPb incompact fluorescent lampbyslurry sampling inductivelycoupledplasmaopticalemissionspectrometry.MicrochemicalJournal96,27-31.

Doughty,D.A.,Wilson,R.H.,Thaler,E.G.,1995.Mercury-glassinteractionsinfluorescentlamps.JournaloftheElectrochemicalSociety142,3542-3550.

Dupont, D., Binnemans, K., 2016. Antimony recovery from the halophosphate fraction in lampphosphorwaste:azero-wasteapproach.GreenChemistry18,176-185.

Durao,W.A.,Jr.,deCastro,C.A.,Windmoller,C.C.,2008.Mercuryreductionstudiestofacilitatethethermal decontamination of phosphor powder residues from spent fluorescent lamps. WasteManagement28,2311-2319.

Dziwinski, E., Szymanowski, J., 1998. Composition of Cyanex® 923, Cyanex® 925, Cyanex® 921 andTOPO.SolventExtractionandIonExchange16,1515-1525.

EuropeanCommission,2010.CriticalrawmaterialsforEU.ReportoftheAd-hocWorkingGroupondefiningcriticalrawmaterials.Availableat:http://www.euromines.org/files/what-we-do/sustainable-development-issues/2010-report-critical-raw-materials-eu.pdf[AccessedMay2016].

European Commission, 2011. Critical Metals in Strategic Energy Technologies. Available at:http://setis.ec.europa.eu/system/files/CriticalMetalsinStrategicEnergyTechnologies-def.pdf[AccessedMay2016]

EuropeanCommission,2014.ReportonthecriticalrawmaterialsforEU.ReportoftheAd-hocWorkingGroup on defining critical raw materials. Available at: https://ec.europa.eu/growth/tools-databases/eip-raw-materials/en/community/document/critical-raw-materials-eu-report-ad-hoc-working-group-defining-critical-raw[AccessedMay2016].

Foust,D.F.,1993.Extractionofmercuryandmercury compounds fromcontaminatedmaterial andsolutions.USPatent5226545.

Francis,T.,Reddy,M.L.P.,2003.ThiosubstitutedOrganophosphinicAcidsasSelectiveExtractantsforMercury(II)fromAcidicChlorideSolutions.SolventExtractionandIonExchange21,109-123.

Greenwood, N.N., Earnshaw, A., 1998. Chemistry of the Elements (2nd Edition). ButterworthHeinemann,Oxford.

Gupta,C.K.,Krishnamurthy,N.,2005.Extractivemetallurgyofrareearths.CRCPress,BocaRaton.

Hobohm, J., Kuchta, K., Krüger, O.,Wasen, S.V., Adam, C., 2016. Optimized elemental analysis offluorescencelampshredderwaste.Talanta147,615-620.

Hull,A.W.,1919.Anewmethodofchemicalanalysis.JournaloftheAmericanChemicalSociety,41(8),1168-1175.

Innocenzi,V.,DeMichelis,I.,Ferella,F.,Veglio,F.,2013.Recoveryofyttriumfromcathoderaytubesandlamps'fluorescentpowders:experimentalresultsandeconomicsimulation.WasteManagement33,2390-2396.

Jang,M.,Hong,S.M.,Park,J.K.,2005.Characterizationandrecoveryofmercuryfromspentfluorescentlamps.WasteManagement25,5-14.

67

Kilbourn, B.T., 1994. A Lanthanide Lanthology: Part 2, M-Z. Molycorp Incorporated. Available at:http://www.phy.davidson.edu/fachome/dmb/RESolGelGlass/Lanthology/Lanthology_M-Z.pdf[AccessedMay2016].

Löfström Engdahl, E., Aneheim, E., Ekberg, C., Skarnemark, G., 2010. Diluent effects in solventextraction.Proceedingsof theFirstACSEPT InternationalWorkshopLisbon,Portugal,31March–2April2010.Availableat:http://www.acsept.org/AIWOproc/AIWO1-PR15-Lofstrom.pdf[AccessedMay2016].

Mattigod, S.V., X. Feng, and G.E. Fryxell, 2009. Mercury separation from concentrated potassiumiodide/iodineleachateusingSelf-AssembledMesoporousMercaptanSupport(SAMMS)technology.

McLennan, S., 1994. Rare earth element geochemistry and the “tetrad” effect. Geochimica etCosmochimicaActa58,2025-2033.

Meera,R.,Francis,T.,Reddy,M.L.P.,2001.Studiesontheliquid–liquidextractionofmercury(II)fromacidicchloridesolutionsusingCyanex923.Hydrometallurgy61,97-103.

Mei, G.J., Rao, P., Matsuda, M., Fujita, T., 2009. Separation of red (Y2O3:Eu3+), blue(BaMgAl10O17:Eu2+)andgreen(CeMgAl10O17:Tb3)rareearthphosphorsbyliquid/liquidextraction.JournalofWuhanUniversityofTechnology-MaterialsScienceEdition24,603-607.

Nakamura, T., Nishihama, S., Yoshizuka, K., 2007. Separation and Recovery Process for Rare EarthMetalsfromFluorescenceMaterialWastesUsingSolventExtraction.SolventExtractionResearchandDevelopmentJapan14,105-113.

Nash, K.L., 1993.A reviewof thebasic chemistry and recent developments in trivalent f-elementsseparations.SolventExtractionandIonExchange11,729-768.

NEMA,2005.Fluorescentandothermercury-containinglampsandtheenvironment.Availableat:https://www.nema.org/Policy/Environmental-Stewardship/Lamps/Documents/Lamp%20Brochure.pdf[AccessedMay2016].

Nilsson,M.,Ekberg,C.,Foreman,M.,Hudson,M.,Liljenzin,J.O.,Modolo,G.,Skarnemark,G.,2006.Separationofactinides(III)fromlanthanides(III)insimulatednuclearwastestreamsusing6,6ʹ-Bis-(5,6-dipentyl-[1,2,4]triazin-3-yl)-[2,2ʹ]bipyridinyl (C5-BTBP) in cyclohexanone. Solvent Extraction and IonExchange,24(6),823–843.

OSHA,1987.Mercuryvaporinworkplaceatmospheres.Availableat:http://www.osha.gov/dts/sltc/methods/inorganic/id140/id140.html[AccessedMay2016].

Otto, R.S.,Wojtalewicz-Kasprzak, A.E., 2011.Method for recovery of rare earths from fluorescentlamps.USPatent7976798B2.

Peppard,D.F.,Mason,G.W.,Lewey,S.,1969.Atetradeffectintheliquid-liquidextractionorderingoflanthanides(III).JournalofInorganicandNuclearChemistry,31,2271-2272.

Quinn, J.E., Soldenhoff, K.H., Stevens,G.W., Lengkeek,N.A., 2015. Solvent extractionof rare earthelementsusingphosphonic/phosphinicacidmixtures.Hydrometallurgy157,298-305.

Rabah,M.A.,2008.Recyclablesrecoveryofeuropiumandyttriummetalsandsomesaltsfromspentfluorescentlamps.WasteManagement28,318-325.

68

Rao,S.R.,2006.Chapter4Hydrometallurgicalprocesses, in:Rao,S.R. (Ed.),ResourceRecoveryandRecyclingfromMetallurgicalWastes.Elsevier,71-108.

Raposo, C., Windmoller, C.C., Durao, W.A., Jr., 2003. Mercury speciation in fluorescent lamps bythermalreleaseanalysis.WasteManagement23,879-886.

Reddy,M.L.P.,Varma,R.L.,Ramamohan,T.R., Sahu,S.K.,Chakravortty,V.,1998.Cyanex923asanextractantfortrivalentlanthanidesandyttrium.SolventExtractionandIonExchange16,795-812.

Rey-Raap, N., Gallardo, A., 2012. Determination of mercury distribution inside spent compactfluorescentlampsbyatomicabsorptionspectrometry.WasteManagement32,944-948.

Ritcey,G.M.,2004.DevelopmentofIndustrialSolventExtractionProcesses, in:Rydberg,J.,Cox,M.,Musikas,C.,Choppin,G.R. (Eds.),Solventextractionprinciplesandpractice,secondedition,revisedandexpanded.MarcelDekker,Inc.,NewYork,277-337.

Ronda,C.R.,Justel,T.,Nikol,H.,1998.Rareearthphosphors:fundamentalsandapplications.JournalofAlloysandCompounds275,669-676.

Shimizu,R.,Sawada,K.,Enokida,Y.,Yamamoto, I.,2005.Supercritical fluidextractionof rareearthelementsfromluminescentmaterialinwastefluorescentlamps.JournalofSupercriticalFluids33,235-241.

Sobral, L.G.S., Santos, R.L.C., Costa, B.R.S., Paiva, F.J.P., 2004. Mercury Removal from ExhaustedFluorescent Lamps: I) The ultimate Oxidative Processes, Proceedings of the 7th InternationalConferenceonMercuryasaGlobalPollutant,Ljubljana,Slovenia,ISSN1408-7073,731-735.

Song,X.,Chang,M.H.,Pecht,M.,2013.Rare-EarthElementsinLightingandOpticalApplicationsandTheirRecycling.JOM65,1276-1282.

Thaler,E.G.,Wilson,R.H.,Doughty,D.A.,Beers,W.W.,1995.Measurementofmercuryboundintheglass envelope during operation of fluorescent lamps. Journal of the Electrochemical Society 142,1968-1970.

TheEuropeanParliamentandtheCounciloftheEuropeanUnion,2003.Directive2002/95/ECoftheEuropeanParliamentandoftheCouncilof27January2003ontherestrictionoftheuseofcertainhazardoussubstancesinelectricalandelectronicequipment.Availableat:http://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A32002L0095[AccessedMay2016].

Tunsu,C.,Petranikova,M.,Gergorić,M.,Ekberg,C.,Retegan,T.,2015.Reclaimingrareearthelementsfromend-of-lifeproducts:Areviewoftheperspectivesforurbanminingusinghydrometallurgicalunitoperations.Hydrometallurgy156,239-258.

Tyler,G.,2004.Rareearthelementsinsoilandplantsystems-Areview.PlantandSoil267,191-206.

UNEnvironmentProgramme,2011.RecyclingRatesofMetals:AStatusReport.Availableat:http://www.unep.org/resourcepanel/Portals/24102/PDFs/Metals_Recycling_Rates_110412-1.pdf[AccessedMay2016].

UN Environment Programme and UN University, 2009. Critical metals for future sustainabletechnologies and their recycling potential. Available at: http://oeko.de/oekodoc/1070/2009-129-en.pdf[AccessedMay2016].

69

USDepartmentofEnergy,2009.LampsstandardsfinalruleTSDAppendix3C(RareEarthPhosphorMarket).Availableat:https://www.yumpu.com/en/document/view/12176112/lamps-standards-final-rule-tsd-appendix-3c-rare-earth-phosphor-/3[AccessedMay2016].

USDepartmentofEnergy,2010.Fluorescentlighting.Availableat:http://energy.gov/energysaver/fluorescent-lighting[AccessedMay2016].

USDepartmentofEnergy,2011.CriticalMaterialsStrategy.Availableat:http://energy.gov/sites/prod/files/DOE_CMS2011_FINAL_Full.pdf[AccessedMay2016].

USEPA, 2012. Rare Earth Elements: A Review of Production, Processing, Recycling, and AssociatedEnvironmental Issues. Available at: http://nepis.epa.gov/Adobe/PDF/P100EUBC.pdf [AccessedMay2016].

USGS,2002.RareEarthElements-CriticalResourcesforHighTechnology.Availableat:http://pubs.usgs.gov/fs/2002/fs087-02/[AccessedMay2016].

Xie,F.,Zhang,T.A.,Dreisinger,D.,Doyle,F.,2014.Acriticalreviewonsolventextractionofrareearthsfromaqueoussolutions.MineralsEngineering56,10-28.

Yang,H.,Wang,W.,Cui,H.,Zhang,D.,Liu,Y.,Chen,J.,2012.Recoveryofrareearthelementsfromsimulatedfluorescentpowderusingbifunctionalionicliquidextractants(Bif-ILEs).JournalofChemicalTechnology&Biotechnology87,198-205.

Yang,F.,Kubota,F.,Baba,Y.,Kamiya,N.,Goto,M.,2013.Selectiveextractionandrecoveryof rareearthmetalsfromphosphorpowdersinwastefluorescentlampsusinganionicliquidsystem.JornalofHazardousMaterials254,79-88.

Zhang, S.G., Yang,M., Liu, H., Pan, D.A., Tian, J.J., 2013. Recovery ofwaste rare earth fluorescentpowdersbytwostepsacidleaching.RareMetals32,609-615.

70

71

APPENDIX1.ANALYTICALANDINTRUMENTALTECHNIQUES

ICP-OES

InductivelyCoupledPlasma-Optical EmissionSpectroscopy is ananalyticalmeasurement techniquedesigned for element quantification at trace levels. The method is characterized by having lowdetectionlimits;lowerthanflameatomicabsorptionspectrometrybutnotaslowasICP-MS.Detectionlimitsintherangeof1-50µg/L(ppb)havebeenreportedbythemanufacturerfortheinstrumentusedintheexperimentspresented(iCAP6500,ThermoFischer).Oneoftheadvantagesofthismethodismulti-elementanalysis,withnocompromiseofprecisionordetectionlimits.

The method requires the analytes to be in solution or as gas. To quantify elements in solids orsediments,adissolutionstepmustfirstbecarriedout.Theresultingsolutionisinjectedintheplasmawith a carrier gas, in this case argon, as aerosol. The sample is atomized in argon plasma at hightemperatures. Upon de-excitation, photons with characteristic wavelength are emitted by theelements present. The intensity of the emitted photons, which is proportional to the amount ofelementspresent, isdetected.Byusinganexternal calibration, theelementspresent in theorigialsamplecanbequantified.

SEM/EDS

Scanning ElectronMicroscopy/EnergyDispersive Spectroscopy is a technique used tomagnify andanalysetheappearanceofmaterials,toobtaindataabouttheelementspresent,andtodeterminetheoccurrenceanddistributionofcomponents.

A focusedbeamofhigh-energyelectrons isdirectedtowardsthetarget.Upon interactionwiththesample,secondaryelectrons,backscatteredelectronsandX-raysareemitted.Theseelectronscangivemorphologicalinformationaboutthesample,whiletheX-raysthatarecharacteristictoeachelementpresentinthesamplecanbeusedforelementalanalysis.

XRD

Hullpointedoutin1919that‘everycrystallinesubstancegivesapattern;thesamesubstancealwaysgivesthesamepattern;andinamixtureofsubstanceseachproducesitspatternindependentlyoftheothers’(Hull,1919).AnX-raybeamhitsthetargetandtheintensityofthereflectedbeamismeasuredbyadetector.Thisresultsinaspectruminwhichthereflectedintensitiesareplottedagainsttheanglesofreflection.SincetheXRDpatternofacrystallinecompoundislikeafingerprintofthatcompound,thismethodcanbeappliedforcharacterizationandidentificationofpolycrystallinephases.

ASiemensDiffraktometerD5000systemwithcopperradiationsourceandscintillationdetectorwasusedinthisstudy.Componentsintheanalysedsampleswereidentifiedbysearch/matchproceduresusingtheDifracEVAsoftwarefromBruker.

72

73

APPENDIX2.MIXER-SETLERS

Industrialsolventextractionofmetalsisoftencarriedoutusingcounter-currentphasecontactorse.g.mixer-settlers. In a counter-current separation process the phase volumes are constant. Theschematicsandprincipleofoperationofamixer-settlerunitarepresentedinFigure34.

Figure34–Schematicsandprincipleofoperationofamixer-settlerunit.

Asthenameimplies,thesystemcomprisesapartthatachievesmixingoftheaqueousandorganicphases,e.g.achamberwithpropeller-assistedmixing,andapartthatallowsforthemixedphasestoseparate.Bycombiningmultiplemixer-settlerunitstogether,multipleseparationstagesareachieved(Figure35).

Figure35–Counter-currentmixer-settlersetupforseparationofmetalions.Dashedarrowsshowtheflowoftheorganicphasewhilecontinuousarrowsshowtheflowoftheaqueousphase.AsimilarsetupwasusedinthepresentstudyforseparationofREEsfromfluorescentlampwasteleachates.

Multiple-stage separation is an important aspectwhen it comes to achieving high product purity.IndividualseparationofREEswithhighpurityoftenrequiresahighnumberofseparationstages,duetothesimilaritiesinchemicalproperties.Theaqueousandorganicphasesarefedinatoppositeendsofthebankofcontactors,whichmaximizesthedrivingforceforextraction(thesoluteconcentrationdifferencebetweenthetwophases)(CoxandRydberg,2004).

74

Mixer-settlershavesomeadvantagesoverothertypesofphasecontactorsusedinindustrialsolventextractionseparations,e.g.pulsecolumn(Ritcey,2004),suchas:

• Highefficiency• Simpleconstruction• Lowmaintenance• Easytoscaleup• Goodphasecontact• Awiderangeofflowratioscanbehandled• Welldefinedstepsallowsimpleprocesscalculations• Theprocessiseasytostartandrestart• Stagescanbeaddedorremovedeasily.

Downsidesrelateto:

• Spacerequirements(sometimesalargefloorareaisneeded)• Theneedforlargeorganicphasevolumes(largeholdups)• Thetimeitmaytakethesystemtoreachsteadystate• Highenergyconsumptionrequirementformixingandpumpingthefeeds insystemswitha

largenumberofstages.

Themixer-settlersetupusedforthegroupseparationofREEswithCyanex923(discussedinsection4.5)ispresentedinFigure36.

Figure36–Themixer-settlersetupusedforthegroupseparationofREEswithCyanex923,discussedinsection4.5.Thesystemconsistedofthreeextractionunits,whichcanbeseenontherightinbothpictures,andfourstrippingunits,locatedontheleft.