Magnetic separation of ilmenite used as an oxygen carrier in fluidized bed combustion

FINAL DEGREE PROJECT Chemical Enginee ing Deg ee MAGNETIC SEPARATION OF ILMENITE USED AS AN OXYGEN CARRIER IN FLUIDIZED BED COMBUSTION Repo and Annexes Au ho : Ignacio Lama ca Ba ba Di ec o : Raúl Bení ez Iglesias Da e: June 2021 iii Magne ic sepa a ion o ilmeni e used as an oxygen ca ie in luidized bed combus ion Bachelo ’s hesis in Chemical Enginee ing Ignacio Lama ca DEPARTMENT OF CHEMISTRY AND CHEMICAL ENGINEERING DIVISION OF ENERGY AND MATERIALS CHALMERS TEKNISKA HÖGSKOLA Go henbu g, Sweden 2021 www.chalme s.se i BACHELOR’S THESIS 2021 Magne ic sepa a ion o ilmeni e used as an oxygen ca ie in luidized bed combus ion Ignacio Lama ca Depa men o Chemis y and Chemical Enginee ing Di ision o Ene gy and Ma e ials CHALMERS UNIVERSITY OF TECHNOLOGY Go henbu g, Sweden 2021 i Magne ic sepa a ion o ilmeni e used as an oxygen ca ie in luidized bed combus ion IGNACIO LAMARCA BARBA © IGNACIO LAMARCA BARBA, 2021. Supe iso : Pa le a Knu sson, Depa men o Chemis y and Chemical Enginee ing Examine : B i -Ma ie S eena i, Depa men o Chemis y and Chemical Enginee ing Bachelo ’s hesis 2021 Depa men o Chemis y and Chemical Enginee ing Di ision o Ene gy and Ma e ials Chalme s Uni e si y o Technology SE-412 96 Go henbu g Telephone + 46 (0)31-772 1000 Go henbu g, Sweden 2021 ii Magne ic sepa a ion o ilmeni e used as an oxygen ca ie in luidized bed combus ion IGNACIO LAMARCA BARBA Depa men o Chemis y and Chemical Enginee ing Di ision o Ene gy and Ma e ials Chalme s Uni e si y o Technology Abs ac This bachelo hesis s udies how oxida ion and educ ion cycles du ing oxygen ca ie aided combus ion (OCAC) ope a ion, as well as ash in e ac ion, a ec he magne ic suscep ibili y o ilmeni e pa icles used as oxygen ca ie s. Magne ically sepa a ed samples o bo om ash collec ed om K a ingen’s ci cula ing luidized boile loca ed in Ö o a (Sweden) we e analysed. The samples we e pa o an OCAC campaign om Ap il 2018, in which was e wood and wood chips we e used as uel. Pa icles om he esh ma e ial, he bo om ash, as well as he ac ions sepa a ed by a magne - he magne ic ac ion (magne ic accep ) and he non-magne ic ac ion (magne ic ejec ) collec ed om di e en days we e mo phologically and chemically cha ac e ized using SEM-EDS. XRD was used o c ys alline compounds iden i ica ion and XRF was used o he analysis o he bulk composi ion o he samples. The a o emen ioned analysis we e complemen ed wi h magne ic suscep ibili y measu emen s o he samples. The sepa a ion o he magne was ound no o be o ally e ec i e as a i ac s (namely eldspa in he magne ic ac ion and inac i a ed ilmeni e in he magne ic ejec ) we e de ec ed in he samples. The esul s indica e ha p olonged esidence ime o he bed ma e ial in he boile inc eased he p esence o CaTi0.8Fe0.2O2.9 en ichmen s in he ash laye o he pa icles om he magne ic accep ac ion. I was obse ed ha ash compounds do no in e e e signi ican ly in he bulk magne ic suscep ibili y. Magne ic suscep ibili y inc eased wi h longe esidence ime, whe e he sugges ed cause behind i was he possible Fe2+ mig a ion owa ds he su ace o ilmeni e pa icles, and i s oxida ion o magne ic Fe oxides – i anomagne i e and hema i e. The esul s om he wo k sugges ha he magne ic suscep ibili y o bed ma e ial inc eases wi h esidence ime in he boile , and so does he ash up ake. Ea ly sepa a ion can esul in esh ma e ial ending in he magne ic ejec ac ion. Based on hese esul s, an op imal ime window o he ex ac ion, sepa a ion, eci cula ion and/o enewal o he bed ma e ial can be de ined ha will likely esul in an inc ease in sepa a ion e iciency o he bed ma e ial. The ewi h, magne ic suscep ibili y measu emen s could be a use ul ool in he op imiza ion and decision making ega ding his ime window. Keywo ds: Oxygen ca ie aided combus ion, ci cula ing luidized bed, oxygen ca ie , ilmeni e, ash in e ac ion, magne ic sepa a ion, magne ic suscep ibili y iii Acknowledgemen s I am g a e ul o ha ing been able o emba k on his p ojec since i has been a g ea oppo uni y o acqui e mo e knowledge han I had e e imagined when I s a ed. I ha e become e y in e es ed in he ield and I hope o keep lea ning mo e abou i in he nea u u e. This Bachelo hesis would ha e no been able wi hou he in ol emen o se e al people ha I would like o ake he ime o hank. Fi s ly, I would like o hank my supe iso , Pa le a Knu sson, o he commi ed guidance and ad ice du ing each one o he s eps o he de elopmen o his epo and o pushing me o achie e my goals. I would also like o hank Robin Faus o his help du ing his p ojec , o eaching me i s -hand he echniques used in he expe imen al pa o he epo , as well as o his company in he labo a o y du ing ha phase o he p ojec . Apa om hei dedica ed assis ance in he p ojec , I also app ecia e hei ca ing help and counselling o e he mon hs I ha e been in Sweden as a newcome . Then, I would like o hank my examine , B i -Ma ie S eena i, o he aluable ad ice o e he las s eps o he elabo a ion o he epo , which has ensu ed a highe s anda d o he quali y o i . My g a i ude also goes owa ds F ed ik Lind and Pa ick Moldenhaue , who p o ided in o ma ion ega ding he magne ic suscep ibili y measu emen s as well as impo an in o ma ion o he indus ial campaign he samples o which ha e been in es iga ed in his epo . In ha line, I would like o hank hem and Imp obed AB o he samples ha we e analysed as well as And eas Schae e o unning he XRF. Las ly, I would like o show my g a i ude owa ds my amily o hei con inuous lo e and suppo h oughou his jou ney. Ignacio Lama ca, Go henbu g, June 2021 ix Con en 1. In oduc ion ....................................................................................................................... 1 1.1 Fluidized bed echnology .......................................................................................... 2 1.2 Oxygen Ca ie s ........................................................................................................ 3 1.3 Ilmeni e...................................................................................................................... 4 1.4 Biomass ash ............................................................................................................... 5 1.5 Reac ions ................................................................................................................... 6 1.6 Alkali up ake and consequences o boile ope a ion wi h OC .................................. 8 1.7 Magne ic sepa a ion ................................................................................................ 12 1.8 Fe magne ic species ................................................................................................. 16 2 Me hods ........................................................................................................................... 18 2.1 SEM-EDS ................................................................................................................ 18 2.2 X- ay Di ac ion (XRD) ......................................................................................... 20 2.3 Magne ic suscep ibili y measu emen s .................................................................... 21 2.4 X- ay Fluo escence (XRF) ...................................................................................... 23 3 Resul s ............................................................................................................................. 24 3.1 SEM-EDS esul s .................................................................................................... 25 3.2 XRD esul s ............................................................................................................. 39 3.3 Magne ic suscep ibili y measu emen s esul s ........................................................ 42 3.4 XRF esul s .............................................................................................................. 43 4 Discussion ....................................................................................................................... 48 5 Fu u e wo k ..................................................................................................................... 53 6 Conclusions ..................................................................................................................... 55 Re e ences ............................................................................................................................... 57 Appendix I. Code used o he samples ................................................................................... 60 Appendix II. Speci ic me hodologies ollowed o SEM-EDS ............................................... 62 Appendix III. Addi ional Figu es o he laye analysis wi h SEM-EDS ................................ 63 Appendix IV. Addi ional Figu es o he XRD scans ............................................................. 64 Appendix V. Addi ional Figu es o he XRF esul s ............................................................. 65 4 combus ion p ocesses. The wo eac o s, he Ai Reac o and he Fuel Reac o , a e luidized bed eac o s whe e oxygen ca ie s a e used as bed ma e ials. Wi h some OCs, a Chemical-Looping wi h Oxygen Uncoupling (CLOU) mechanism is also p esen along wi h egula CLC. CLOU is based on he same p inciple as CLC bu he e is an O2 elease om he oxygen ca ie o he gas phase. The O2 eleased goes on o eac wi h he uel, esul ing in as e oxida ion han wi h he egula he e ogeneous eac ion be ween he uel and he solid OC. Howe e , ilmeni e is conside ed a non-CLOU OC so he CLOU p ocess will no be u he discussed in his epo [12]. In OCAC he OC is no s ic ly subjec o an oxidizing o educing a mosphe e like in CLC, as i is used o eplace he bed ma e ial o a con en ional CFB. Thus, he oxida ion and he educ ion do no ake place in sepa a e eac o s and he boile p esen s milde local educing and oxidizing en i onmen s compa ed o CLC. Tha is why he OC has o be eac i e in con inuously a ying oxidizing/ educing condi ions. An oxygen ca ie can ei he be syn he ically manu ac u ed o i can be ex ac ed om a na u al o e o om an indus ial was e s eam. A na u al oxygen ca ie is ound di ec ly in he na u e and does no equi e ex ensi e ea men apa om c ushing o sie ing. Manu ac u ed ma e ials a e based on na u al ma e ials ha a e subjec ed o ea men s like, o example, inco po a ion o addi i es and hea ea men among o he s. Manu ac u ed o syn he ic ma e ials end o be mo e expensi e han na u al/was e ma e ials. Na u al and was e s eam ma e ials a e inc easing in impo ance o use as OCs. This can be con i med by he da a published by Lyng el e al. [12], ha shows ha he e was a 68% inc ease in he ope a ion ime o na u al o es and was e ma e ials as OCs du ing he yea s 2014-2018. Tha means ha a he end o 2018 he sha e o use o hese OCs had doubled i compa ed o he sha e in 2014. This esea ch akes in o accoun a o al o 11338 hou s o ope a ion in o al sp ead o e 212 publica ions [12]. Ilmeni e is one o he mos widely s udied and used OCs wi h solid uels and i is indeed he mos ecu en OC inside he na u al/was e ca ego y [9, 11]. In OCAC, he OC can be deac i a ed and pa o he ma e ial can be los as ines due o a i ion, which means ha pa o he bed needs o be egene a ed ei he wi h new bed ma e ial o eci cula ed bed ma e ial ha has unde gone a ea men o sepa a ion. This leads o gene a ing was e ma e ial lows and inc eases he o e all cos o he p ocess i expensi e OC ma e ials a e used as bed ma e ial. Tha is why he a ge o OC esea ch and ope a ion a e al eady exis ing low-cos ma e ials a ailable in la ge quan i ies. Ilmeni e ul ils he equi emen s o a good OC and ha jus i ies he inc ease in esea ch s udies in ol ing his o e du ing ecen yea s [11]. 1.3 Ilmeni e Ilmeni e is a na u ally occu ing mine al composed o he ilmeni e phase FeTiO3 which can be oxidized o Fe2TiO5 (pseudob ooki e). Ilmeni e anspo s oxygen by changing he deg ee o oxida ion o i on om 3+, as in pseudob ooki e, o 2+, as i is in ilmeni e. While unde going his change i oxidizes he uel in he eac o . Because o his edox eac ion, Fe is conside ed o be he ac i e oxygen ans e ing elemen while Ti is no ac i e in he oxygen ans e [12, 9]. Ilmeni e exis s in he o m o sand o ock ilmeni e. Bo h sha e he same chemical composi ion ( he main phase is FeTiO3) and he main di e ence is he pa icle size and mo phology. Sand ilmeni e pa icles ha e smoo h ounded shapes due o being exposed o e osion, a i ion and o he kinds o na u al wea he ing. Ins ead, ock ilmeni e pa icles ha e sha p edges because he ock comes om he di ec mining and c ushing o he o e. Sand ilmeni e some imes shows a highe con en o TiO2 due o na u al oxida ion and dissolu ion o Fe du ing wea he ing [15]. 5 Usually, i is mo e common o ind ock ilmeni e han sand ilmeni e in OC applica ions because i o en has a highe oxygen ans e ing capaci y. Tha comes om he ac ha i is common o ind mo e Fe in he ock ilmeni e pa icles han in he sand coun e pa [11]. Fuels also ake a huge pa in how he combus ion p ocess wi h an OC de elops. Among all ypes o uel, gaseous uels a e he mos con enien ega ding hei handling and homogenei y. These ypes o uels can eac di ec ly wi h he OC in a as gas-solid eac ion and can also ac as a luidizing agen . One example o gaseous uels is syngas. Fo he solid uels, in o de o eac wi h he OC, he ola ile pa o he solid uel has o be eleased. The eac ion ha akes place is also a gas-solid eac ion bu he addi ional s ep o ola iliza ion adds complexi y o he p ocess. The emaining solid cha s go h ough a simila p ocess as he solid uels ha need o be combus ed [11]. 1.4 Biomass ash Du ing he combus ion o he uel, bed ma e ial and lue gases a e no he only species p esen in he eac o . Ashes a e he emaining ac ions om he uel a e combus ion. Biomass ash is a complex he e ogeneous mix u e o mainly ypes o ino ganic ma e bu also o a lesse deg ee o ganic ma e . In he boile he e a e wo ypes o ashes - ly and bo om ashes. Fly ashes consis o he ine pa icula es le a e uel con e sion ha lea e om he uppe pa o he boile en ained by he lue gases. They a e usually sepa a ed by he cyclone and he il e and can p o oke co osion and ouling i hey ge deposi ed on he hea equipmen ha is pa o he downs eam p ocessing o he lue gases [11]. Bo om ashes consis o bed ma e ial and ash componen s ha s ay in he dense pa o he bed and hey a e he main con ibu o s o bed agglome a ion and de luidiza ion. The main bo om ash componen s a e silicon, alkali (mos ly K and Na) and alkaline ea h (mos ly Ca and Mg) in he o m o oxides, ca bona es, chlo ides, hyd oxides and sulpha es. Ashes can also inco po a e elemen s such as aluminium, i anium, sulphu o phospho us. The composi ion o biomass ashes depends on h ee ac o s: he ype o biomass and condi ions in which he biomass esou ce is ob ained and s o ed, he condi ions in which he combus ion akes place and bo h he anspo and s o age condi ions o he ashes. I is widely accep ed ha wood and woody biomass yields he minimum amoun o ashes compa ed o he o he g oups, being animal and was e lows like municipal solid was e he esou ces ha ha e a highe ash yield (i can be up o 50-70%) [10]. When e e ing o biomass o i s ashes, elemen al concen a ions a e a aluable measu e o he composi ion o a sample. Depending on he concen a ion o a ce ain elemen , i can be classi ied as a majo (>1%), mino (1-0.1%) and ace (<0.1%) elemen in he sample. Following his classi ica ion, ypically he majo elemen s in biomass a e C>O>H>N>Ca>K and he mino elemen s a e Si>Mg>Al>S>Fe>P>Cl>Na. O cou se, he ash composi ion a ies depending on he sou ce o he biomass bu ha classi ica ion is gene ally accep ed, along wi h he inclusion o Mn and Ti as mino elemen s. The composi ion o he biomass ashes a ies depending on he ype o biomass ha was used o combus ion bu i has been ound ha wood and woody biomass shows a low concen a ion o Cl, K, Na and S while being ich in Ca [10]. Ilmeni e eac s wi h he main ash elemen s, K and Ca, o ming new compounds ei he while di using in o he bed ma e ial pa icles (such as e.g. KTi8O16) o while s aying a ached o he pa icle su ace (such as e.g. Ca(Ti0.7Fe0.3)O2.5) [16]. The o ma ion o hese compounds can lead 6 o K cap u e which p e en s he o ma ion o gaseous K compounds ha can co ode he hea exchange equipmen p esen in he downs eam p ocessing o he lue gases. Tha ype o co osion leads o ex ensi e main enance cos s. The e o e, i ansla es in o a g ea loss o bo h p oduc ion and money. The mechanism o eac ion o o m he lis ed i ana e compounds is ha d o assess and emains unclea . 1.5 Reac ions The main ole o he OC in an OCAC applica ion is o ans e oxygen om he oxygen ich si es o he oxygen lean si es o he eac o . The chemical mechanism behind ha is based on edox eac ions be ween he oxygen ca ie and i s en i onmen . 1.5.1 Ilmeni e eac ions wi h o ganic cons i uen s om he uel The OC mus be i s oxidized in an oxygen ich loca ion (e.g. nea he ai inle o he combus ion chambe ) by ai : 𝑀𝑒𝑥𝑂𝑦−1 +1 2𝑂2→ 𝑀𝑒𝑥𝑂𝑦 (R1) In he case o ilmeni e, he oxida ion occu s ollowing he nex eac ion: 2𝐹𝑒𝑂 · 𝑇𝑖𝑂2(𝑠)+1 2𝑂2(𝑔)→𝐹𝑒2𝑂3·𝑇𝑖𝑂2(𝑠) (R2) And hen, he OC is educed a an oxygen lean loca ion (e.g. nea he uel inle ). The uel will eac o o m ola ile compounds and cha ( hese las o ganic compounds a e also con e ed o CO) [11, 14]. 𝐵𝑖𝑜𝑚𝑎𝑠𝑠 (𝐶𝑎𝐻𝑏𝑂𝑐)→ 𝑣𝑜𝑙𝑎𝑡𝑖𝑙𝑒𝑠 (𝐶𝑖𝐻𝑗)+ 𝑐ℎ𝑎𝑟 (𝐶𝑥𝐻𝑦𝑂𝑧) (R3) 𝑐ℎ𝑎𝑟(𝑠) + 𝐻2𝑂(𝑔) → 𝐻2(𝑔) + 𝐶𝑂(𝑔) (R4) 𝑐ℎ𝑎𝑟(𝑠) + 𝐶𝑂2(𝑔) → 2𝐶𝑂(𝑔) (R5) Vola iles a e in he o m o nCiHj gaseous hyd oca bons and along wi h CO hey eac wi h he OC in a edox eac ion ha yields CO2. Du ing ha eac ion he OC is educed. 𝐻2(𝑔) + 𝐶𝑂(𝑔) + 𝑛𝐶𝑖𝐻𝑗(𝑔) + 𝑀𝑒𝑥𝑂𝑦(𝑠) → 𝐶𝑂2(𝑔) + 𝐻2𝑂(𝑔) + 𝑀𝑒𝑥𝑂𝑦−1(s) (R6) 7 Du ing his eac ion he OC eac s wi h he ola ilized hyd oca bons. Howe e , inside he eac o he e is also CO ha comes om he gasi ica ion o ola iles and cha s, and i can also yield CO2 when oxidized. Taking ilmeni e as he OC o choice, he eac ion should be he ollowing. 𝑥𝐹𝑒2𝑂3·𝑇𝑖𝑂2(𝑠)+ 𝑎𝐶𝑂(𝑔)+ 𝑏𝐶𝑖𝐻𝑗(𝑔) → 2𝑥𝐹𝑒𝑂 · 𝑇𝑖𝑂2(𝑠)+𝑐𝐶𝑂2(𝑔)+ 𝑑𝐻2𝑂(𝑔) (R7) Whe e pseudob ooki e o med in he oxygen ich sec ions is educed o ilmeni e again [14]. Du ing he oxida ion o ilmeni e, mul iple s udies ha e ound ha an Fe2O3 laye is o med on he su ace o he pa icles and ha he oxidizing condi ions p omo e he sepa a ion be ween Fe and Ti inside ilmeni e pa icles [18, 19]. This Fe oxide laye is o med as a esul o Fe2+ and Ti4+ di usion owa ds egions wi h high oxygen pa ial p essu es, such as he su ace o he pa icle. While Fe2+ and Ti4+ a e di using owa ds he su ace o he pa icle, oxygen molecules a e di using inside he pa icles. Acco ding o Knu sson e al. [18], Fe2+ di uses as e in FeTiO3 han Ti4+. When eaching he su ace, Fe2+ ions a e oxidized o Fe3+ as in Fe2O3 and ha is how he hema i e laye is de eloped. D. Rao e al. [19] s a e ha unde 770 deg ees Fe2O3 is o med on he su ace o he pa icle. Since O2 has o di use h ough Fe2O3 u he eac ion occu s a he Fe2O3-FeTiO3 in e ace. Then, be ween 770 deg ees and 900 deg ees Fe2Ti3O9 (pseudo u ile) is o med and i on di uses h ough i o he a ailable g ain bounda ies, c acks o he su ace, whe e he oxygen pa ial p essu e is highe [19]. Since du ing oxida ion a e en highe empe a u es pseudob ooki e (Fe2TiO5) is also o med, o p esen Fe-Ti seg ega ion he mobili y o he Fe ca ions h ough he s uc u e has o be high, due o he ac ha he Fe:Ti a io is highe o Fe2TiO5 han o FeTiO3. O he han he sugges ed mechanism, i is possible o ind o he mechanisms o his Fe-Ti seg ega ion in he li e a u e as i has been desc ibed in mul iple ways. The ollowing eac ion is a sugges ed eac ion o he o ma ion o pseudo u ile om ilmeni e: 2𝐹𝑒𝑂 · 𝑇𝑖𝑂2(𝑠)+ 𝑇𝑖𝑂2(𝑠)+1 2𝑂2(𝑔) → 𝐹𝑒2𝑇𝑖3𝑂9(𝑠) (R8) The compound o med is pseudo u ile. Pseudo u ile, acco ding o Fu e al. [20], has a sligh ly lowe la ice densi y han ilmeni e which could acili a e he mig a ion o Fe2+ h ough he s uc u e, as obse ed by Co co an e al. [21]. This allows mo e o ma ion o hema i e (Fe2O3) and he mig a ion o Fe con en ou wa ds he pa icle [21]. Recen s udies s a e ha i is p ecisely his Fe mig a ion ou wa ds he cause o he phase sepa a ion be ween Fe2O3 (ou e pa o he pa icle) and TiO2 (inne pa o he pa icle) [16]. The Fe mig a ion owa ds he su ace imp o es con ac wi h ai . Tha e ec , combined wi h he ac ha Fe is he ac i e elemen in oxygen ans e ing leads o an inc ease in oxygen ans e ing capaci y. Howe e , a i ion is happening wi h o wi hou Fe on he su ace. The p oblem wi h ha is ha while Fe mig a es owa ds he su ace, Fe a i ion also inc eases, educing he OC abili y as he ac i e elemen is los . This phenomenon educes he o e all li e ime o he OC pa icles. Fe mig a ion also inc eases he po osi y o he pa icles, inc easing he pa icle/ai eac ion su ace. Tha po osi y inc ease is only co ela ed wi h an inc ease in OC abili y in he cases whe e a i ion is no dominan and he e is s ill enough Fe in he s uc u e [11]. 8 Fe ends o mig a e o a eas wi h a high oxygen pa ial p essu e. In he case o sand ilmeni e, his mig a ion ends o be mo e in ense owa ds ce ain c acks and ca i ies o he pa icle, whe e s uc u al de ec s ac as passages o oxygen di usion. Howe e , his is no he expec ed beha iou o ock ilmeni e. Since ewe ca i ies a e o med in hese pa icles i is mo e common o see Fe mig a ing owa ds he su ace, as i is he nea es loca ion wi h a conside able oxygen po en ial [11]. Ano he possible scena io o he obse ed Fe-laye a he ou e mos su ace is ha a o med ou e ash laye ( he o ma ion o his laye will be u he explained la e on) can be wo n o due o a i ion and as a esul , he Fe laye is p esen on he su ace. An example o his has been obse ed in esea ch ca ied ou by S aničić e al. [22]. 1.5.2 Ilmeni e eac ions wi h ino ganic cons i uen s om he uel (ashes) Some o he mos impo an eac ions be ween ilmeni e and ash compounds a e he ones ha o m po assium i ana es inside he co e o he OC pa icles (ilmeni e pa icles) due o po assium di usion inwa ds. This eac ion happens due o in e ac ion o he TiO2 co e (le a e he Fe di usion ou wa ds in oxida ion condi ions ha has been explained p e iously) wi h po assium- based sal s o o he compounds such as po assium oxides. Fo ins ance, KCl has been chosen as he po assium sal o he sugges ed eac ion bu o he ones could ha e been chosen oo. 8𝑇𝑖𝑂2(𝑠)+ 𝐾𝐶𝑙(𝑠)→𝐾𝑇𝑖8𝑂16(𝑠)+1 2𝐶𝑙2(𝑔) (R9) When he K compounds in he combus ion chambe (e.g. KCl) a e no aken up by an addi i e hey can lea e he combus ion chambe wi h he lue gases. Once eleased, hey can be ha m ul o he hea exchanging equipmen so i is an ad an age i K is aken up by he OC. Thus, he ongoing esea ch ocusing on how o make ilmeni e abso b mo e K and he limi s o ha p ocess is bene icial o p e en ing co osion [14]. In a simila in e ac ion, Ca compounds can be adhe ed o he su ace o he s uc u e o he ilmeni e pa icle whe e eac ion akes place and Ca2+ is inco po a ed in o he i ana e s uc u e by eac ing wi h TiO2 o o m CaTiO3 [16]. The ollowing sugges ed eac ion desc ibes he o ma ion o CaTiO3 s a ing om CaO. 𝐶𝑎𝑂(𝑠)+𝑇𝑖𝑂2(𝑠)→ 𝐶𝑎𝑇𝑖𝑂3(𝑠) (R10) 1.6 Alkali up ake and consequences o boile ope a ion wi h OC Se e al s udies ha e documen ed how his heo e ical se o eac ions a ec he ac i i y, he mo phology and mechanical s abili y o he ilmeni e pa icles when ac ing as bed ma e ial and he chemical eac ions ha ilmeni e unde goes wi h he main ash compounds [12, 13, 16]. Acco ding o published da a, when exposed o combus ion condi ions, Fe mig a es owa ds he su ace o he pa icle, o ming an i on- ich laye . This laye is ei he a uni o m laye o i appea s as i on- ich clus e s o islands a loca ions whe e he oxygen pa ial p essu e is highe . Fo he same combus ion condi ions, Co co an e al. [16], ound ha he po assium and calcium compounds o he ashes eac wi h he ilmeni e pa icle o ming new compounds. On one hand, 9 po assium compounds o m wha is hough o be KTi8O16, a esul o K di using in o he pa icle co e and eac ing wi h TiO2. On he o he hand, a Ca- ich double laye is o med a ound he Fe- en iched laye ha can be ound on he pa icle. This double laye consis s o an ou e laye ha accumula es on he su ace o he pa icle and an inne laye ha p esen s a mo e homogeneous composi ion han he ou e one. The ou e Ca- ich laye inco po a es o he ash compounds (such as Si and P). The inne laye is he esul o he o ma ion o a new compound, Ca(Ti0.7Fe0.3)O2.5 (o CaTiO3 in sand ilmeni e), which is o med when Ca2+ mig a es inside he pa icle and eac s wi h he inne ilmeni e s uc u e. The o ma ion o his Ca double laye is p esen ed in a Co co an e al. esea ch [16]. In he s udy o Co co an e al. [16], ilmeni e was es ed in he Chalme s CFB esea ch boile as an oxygen ca ie in he bed eac o . The uel used was mainly wood chips, he mean ope a ion empe a u e was 850 °C and he es wen on o a span o 6 weeks. A e ha pe iod, he pa icles wi h a highe esidence ime we e chosen due o heo e ically being he mos ep esen a i e o he deg ee o in e ac ion wi h ash compounds [16]. To s udy agglome a ion and he mechanical s abili y o he pa icles, SEM-EDX and XRD analysis we e pe o med on hese pa icles. A shi owa ds la ge pa icle diame e s and a educ ion in he numbe o he ines pa icles was ound a e he expe imen . The i s e ec obse ed a e he expe imen could be caused by laye o ma ion, an inc ease in he pa icle’s po osi y (and he eby inc ease o he olume o he pa icles) and he inco po a ion o ash compounds. Ano he possibili y could ha e been he agglome a ion o pa icles, al hough agglome a ion was no isually obse ed. The educ ion o he numbe o ine pa icles could be explained by he en aining o he ines pa icles in he lue gases. When i comes o he mo phology o he pa icles, i was obse ed ha sand ilmeni e pa icles had inc eased he numbe o ca i ies and had unde gone clus e and c acks o ma ion. I was also no ed ha b igh spo s appea ed on he pa icles’ su aces. These spo s we e ich in i on while he emaining da ke spo s we e simila o he composi ion o he ini ial esh ilmeni e. I on- ich clus e s ended o appea on he su ace, o en close o c acks. The heo y ha explains his ac is ha he ca i ies a e sho passages o oxygen di usion and ha inc eases he oxygen po en ial locally [16]. When no ca i ies we e o med (a ea ly s ages o exposu e) EDS e ealed ha a laye had o med. Tha laye had a simila composi ion o ha o he b igh spo s on he su ace ( ich in i on). The inne pa o he pa icles had also a simila composi ion o ha o he ini ial esh ilmeni e. Smoo he pa s o he su ace we e obse ed o be ich in magnesium and calcium and i was no ed ha hese a eas had become poo e in i on and i anium compa ed o esh ilmeni e [16]. 10 Figu e 1. Backsca e ed Elec on Mic og aphs o ilmeni e pa icles used as bed ma e ial (51, 147 and 364 h). Fe ich smoo h clus e s can be seen in he 147h op ow o e iew while he Fe ich laye can be seen in he 147h bo om ow c oss-sec ion o he pa icle [16]. Wi h wha had been obse ed, some conclusions we e d awn ega ding he kine ics o he ash – ilmeni e in e ac ions and he di usion o he laye s and hei hickness. Co co an e al. [16] obse ed ha an ou e Ca- ich laye is o med on he su ace o he pa icle, and i accumula es wi h ime, he eby becoming hicke . The concen a ion o Ca in he laye does no change signi ican ly bu he hickness o he laye does. Also, Ca2+ mig a es inwa ds h ough he i ana e s uc u e, esul ing in an inne laye ha can be dis inguished om he ou e laye . Th ee hypo heses we e o mula ed o explain he o ma ion o he inne Ca-en iched laye : 1) Ca is deposi ed om he ash on he su ace while he pa icle g ows in diame e , emaining “locked” inside he ash laye ; 2) Ca is deposi ed om he ash and mig a es inwa ds he pa icle and 3) Ca exis ing in he esh ilmeni e mig a es ou wa ds. The expec ed main sou ce o Ca is he ash because o he hickness o he su ace laye . This expec a ion does no align well wi h he hi d hypo hesis. While he mig a ion inwa ds could explain he p esence o he in e nal laye s, i does no align well wi h he ac ha P and Si we e ound a he ex e nal pa s o he Ca laye . These elemen s (P and Si) would mig a e inwa ds a di e en a es. This las ac was no p o en due o no being he main pu pose o he in es iga ion [16]. P and Si compounds a e no o iginally ound on esh ilmeni e pa icles. Thus, he inding o P and Si compounds in he laye sugges ed ha bo h elemen s we e deposi ed as ash componen s di ec ly ou side o he pa icle. Finally, i was sugges ed a mechanism ha combines hypo hesis 1) and hypo hesis 2). In his mechanism, Ca compounds o m a laye ou side o he pa icle ha g ows ou wa ds and also Ca mig a es inwa ds, eac ing wi h TiO2 om he in e nal s uc u e o he pa icles and o ming CaTiO3 o simila Ca- i ana es [16]. 11 Figu e 2. Illus a ion o he g ow h o he Ca laye in ilmeni e pa icles and i s mig a ion inwa ds a e se e al ope a ion hou s (27 h, 147 h and 364 h om le o igh espec i ely). Images ob ained as Ca EDS-in ensi y maps [16]. A simila analysis o K showed ha i was no p esen in he Ca su ace laye . Howe e , i also o med ou side o he pa icle and mig a ed inwa ds, un il eaching he limi o 2% in a omic composi ion in he pa icle. Tha limi was he K up ake limi a ion in he exac condi ions o ha expe imen . Acco ding o expe imen al esul s, he pa icle was conside ed o be ich in po assium (and he e o e ha ing eached he limi ) a e 83 hou s o ope a ion. Figu e 3. Illus a ion o he mig a ion o K in ilmeni e pa icles and i s mig a ion inwa ds a e se e al ope a ion hou s (27 h, 51 h and 147 h om le o igh espec i ely). Images ob ained as K EDS-in ensi y maps [16]. Al hough i was no he goal o he esea ch, i was sugges ed ha c acks could be spo s o mig a ion inwa ds o Ca and K compounds. The main di e ence be ween he Ca and he K compounds o med on he su ace is ha Ca compounds end o accumula e on he su ace and hen mig a e inwa ds wi h ime, while K compounds do no only accumula e on he su ace. I is ega ded ha he mig a ion o K inwa ds is as e han Ca mig a ion, al hough he mig a ion o K is p omo ed by he p esence o Ca on he su ace [16]. Bo h K and Ca eac wi h TiO2 o o m K2Ti3O7/ K2Ti6O13 and CaTiO3 bu his las compound is he mos s able among he h ee. This s abili y educes he ee TiO2 on he su ace and p omo es he mig a ion o K inwa ds [16]. Ano he s udy by Co co an e al. [21] also cha ac e ized he shi in composi ion om ilmeni e when in con ac wi h bo om ashes du ing OCAC ope a ion. In ha s udy, some samples we e analysed be o e, du ing and a e OCAC ope a ion wi h bed ma e ial eplacemen wi h ilmeni e. Fi s , 20% o he silica sand was eplaced wi h ilmeni e and a e wo days o ope a ion, app oxima ely 40% o he bed ma e ial was ilmeni e. Elemen al analysis o he samples ob ained show how he Fe weigh % diminishes wi h ime while Ti weigh % s ays qui e cons an and Ca, K, Si, P and Mn weigh % inc ease [21]. 12 Ano he in e es ing iew is he global s andpoin ega ding wha happens o he weigh o ilmeni e du ing ope a ion, and how ha can be e lec ed by he obse ed di e ences in weigh be o e and a e i s use as bed ma e ial. F om his s andpoin , Hildo e al. [14], obse ed wi h he help o TGA ha when an ilmeni e sample was oxida ed i gained weigh whe eas in educing condi ions pa o he weigh was los . In ac , he a e a which weigh was gained in oxida ing condi ion was as e han he a e o losing weigh in educing condi ions [14]. I is expec ed, hen, o see a weigh gain when ope a ing wi h ilmeni e as he OC and bed ma e ial a he same ime. This weigh gain, hough, is expec ed o be slowed down due o he g ea loss o oxygen and sal s such as sulphu sal s and KOH wi h he lue gases. The amoun o weigh loss expec ed a ies depending on he educing condi ions o he eac o and he d yness o he in e nal en i onmen . Rega ding s ic ly he bed ma e ial and conside ing he inle and ou le s eams o he boile , i is expec ed o obse e a weigh gain due o ash laye build-up. Especially in he p esence o s eam, KOH is o med and e apo a ed, hus p o oking ewe in e ac ions be ween ash K compounds and ilmeni e pa icles. Unde hese condi ions, hen, i is expec ed less K % in he sample han in absence o mois u e. Rega ding S compounds ha can be ans e ed on o he ilmeni e, a weigh loss was expec ed unde we and d y condi ions due o he apo iza ion as SO2 o H2S [14]. 1.7 Magne ic sepa a ion Since in he middle o he 18 h cen u y Michael Fa aday disco e ed ha a subs ance in he way o a magne ic ield inc eases o dec eases ( o some ex en ) he lux densi y o he magne ic ield (B) when passing h ough i , a b oad ange o applica ions ha s em om his phenomenon ha e been disco e ed and applied. Magne ic sepa a ion is one o hose applica ions. I is based in he ac ha ma e ials can be sepa a ed based on hei di e en magne ic p ope ies. The i s widely ex ended comme cial applica ion o he p ocess came in he la e 1860s, when i was in oduced in he sepa a ion o i on om b ass [23]. This p ocess, howe e , has been used in many indus ies since hen. Some examples a e he magne ic aps om ood indus y ha exclude me al o eign bodies ha could be a ood sa e y haza d om ce ain ood p oduc s, he magne ic sepa a o s used o was e managemen in ecycling cen es o i s use in was ewa e ea men in s eel indus y and in powe plan s [24]. The e is a wide a ie y o magne ic sepa a o s in he di e en indus ies ha equi e his echnique, bu he mos common sepa a o s a e he ollowing ones: • Magne ic pulleys: Elonga ed cylinde s suppo ed by a sha ha cons i u es he head pulley o con eyo s ha anspo ma e ial in he plan . The pulleys ha e magne s ins alled o a ached o hem so ha hey p o ide a cons an magne ic ield a ound he whole ci cum e ence. • Suspended magne s: Pe manen magne s o elec omagne s pu posely suspended o e a con eyo bel o a eede ha anspo s ma e ial. • Magne ic d ums: A magne assembly is buil inside a o a ing d um cylinde ha d aws he magne ic pa icles on o i s su ace and ca ies hem ou side o he ange o in luence o he magne . • Pla e magne s: Magne s designed in he o m o pla es a e placed down a chu e om which ma e ial alls. The magne ically sepa a ed ma e ial a ached o he pla e has o be emo ed pe iodically om i s su ace. 13 • G a e magne s: Se e al magne ic ods o ubes a e disposed inside a suppo ame h ough which he ma e ial is supposed o low ( he ma e ial is usually ee alling owa ds he magne ) [23]. Figu e 4. Illus a ion o con en ional indus ial magne ic sepa a o s: 1. Magne ic pulley; 2. Suspended magne s; 3. Magne ic d ums; 4. Pla e magne ; 5. G a e magne s. A anged om he o iginal sou ce [23]. Magne ic pulleys o bel sepa a o s a e ecu en in many indus ies because hese magne s a e easy o ins all, ha e a low ini ial cos and a low cos o ope a ion (i he magne s a e pe manen , such as a e ea h ones), and a e able o accomplish con inuous and au oma ic emo al o magne ic ma e ials. Howe e , hese sepa a o s a e s ill a highly empi ical echnology. The aim o he sepa a ion is o ob ain he igh quali y o he concen a e wi h he igh eco e y, and ha is ixed by he use wi h an adjus able non-magne ic spli e [25]. The d i ing o ce in he sepa a ion is he magne ic o ce (Fm), ha compe es wi h ex e nal o ces such as g a i a ional o ce (Fg), he hyd odynamic d ag o ce (Fd), ac ion o ce (F ) and ine ia (Fi), depending on he ype o he ope a ion. Wha e e he dynamic sys em is o he p ocess, i needs o be made su e ha he magne ic o ce o he magne ic ma e ials ha need o be eco e ed is highe han he compe ing o ces. Else, i will all wi h he non-magne ic ma e ial. In o de o do ha , se e al hings need o be adjus ed, such as he pulley e olu ions in a magne ic pulley, as o e ly high e olu ions could cause an unwan ed cen i ugal e ec ha diminishes he pe o mance o he sepa a o . Pa icle size also has a lo o do wi h he deg ee o eco e y achie ed. Usually, he magne ic pulleys a e used in sys ems ha ha e pa icles o a wide a ie y o diame e s and ha is why he spli e be ween magne ic ac ion and non-magne ic ac ion is adjus able [20, 21]. In indus ial-scale boile s such as he K a ingen’s 115 MW h CFB boile loca ed in Ö o a, i is common o ex ac samples o he bed ma e ial and magne ically sepa a e hem o in oduce he magne ic ac ion again in he boile . This is handled by a oll bel magne loca ed a ound wo cylinde s ha o a e con inuously o displace he en i ies placed on he bel . The magne ic sepa a o ac s as a con eyo bel o he samples un il hey app oach one o he ex emes, whe e 20 2.2 X- ay Di ac ion (XRD) The p inciple o X- ay di ac ion (XRD) is based on he di e en c ys al s uc u es o he subs ances and how hey di ac X- ay wa es. Due o he di e en dis ibu ion o a oms in he cells o he c ys al o mine als such as, o example, ilmeni e, he inciden X- ay wa e can be di ac ed in o many di e en di ec ions. The angles and in ensi y o he esul ing adia ion can be measu ed by a de ec o whe e he signal, in e ms o in ensi y s. angle o di ac ion, o ms a di ac og am om which in o ma ion abou he s uc u e and dis ibu ion o he a oms in he c ys al s uc u e can be ex ac ed. The undamen als behind he X- ay di ac ion a e ha X- ays, which ep esen elec omagne ic adia ion wi h wa eleng hs in he nanome ic scale, can in e ac wi h he elec ons o he ma e ial. In e e ence occu s when he sepa a ion o he a oms o he sample is compa able in size o he wa eleng h o he X- ays. The in e e ence is conside ed cons uc i e when he wo X- ay di ac ed wa es a e in phase wi h each o he . When ha happens, he ampli ude o he wa e ha esul s adds up while he wa eleng h s ays he same. Else, i he wa es a e no in phase, he in e ac ion is conside ed des uc i e o pa ially des uc i e and he ou come is an X- ay wa e ha has a diminished ampli ude compa ed o he o he o iginal wo di ac ed wa es. In ac , i he in e e ence is ully des uc i e he ampli ude o he wa es ha esul om he in e ac ion is null. Figu e 7. Illus a ion o he di ac ion o X- ays in wo planes o a c ys allog aphic s uc u e. The ype o in e e ence is desc ibed by he B agg’s law. B agg’s law is as ollows: 𝑛 · 𝜆 = 2 · 𝑑 · sin (𝜃) whe e n is an in ege (1, 2, 3 …) and i is he o de o e lec ion o he sca e ed wa e, λ is he wa eleng h o he inciden X- ay beam, d is he dis ance be ween wo consecu i e planes o he molecula la ice and θ is he inciden angle o he X- ay. I he di ac ed wa e ollows B agg’s law he in e e ence is cons uc i e. The classic ins umen a ion o XRD is an X- ay sou ce ( ends o be an X- ay ube) ha emi s ocused X- ay beams owa ds he sample a a θ inciden angle. Tha angle is inc eased wi h ime so he de ec o ecei es a wide spec um o sca e ed wa es based on he angle o incidence o he sou ce beam. The sca e ed wa es can be analysed by a ansduce ha coun s pho ons. The di ac ion pa e n can be compa ed o he di ac ion pa e n o e e ence ma e ials a ailable in a da abase. Based 21 on he coincidence o di ac ion pa e ns, he c ys alline compounds p esen in he sample can be iden i ied [35]. The in o ma ion ob ained by XRD is he dis ibu ion and dis ances be ween he a oms inside he c ys als and he leng h and ype o hei bonds, as well as he p esence o any impu i y in he sample. I gi es in o ma ion abou he o ganiza ion o packing o he a oms, which allows o ma ch he iden i ied s uc u es o such ha ha e a simila o ganiza ion o a oms. No mally, hese pa e ns a e compa ed o hose o a da abase and he s uc u es a e iden i ied based on he compa ison be ween he main peaks o he pa e ns and he index o adjus men o he da a om he da abase. XRD was pe o med in a Siemens D8 di ac ome e . The samples we e g ound p io o he analysis and he pa e ns we e compa ed o he da abase “C ys allog aphy Open Da abase – REV212673” ha he so wa e DIFFRAC.EVA p o ides. 2.3 Magne ic suscep ibili y measu emen s In his epo , an analysis o he magne ic suscep ibili y o he sample was done by a Ba ing on MS2B magne ic suscep ibili y senso . This senso is connec ed o he MS3 Ba ing on de ice ha connec s o a compu e in o de o ans e he da a o he Ba so compu e p og am. The ela ion be ween he pe meabili y o ai ( he senso app oxima es i o he pe meabili y o he oid) and he pe meabili y o a specimen can gi e he alue o he bulk magne ic suscep ibili y o a sample, acco ding o he p e iously in oduced Equa ion 3. The MS2B sys em measu emen elies on he p inciple ha any change in he pe meabili y o a co e causes a change o induc ance in a wound induc o . Ba ing on senso s c ea e a weak magne ic ield om an al e na ing cu en (AC) and he equency o oscilla ion is de e mined by he induc ance o he sys em. A he same ime, he induc ance is dic a ed by he pe meabili y o he sys em. When he e is no sample in he senso , he magne ic pe meabili y o ai , which he senso app oxima es o he acuum pe meabili y (μ0), de e mines he induc ance, and a he same ime he equency o oscilla ion o he al e na ing cu en . When he sample is placed, he e is a change in he oscilla ion o he al e na ing cu en ha he senso uses o calcula e he change in induc ance and he e o e he di e ence in magne ic pe meabili y. Wi h ha alue, he magne ic suscep ibili y can be calcula ed acco ding o Equa ion 3. This de ice can measu e suscep ibili ies o samples ha a e ca e ully placed in 12 ml. con aine s. The MS2B Senso in pa icula , is a Dual F equency Senso , which means ha i can measu e he suscep ibili y a wo di e en equencies o he gene a ed magne ic ield. Thus, i has a Low F equency (LF) mode (0.46 kHz) and a High F equency (HF) mode (4.6 kHz). The esul s in HF mode a e used o compa ing wi h he LF esul s and i hey do no ma ch, i is an indica o o ul a ine e imagne ic pa icles wi h a diame e sho e han 0.03 μm. This is no expec ed in he samples analysed so he esul s ha e only been aken in LF mode. Since he e is a wide a ie y o magne ic suscep ibili y senso s and some senso s call o di e en me hodologies, he me hodology ollowed o he suscep ibili y measu emen s in he p esen epo is u he explained: 1. Fi s , a d i es is conduc ed o 15 minu es wi h he magne ic suscep ibili y senso . This wa ms up he senso s and p epa es hem o he measu emen s. 2. The calib a ion sample is measu ed o check whe he he senso is ob aining measu emen s ha a e exac enough. The measu ed alue is compa ed o he nominal alue p o ided by he manu ac u e o he senso .* 22 3. The 12 ml. emp y con aine is weigh ed in an analy ic balance and he measu emen is w i en down. 4. The suscep ibili y o he emp y con aine is measu ed be o e s a ing o measu e a di e en sample.* The eason behind i is ha some dus om o he samples always accumula es inside and in o de o make su e ha i is no in e e ing o ouling he ollowing measu emen s i has o be checked ha i s suscep ibili y is se e al o de s o magni ude lowe han ha o he sample. 5. Du ing he anspo o he samples o Chalme s uni e si y om Ö o a, seg ega ion be ween hea y and ligh elemen s/compounds migh ha e occu ed. I is possible ha he hea y compounds end o dis ibu e nea he bo om o he sample bo le whe eas he ligh e compounds may loa o he op o he bo le. To p e en ha , be o e ex ac ing he sample om i s bo le i is s i ed so ha a ep esen a i e amoun can be aken. 6. The con aine is illed up wi h he sample ha is going o be measu ed. I mus be illed up wi h sample comple ely un il i can be sc aped o he op o he po wi h a spa ula. Since he ollowing weigh measu emen will be used o calcula ing he pou ed bulk densi y** i has o be done wi h cau ion and wi hou apping he con aine by acciden . 7. The po con aining he sample is weigh ed again in an analy ical balance. The mass o he sample ha con ains can be ob ained by sub ac ing he weigh o he emp y po o his las weigh measu emen . 8. An Indi idual Tes is selec ed in he Ba so so wa e. 9. The op ions selec ed o he measu emen a e a 5 second Blank es ollowed by a 10 second sample measu emen and ano he 5 second Blank es . Fo he Blank es no sample should be placed in he suscep ibili y senso and only a e he i s Blank measu emen he sample should be placed inside he MS2B senso . 10. S eps 5 and 6 a e epea ed h ee imes in o de o ha e mo e han one alue and see he de ia ion o he measu emen s. * The s eps ha a e ma ked wi h his symbol ollow he same suscep ibili y measu emen s ha ollow s eps 8 and 9. ** The pou ed bulk densi y e e s o he densi y o he sample ha is ob ained by in oducing he sample inside o he con aine wi hou mo ing o apping i . I i is apped, he ee space p esen be ween he pa icles will be illed wi h o he pa icles and hen he olume dec eases. This me hod was chosen due o i s quickness, ep oducibili y and sca ce equi emen s o equipmen ha makes i sui able o he indus ial s age o measu emen s. The alue ob ained wi h he senso is he olume ic magne ic suscep ibili y o he bulk o he sample. Howe e , no e e y sample has he same mass and no e e y measu emen p esen in he bibliog aphy is measu ed in 12 ml con aine s. Since he olume ic magne ic suscep ibili y alue depends on he mass and olume o he sample measu ed, i has o be di ided by he bulk densi y o he sample in o de o be compa able wi h o he sample measu emen s and wi h da a om he bibliog aphy. In he p esen epo , he alue o he densi y co esponds o he pou ed bulk densi y. I calcula ed ollowing Equa ion 4, he mass speci ic magne ic suscep ibili y is ob ained: 𝜒𝑚𝑎𝑠𝑠 =𝜒 𝜌 (𝑚3 𝑘𝑔) (E4) Whe e χmass is he mass speci ic magne ic suscep ibili y, χ he magne ic suscep ibili y and ρ is he bulk densi y o he sample ha is measu ed. Fo con enience, om his poin on, in he epo he 23 “mass speci ic magne ic suscep ibili y” will be e e ed as “magne ic suscep ibili y” as i is he alue ha will be compa ed. 2.4 X- ay Fluo escence (XRF) X- ay luo escence (XRF) is a non-des uc i e analy ical echnique used o he quan i ica ion o he bulk elemen al composi ion o ma e ials. The p inciple behind i s ems om he use o p ima y high-ene gy X- ays o gamma ays o exci e he co e-le el elec ons om he sample owa ds an exci ed s a e. The hole le by ha exci ed elec on will be e en ually illed by an elec on om a highe ene gy s a e ha decays and ha esul s in he emission o a cha ac e is ic seconda y X- ay. These emi ed seconda y X- ays (also known as luo escen X- ays) ha e a lowe ene gy han he inciden p ima y ones. The ene gy o he seconda y X- ays is cha ac e is ic o e e y elemen as i gi es in o ma ion abou he ene gy le els o he a omic species and hei di e ences in ene gy [36]. Figu e 8. Illus a ion o he inciden X- ay adia ion hi ing he a om and he esul an X- ay luo escence emission [37]. An X- ay de ec o can con e he X- ay ene gy in o a ol age ha can be p ocessed and hen a spec og am can be ob ained om i . The me hod allows o quan i ica ion o he elemen al composi ion in he sample since he in ensi y o he seconda y X- ays is p opo ional o he abundance o he elemen s in he ma e ial [38]. The undamen als behind he echnique a e simila o SEM-EDS, bu he inciden adia ion di ec ed owa ds he sample in each echnique is di e en . SEM-EDS uses a ocused elec on beam while XRF uses sho wa eleng h X- ays. In some cases, bo h echniques e en use he same de ec o . Howe e , gene ally, XRF is able o gi e deepe in o ma ion since i ypically de ec s elemen s a he ppm le el, while EDS usually has a de ec ion limi a he 0.1% le el (i can a y depending on he elec on beam accele a ion ol ages applied) [39]. XRF allows he iden i ica ion o he di e en elemen s p esen in he bulk phase o a sample and hei quan i ica ion in a omic concen a ion. This way, i can accu a ely es ima e he composi ion o he sample. Wi h XRF i is possible o analyse he chemical composi ion o he bulk phase, no only speci ic spo s, lines o in ensi y maps as i happens wi h EDS. 24 XRF was pe o med wi h a PANaly ical Axios spec ome e and i was made su e ha he samples co e ed he whole bo om o he suppo so ha no holes we e p esen when he analysis ook place. To do so, a weigh o app oxima ely 4-5 g was in oduced in he sample holde . 3 Resul s In his sec ion, he esul s o he analy ic me hods and he magne ic suscep ibili y measu emen s a e p esen ed. The ou comes o each me hod a e p esen ed he e as well as some obse a ions while he connec ions be ween he esul s can be ound in he Discussion sec ion. The i s hing ha was done was c ea ing a code o name all he samples ha we e ini ially conside ed in an easy way. Mo e in o ma ion abou he code and i s meaning can be ound in Appendix I. SEM-EDS esul s will be p esen ed in he i s place, ollowed by XRD, magne ic suscep ibili y measu emen s and XRF espec i ely. I was decided o in es iga e abou he di e ences in he magne ically sepa a ed ac ions o an indus ial sample (Ö o a samples) and how ha can be ela ed o a unc ion o a ailable ee Fe and in e ac ions wi h he ash compounds. The main ques ions o in e es ha ollow ha analysis a e how his magne ic sepa a ion is handled in an indus ial size boile and i he p ocess o eci cula ion o ilmeni e can be imp o ed. Tha is why he e o s we e ocused on he bo om ash samples om K a ingen’s CFB boile loca ed in Ö o a (Skåne, Sweden) ha appea in he Appendix I. The analysed samples a e om a h ee-week campaign ha was un in K a ingen’s CFB boile om he 4 h o Ap il o he 22nd o Ap il o he yea 2018. The CFB boile has a nominal he mal capaci y o 115 MW h and i is no mally ope a ed wi h qua z sand bu in ha campaign, he bed ma e ial was eplaced o oxygen-ca ying ock ilmeni e ha comes om he Tellnes mine owned by Ti ania (in No way). The uel used was a mix o was e wood and wood chips. When analysing samples ha come om an indus ial boile se e al hings ha e o be aken in o accoun . The uel ha is ed a he Ö o a plan is e y he e ogeneous. Tha means ha i should be expec ed o ind a qui e la ge numbe o di e en elemen s in he SEM-EDS and XRF analysis. In indus ial acili ies he composi ion o he uel is a ying also om one ime poin o ano he and be ween he plan s. Also, when a change o uel o ope a ional condi ions is done, a change in he ac ual bed ends o happen wi h a delay o one o a ew days ( his delay is no homogeneous in ime). Tha same phenomenon happens i one models he boile as a CSTR, whe e a change in he eed p o okes a change in he in e nal condi ions o he eac o . Tha change can be ex ended in ime and one o he a iables ha dic a es how as /slow i appea s is he amoun o inle ma e ial compa ed o he ma e ial p esen in he ac ual eac o . In he case o he Ö o a boile , he amoun o ma e i con ains is qui e big as i has a capaci y o 60 ons o bed ma e ial unde no mal ope a ion. The loca ion in he combus o om whe e he samples o bo om ash a e ex ac ed can a y om one combus ion uni o ano he . Tha means ha he bo om ash may be ex ac ed om a ying loca ions in boile s o di e en CHP plan s. In he case o he Ö o a plan , i is ex ac ed om he bo om o he boile . The plan condi ions di e om one acili y o ano he and he e o e 25 conclusions should be made p edominan ly o he s udied case a Ö o a and da a should be ca e ully used o d aw mo e gene al conclusions. All o hese conside a ions imply ha he esul s ob ained should be ini ially ela ed o he Ö o a plan and speci ically o he campaign ha was held o he mon h o Ap il ( om he 4 h o Ap il o he 22nd o Ap il) o 2018. Tha does no necessa ily mean ha a gene al end o a ep esen a i e phenomenon canno be obse ed o pushed o wa d in o he powe plan s bu a he ha he limi a ion should be aken in o conside a ion. I would ha e been in e es ing o e alua e he ly ashes om he Ö o a combus o o check i he e is a i ion o Fe going on and i so, when does i s a . Howe e , he e we e no a ailable ly ash samples om Ö o a. A his poin , i was decided o use he a ailable esou ces and a simpli ica ion was p oposed. Since he a ailable ly ash samples we e om a campaign un on he Chalme s 12 MW h CFB boile du ing he win e o 2014, i was assumed ha he semi- indus ial-size boile can be simila o he ully indus ial-size acili y loca ed a Ö o a. The uel used in ha campaign was wood chips and hei composi ion can be checked in Co co an e al [21]. The indings unde his assump ion will need o be e isi ed in u u e wo k as he assump ion will no be p o en ue in his p ojec . This assump ion is done, hus, o explana o y pu poses. 3.1 SEM-EDS esul s SEM-EDS was used bo h o imaging and chemical analyses. Fo imaging, se e al mic og aphs o he su ace o he sample we e aken. They we e used o ollow he mo phology de elopmen o he di e en samples. Fo chemical analysis, EDS poin analysis we e pe o med on selec ed loca ions. All o he analysis esul s a e no malized on a C and O ee basis. This no malisa ion is equi ed because C and O a e common elemen s and hus hei quan i ica ion is no possible. The esul s ha e o be analysed, hen, wi hou aking hem in o accoun . Fi s , a p elimina y poin analysis was pe o med on selec ed pa icles om he Ö o a samples o see i he e was some hing ha s ood ou om he samples. When obse ing he samples om Ö o a sepa a ed by he magne ic sepa a o , he ollowing obse a ions we e made: 1) Mo phology o he pa icles, 2) simila i ies and di e ences in chemical composi ion and 3) a i ac s, pa icles ha a e no expec ed in each ac ion, such as eldspa in he magne ic ac ion and ilmeni e in he magne ic ejec . A e ha , he sha e o a i ac s and ilmeni e pa icles in each sample was quan i ied hanks o poin analysis wi h EDS. The su aces o selec ed a i ac s and an ilmeni e pa icle om he magne ic ac ion Ö -0417-MF we e also analysed wi h poin analysis in EDS. Las ly, SEM-EDS was also used o a emp ing o see i a i ion o Fe happened in he Chalme s boile campaign in 2014. The samples analysed o his pu pose a e he pa icles om he cyclone o he same boile . 26 Figu e 9. BS SEM mic og aphs o he samples: F esh ilmeni e (Top le ), Ö -0417-MF (Top igh ), Ö -0416-RF (Bo om le ) and Cyk-1125-OCAC (Bo om igh ). A e pe o ming p elimina y poin analysis on selec ed pa icles o he Ö o a samples and compa ing hem o he F esh ilmeni e sample as well as he ly ash samples om he cyclone, se e al indings we e no ed: 1. The esh ilmeni e pa icles al eady con ain Mg (mainly in he o m o single eldspa pa icles o ilmeni e pa icles a ached o eldspa pa icles) and ha seems o be he main sou ce o Mg o he samples ex ac ed om he boile . The Ö o a samples also ha e hese eldspa pa icles, e en in he magne ic ac ion. This las inding especially s ands ou as eldspa is no magne ic. 2. The e a e ilmeni e pa icles in he magne ic ejec ac ion. Tha also s ands ou as ilmeni e, a e unde going oxida ion, ends o ha e o med Fe oxides ha show magne ic p ope ies, oxides such as magne i e o hema i e. 3. A e mo e ope a ion ime, he concen a ions o Ca and K ha e e ec i ely isen. Tha is, he poin analysis shows ha he pa icles om Ap il 17 ha e highe concen a ions o hese wo elemen s han he samples om Ap il 9. 27 4. As i can be seen in Figu e 9, pa icles in he cyclone sample a e smalle compa ed o bo om ash (as i should be o ly ash) and con ain a lowe concen a ion o K. In o de o answe he ques ion ega ding he p esence o eldspa pa icles in he magne ic ac ion, wo hings we e conside ed. Fi s , i has o be quan i ied he sha e o hese eldspa pa icles in he esh ilmeni e and he Ö o a samples o make su e abou he o igin o he pa icles and whe he i happens consis en ly in he samples. I also needs o be made su e whe he wha was obse ed wi h SEM-EDS is ep esen a i e o he samples. Second, an ini ial hypo hesis was aised. The hypo hesis was ha he eldspa pa icles ha appea in he magne ic ac ion a e co e ed in some ype o Fe ich laye ha has magne ic p ope ies and ha is why hey end up being d agged on o he magne ic ac ion con aine when hey a e sepa a ed om he blend. Simila conside a ions we e aken when app oaching he p esence o ilmeni e in he magne ic ejec . The ini ial hypo hesis, in his case, is ha he pa icle has ei he los he magne ic Fe oxides ha may mig a e o he su ace due o a i ion, ha he pa icle has no been oxidized enough o ha he ash laye somehow is so hick ha i s ongly a ec s he magne ic suscep ibili y o he pa icle. To quan i y he ilmeni e and eldspa sha e o hese pa icles, 6 o e iews we e aken wi h SEM o he ollowing samples immobilized in epoxy: F esh, Ö -0409-BA, Ö -0416-RF, Ö -0417- MF, Ö -0417-BA. The o e iews a e SEM mic og aphs aken wi h he lowes magni ica ion possible (400X) in he able op SEM ha was used du ing he expe imen al pa . These mic og aphs we e used o iden i y he sha e o he ype o pa icles in each s udied ep esen a i e sample. In o de o no in oduce bias when choosing he loca ions o he o e iews, hey we e always aken in he same 6 posi ions. A u he explana ion o how ha was done can be ound in he Appendix II. To app ecia e he dis ibu ion o he pa icles in each sample, poin analysis wi h SEM-EDS was pe o med on he pa icles wi h a diame e bigge han 100 μm in he o e iews o each sample. Based on he elemen concen a ions o he pa icles, hey we e classi ied as ilmeni e, eldspa , ilmeni e-like pa icles and o he pa icles, and a sha e o each ype o pa icle was calcula ed o each o e iew. Tha classi ica ion would allow o see he de ia ions in he bulk composi ion o he samples and hei sha es o each o he pa icles o in e es (ilmeni e and eldspa mainly). A e OCAC ope a ion, ilmeni e pa icles ha ha e eac ed wi h he uel ashes can ha e a qui e di e se composi ion. Tha complica es he classi ica ion o he pa icles. The way i has been done is ha i he pa icle shows a conside able a omic % o bo h Fe and Ti, bu also ash elemen s like Ca, K o Si, he pa icle can be conside ed as an ilmeni e-like pa icle o de i ed om i . I he pa icle has a p edominan ly high concen a ion o Fe and Ti compa ed o he elemen s p esen in ash compounds, hen i is classi ied as an ilmeni e pa icle. I a pa icle is no classi ied as nei he ilmeni e-like, ilmeni e o eldspa pa icle, hen i is aken in o accoun as “o he pa icles”. The ollowing Table 2 shows he numbe o ilmeni e, ilmeni e-like and eldspa pa icles in each one o he 6 o e iews aken om he samples and he sum o he numbe o pa icles o all o e iews o each sample and he co esponding sha es. I can be seen ha he sum o ilmeni e, eldspa and ilmeni e-like pa icles does no equal he o al numbe o pa icles ha we e accoun ed o . Tha is because he e we e pa icles ha could no be classi ied as nei he o he h ee bu hey we e accoun ed o he o al numbe o pa icles due o ha ing a diame e la ge han 100 μm. 28 Table 2. Sha es o ilmeni e, eldspa and ilmeni e-like pa icles in he o e iews o he F esh ilmeni e sample and ou Ö o a samples. F esh* Ö -0416-RF** O e iew 1 2 3 4 5 6 SUM 1 2 3 4 5 6 SUM Numbe o ilmeni e pa icles 14 19 18 11 20 18 100 2 2 1 2 2 3 12 Numbe o eldspa pa icles 0 0 0 0 0 2 2 10 16 15 14 7 13 75 Numbe o mixed pa icles* Numbe o ilmeni e-like pa icles** 1 1 0 2 1 0 5 0 0 0 0 0 0 0 Numbe o pa icles 15 20 18 13 21 21 108 12 18 16 16 9 16 87 % o ilmeni e 93 95 100 85 95 86 93 17 11 6 13 22 19 14 % o eldspa 0 0 0 0 0 10 2 83 89 94 88 78 81 86 % o mixed pa icles 7 5 0 15 5 0 5 0 0 0 0 0 0 0 % o ilmeni e o mixed pa icles* /ilmeni e-like pa icles** 100 100 100 100 100 86 97 17 11 6 13 22 19 14 Ö -0417-MF Ö -0409-BA O e iew 1 2 3 4 5 6 SUM 1 2 3 4 5 6 SUM Numbe o ilmeni e pa icles 1 0 0 0 0 0 1 2 1 3 1 0 2 9 Numbe o eldspa pa icles 3 3 3 0 3 2 14 3 6 3 4 6 5 27 Numbe o ilmeni e-like pa icles 7 11 11 13 8 9 59 4 8 6 7 8 6 39 Numbe o pa icles 12 16 16 17 12 12 85 16 18 15 17 17 14 97 % o ilmeni e 8 0 0 0 0 0 1 13 6 20 6 0 14 9 % o eldspa 25 19 19 0 25 17 16 19 33 20 24 35 36 28 % o ilmeni e-like pa icles 58 69 69 76 67 75 69 25 44 40 41 47 43 40 % o ilmeni e o ilmeni e like pa icles 67 69 69 76 67 75 71 38 50 60 47 47 57 49 Ö -0417-BA O e iew 1 2 3 4 5 6 SUM Numbe o ilmeni e pa icles 0 0 0 0 0 0 0 Numbe o eldspa pa icles 2 2 0 4 3 0 10 Numbe o ilmeni e-like pa icles 8 9 14 8 14 16 69 Numbe o pa icles 11 12 16 15 18 17 89 % o ilmeni e 0 0 0 0 0 0 0 % o eldspa 18 17 0 27 17 0 12 % o ilmeni e like pa icles 73 75 88 53 78 94 78 % o ilmeni e o ilmeni e like pa icles 73 75 88 53 78 94 78 Ou o he sha es he ollowing in o ma ion can be ex ac ed: • The blend Ö -0409-BA sample has a subs an ial sha e o pa icles ha a e no ilmeni e nei he eldspa (i.e. pa icles whe e he only wo main elemen s a e Si and Ca) so he sha e o ilmeni e, eldspa o ilmeni e-like pa icles is lowe consequen ly. 29 • The sha e o ilmeni e (only accoun s o he ilmeni e pa icles ha ha e a simila composi ion o ha o he esh ilmeni e, i.e. wi hou ha ing majo ash compounds in hei s uc u e) ound in he magne ic ejec is bigge han he one ound in he magne ic ac ion (Ö -0417-MF) and blend (Ö -0417-BA). Howe e , ha does no ake in o accoun he numbe o ilmeni e-like pa icles o pa icles de i ed om ilmeni e, which added up o he % o ilmeni e s ands o a bigge sha e in he magne ic sepa a ion blend (bo h om day 9 and day 17) and ac ion samples. Tha seems o indica e ha he pa icles ha eac mo e wi h he ash compounds end up in he magne ic ac ion a he han in he ejec . • The eldspa sha e o he magne ic ac ion Ö -0417-MF (17%) is qui e high o being he magne ic accep as eldspa should all wi h he non-magne ic ac ion. This numbe is also highe han he eldspa sha e o he Ö -180417-BA (12%) which is he blend be o e magne ic sepa a ion. I he magne ic sepa a ion is being handled co ec ly, he bulk magne ic ac ion should ha e a lowe % o eldspa han he bulk blend be o e sepa a ion. This phenomenon can be explained mainly because o he he e ogenei y o he Ö o a samples. Gi en ha he e ogeneous mix u e, ob aining a ep esen a i e ac ion om he bo om ash as a whole and hen aking a ep esen a i e o e iew may esul in his ype o de ia ions. The % di e ence be ween bo h samples is only 5%, which conside ing he o de o magni ude o he pa icles accoun ed o (85 and 89 o Ö -0417-MF and Ö -180417-BA espec i ely) is a low disc epancy o a leas i is no abe an . In o de o co ela e he sha e in o ma ion o he hypo heses ha we e p e iously sugges ed, an analysis o he laye s o he ilmeni e pa icles o he magne ic ejec and he esh ilmeni e was pe o med. To ep esen he a ia ion o he a omic % close o he edge o he pa icles, se e al poin s o each pa icle we e analysed wi h EDS and ep esen ed in he ollowing Figu es 10 o 12. The successi e poin analysis pe o med in he same pa icle we e done in a way ha hey ollowed a s aigh line. Also, he i s poin analysis was done s a ing om he poin a hes o he su ace and each poin analysis ha ollowed was pe o med close o he su ace o he pa icle. Tha is why i should be no ed ha in hese Figu es whe e a omic % is ep esen ed agains he loca ion o he pa icle, he X axis is inc easingly close o he edge o he pa icle (i.e. he highes numbe in he axis ep esen s he edge o he pa icle). The Fe:Ti a io is also ep esen ed o see i he e is any di e ence in Fe dis ibu ion along he pa icle. As an example o how he poin s we e aken in he sample, he nex Figu e 10 (le ) also illus a es he analysis o he su ace o an ilmeni e pa icle om he esh ilmeni e sample. Fi s , he su ace o an ilmeni e pa icle om he esh ilmeni e is compa ed o one om an ilmeni e pa icle om he magne ic ejec (Ö -0416-RF), so ha i is possible o obse e di e ences ha may ha e o do wi h he loss o magne ism. The dis ibu ion o Fe, Ti and Si elemen al concen a ions as well as he Fe:Ti a io om F esh ilmeni e along he line o poin s is also ep esen ed in Figu e 10 ( igh ). The same ep esen a ion is done in Figu es 12 and 31 bu o ash elemen s Al, Ca, Na and K. This is done so ha i can be seen how Fe, Ti and ash elemen s dis ibu e along he pa icle, which may indica e i an ash laye is de ec ed in he ejec pa icles and i a case o Fe mig a ion can be done. In case a laye is de ec ed, i will be de ec ed by compa ing wi h he esh ilmeni e dis ibu ion, ha does no p esen a laye . 36 The pa icle analysis was pe o med in wo di e en ilmeni e pa icles om he magne ic ac ion Ö -0417-MF. Figu es 20 and 22 show a BS-SEM mic og aph o he pa icle and i s su ace o pa icle 1# and 2# espec i ely, while Figu es 21 and 23 show hei espec i e Fe:Ti a io and he Ca concen a ion dis ibu ion along he su ace o he pa icle. Figu e 20. BS-SEM mic og aph o ilmeni e pa icle #1 (le ) om he magne ic ac ion (Ö -0417-MF) and su ace analysed ( igh ) in Figu e 21. Figu e 20 ( igh ) shows he su ace o he pa icle, which appea s o be mo e he e ogeneous han he co e o he pa icle. Figu e 21. Fe:Ti a io and elemen al concen a ion o Ca in ilmeni e pa icle #1 om Ö - 0417-MF. The a io and he concen a ion a e plo ed om a loca ion nea he co e o he pa icle (1) o he su ace o he pa icle (9). Figu e 21 appea s o show ha whe e Ca concen a ion is high, he a io be ween Fe and Ti dec eases and ice e sa. The e is an inc ease in Fe concen a ion a ound he poin s 6 and 7 ha could be indica ing he p esence o an Fe laye . The Fe:Ti a io ela i e inc ease be ween poin 1 ( he closes o he pa icle co e) and he maximum Fe:Ti a io is 201%. Also, he inc ease in Ca 0 0,5 1 1,5 2 2,5 3 0 5 10 15 20 25 30 123456789 Fe:Ti A omic % Ö -0417-MF ilmeni e pa icle #1 Calcium Fe:Ti 37 o he closes poin s o he su ace o he pa icle (poin s 7, 8 and 9) could be signalling he o ma ion o a Ca double laye ha was obse ed in ea lie bibliog aphy. This will be u he discussed in he Discussion sec ion. Figu e 22. BS-SEM mic og aph o ilmeni e pa icle #2 (le ) om he magne ic ac ion (Ö -0417-MF) and su ace analysed ( igh ) in Figu e 23. Figu e 22 ( igh ) also shows an i egula su ace o he ilmeni e pa icle #2 ha seems o be mo e he e ogeneous in composi ion han i s co e. Figu e 23. Fe:Ti a io and elemen al concen a ion o Ca in ilmeni e pa icle #2 om Ö - 0417-MF. The a io and he concen a ion a e plo ed om a loca ion nea he co e o he pa icle (1) o he su ace o he pa icle (13). As i can be obse ed in Figu e 23, i appea s again ha Fe:Ti a io shows highe alues whe e Ca has a highe elemen al concen a ion and ice e sa. Howe e , in his case, i can be no ed ha 0 0,5 1 1,5 2 2,5 3 0 5 10 15 20 25 30 35 12345678910 11 12 13 Fe:Ti A omic % Ö -0417-MF Ilmeni e pa icle #2 Calcium Fe:Ti 38 he loca ion whe e he Fe:Ti a io ises is close o he su ace han i is o pa icle #1. The Fe:Ti a io ela i e inc ease be ween poin 1 and he poin whe e he Fe:Ti a io is maximum is 181.5%. I can also be app ecia ed how when looking a he wo closes poin s o he su ace, pa icle #2 does no show an inc ease in Ca elemen al concen a ion as signi ican as pa icle #1 does. This could indica e ha some ash compounds ha a e o med on he su ace a e being wo n o due o a i ion and i will be u he discussed in he Discussion sec ion. To obse e Fe a i ion, wo ly ash samples om he cyclone o Chalme s boile we e analysed. As i has been said, his analysis canno p o e ha a i ion was happening in he Ö o a campaign, only in he Chalme s boile . Howe e , i is le o u u e wo k o see i bo h boile s showed his phenomenon and i he assump ion ha hey can be assimila ed in ha sense is co ec . Since ly ash consis s o smalle pa icles han bo om ash, ollowing he me hod o aking o e iews wi h SEM-EDS mic og aphs and classi ying he pa icles wi h a diame e bigge han 100 μm was no a good idea. In ac , he majo i y o he obse ed pa icles in he ly ash ha we e bigge han 100 μm we e ilmeni e pa icles. Wha was done ins ead, was only ying o loca e smalle pa icles ha migh gi e in o ma ion on whe he a i ion was going on. In ha sense, only selec ed pa icles we e analysed, as he esea ch o he composi ion o he ly ash was a om he objec i e o his p ojec . Wi h his analysis some ines ich in Fe we e disco e ed, which means ha a i ion o Fe was happening o some ex en . The mic og aph o one o hese pa icles and i s composi ion a e ep esen ed in Figu e 24 and Table 3 espec i ely. Figu e 24. BS-SEM mic og aph o selec ed Fe ine om he ly ash sample Cyk-1125- OCAC. The highligh ed pa icle is a ine om he oxygen ca ie pa icles and wi h poin analysis SEM- EDS i s a omic concen a ion was analysed. 39 Table 3. No malised elemen al concen a ion o he highligh ed pa icle in Figu e 24 ob ained wi h poin analysis SEM-EDS. Elemen A omic Concen a ion (%) I on 49.1 Magnesium 25.1 Ti anium 11.2 Calcium 10.1 Manganese 2.1 Silicon 1.5 Po assium 1 I can be seen ha he pa icle is indeed ich in Fe and ha could mean ha ilmeni e pa icles a e unde going Fe a i ion om hei su aces in he Chalme s boile . Since only a couple o pa icles we e selec ed, his analysis could be no ep esen a i e o he o ali y o he bed and he e o e a i ion could be happening only o a small sha e o pa icles. I has been ound ha he e a e Fe ines in he ly ash o he Chalme s boile and ha could mean ha a leas a small numbe o pa icles a e unde going a i ion, and he same could be happening in K a ingen’s boile . 3.2 XRD esul s In o de o iden i y he c ys alline compounds ha may make hese laye s magne ic, XRD was pe o med on he same samples om Ö o a as he SEM-EDS. I can be checked which samples we e analysed wi h each echnique in he Appendix I. XRD was used o de e mine which c ys alline compounds a e p esen in he samples. The esul o applying he echnique a e he XRD scans ha a e shown in Figu e 25. Figu e 25. XRD scans o all lis ed samples. 40 The indi idual no malised scans (whe e he in ensi y – I in he e ical axis has been di ided by he maximum in ensi y – I0) can be ound in Appendix IV. The ollowing Table 4 is a summa y o he main c ys alline compounds de ec ed in each o he scans shown in Figu e 25 and hei ela i e p esence, based on he ela i e peak in ensi y be ween he di e en scans. Table 4. C ys alline compounds de ec ed wi h XRD Analysis and hei ela i e p esence based on he ela i e peak in ensi y. In his able ++ indica es e y high peak in ensi y, + indica es high peak in ensi y, - indica es low peak in ensi y and – indica es e y low peak in ensi y. Pseudob ooki e (Fe2TiO5) Ilmeni e (FeTiO3) Magne i e / Ti anomagne i e (Fe3O4/Fe2.8Ti0.2O4) K- eldspa / Mic ocline (KAl0.93O8Si3.07) Ca- i ana e (CaTi0.8Fe0.2O2.9) Ö o a 0409 BA ++ + + - + Ö o a 0409 MF ++ + + -- + Ö o a 0416 RF ++ + -- + -- Ö o a 0417 BA ++ + ++ - ++ Ö o a 0417 MF ++ + ++ -- ++ Based on he iden i ied peaks, i has been de e mined ha pseudob ooki e (Fe2TiO5) is p esen in e e y sample and in simila quan i ies. Ö -0416-RF is he sample ha shows less pseudob ooki e. I should be no ed ha he mine al phase ilmeni e (FeTiO3) has simila peaks as he oxidized phase and can hus o e lap wi h pseudob ooki e peaks. Howe e , ilmeni e phase is no expec ed o be in a signi ican ly highe quan i y han pseudob ooki e a e a subs an ial oxida ion pe iod as he one ha happens in he combus o . Bo h phases could be p esen e en hough i could be assumed ha pseudob ooki e is mo e dis ibu ed owa ds he su ace o he pa icle (as ha egion is oxidized as e ), while ilmeni e phase is dis ibu ed owa ds he co e o he pa icle. Pseudob ooki e also p esen s a sligh ly highe co ela ion o he da a o he used da abase (C ys allog aphy Open Da abase – REV212673) han ilmeni e. Thus, an assump ion can be made ha he iden i ied phase is p ima ily pseudob ooki e. The e is a s ong indica o ha magne i e is p esen bo h in he blend be o e sepa a ion and in he magne ic ac ion o he samples. In o he wo ds, Ö -0416-RF shows a low numbe o coun s whe e he es o he samples show magne i e peaks. Hema i e sha es he main peaks wi h magne i e and ha is why, as i happens wi h ilmeni e and pseudob ooki e, i is ha d o ela i ely quan i y how much hema i e he e is in he samples. Magne i e s ill shows a conside ably highe deg ee o co ela ion o he used da abase han hema i e and ha is why i is ega ded as he main magne ic componen o he samples. To be able o obse e he e ec o ope a ion ime and magne ic sepa a ion, he no malised scans ha e been pai ed by ime poin o he sample in Figu e 26 and by ype o sample (BA, MF, RF) in Figu e 27. 41 Figu e 26. No malised XRD (I/I0) scans pai ed by ime poin . I can be seen how he e is no a signi ican di e ence be ween he blend and he magne ic ac ion ( o bo h days conside ed Ap il 9 and 17) when i comes o he peaks ha indica e he p esence o magne i e. Addi ionally, some Fe py oxenes (FeSiO3 / CaFeSi2O6) seem o be p esen in he samples om Ap il 17 and Mg py oxene (MgSiO3) could be p esen in he magne ic ejec (Ö - 0416-RF). These py oxenes come mos p obably om he ashes. Figu e 27. No malised XRD (I/I0) scans pai ed by ype o sample (BA, MF, RF). Howe e , in Figu e 27 i can be obse ed how bo h Ö -0417-MF and Ö -0417-BA ha e s onge peaks ha ela e o magne i e han hei ea ly campaign coun e pa s. Tha could sugges ha , a leas o his campaign, a longe esidence ime inc eases he oxida ion o Fe om ilmeni e owa ds magne i e and hema i e. Ano he obse a ion when i comes o he esidence ime is ha a Ca- i ana e (CaTi0.8Fe0.2O2.9) also inc eases i s concen a ion wi h ime in bo h he magne ic ac ion and he blend. In ha sense, 42 he samples om Ap il 17 show a highe signal o his Ca- i ana e han he samples om Ap il 9. A simila compound was al eady p esen in ea lie bibliog aphy (see Co co an e al. [16]) and he analysis shows ha i can be ound wi h high possibili y bo h in he magne ic ac ion and he blend o ea ly o la e ope a ion days. Rega ding he magne ic ejec (Ö -0416-RF), i shows a lowe amoun o pseudob ooki e compa ed o he es o he samples bu he e a e h ee s ong peaks ha can be seen a 27.5°, 41.6° and 50.6° ha a e no as s ong on he o he samples. Those peaks indica e he p esence o mic ocline (K eldspa ). The same peaks can be seen wi h he blends and he magne ic ac ion bu he heigh o he peaks is conside ably lowe and i can be obse ed ha he magne ic ac ion has a dec ease in he K- eldspa peak in ensi y. This las dec ease is he mos no able di e ence be ween he blend and he magne ic ac ion (conside ing each pai wi h he same da e o sampling). 3.3 Magne ic suscep ibili y measu emen s esul s A e he SEM-EDS and he XRD analysis, he magne ic suscep ibili y measu emen s we e done o all o he samples om Ö o a om Ap il 9, 16 and 17 ha appea in Table 6 as well as o F esh Ilmeni e. As sugges ed by esea che s p e iously using he cu en me hod o magne ic suscep ibili y measu emen , he ob ained suscep ibili y was di ided by he pou ed densi y o he sample as well as by i s mass. Tha is because some imes he mass speci ic magne ic suscep ibili y migh no be compa able o o he measu ed olumes o he da a o o he esea che s [42]. The esul s a e summa ized in he ollowing able: Table 5. Magne ic suscep ibili ies measu ed o 5 samples ob ained om Ö o a and a F esh ilmeni e sample. Sample Mass speci ic magne ic Suscep ibili y (m3/kg) Weigh speci ic magne ic suscep ibili y (g-1) Mass o he sample (g) Ö o a-0417-MF 1.38·10-4 1.15·10-2 25.32 Ö o a-0417-BA 1.20·10-4 1.00·10-2 24.26 Ö o a-0409-MF 7.68·10-5 6.40·10-3 22.52 Ö o a-0409-BA 5.91·10-5 4.92·10-3 23.39 Ö o a-0416-RF 1.82·10-5 1.51·10-3 20.88 F esh ilmeni e 3.36·10-6 2.80·10-4 29.03 As i can be seen, he samples a e o de ed in Table 5 based on a dec easing magne ic suscep ibili y. F om he able, he magne ic ac ion o one ope a ion day (Ap il 9 o Ap il 17) can be compa ed wi h he blend o he same day and hey ha e a magne ic suscep ibili y ha is qui e close be ween hem (13.24% and 23.12% o ela i e di e ence espec i ely). I can be obse ed ha he di e ence be ween any o he pai o samples suscep ibili ies is highe . This obse a ion was somewha expec ed because he magne ic suscep ibili y is an addi i e p ope y and he compounds ha a e mo e magne ic can ha e a suscep ibili y which is se e al o de s o magni ude highe han ha o o he compounds ha a e less magne ic (as desc ibed p e iously in he Magne ic sepa a ion sec ion o he In oduc ion). Thus, i he magne ic 43 compounds accoun o a bigge suscep ibili y and hey a e bo h in he blend and in he ac ion, a simila suscep ibili y will be measu ed. F esh ilmeni e is he less magne ically suscep ible sample (e en less han he magne ic ejec ). Tha p o es ha in o de o achie e magne ic sepa a ion, i is equi ed o ac i a e he ilmeni e wi h se e al oxidizing/ educing cycles. 3.4 XRF esul s XRF was he echnique o choice o e-e alua e wha had been obse ed wi h SEM-EDS and o y o link he chemical composi ion wi h he c ys alline compound in o ma ion ha XRD p o ides and he magne ic suscep ibili y measu emen s om he Ba ing on senso . XRF has an ad an age, i allows o mul iple samples o be analysed simul aneously which enables he ob en ion o esul s o di e en samples in a sho pe iod o ime. Tha is why he samples chosen o unde go his echnique we e he same 5 Ö o a samples as o he o he ou echniques (BS SEM, SEM-EDS, XRD, magne ic suscep ibili y measu emen s) bu wi h he addi ion o wo o he Ö o a samples om a di e en ime poin (Ö -0413-BA and Ö -0413- MF). The in o ma ion om hese samples is shown in Table 6 in he Appendix I. The samples analysed wi h XRF we e: Ö -0409-BA, Ö -0409-MF, Ö -0413-BA, Ö -0413-MF Ö -0416-RF, Ö -0417-BA and Ö -0417-MF. In o de o ha e an idea o how much o he Fe was on he su ace o he pa icles, i was decided o g ind e e y sample and analyse wice each sample, one ha had been g ound and ano he one ha had no been g ound. The no a ion o dis inguish be ween hem is “GR” o he g ound samples and no addi ion o he non-g ound ones. The eason o g inding is ha he signal dep h o XRF is limi ed o 1-100 μm in o he pa icles so i he su ace o he pa icles is made ou o a Ca laye , XRF analysis will ha e a s onge signal coming om his elemen a he han om he elemen s loca ed close o he co e o he pa icle (i.e. Ti and Fe). When g inding he samples, hese laye s all o and he co e is made accessible o measu emen wi h XRF. In he magne ic ejec (Ö -0416-RF) he amoun o Al, Si and K was pa icula ly ollowed o see i i had any co ela ion o he high amoun s o eldspa obse ed in he magne ic ejec wi h BS SEM and SEM-EDS. T acking he amoun o Fe in e e y sample was also impo an o be able o unde s and wha is happening wi h ha elemen when ilmeni e pa icles s ay mo e ime inside he Ö o a boile . As XRF akes in o accoun he composi ion o he bulk phase, and no se e al pa icles like in SEM-EDS, he obse a ions om SEM-EDS could gain s eng h o lose i (o cou se, also aking in o accoun XRD esul s). The da a ob ained om he XRF we e spec og ams wi h which he XRF so wa e can quan i y he elemen al composi ion measu ed in A omic concen a ion. Those concen a ions needed o be no malised wi hou O and C. In he sample Ö -0413-MF GR an abno mal amoun o Ce was de ec ed and he elemen was excluded du ing he p ocessing o he da a. A e he no malisa ion and he adjus men s we e done, a lis o elemen s anging om 22 o 25 elemen s was ob ained. Ou o hose elemen s, only he elemen s ha ei he had a concen a ion highe han 1% o we e p esen in he SEM-EDS/XRD analysis we e aken in o accoun . The elemen s ha mee he condi ions a e he same elemen s ha we e ound in he SEM-EDS analysis 44 ha was done o he o e iews om 5 Ö o a samples. Tha is: Al, Ca, Fe, K, Mg, Mn, Na, P, S, Si, Ti and Zn. Seeking o see whe he he g inding has any e ec on he composi ion shown by XRF, each pai o g ound and non-g ound sample is ep esen ed in he same g aph. Figu e 28. XRF elemen al dis ibu ion compa ison be ween he g ound agains he non- g ound samples. The Figu e ha shows he XRF elemen al dis ibu ion compa ison be ween he g ound and non- g ound Ö -0413-MF sample can be ound in Figu e 34 o he Appendix V. The i s hing ha can be obse ed om he Figu e 28 plo s is he ac ha he di e ence be ween he Ca concen a ion om a g ound sample and a non-g ound sample inc eases wi h ime. When looking a he samples om Ap il 9 and Ap il 13, hey di e a maximum o 3.71 pe cen age uni s, whe eas when looking a he magne ic ejec om Ap il 16 (Ö -0416-RF), he blend om Ap il 17 (Ö -0417-BA) and he magne ic ac ion om Ap il 17 (Ö -0417-MF) he di e ence inc eases up o 5.98%, 8.09% and 10.23% espec i ely. 45 This could mean ha Ca is wo n o in he p ocess o g inding and while a highe amoun o Fe and Ti is de ec ed (which is he case), he concen a ion o Ca dec eases. I ha is he case, i would indica e ha he e was no mig a ion o Ca inwa ds o a leas i was no e y signi ican . The exac opposi e o wha happens wi h Ca can be a gued o Ti. I can be seen how when g inding a highe % o Ti is de ec ed. Howe e , his di e ence is ini ially low (0.16 and 0.48 o he Ö -0409-BA and Ö -0409-MF samples espec i ely) and inc eases o e ime (3.03 and 9.23 o Ö -0417-BA and Ö -0417-MF espec i ely). This could mean ha Ti, ha acco ding o he bibliog aphy does no mig a e owa ds he ou e pa s o he pa icle, does no mig a e and a highe quan i y o i is e ealed when he Ca laye is wo n o . Finally, i is ele an o look a how he Fe % changes wi h o wi hou g inding. Fo he samples om Ap il 9 and Ap il 13, i can be seen ha i does no make a big di e ence as he pe cen age di e ence be ween he g ound and he non-g ound samples is low (up o 0.7% in he case o he Ap il 13 blend - Ö -0413-BA). A conside able inc ease in his di e ence in concen a ion comes in he Ö -0417-MF, whe e Fe inc eases in he bulk composi ion by a o al o 7.4% when g inding. This also could be e idence ha he e is a Ca laye co e ing he Fe and Ti om he ilmeni e pa icles. As i can be seen, he blends om Ap il 9, 13 and 17 do no show a high di e ence be ween he g ound samples and he non-g ound ones. The same can be s a ed abou he magne ic ac ions om Ap il 9 and Ap il 13. Tha could mean ha he e is no Ca laye o a slim one in hose samples. Tha is cohe en conside ing ha he Ap il 9 and Ap il 13 samples a e om an ea ly ope a ion day in he campaign so he ac i a ion o he ilmeni e and i s mig a ion migh s ill be s a ing. I is also cohe en wi h he ac ha he blend samples p esen an a e age o he elemen s om all he species p esen in he bo om ash and hus any possible di e ence ha could be obse ed is smoo hed ou by i s di e se composi ion. Tha could be a possible explana ion o he low di e ence in Fe % in Ö -0417-BA in compa ison o he high di e ence shown o Ö -0417-MF. Rega ding he ejec (Ö -0416-RF), e en i i is om a la e ime poin , i is no expec ed o ha e as many ilmeni e pa icles wi h Fe laye s in i s s uc u e in compa ison o o he ash compounds. The di e ence in Fe % in his sample is conside ably lowe han o Ö -0417-MF bu i is highe han o Ö -0413-MF. This could mean ha he e is an ash laye in he ejec pa icles bu i is no as hick as he magne ic ac ion one. The compa ison be ween g ound and non-g ound samples has gi en in o ma ion on he Ca laye , bu he e is mo e in o ma ion ha can be ex ac ed om he elemen al composi ion compa ison be ween di e en samples. 52 E en hough Fe mig a ion has no been p o ed by all echniques, a sugges ed explana ion o he inc ease in magne ic suscep ibili y in his epo is ha Fe mig a es owa ds he su ace o he ock ilmeni e pa icle, whe e he oxygen pa ial p essu e is highe , and hen is oxidized in o hema i e and magne i e, which show highe magne ic suscep ibili ies han esh ilmeni e. Rega ding he magne ic sepa a ion e iciency, i should be aken in o accoun ha i no enough esh ma e ial is ed o no enough bo om ash is magne ically sepa a ed, ilmeni e pa icles will s a o add mo e and mo e ash compounds o i s s uc u e. The p oblem is ha hose pa icles, no ma e how much ash hey ha e inco po a ed in he o m o Ca- i ana es, Fe py oxenes o o he compounds, will become mo e and mo e magne ically suscep ible. A leas ha can be s a ed when conside ing he span o ime om Ap il 9 o Ap il 17 om ha exac Ö o a campaign. The pa icles will become mo e suscep ible e en i o e all Fe concen a ion is dec easing (i is no possible o say ha Fe is being los he e due o a i ion as i has no been p o en). Supposing Fe also mig a es owa ds he su ace in he ilmeni e pa icles om Ö o a and he pa icles a e unde going a i ion, as long as new ee Fe keeps being oxidized o magne i e hose pa icles will s ill be magne ically sepa a ed. This can be posi i e i he posi i e e ec s o la ge esidence imes and ash up ake a e conside ed, like educ ion o he co osion due o K up ake. Howe e , i can also be nega i e i he nega i e impac is highligh ed, like he inc ease in Fe a i ion due o pa icle po osi y inc easing. This means ha he e is a ime window ega ding he ex ac ion in which he magne ic sepa a ion has a posi i e impac and inc eases he e iciency o he p ocess when conside ing ash up ake and oxygen ca ying capaci y. The sepa a ion should be pe o med once ilmeni e has been ac i a ed and is suscep ible enough o be sepa a ed. The ac i a ion ime is dependen on he uel and he ope a ional condi ions o he boile . I should also be ho oughly combined wi h egene a ion o he ma e ial so ha i is made su e ha he magne ic sepa a ion is no only sepa a ing ilmeni e pa icles wi h lowe and lowe oxygen ca ying abili y (once a i ion s a s o happen) wi h hick ash laye s a ached o hem. The sepa a ion should no be delayed un il hose e ec s s a o ha e a nega i e impac on he o e all e iciency o he oxygen ca ie bed ma e ial inside he combus o . I he amoun o esh ilmeni e added is excessi e o he magne ic sepa a ion is handled be o e he esh ma e ial pa icles ha e been ac i a ed, a conside able amoun o esh ilmeni e pa icles could go in o he ejec due o no showing he magne ic suscep ibili y needed in o de o be sepa a ed. I ha is he case, esh ma e ial could be los be o e being used. This exac hing would explain why he e we e some big ilmeni e pa icles wi h ew ash elemen s in hei composi ion in he sha es o he magne ic ejec . These ilmeni e pa icles we e he a i ac s ha we e p esen ed in he ejec and we e no iced wi h he sha e classi ica ion o pa icles wi h SEM-EDS. Those pa icles esembled esh ilmeni e in hei composi ion and ha is because mos p obably hey we e ilmeni e pa icles ha had been ecen ly added be o e he poin o ex ac ion. Those pa icles had no been ac i a ed and he e o e hey did no ha e oxygen ca ying oxides such as magne i e o hema i e in hei s uc u e. These pa icles p obably had a sligh ly highe magne ic suscep ibili y han esh ilmeni e and ha is why hey could no make i o he magne ic ac ion. Tha way, magne ic sepa a ion has o be imed co ec ly in o de o no lose unexploi ed ma e ial and o make su e ha he one ha is being used allows o he bes pe o mance possible. The e is a p e e ed ime window ega ding he ex ac ion ha could be heo e ically op imized by inco po a ing magne ic suscep ibili y measu emen s in he plan in o de o ha e mo e insigh on when i is adequa e o apply he magne ic sepa a ion and consequen eci cula ion o bed ma e ial o when i is be e o eed mo e esh ilmeni e. The magne ic suscep ibili y me hod used in his 53 epo is non-in asi e and is based on manual ex ac ion o samples bu i could be au oma ed which could make i i ing o applica ion in indus y. Fu he mo e, in his epo , i has been obse ed ha along wi h ac i a ion o he pa icles, a highe magne iza ion is accomplished. Tha happens, p esumably, un il Fe a i ion o he pa icles p o okes he dec ease o oxygen ca ying capaci y. I ha is he case, an in e es ing ollow-up o his epo would be he in es iga ion o he connec ion be ween oxygen ca ying capaci y and magne ic suscep ibili y measu emen s. The in es iga ion on bulk magne ic suscep ibili y as a p edic o o an indica o o oxygen ca ying capaci y could be e y help ul owa ds accomplishing highe e iciency o magne ic sepa a ion in OCAC. 5 Fu u e wo k The i s hing ha can be p oposed as u u e wo k is he epe i ion o he measu emen s and cha ac e iza ions done in his epo . As i was said when in oducing he expe imen al pa , he boile loca ed a Ö o a has a high capaci y (60 onnes) and in compa ison, he numbe o pa icles analysed wi h XRF and magne ic suscep ibili y was small. A ye smalle numbe o pa icles we e analysed wi h XRF and SEM-EDS, and he e o e, he conclusions d awn o his small numbe o pa icles could no ep esen he en i e sample o wha had happened in he boile . I was assumed ha he combina ion o he me hods used in his epo could adequa ely ep esen he whole, bu i needs o be e i ied wi h u he expe imen a ion. Especially he conclusions ha we e d awn based o pa icle analysis wi h SEM-EDS will need o be e isi ed o check ha hey we e ep esen a i e. O he analysis echniques o e en o he modes o ope a ion o he echniques used in his epo could be used in o de o ob ain mo e in o ma ion abou he samples. SEM-EDS, o example, could be used in he mode ha ob ains in ensi y maps o gain in o ma ion abou whole pa icles and no only pa s o hem. Tha could allow o easie cha ac e iza ion o he pa icles and a be e unde s anding o how he di e en elemen concen a ions a e dis ibu ed h ough he bed ma e ial pa icles. Ano he possible imp o emen ha could be done in u u e wo k is he addi ion o mo e magne ic ejec samples, as he p esen epo only ea ed one. The sample was indeed di e en o he o he samples, so he magne ic sepa a ion had e ec i ely been applied and i can be app ecia ed bu he compa ison wi h o he magne ic ejec s would gi e mo e in o ma ion abou he e iciency o he sepa a ion and he condi ions o he pa icles ha end up in he ejec . The samples analysed only wi h XRF (Ö -0413-BA and Ö -0413-MF) could be analysed wi h XRF, BS SEM, SEM-EDS and hei magne ic suscep ibili ies could be measu ed. This was no done due o ime cons ain s bu i could be done o complemen he in o ma ion o he Ap il 9 and Ap il 17 ime poin s wi h ano he ime poin ha is equally dis an o bo h. I ob ained, he ly ash samples om he Ö o a acili y could also be analysed o check he Fe a i ion and when does i appea . As commen ed al eady in he Discussion, he eldspa pa icles p esen in he magne ic accep show a Fe-Ti-Ca laye whose p ope ies (magne ic suscep ibili y and oxygen ca ying capaci y) would be in e es ing o see. The alkali up ake o he pa icles om Ö o a could be in es iga ed as XRF showed a qui e s able bulk concen a ion o he pa icles ac oss he samples excep o he magne ic ejec . K in ensi y 54 maps could be aken o he pa icles o see he K dis ibu ion in he pa icle, as he p esence o eldspa in he bulk phase o he samples complica es d awing conclusions wi h XRF. An in e es ing con inua ion o he epo would be measu ing he oxygen ca ying capaci y o he samples wi h TGA (The mog a ime ic Analysis) o in a small boile in o de o co ela e i o wha has been obse ed h ough he epo . As i has been said in he Discussion, ano he po en ially in e es ing ou e o ollow a e his p ojec would be looking in o he co ela ion be ween bulk magne ic suscep ibili y and he oxygen ca ying capaci y o he pa icles. The po en ial use o magne ic suscep ibili y as a p edic o o indica o o he oxygen ca ying capaci y could be e y p omising i p o en. A model could be adjus ed be ween he wo a iables and he co ela ion coe icien (R2) could be ob ained. I one would like o ake i a s ep u he , ha model could be ained wi h da a om he egula ex ac ions o he plan applying Machine Lea ning, so ha he esul ing model could help in decision making ega ding he p ope ime o ex ac ion o ma e ial, magne ic sepa a ion and eci cula ion. Las bu no leas , he in es iga ion conduc ed in his epo could be done wi h o he pa icles om o he indus ial plan s o o he campaigns om he same CHP plan so ha he obse a ions could be compa ed o maybe gene al ends could be obse ed. 55 6 Conclusions In his Bachelo hesis, he magne ically sepa a ed samples om an OCAC campaign in K a ingen’s CHP plan loca ed in Ö o a ha e been in es iga ed. The blend be o e sepa a ion, he magne ic ac ion and he magne ic ejec ha e been subjec ed o a mo phological and chemical cha ac e iza ion wi h BS SEM and SEM-EDS. The c ys alline compounds p esen in each sample ha e been iden i ied wi h XRD and hei bulk composi ion has been cha ac e ized wi h XRF. In o de o co ela e he chemical composi ion o he samples wi h he esidence ime in he boile and hei magne ic suscep ibili y, magne ic suscep ibili y measu emen s ha e been also ca ied ou . A classi ica ion o pa icles analysed in each sample wi h SEM-EDS ga e he sha e o ilmeni e pa icles, eldspa pa icles, pa icles ha de i ed om ilmeni e o had a high ash elemen s concen a ion and o he pa icles. Feldspa pa icles ha come om he uel we e obse ed as a i ac s in he magne ic ac ion, co e ed in an Fe-Ti-Ca laye , acco ding o SEM-EDS analysis. Ope a ion ime allowed he o ma ion o Ca laye s in he ilmeni e pa icles o he magne ic ac ion. SEM-EDS analysis o wo pa icles om he sample showed he o ma ion o a double laye a ound a possible Fe laye in one o he pa icles and a single inne laye below ano he possible Fe laye on he o he pa icle. The mos p obable explana ion o he di e ence be ween he wo laye s was a i ion o ces ha wo e o he ou e ash laye o he pa icle wi h only an inne one. The mechanism o mig a ion o he Ca2+ o o m such inne laye s could be assimila ed o he one p esen in p e ious bibliog aphy, as acco ding o XRD analysis he compound o med is CaTi0.8Fe0.2O2.9, simila o o he Ca- i ana es epo ed in he bibliog aphy. The p oposed mechanism consis s in Ca deposi ion in he su ace and di usion inwa ds o eac wi h TiO2 and be inco po a ed in he ilmeni e s uc u e as he iden i ied Ca- i ana e. Also, wi h an inc ease in ope a ion ime (i.e. in a e age esidence ime o he bed ma e ial) an inc ease in mass speci ic magne ic suscep ibili y was obse ed o he magne ic blend and he magne ic ac ion. Tha is mainly due o he o ma ion o magne ic Fe oxides such as i anomagne i e ( e imagne ic) and hema i e (an i e omagne ic), iden i ied wi h XRD. A mechanism o he o ma ion o hese oxides was sugges ed. The mechanism inds i s suppo in he SEM-EDS inding o possible Fe laye s in he su ace o ilmeni e pa icles om he magne ic ac ion o he sample wi h he la es ime poin . This mechanism should be e-e alua ed wi h o he echniques. The p oposed mechanism o o ma ion o he oxides s a s wi h Fe2+ mig a ion owa ds he su ace o he pa icle, whe e he oxygen pa ial p essu e is highe , and whe e i is oxidized o o m magne i e and i u he oxidized i o ms hema i e. The magne ic suscep ibili y o he ilmeni e pa icles seems o inc ease wi h esidence ime and an ini ial s age o ac i a ion is equi ed in o de o p esen he Fe oxides ha show he suscep ibili y needed o magne ic sepa a ion. A p e e ed ime window o he sepa a ion was p esen ed. An ea ly sepa a ion and eci cula ion will esul in ilmeni e pa icles showing low magne ic suscep ibili ies, simila o he ones measu ed o esh ilmeni e (3.36·10-6 m3/kg), which leads o he pa icles ending in he magne ic ejec . A la e sepa a ion and eci cula ion will esul in ilmeni e pa icles becoming inc easingly magne ic, e en hough he oxygen ca ying capaci y is being diminished due o Fe a i ion and hick ash laye s a e being o med a ound he pa icle. The e o e, he p ocess o sepa a ion and ecycling has o be imed co ec ly in o de o no lose esh ma e ial o eci cula e ma e ial wi h low oxygen ca ying capaci y. Magne ic suscep ibili y 56 measu emen s a e sugges ed as a use ul ool in decision making ega ding he adequa e ime o he sepa a ion and i s consequen eci cula ion. 57 Re e ences [1] NASA’s Global Clima e Change, “O e iew: Wea he , Global Wa ming and Clima e Change.” h ps://clima e.nasa.go / esou ces/global-wa ming- s-clima e-change/ (accessed May 11, 2021). [2] WWF Aus alia, “Impac s o global wa ming.”, [Online]. A ailable: h ps://www.ww .o g.au/wha -we-do/clima e/impac s-o -global-wa ming#gs.12e 9 (accessed May 11, 2021). [3] U.S. Ene gy In o ma ion Adminis a ion, “Ene gy and he en i onmen explained - Whe e g eenhouse gases come om,” 2020, [Online]. A ailable: h ps://www.eia.go /ene gyexplained/ene gy-and- he-en i onmen /whe e-g eenhouse- gases-come- om.php (accessed May 21, 2021). [4] P. F iedlings ein e al., “Global Ca bon Budge 2020,” Ea h Sys em Science Da a, ol. 12, no. 4, pp. 3269–3340, 2020, doi: 10.5194/essd-12-3269-2020. [5] C. Le Qué é e al., “Fossil CO2 emissions in he pos -COVID-19 e a,” Na u e Clima e Change, ol. 11, no. Ma ch, 2021, doi: 10.1038/s41558-021-01001-0. [6] P. Ag eemen and U. Na ions, “Pa is ag eemen ,” 2015, [Online]. A ailable: h ps://un ccc.in / iles/essen ial_backg ound/con en ion/applica ion/pd /english_pa is_a g eemen .pd . [7] R. Rohde, “Be keley Ea h Janua y 2021 Tempe a u e Upda e,” Be keley Ea h, [Online]. A ailable:. h p://be keleyea h.o g/janua y-2021- empe a u e- upda e/#:~: ex =Global Summa y& ex = (accessed Ap . 21, 2021). [8] T. Mau i sen and R. Pincus, “Commi ed wa ming in e ed om obse a ions,” ol. 7, no. July, pp. 1–5, 2017, doi: 10.1038/NCLIMATE3357. [9] S. V Vassile , D. Bax e , and C. G. Vassile a, “An o e iew o he beha iou o biomass du ing combus ion: Pa I. Phase-mine al ans o ma ions o o ganic and ino ganic ma e ,” Fuel, ol. 112, pp. 391–449, 2013, doi: 10.1016/j. uel.2013.05.043. [10] S. V Vassile , C. G. Vassile a, Y. Song, W. Li, and J. Feng, “Ash con en s and ash- o ming elemen s o biomass and hei signi icance o solid bio uel combus ion,”, Fuel, ol. 208, pp. 377–409, 2017, doi: 10.1016/j. uel.2017.07.036. [11] A. Gyllén, Oxygen ca ie aided combus ion : Implemen a ion o oxygen ca ie s o exis ing indus ial se ings. 2019. [12] A. Lyng el , A. B ink, Ø. Langø gen, T. Ma isson, M. Rydén, and C. Linde holm, “11,000 h o chemical-looping combus ion ope a ion — Whe e a e we and whe e do we wan o go?,” In e na ional Jou nal o G eenhouse Gas Con ol, ol. 88, no. May, pp. 38–56, 2019, doi: 10.1016/j.ijggc.2019.05.023. [13] F. S ö ne , F. Hildo , H. Leion, M. Ze enho en, L. Hupa, and M. Rydén, “Po assium Ash In e ac ions wi h Oxygen Ca ie s S eel Con e e Slag and I on Mill Scale in Chemical-Looping Combus ion o Biomass-Expe imen al E alua ion Using Model Compounds,” Ene gy and Fuels, ol. 34, no. 2, pp. 2304–2314, 2020, doi: 10.1021/acs.ene gy uels.9b03616. [14] F. Hildo , M. Ze enho en, A. B ink, L. Hupa, and H. Leion, “Unde s anding he in e ac ion o po assium sal s wi h an ilmeni e oxygen ca ie unde d y and we condi ions,” ACS Omega, ol. 5, no. 36, pp. 22966–22977, 2020, doi: 10.1021/acsomega.0c02538. 58 [15] D. Filippou and G. Hudon, “I on emo al and eco e y in he i anium dioxide eeds ock and pigmen indus ies,” JOM, ol. 61, no. 10. pp. 36–42, 2009, doi: 10.1007/s11837- 009-0150-3. [16] A. Co co an, P. Knu sson, F. Lind, and H. Thunman, “Mechanism o Mig a ion and Laye G ow h o Biomass Ash on Ilmeni e Used o Oxygen Ca ie Aided Combus ion,” Ene gy and Fuels, ol. 32, no. 8, pp. 8845–8856, 2018, doi: 10.1021/acs.ene gy uels.8b01888. [17] C. Guizani, M. Jegui im, S. Valin, L. Limousy, and S. Sal ado , “Biomass cha s: The e ec s o py olysis condi ions on hei mo phology, s uc u e, chemical p ope ies and eac i i y,” Ene gies, ol. 10, no. 6, 2017, doi: 10.3390/en10060796. [18] P. Knu sson and C. Linde holm, “Cha ac e iza ion o ilmeni e used as oxygen ca ie in a 100 kW chemical-looping combus o o solid uels,” Applied Ene gy, ol. 157, pp. 368–373, 2015, doi: 10.1016/j.apene gy.2015.05.122. [19] D. Bhogeswa a Rao and M. Rigaud, “Kine ics o he oxida ion o ilmeni e,” Oxida ion o Me als, ol. 9, no. 1, pp. 99–116, 1975, doi: 10.1007/BF00613496. [20] X. Fu, Y. Wang, and F. Wei, “Phase ansi ions and eac ion mechanism o ilmeni e oxida ion,” Me allu gical and Ma e ials T ansac ions A: Physical Me allu gy and Ma e ials Science, ol. 41, no. 5, pp. 1338–1348, 2010, doi: 10.1007/s11661-010-0173- y. [21] A. Co co an, J. Ma inko ic, F. Lind, H. Thunman, P. Knu sson, and M. Seemann, “Ash p ope ies o ilmeni e used as bed ma e ial o combus ion o biomass in a ci cula ing luidized bed boile ,” Ene gy and Fuels, ol. 28, no. 12, pp. 7672–7679, 2014, doi: 10.1021/e 501810u. [22] I. S aničić, “Fa e o T ace Elemen s in The mochemical Con e sion o Was e Fuels Using Oxygen Ca ie s,” p. 24, 2021, [Online]. A ailable: h ps:// esea ch.chalme s.se/en/publica ion/523712%0Ah ps:// esea ch.chalme s.se/publ ica ion/523712/ ile/523712_Full ex .pd . [23] W. J. B onkala, “Magne ic sepa a ion,” in Ullmann’s Encyclopedia o Indus ial Chemis y, Wisconsin: Wiley, 2000, pp. 133–144. [24] Powde P ocess.ne , “Indus ial Magne aps and sepa a o s o powde - A guide o magne ic sepa a ion.”, [Online]. A ailable: h ps://www.powde p ocess.ne /Equipmen s h ml/Magne s.h ml (accessed May 15, 2021). [25] J. S oboda, “Magne ic Sepa a ion,” in Encyclopedia o Ma e ials: Science and Technology, Else ie , 2005, pp. 1–7. [26] J. A. Obe eu e , “Magne ic Sepa a ion: A Re iew o P inciples, De ices, and Applica ions,” IEEE T ansac ions on Magne ics, ol. 10, no. 2, pp. 223–238, 1974, doi: 10.1109/TMAG.1974.1058315. [27] C. Ki el and S. Johnson, In oduc ion o Solid S a e Physics, 8 h Edi ion Be keley: John Wiley & Sons Inc, 2004. [28] W. D. Callis e and D. G. Re hwisch, Ma e ials Science and Enginee ing: An In oduc ion, 8 h Edi ion Wiley, 2010. [29] J. Dea ing, En i onmen al Magne ic Suscep ibili y - Using he Ba ing on MS2 sys em, 2nd Edi ion Ba ing on, 1999. [30] A. Gup a and D. Yan, Mine al P ocessing Design and Ope a ions, 2nd Edi ion Else ie , 2016. 59 [31] J. D. S eenkamp and P. C. Pis o ius, “Re lec ions on ilmeni e oas ing and magne ic sepa a ion,” in Hea y mine als con e ence p oceedings, 2005, no. July 2017, pp. 133– 141. [32] F. Ka lsson, “Magne ic sepa a ion o oxygen ca ie s om bed ash in hea and powe plan s,” 2019. [33] M. Ezzahmouly e al., “Mic o-compu ed omog aphic and SEM s udy o po ous bioce amics using an adap i e me hod based on he ma hema ical mo phological ope a ions,” Heliyon, ol. 5, no. 12, p. 3, 2019, doi: 10.1016/j.heliyon.2019.e02557. [34] J. I. Golds ein e al., Scanning Elec on Mic oscopy and X- ay Mic oanalysis, 3 d Edi ion New Yo k: Sp inge , 2003. [35] S. E. Dann, Reac ions and Cha ac e iza ion o Solids. Camb idge: Royal Socie y o Chemis y, 2000. [36] A. H. Simon, Spu e P ocessing. Else ie Inc., 2018. [37] The moFishe Scien i ic - Ask A Scien is S a , “Wha is XRF (X- ay Fluo escence) and How Does i Wo k?,” 2020, [Online]. A ailable: h ps://www. he mo ishe .com/blog/ask-a-scien is /wha -is-x -x- ay- luo escence-and- how-does-i -wo k/ (accessed May 13, 2021). [38] A. Ba houm, M. L. Ga cía-Be ancou , H. Rahie , and G. Van Assche, Physicochemical cha ac e iza ion o nanoma e ials: Polymo ph, composi ion, we abili y, and he mal s abili y. Else ie Inc., 2018. [39] Yale Uni e si y Wes Campus Ma e ials Cha ac e iza ion Co e, “XRF - P inciple,” 2021, [Online]. A ailable: h ps://ywcma sci.yale.edu/galle y/x /p inciple (accessed May 13, 2021). [40] C. J. Dixon, “The Tellnes Ilmeni e Deposi - No way,” in A las o Economic Mine al Deposi s, Do d ech : Sp inge , 1979, pp. 108–109. [41] L. G. Wood u , G. M. Bedinge , and N. M. Pia ak, “Ti anium Chap e T o C i ical Mine al Resou ces o he Uni ed S a es — Economic and En i onmen al Geology and P ospec s o Fu u e Supply P o essional Pape 1802 – T U.S. Depa men o he In e io ,” U.S. Depa men o he In e io - U.S. Geological Su ey, no. P o essional Pape 1802–T, p. 26, 2017, doi: h ps://doi.o g/10.3133/ pp1802T. [42] “Pe sonal communica ion wi h esea che s om Imp obed AB.”, 2021. 60 Appendix I. Code used o he samples The ollowing Table 6 (in he nex page) shows all he samples ha we e analysed wi h each echnique along wi h whe e hey we e ob ained om and he ime in which hey we e ex ac ed om he boile . A code has been designed o a conciseness ma e . The code consis s o he ollowing e ms: 𝐴𝐵𝐶 − 𝑀𝑀𝐷𝐷 − 𝑇𝑦𝑝𝑒 Whe e ABC is based on he i s h ee le e s o he name o he place whe e he sample was ex ac ed, MMDD is he da e o imes amp o ob en ion o he sample (mon h and day) and Type consis s o addi ional in o ma ion, e.g. he indica ion on whe he he sample is he blend ha has no unde gone magne ic sepa a ion (BA), he magne ic ac ion (MF) o he magne ic ejec ac ion (RF) o he bo om ash (BA) sample. On he da e o ob en ion he yea is omi ed because he samples o in e es a e om he same o igin (Ö o a) and since all o hem sha e he same yea o ope a ion, adding he yea does no gi e any new in o ma ion. The no a ion o he addi ional in o ma ion s ands o he ollowing: • BA: Bo om Ash collec ed di ec ly om he boile and ha has no unde gone magne ic sepa a ion. I is he blend ha would go in o he magne ic sepa a o . • MF: Magne ic ac ion ob ained om he in oduc ion o he bo om ash blend (BA) in he boile . • RF: Rejec ac ion ob ained om he in oduc ion o he bo om ash blend (BA) in he boile . • OCAC: I jus iden i ies ha he sample has been used o OCAC in he Chalme s boile . 61 Table 6. Lis o all he samples analysed wi h SEM-EDS echniques. Code Name o he sample Place o o igin Times amp o ob en ion o he sample SEM- EDS XRD χ measu emen s XRF F esh ilmeni e F esh ilmeni e Ti ania mine, No way (Chalme s) Season 2016 (sp ing) X X Ö -0409- BA Ö o a 180409 Ö o a Imp obed 2018-04-09 X X X X Ö -0409- MF Ö o a 180409 magne ic ac ion Ö o a Imp obed 2018-04-09 X X X X Ö -0416- RF Ö o a 180416 magne ic ejec Ö o a Imp obed 2018-04-16 X X X X Ö -0417- BA Ö o a 180417 Ö o a Imp obed 2018-04-17 X X X X Ö -0417- MF Ö o a 180417 magne ic ac ion Ö o a Imp obed 2018-04-17 X X X X Ö -0413- BA Ö o a 180413 Ö o a Imp obed 2018-04-13 X Ö -0413- MF Ö o a 180413 Ö o a Imp obed 2018-04-13 X Cyk- 1117- OCAC Sekcyklon 141117 OCAC 7.09 Chalme s Boile 2014-11-17 X Cyk- 1125- OCAC Sekcyklon 141125 OCAC 6.40 Chalme s Boile 2014-11-25 X