1
Scale-up modelling and li e cycle assessmen o 1
elec ochemical oxida ion in was ewa e ea men 2
Sa a Feijoo1,*, So ía Es é ez2, Mohammad eza Kamali1, Ra Dewil1,3, and Ma ía Te esa 3
Mo ei a2
4
1 KU Leu en, Depa men o Chemical Enginee ing, P ocess and En i onmen al Technology Lab, 2860 Sin -Ka elijne-5
Wa e , Belgium 6
2 Uni e sidade de San iago de Compos ela, Depa men o Chemical Enginee ing, CRETUS, 15782 San iago de 7
Compos ela, Spain 8
3 Uni e si y o Ox o d, Depa men o Enginee ing Science, Pa ks Road, Ox o d, OX1 3PJ, Uni ed Kingdom 9
* Co esponding au ho : Sa a Feijoo, [email p o ec ed] 10
Keywo ds— Elec ochemical Ad anced Oxida ion P ocesses (eAOPs), Li e Cycle Assessmen (LCA), 11
was ewa e ea men , scale-up modelling, mic opollu an s 12
Abs ac 13
The need o imp o e cu en was ewa e ea men s o ensu e a clean and sus ainable wa e 14
supply is an unques ionable con empo a y challenge. I is he e o e essen ial o acili a e 15
knowledge ans e be ween esea ch ins i u ions and he indus y by de eloping no el 16
echnologies o a p oo -o -concep s age, demons a ing bo h ea men e iciency and 17
compliance wi h en i onmen al c i e ia. This s udy has combined p ocess modelling o he 18
design o an elec ochemical Ad anced Oxida ion P ocess (eAOP) o emo e ca bamazepine 19
(CBZ) om was ewa e wi h he iden i ica ion o he en i onmen al impac s associa ed wi h i s 20
ope a ion. A comp ehensi e se o scena ios conside ing se e al eac o designs and ope a ing 21
condi ions p o ides he assessmen amewo k o iden i y he in luence o di e en p ocess 22
a iables on he en i onmen al p o ile o he pilo -scale eAOP. The mos sus ainable ea men 23
co esponds o he ope a ion o a s anda dised modula eac o in ba ch mode, especially when 24
he was ewa e has a low concen a ion o sca enge s, such as o he ions, o ganics o 25
pollu an s. Ne e heless, in all scena ios e alua ed, he main en i onmen al ho spo was 26
a ibu ed o he elec ical ene gy consumed by he auxilia y pumps a he han he 27
elec ochemical eac o i sel . In compa ison o o he AOPs, ou sys em showed conside ably 28
lowe impac s in he global wa ming po en ial (GWP) ca ego y, wi h a minimum o 7.6 kg CO2 29
eq pe g CBZ emo ed o he mos p omising scena io. This demons a es he implemen a ion 30
po en ial o eAOPs as well as he impo ance o da a om scaled-up expe imen s, whe e 31
op imisa ion should ocus on mi iga ing he impac s o ene gy-in ensi e pieces o equipmen . 32
Abb e ia ions: API – Ac i e pha maceu ical ing edien , B – Ba ch mode, BDD – Bo on-doped diamond, C – Con inuous mode / 33
Chemical, CBZ – Ca bamazepine, CECs – Con aminan (s) o eme ging conce n, COD – Chemical oxygen demand, CSTR – Con inuous 34
s i ed ank eac o , E – Ene gy, eAOP(s) – Elec ochemical ad anced oxida ion p ocess(es), EC – Enhanced conduc i i y, FB – Fed-35
ba ch mode, FE – F eshwa e eu ophica ion, FRS – Fossil esou ce sca ci y, FU – Func ional uni , GAC – G anula ac i a ed ca bon, 36
GWP – Global wa ming po en ial, H – High con en , L – Low con en , LCA – Li e cycle assessmen , M – Mul icomponen , ME – 37
Ma ine eu ophica ion, NF – Nano il a ion, RL – Regula o y limi s, SDG – Sus ainable de elopmen goal, SPF – Sola pho o-Fen on, 38
SW – Syn he ic was ewa e , TA – Te es ial acidi ica ion, TET – Te es ial eco oxici y. 39
2
1 In oduc ion 40
In 2020, app oxima ely 2 billion people lacked sa ely managed d inking wa e , 2.3 billion people 41
su e ed om poo hygiene and up o 3.6 billion people did no ha e access o basic sani a ion, 42
which has aised some conce n abou he accomplishmen o Sus ainable De elopmen Goal 43
(SDG) No. 6 on Clean Wa e and Sani a ion by 2030 [1]. One o he oo causes is he occu ence 44
o con aminan s o eme ging conce n (CECs) in was ewa e ea men plan e luen s, which 45
ep esen s a majo issue no only o human heal h bu also o ecosys ems [2–5]. Ca bamazepine 46
(CBZ) is one o hese con aminan s since i is a pha maceu ical poo ly emo ed by con en ional 47
biological ea men (i.e., emo al e iciency is ypically lowe han 10%) [6–8]. In ac , due o 48
i s widesp ead consump ion and ecalci an na u e, CBZ has been ecen ly ound o be he mos 49
ecu ing ac i e pha maceu ical ing edien (API) in i e basins wo ldwide [9]. 50
One o he solu ions o his p oblem is he de elopmen no el was ewa e ea men s ha 51
p e en he elease o pollu an s h ough hei e ec i e deg ada ion, as is he case o Ad anced 52
Oxida ion P ocesses (AOPs). AOPs a e an ex ensi e amily o ea men s comp ising ozona ion, 53
he e ogeneous and homogeneous (pho o)ca alysis, Fen on and Fen on-like p ocesses, and 54
elec ochemical-, ul asound-, mic owa e- o gamma- adia ion ea men s as well as any o 55
hei combina ions [10]. Among hese a ious echnologies, elec ochemical Ad anced 56
Oxida ion P ocesses (eAOPs) ha e ecei ed signi ican a en ion in ecen yea s [11–13]. They 57
a e commonly used as e ia y was ewa e ea men s, d i ing pollu an deg ada ion h ough 58
di ec and indi ec oxida ion pa hways by elec ochemically gene a ing highly eac i e oxida i e 59
species, mainly hyd oxyl (•OH) and sul a e (SO4•−) adicals [14–16]. Elec ochemical AOPs allow 60
o high deg ada ion e iciencies and eac ion a es unde mild condi ions, while showing no o 61
limi ed dependence on chemical addi ion [2, 17]. O he ad an ages a e hei e sa ili y, ease o 62
p ocess in eg a ion and sa e ope a ion [13, 18]. 63
When implemen ing an eAOP, he selec ion o he elec ode ma e ial and he p ecu so species 64
o he oxida i e adicals a e key ac o s in luencing he o e all ea men e iciency and 65
selec i i y [14]. Bo on-doped diamond (BDD) elec odes a e o pa icula in e es o 66
was ewa e applica ions, as hey ha e demons a ed high e iciency in he gene a ion o 67
oxida i e species and deg ada ion o se e al con aminan s, as well as high conduc i i y, 68
s abili y, O2 o e po en ial and du abili y [17, 19, 20]. Despi e he signi ican ene gy consump ion 69
associa ed wi h elec ochemical ea men s [15], he in si u adical gene a ion o e ed by he 70
BDD ma e ial om wa e molecules [21] and ionic species such as sul a e ions [22, 23] has a 71
high added alue o indus ial implemen a ion, gi en ha hey a e al eady a ailable in 72
was ewa e s eams [24, 25]. The absence o addi ional chemicals can minimise no only he 73
o e all consump ion o aw ma e ials bu also he gene a ion o seconda y was e s eams and 74
hence he associa ed en i onmen al impac s [10]. Consequen ly, achie ing SDG No. 6 while 75
aiming o sus ainable and ca bon neu al p ocesses is essen ial o p o ide a a - eaching 76
solu ion. In his ega d, he Li e Cycle Assessmen (LCA) me hodology is a use ul esou ce o 77
e alua ing he en i onmen al iendliness o no el eAOPs. None heless, as o June 2022, a 78
Scopus sea ch o s udies applying LCA me hodology o elec ochemical oxida ion in 79
was ewa e ea men e ie ed 281 documen s, o which only 8 publica ions speci ically 80
included an elec o-oxida ion sys em (Table C.1). Among hese s udies, none we e dedica ed o 81
he emo al o pha maceu icals, 5 we e applied o syn he ic o eal was ewa e ma ices, and 82
3
only wo conside ed was ewa e olumes a a la ge scale. Consequen ly, he e is a signi ican 83
knowledge gap on he en i onmen al implica ions o eAOPs in was ewa e ea men . 84
In ou p e ious wo k [26], wo p elimina y design conside a ions o he implemen a ion o a 85
BDD-based eAOP as a e ia y was ewa e ea men we e add essed: he e ec s o he 86
was ewa e composi ion and he eac o mode o ope a ion. A compa a i e assessmen o 8 87
di e en scena ios, bo h in e ms o CBZ deg ada ion and elec ical ene gy consump ion pe uni 88
o e ec i e ope a ion ime, e ealed ha he compe i ion eac ions aking place due o 89
was ewa e componen s could be mi iga ed when ope a ing in ed-ba ch mode, since a 2.1-90
old inc ease in CBZ deg ada ion and a 60% educ ion in ene gy consump ion we e achie ed wi h 91
espec o a con en ional ba ch ope a ion. Simila ly, ope a ing in con inuous a he han ba ch 92
mode esul ed in signi ican ene gy sa ings (app oxima ely 19%) o simila deg ada ion 93
e iciency. Howe e , in o de o u he e alua e he ad an ages and disad an ages o each o 94
he in es iga ed scena ios, i is essen ial o ake in o accoun hei po en ial en i onmen al 95
impac when applied on a la ge scale. In ac , mos con en ional municipal was ewa e 96
ea men plan s al eady in ol e high ene gy consump ion om he g id due o all he machine y 97
in ol ed [27], leading o a signi ican ca bon oo p in (i.e., 23–432 kg CO2 pe popula ion 98
equi alen ) [27–29], o which app oxima ely 70% was a ibu ed o he indi ec emissions om 99
ene gy equi emen s [29]. 100
To ill he knowledge gap on he en i onmen al pe o mance o pilo -scale eAOPs o he 101
emo al o pha maceu icals om seconda y was ewa e e luen s, his s udy ocused on 102
conduc ing a echno-en i onmen al analysis including he ollowing: 103
(i)
De elopmen o a scale-up model o ansla e he labo a o y esul s in o a pilo -scale 104
ope a ion. To his end, wo eac o con igu a ions ha e been conside ed: a s anda dised 105
modula eac o and a e ical pla e s i ed ank eac o . 106
(ii)
Quan i ica ion o he en i onmen al p o ile o he eAOP by LCA me hodology unde 107
mul iple expe imen al condi ions. Mo e speci ically, he in luence o he eac o 108
con igu a ion, he mode o ope a ion (namely, ba ch, ed-ba ch and con inuous), he 109
was ewa e ma ix (conside ing a ious composi ions o pu e and syn he ic was ewa e , 110
di e en amoun s o oxidising species and he possible p esence o addi ional 111
pollu an s) and he po en ial o e sizing e ec ha e been e alua ed. 112
2 Me hodology 113
2.1 Expe imen al scena ios 114
The mic opollu an deg ada ion expe imen s we e ca ied ou using a BDD elec o-oxida ion 115
sys em, as p e iously desc ibed by Feijoo e al. (2022) [26]. Bo h single and mul icomponen 116
sys ems we e p ima ily aimed a CBZ emo al, whe e a di e se se o concen a ions o 117
oxida i e and sca enge species we e in es iga ed. The eac o ope a ing modes included in he 118
compa a i e analysis we e ba ch, ed-ba ch and con inuous. As a esul , he ollowing 8 119
scena ios we e e alua ed in he echno-en i onmen al analysis: 120
4
•
Scena io o “Regula o y Limi s o Sul a es and Ni a es Conduc ed in Ba ch Mode (RL-121
B)”: CBZ deg ada ion was ca ied ou in ba ch mode and in a pu e wa e ma ix 122
con aining he concen a ion limi s o ni a e (50 mg/L) and sul a e (250 mg/L) ions as 123
de ined by hei espec i e EU di ec i es [30, 31]. 124
•
Scena io o “Enhanced Conduc i i y Medium Conduc ed in Ba ch Mode (EC-B)”: an 125
ex ension o he RL-B scena io assumed ha ni a e and sul a e concen a ions we e 126
highe han he egula o y limi s, a 100 mg/L and 500 mg/L, espec i ely. 127
•
Scena io o “Enhanced Conduc i i y Medium in Syn he ic Was ewa e wi h Low O ganic 128
Load Conduc ed in Ba ch Mode (ECSWL-B)”: CBZ deg ada ion was pe o med in ba ch 129
mode and in he p esence o he enhanced ni a e and sul a e concen a ions as in he 130
EC-B scena io. The ea ed wa e ma ix consis ed o a syn he ic seconda y e luen wi h 131
low concen a ions o o he o ganics and ions (COD: 25.2 mg/L, o al N: 5.0 mg/L, o al 132
P: 0.5 mg/L, alkalini y: 2.5 mg/L). 133
•
Scena io o “Enhanced Conduc i i y Medium in Syn he ic Was ewa e wi h Highe 134
O ganic Load Conduc ed in Ba ch Mode (ECSWH-B)”: his a ia ion o he ECSWL-B 135
scena io consis ed o he deg ada ion o CBZ in a syn he ic was ewa e ma ix wi h a 136
high ion and o ganic composi ion (COD: 50.4 mg/L, To al N: 10.0 mg/L, To al P: 0.9 mg/L, 137
Alkalini y: 4.9 mg/L). 138
•
Scena io o “Enhanced Conduc i i y Medium in Syn he ic Was ewa e wi h Low O ganic 139
Load Conduc ed in Fed-Ba ch Mode (ECSWL-FB)”: his modi ica ion o he ECSWL-B 140
scena io consis ed o ed-ba ch ope a ion, whe e CBZ spikes we e added a he beginning 141
o each 60 min cycle o a o al o 6 cycles o euse sul a e and ni a e species al eady 142
p esen in he was ewa e . 143
•
Scena io o “Enhanced Conduc i i y Medium in Syn he ic Was ewa e wi h Low O ganic 144
Load Conduc ed in Con inuous Mode (ECSWL-C)”: he ECSWL-B was adap ed o a 145
con inuous ope a ion, ha is, wi h con inuous inle and ou le lows se o 25 mL/min, 146
leading o an a e age esidence ime o 30 min. 147
•
Scena io o “Mul icomponen Sys em in Syn he ic Was ewa e wi h Low O ganic Load 148
Conduc ed in Ba ch Mode (MSWL-B)”: his a ia ion o he ECSWL-B scena io consis ed 149
o he simul aneous deg ada ion o CBZ wi h addi ional mic opollu an s, including 150
ca eine, diclo enac and sul ame hoxazole. 151
•
Scena io o “Mul icomponen Sys em in Syn he ic Was ewa e wi h Low O ganic Load 152
Conduc ed in Fed-Ba ch Mode (MSWL-FB)”: his modi ica ion o he MSWL-B scena io 153
was conduc ed in ed-ba ch mode o 6 cycles o 60 min wi h mul icomponen spikes. 154
2.2 Selec ed eac o designs 155
A e conduc ing a e iew o a ailable con igu a ions o pilo -scale BDD elec ochemical 156
eac o s, i was obse ed ha a la ge numbe o s udies conside ed comme cial DiaCell® uni s 157
5
[32–38], il e p ess low cells [38–41], mul ielec ode s acks wi h a se pen ine a ay [39–41], 158
o a e ical elec ode pla e a angemen in a s i ed ank eac o [39, 42, 43]. In his s udy, he 159
wo designs selec ed we e (i) a s anda dised modula eac o inspi ed by he DiaCell® uni s and 160
(ii) a ully cus omised e ical pla e s i ed ank eac o (Fig. 1). Bo h con igu a ions a e 161
commonly epo ed in he li e a u e, easible o scale up and signi ican ly di e en om each 162
o he in e ms o a ea, geome y and dis ance be ween he elec odes. 163
164
(a) S anda dised modula eac o . (b) Ve ical pla e s i ed ank eac o . 165
Figu e 1: Schema ics o he (a) s anda dised modula eac o and (b) e ical pla e s i ed ank eac o . Fo hei 166
ed-ba ch ope a ion, a dosing pump is added. 167
2.2.1 S anda dised modula eac o design 168
The s anda dised modula eac o con igu a ion was based on he DiaCell® 1001 elec ochemical 169
cell [34, 36]. I comp ises mul iple compa men s cons i u ed by wo BDD anodes and one 170
s ainless s eel ca hode wi h an in e elec ode dis ance o 1 mm, leading o a o al o 10 anodes 171
and 5 ca hodes pe cell. S anda d shapes o he elec odes a e ci cula , wi h a su ace a ea o 172
70 cm2 and monopola connec ions (Fig. 1a). Du ing i s ope a ion, a p ocess ank is loaded wi h 173
he seconda y was ewa e o be ea ed, and i needed, addi ional chemicals a e added. 174
A e wa ds, he con en o he ank is con inuously s i ed and ed o he s anda dised modula 175
eac o , whe e i is dis ibu ed be ween i e compa men s in pa allel. The sys em ope a es in 176
eci cula ion mode, meaning ha he o al olume o was ewa e emains cons an and is 177
eci cula ed un il he desi ed deg ada ion is a ained. Consequen ly, his eac o design is 178
applicable o ba ch and ed-ba ch ope a ions. Finally, he ea ed e luen is accumula ed in 179
he p ocess ank and discha ged. 180
2.2.2 Ve ical pla e s i ed ank eac o design 181
The e ical pla e s i ed ank eac o consis s o a se o pa allel monopola elec odes ha a e 182
ully imme sed in he bulk o he eac o (Fig. 1b). The numbe o elec ode pai s as well as hei 183
size and a angemen a e e sa ile pa ame e s, and hence, any eac o design can be 184
implemen ed. To a oid any damage o he elec odes du ing ope a ion, s i ing inside he 185
eac o is p omo ed by he inle and he eci cula ion pump lows. In addi ion, he elec ode 186
channels can con ain an ine polyme mesh and o he u bulence p omo e s o imp o e mass 187
ans e . I is assumed ha he cu en densi y and ol age a e uni o mly dis ibu ed ac oss he 188

6
cell. This ype o se up allows o ei he a ba ch, ed-ba ch o con inuous ope a ion wi h 189
eci cula ion. 190
2.3 Gene al scale-up conside a ions 191
Based on he collec ed expe imen al da a om labo a o y expe imen s in a 1 L elec ochemical 192
cell, he scale-up a ge was o model he s eady-s a e condi ions in a 100 L eac o illed up o 193
75% o i s capaci y and whe e 90% CBZ deg ada ion can be a ained. The scale-up me hodology 194
consis ed o analysing he expe imen al esul s based on he eac ion kine ics, elec ical 195
consump ion and ea men capaci y. This enabled mass and ene gy balances o be pe o med 196
a he pilo scale, wi h he equi ed pieces o equipmen (i.e., elec ochemical cell and associa ed 197
pumps) designed acco dingly. Rele an scale-up conside a ions a e de ined in he ollowing 198
subsec ions. 199
2.3.1 Common design condi ions 200
To compa e scena ios unde he same ime e e ence, all eac o designs we e simula ed o 201
ope a e o 1 day (i.e., 24 h). The numbe o ba ch and ed-ba ch expe imen s du ing ha ime 202
o achie e 90% emo al o CBZ we e calcula ed conside ing he e ec i e eac ion imes obse ed 203
expe imen ally. In addi ion, a o al o 25 min was conside ed pe expe imen o accoun o 204
p epa a ion, cha ge and discha ge ac i i ies. 205
The s a ing concen a ions o he di e en chemicals in ol ed we e assumed o be he same as 206
in he expe imen s a he lab scale, gi en ha hey a e independen o he eac o ype and size. 207
The e o e, hei o al ini ial mass was di ec ly p opo ional o he scaled-up eac o olume. Fo 208
he addi ion o sul a e and ni a e ions, only he di e en ial concen a ions wi h espec o he 209
egula o y limi s we e conside ed as inpu chemicals in he LCA in en o y, gi en ha i is 210
plausible ha he egula o y limi s may al eady be ound in he in luen was ewa e . In he case 211
o ed-ba ch ope a ion, i was assumed ha a concen a ed s eam o 200 mg/L CBZ was used 212
o he spikes o gua an ee ha olume a ia ions a e hei addi ion du ing 1 day o ope a ion 213
would no yield o mo e han an o e all 10% inc ease. 214
2.3.2 Mass balance assump ions 215
Since he kine ic cons an s (k, h−1) we e de e mined om lab-scale expe imen s, a co ec ion 216
ac o was applied o es ima e he inal CBZ concen a ions in he pilo -scale s anda dised 217
modula eac o . The need o a co ec ion ac o in his speci ic eac o con igu a ion a ises 218
om he di e ences in he numbe o elec odes and subsequen elec oac i e a eas be ween 219
he lab-scale eac o used and he scaled-up design. These di e ences lead o dis inc a ea- o-220
olume a ios, and he e o e, he a ia ion in kine ic cons an s has been es ima ed acco dingly. 221
As shown in Eq. 1, k is ela ed o he mass ans e coe icien (km, m/h), a pseudo- i s o de 222
kine ic cons an ela ed o he ac i i y o ino ganic oxidan s (ki, h−1), he elec oac i e a ea (A, 223
m2) and he eac o olume (V, m3) [34, 44]. Assuming ha ki is negligible in ou sys em as 224
oxidan s a e p esen in excess and ha km emains cons an wi h inc easing scale, he obse ed 225
7
kine ics a e a ec ed by he A/V a io. Consequen ly, he kine ic a e cons an s in he ba ch and 226
ed-ba ch scaled-up s anda dised modula eac o (kscale, h−1) we e calcula ed as shown in Eq. 227
2, whe e A and Ascale a e he elec oac i e a eas (m2) a he lab and pilo scales, espec i ely, 228
and V and Vscale a e he olumes (m3) o ea ed was ewa e a he lab and pilo scales, 229
espec i ely. 230
𝐶𝐶𝐶𝐶𝐶𝐶 =𝐶𝐶𝐶𝐶𝐶𝐶0∙𝑒𝑒(−𝑘𝑘∙𝑡𝑡)=𝐶𝐶𝐶𝐶𝐶𝐶0∙𝑒𝑒�−�𝐴𝐴
𝑉𝑉∙𝑘𝑘𝑚𝑚+𝑘𝑘𝑖𝑖�∙𝑡𝑡� (1)
𝑘𝑘𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 =𝑘𝑘 ∙𝐴𝐴𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑉𝑉𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
�
𝐴𝐴𝑉𝑉
� (2)
Simila ly, he modelling o he con inuous ope a ion was based on he de ini ion o an ideal
231
con inuous s i ed ank eac o (CSTR) (Eq. 3), whe e X is he con e sion o he a ge pollu an 232
ob ained expe imen ally and τ is he esidence ime (h). A e subs i u ion o common e ms 233
wi h Eq. 2, he con e sion in he scaled-up s anda dised modula eac o (Xscale) was ob ained 234
om Eq. 4, whe e F and Fscale a e he low a es (m3/h) o ea ed was ewa e a he lab and 235
pilo scales, espec i ely. Gi en ha a a ge o 90% CBZ emo al was selec ed, Eq. 4 was used 236
o e ie e he equi ed low a e a he pilo scale [34, 45]. 237
𝑘𝑘𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 =𝑋𝑋
1−𝑋𝑋 ∙1
𝜏𝜏 (3)
𝑋𝑋𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
1−𝑋𝑋𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 =𝑋𝑋
1−𝑋𝑋∙𝐴𝐴𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝐹𝐹𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
�
𝐴𝐴𝐹𝐹
� (4)
Rega ding he e ical pla e s i ed ank eac o , he expe imen al kine ic cons an s we e used
238
in he mass balance since he a ea- o- olume a io was conside ed cons an . Fo o he 239
was ewa e componen s, i was assumed ha hey we e p esen in excess and ha a ia ions 240
in concen a ion du ing he ea men we e negligible. In addi ion, he consump ion o NaOH o 241
neu alise acidic ou le s eams be o e discha ge was also calcula ed a he pilo scale and 242
included in he mass balance. 243
2.3.3 Ene gy balance assump ions 244
The limi ing cu en densi y (jlim, A/m2) o each ea men was es ima ed based on he model 245
de eloped by Panizza e al. (2001) de ined in Eq. 5, whe e F is he Fa aday cons an (C/mol), km 246
is he a e age mass anspo coe icien in he elec ochemical cell (m/s) and COD is he 247
chemical oxygen demand exp essed in mol O2/m3 [46]. The mass anspo coe icien (km) a 248
he pilo scale was es ima ed acco ding o he co ela ions as a unc ion o he low a e 249
p oposed by Anglada e al. (2009) [47], wi h a maximum alue o app oxima ely 1.7·10−5 m/s 250
o a 10 L/min low. 251
𝑗𝑗𝑠𝑠𝑙𝑙𝑙𝑙 = 4 ∙𝐹𝐹 ∙𝑘𝑘𝑙𝑙∙𝐶𝐶𝐶𝐶𝐶𝐶 (5)
As a esul , scena ios in ol ing was ewa e wi h low and high concen a ions o o ganics and
252
8
o he ionic species showed es ima ed limi ing cu en densi ies o 5.2 and 10.3 A/m2, 253
espec i ely. Gi en ha expe imen s we e pe o med a highe cu en densi ies, i can be 254
concluded ha elec ochemical oxida ion is unde mass anspo con ol and ha pollu an and 255
COD emo al ollow an exponen ial end. 256
The elec ical ene gy consump ion by he pilo -scale pumps (Ppump, kWh), which depends on he 257
supplie ca alogue nominal powe (Pn), was de e mined based on he modelled ope a ion ime 258
( , h), as shown in Eq. 6. The ope a ion ime o he eci cula ion, inle and ou le pumps 259
co esponded o he ac ual eac o ope a ion, whe eas o he pumps dedica ed o indi idual 260
cha ge, discha ge and dosing ope a ions, i was calcula ed as he ime equi ed o anspo a 261
scaled-up olume o liquid (Vscale, m3) a a speci ic low a e (Fscale, m3/h), as shown in Eq. 7. 262
𝑃𝑃𝑝𝑝𝑝𝑝𝑙𝑙𝑝𝑝 =𝑃𝑃𝑛𝑛∙𝑡𝑡 (6)
𝑡𝑡=𝑉𝑉𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
𝐹𝐹𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 (7)
Fo ba ch and ed-ba ch ope a ions in bo h eac o con igu a ions, cen i ugal pumps used o
263
cha ge/discha ge ope a ions we e assumed o be simila o he model KPM 50 by Spe oni S.p.A., 264
which has a Pn o 0.37 kW and can ope a e be ween 5 and 30 L/min [48]. The selec ed low a e 265
o he cha ge/discha ge pumps was 15 L/min o minimise ime losses du ing he 1-day 266
ope a ion. Du ing he elec ochemical ea men , he same pump ype was conside ed in bo h 267
eac o s o d i e a con inuous eci cula ion a 10 L/min o ensu e a Reynolds numbe highe han 268
500 [49]. 269
The dosing pump o he ed-ba ch ope a ion in bo h eac o con igu a ions was assumed o be 270
simila o he Model A pe is al ic pump by Redox.me, which has a Pn o 0.04 kW and can ope a e 271
be ween 0.07 and 380 mL/min [50]. The selec ed low a e was app oxima ely 5 mL/min. 272
Fo con inuous ope a ion, which is only applicable o he e ical pla e s i ed ank eac o , 273
model KPM 50 was also selec ed o he eci cula ion pump, whe eas he WT600F-65/KZ25 274
model by Golande Pump was chosen as he inle /ou le pump, wi h a Pn o 0.2 kW and a low 275
a e window be ween 0.25 and 6 L/min [51]. The selec ion o hese pumps is jus i ied based on 276
an in-dep h analysis ega ding he scaled-up low a es needed, since he low a e di ec ly 277
in luences he mass ans e and he esidence ime inside he eac o , and hence, he o e all 278
con e sion and he elec ical ene gy consump ion. Gi en ha i was desi ed o ensu e a 279
eci cula ion low a e wi h a Reynolds highe han 500, an app oxima e eci cula ion a io o 10 280
was equi ed (Fig. A.1a). Based on ha a io, he e ec o he selec ed in luen low a e on he 281
en i e elec ochemical sys em was in es iga ed. As depic ed in Fig. A.1b, an in luen low a e 282
o 0.75 L/min was equi ed o an o e all 90% con e sion, meaning ha he eci cula ion low 283
a e had o be app oxima ely 8.2 L/min. Bo h hese low a es can be deli e ed wi h he selec ed 284
pumps, and he e o e, hei ca alogue nominal powe allowed o a sui able ene gy es ima ion. 285
9
2.4 LCA amewo k 286
2.4.1 Goal and scope 287
The goal o he Li e Cycle Assessmen (LCA) s udy was o e alua e he en i onmen al p o ile o 288
he pilo -scale elec ochemical oxida ion o CBZ when se e al seconda y was ewa e 289
composi ions and eac o con igu a ions we e in ol ed. The e o e, a en ion was paid o he 290
ope a ion s age, and scena ios we e e alua ed om a ga e- o-ga e pe spec i e. Tha is, he 291
ope a ion o he elec ochemical eac o and i s associa ed pumps was conside ed, whe eas he 292
impac s ela ed o cons uc ion, decommissioning, ups eam and downs eam p ocesses we e 293
excluded. The analysis consis ed o an a ibu ional LCA ollowing ISO s anda ds 14040:2006 294
and 14044:2006 [52, 53]. 295
2.4.2 Assessmen me hod 296
The LCA was pe o med using he ReCiPe MidPoin (H) V1.06/Wo ld (2010) and EndPoin (H/H) 297
V1.06/Wo ld (2010) me hods [54] in SimaP o 9.3.0.2. so wa e [55]. The ollowing impac 298
ca ego ies we e selec ed as hey a e ep esen a i e o ene gy, oxici y and wa e e ec s: global 299
wa ming po en ial (GWP), e es ial acidi ica ion (TA), eshwa e eu ophica ion (FE), ma ine 300
eu ophica ion (ME), e es ial eco oxici y (TET) and ossil esou ce sca ci y (FRS). Addi ional 301
esul s on o he impac ca ego ies can be ound in he Supplemen a y Ma e ial, Appendix B. The 302
unc ional uni (FU) selec ed was 1 mg o CBZ emo ed pe cubic me e o was ewa e ea ed 303
du ing one day o ope a ion, and hence, i s uni s a e mg/(m3·day). Based on he di e en eac o 304
con igu a ions modelled a he pilo scale and he es ima ed in en o ies (Tables A.4 and A.5), 305
he ollowing en i onmen al analyses we e conduc ed: 306
(i)
To elucida e he in luence o he eac o ope a ing mode, he esul s o he scale-up 307
modelling o all expe imen al scena ios we e analysed in e ms o chemical and ene gy 308
equi emen s pe FU in he e ical pla e s i ed ank eac o . Based on hese esul s, a 309
benchma k on he en i onmen al p o iles o he ba ch, ed-ba ch and con inuous modes 310
was conduc ed. To his end, he ECSWL-B, ECSWL-FB and ECSWL-C scena ios we e 311
compa ed. The esul s epo ed co espond o he midpoin assessmen me hod. 312
(ii)
To disce n he e ec o he was ewa e ma ix, a benchma k on he en i onmen al p o iles 313
when di e se in luen composi ions a e ea ed in he s anda dised modula eac o in 314
ba ch mode was conduc ed. To his end, he RL-B, EC-B, ECSWL-B, ECSWH-B and MSWL-315
B scena ios we e compa ed. The esul s epo ed co espond o he midpoin assessmen 316
me hod. 317
(iii)
To de e mine he in luence o he eac o con igu a ion, a benchma k on he en i onmen al 318
p o iles o he s anda dised modula eac o and he e ical pla e s i ed ank eac o 319
ope a ed in ba ch mode was conduc ed. To his end, he RL-B, EC-B, ECSWL-B, ECSWH-320
B and MSWL-B scena ios we e compa ed o bo h eac o ypes. The esul s epo ed 321
co espond o he midpoin and endpoin assessmen me hods. 322
(i )
The o e sizing e ec was analysed o he s anda dised modula eac o . Gi en ha his 323
16
3.2.2 Benchma k o was ewa e composi ions 479
To elucida e he en i onmen al e ec s o he in luen was ewa e composi ion, he scena ios 480
ope a ed in ba ch mode (i.e., RL-B, EC-B, ECSWL-B, ECSWH-B and MSWL-B) we e analysed 481
unde he s anda dised modula eac o con igu a ion. Thei ela i e con ibu ions ac oss he 482
selec ed LCA midpoin ca ego ies a e shown in Fig. 5. I can be obse ed ha he con ibu ions 483
o he di e en scena ios a e uni o m ac oss all impac ca ego ies, wi h he scena ios wi h he 484
mos complex was ewa e ma ices (i.e., MSWL-B and ECSWH-B) being he p edominan ones. 485
The scena ios in pu e wa e (i.e., RL-B and EC-B) co esponded o less han 22% o he impac 486
o he mul icomponen sys em MSWL-B. In addi ion, be ween hose wo, he addi ion in scena io 487
EC-B o sul a e and ni a e species abo e he egula o y limi s con ibu ed o an o e all 488
educ ion o e he RL-B scena io o all ca ego ies, excep o e es ial eco oxici y (TET), whe e 489
he ela i e impac was 0.1% highe . This is due o he enhanced deg ada ion kine ics by 490
inc easing he amoun o oxida i e adical sou ces, which ansla es in o a educed ope a ion 491
ime and hence a lowe ene gy consump ion (Table A.4). None heless, a pu e wa e -based 492
ope a ion di e s om wha in p ac ice a was ewa e ea men plan will be dealing wi h. 493
The e o e, scena ios ECSWL-B, ECSWH-B and MSWL-B a e mo e in e es ing om an 494
implemen a ion pe spec i e. F om hei ela i e di e ences, i can be a gued ha inc easing he 495
was ewa e ma ix complexi y also nega i ely a ec ed he en i onmen al p o ile o he 496
ea men (i.e., an inc ease o 186-264% be ween ECSWL-B and MSWL-B in all ca ego ies), gi en 497
ha he mo e compe i ion eac ions aking place, he slowe he CBZ deg ada ion and he highe 498
he ene gy consump ion o he same emo al a ge . In ac , he main con ibu o o he 499
en i onmen al impac s o hese h ee scena ios was he elec ici y a ibu ed o he eci cula ion 500
pump (Fig. 6), accoun ing o 51-74% in he ca ego ies o global wa ming po en ial (GWP), 501
e es ial acidi ica ion (TA), eshwa e eu ophica ion (FE) and ossil esou ce sca ci y (FRS). 502
Rega ding e es ial eco oxici y (TET), he con ibu ion o he eci cula ion pump was sligh ly 503
lowe , al hough p edominan (i.e., 36-49%). Sodium ni a e was he main con ibu o in he 504
ma ine eu ophica ion (ME) ca ego y, accoun ing o 59-62% o he o e all impac . 505
3.2.3 Benchma k o eac o con igu a ions 506
Rega ding he en i onmen al p o ile o he di e en eac o con igu a ions in ba ch mode, i was 507
obse ed ha he e ical pla e s i ed ank eac o consis en ly p esen ed highe LCA midpoin 508
impac alues han he s anda dised modula eac o (app oxima ely 23-54% highe ) ega dless 509
o he expe imen al scena io and he ca ego y conside ed (Fig. 7). An inc easing end in impac 510
alues was also obse ed wi h ega d o he was ewa e ma ix complexi y, as p e iously 511
elucida ed. Only he ECSWH-B and MSWL-B scena ios displayed he same impac in he ma ine 512
eu ophica ion (ME) ca ego y o bo h eac o s, wi h less han a 3% di e ence. 513
The con ibu ions o he di e en ca ego ies o he e ical pla e s i ed ank eac o (Fig. B.6) 514
we e analogous o hose men ioned abo e o he s anda dised modula eac o (Fig. 5). 515
The e o e, he benchma k be ween bo h eac o ypes was conduc ed om an endpoin 516
pe spec i e, as i led o mo e accen ua ed di e ences among he expe imen al scena ios in 517
ba ch mode. As depic ed in Fig. 8, he single sco e indica o alloca ed o he consump ion o 518
chemicals was e y simila o bo h eac o s ega dless o he scena io unde conside a ion. In 519
addi ion, i s alue showcased a mild inc ease wi h inc easing was ewa e complexi y. On he 520

17
o he hand, highe sco es and mo e e iden di e ences we e ound ega ding ene gy 521
equi emen s. Fo scena ios ea ing pu e wa e ma ices (i.e., RL-B and EC-B), he e ical pla e 522
s i ed ank eac o sco ed 30-33% highe han he s anda dised modula eac o . As mo e 523
compounds we e ound in he in luen and hence igge ed compe i ion eac ions ha hinde ed 524
he deg ada ion kine ics, his di e ence was app oxima ely 44-49% highe o he e ical pla e 525
s i ed ank eac o , co esponding o scena ios ECSWL-B, ECSWH-B and MSWL-B. 526
527
Figu e 5: En i onmen al benchma k in he selec ed LCA midpoin ca ego ies o he s anda dised modula eac o 528
ac oss scena ios ope a ed in ba ch mode as de ined in Sec ion 2.1. GWP: global wa ming po en ial, TA: e es ial 529
acidi ica ion, FE: eshwa e eu ophica ion, ME: ma ine eu ophica ion, TET: e es ial eco oxici y, FRS: ossil 530
esou ce sca ci y. 531
532
Figu e 8: Single Sco e Indica o s om LCA endpoin analysis o he s anda dised modula eac o and he e ical 533
pla e s i ed ank eac o con igu a ions ope a ed in ba ch mode as de ined in Sec ion 2.1. 534
18
535
Figu e 6: En i onmen al con ibu ions in he selec ed LCA midpoin ca ego ies o he s anda dised modula eac o 536
in scena ios ECSWL-B, ECSWH-B and MSWL-B as de ined in Sec ion 2.1. Da a labels co espond o he con ibu ion 537
o he ho spo pe scena io and ca ego y. GWP: global wa ming po en ial, TA: e es ial acidi ica ion, FE: 538
eshwa e eu ophica ion, ME: ma ine eu ophica ion, TET: e es ial eco oxici y, FRS: ossil esou ce sca ci y. 539
540
Figu e 7: En i onmen al benchma k in he selec ed LCA midpoin ca ego ies o he s anda dised modula eac o 541
and he e ical pla e s i ed ank eac o ac oss scena ios ope a ed in ba ch mode as de ined in Sec ion 2.1. GWP: 542
global wa ming po en ial (kg CO2 eq), TA: e es ial acidi ica ion (kg SO2 eq), FE: eshwa e eu ophica ion (kg P 543
eq), ME: ma ine eu ophica ion (kg N eq), TET: e es ial eco oxici y (kg 1,2-DCB eq), FRS: ossil esou ce sca ci y 544
(kg oil eq). 545
19
3.2.4 O e sizing e ec 546
As o he s anda dised modula eac o con igu a ion, he concep o o e sizing e ec is 547
in oduced he e o accoun o he misma ch be ween he numbe o anodes comme cially 548
a ailable and he numbe o anodes ac ually needed o achie e a desi ed CBZ emo al a e. This 549
gap occu s because his eac o con igu a ion is composed o ixed s acks o 10 anodes, which 550
in p ac ice will lead he ound-up in he numbe o anodes o he closes en h mul iple. 551
Consequen ly, he ene gy consump ion and in es men cos s o elec ochemical ea men may 552
also ine i ably inc ease wi hou being e ec i ely exploi ed. The e ec o o e sizing was 553
analysed o he RL-B scena io by assuming ha he anodes can ope a e wi h di e en e ec i e 554
su ace a eas (Fig. 9). A e calcula ing he equi ed numbe o anodes o each e ec i e su ace 555
and ounding i o he nea es en h mul iple, he numbe o eac o s equi ed was ob ained. 556
Depending on he di e ences be ween he heo e ically equi ed ene gy consump ion o a gi en 557
e ec i e anode a ea and he ac ual ene gy consump ion due o he equi ed numbe o eac o s 558
( ha is, he gap be ween he s aigh and he do ed lines), he o e sizing may ange om 0.95% 559
o 54.3% imbalance. The smalles gap co esponded o 65% e ec i e a ea, and he la ges o 560
10%. In he case o he 90% e ec i e anode a ea ha was conside ed o scale up, a gap o 17.7% 561
was obse ed. 562
In o de o elucida e he e ec o o e sizing on he en i onmen al p o ile o he di e en eac o 563
con igu a ions, he 17.7% in insic excess ound in he s anda dised modula eac o was applied 564
o he e ical pla e s i ed ank eac o . Unde he EC-B scena io, a benchma k in e ms o LCA 565
midpoin and endpoin ca ego ies o he h ee esul ing eac o con igu a ions was conduc ed 566
(Fig. 10 and Fig. 11). As shown in Fig. 10, he o e sized e ical pla e s i ed ank eac o 567
p esen ed he la ges con ibu ion o all LCA midpoin ca ego ies, wi h a di e ence o 1-3% o e 568
he e ical pla e s i ed ank eac o and 20-36% o e he s anda dised modula eac o . 569
Rega ding he LCA endpoin ca ego ies (Fig. 11), he sco es o Human Heal h and Ecosys ems 570
ca ego ies we e he mos subs an ial, wi h he Resou ces ca ego y epo ing sco es up o 2 571
o de s o magni ude lowe . Fo all endpoin ca ego ies, be ween 63 and 72% o he indi idual 572
sco es we e a ibu ed o ene gy consump ion. Acco ding o his analysis, he o e sized e ical 573
pla e s i ed ank eac o was again he mos impac ul con igu a ion, ollowed closely by i s 574
non-o e sized e sion. Simila o he ou comes o he midpoin benchma k, he 17.7% o e sizing 575
did no lead o a d ama ic inc ease in endpoin sco es, as he e was a di e ence o 576
app oxima ely 2% in he Human Heal h and Ecosys ems ca ego ies and a di e ence o 25% o 577
Resou ces, al hough he la e ca ego y con ibu ed less o he o e all endpoin damage. In bo h 578
he midpoin and endpoin analyses, he s anda dised modula eac o was he con igu a ion 579
wi h he lowes en i onmen al impac , al hough inhe en ly o e sized, mainly due o he lowe 580
ene gy consump ion o he elec odes (Table A.4) ela i e o he e ical pla e s i ed ank 581
eac o (Table A.5). This educ ion o igina ed no only om he di e en anode a eas in ol ed 582
bu also om he dis ances be ween elec odes (being conside ably lowe o he modula 583
eac o ), which a ec ed he calcula ion o he po en ial di e ence a he pilo scale (Vdi ), as 584
discussed in Sec ions 3.1.1 and 3.1.2. 585
20
586
Figu e 9: Co ela ions be ween ene gy consump ion and he numbe o eac o s wi h espec o he anode e ec i e 587
a ea in he RL-B scena io. 588
589
Figu e 10: En i onmen al benchma k in he selec ed LCA midpoin ca ego ies o he EC-B scena io ope a ed unde 590
di e en eac o con igu a ions, including he o e sizing e ec . GWP: global wa ming po en ial, TA: e es ial 591
acidi ica ion, FE: eshwa e eu ophica ion, ME: ma ine eu ophica ion, TET: e es ial eco oxici y, FRS: ossil 592
esou ce sca ci y. 593
21
594
Figu e 11: En i onmen al sco es in he LCA endpoin ca ego ies o he EC-B scena io ope a ed unde di e en 595
eac o con igu a ions, including he o e sizing e ec . 596
4 T ea men selec ion and li e a u e compa ison 597
When e alua ing he mos sus ainable mode o ope a ion, he i s inding o his wo k was ha 598
con inuous ope a ion showed he highes chemical consump ion, gi en ha i s ni a e and 599
sul a e equi emen s we e abo e he egula o y limi s, and hus, hese ionic species needed o 600
be con inuously added o he was ewa e in luen . On he o he hand, he ed-ba ch ope a ion 601
could minimise such chemical consump ion wi h a 74-93% educ ion, while he ba ch 602
pe o mance anged in be ween ega dless o he was ewa e ea ed. Rega ding elec ical 603
ene gy consump ion, he ed-ba ch ope a ion was he one wi h he highes ene gy demand, 604
especially when ea ing was ewa e wi h mul iple pollu an s. This was due no only o he 605
ene gy equi ed by he ope a ion o he eci cula ion pump bu also o i s lowe was ewa e 606
olume capaci y and nega i ely a ec ed deg ada ion e iciency by compe i ion kine ics. Ene gy-607
wise, bo h con inuous and ba ch ope a ions ea ing complex was ewa e ma ices we e 608
compa able e en i he ene gy consump ion was dis ibu ed di e en ly ac oss pump ypes. 609
Howe e , when he in luen was ewa e con ained a lowe con en o o he ions and o ganics, 610
he ba ch ope a ion s ood ou as he mos en i onmen ally iendly solu ion ac oss all LCA 611
impac ca ego ies, esul ing om i s lowe ene gy consump ion and ega dless o he chemical 612
addi ions. The e o e, om an en i onmen al pe spec i e, he decision-making p ocess ega ding 613
he mode o ope a ion comes down o he ene gy- ela ed impac s, whe e ba ch is he mos 614
sus ainable op ion, ollowed by con inuous and ed-ba ch. I he addi ion o chemicals we e o 615
be a oided by ope a ing wi h was ewa e in luen s wi h enough sul a e and ni a e 616
composi ions, he ba ch ope a ion would s ill be he leading condi ion, and he o e all ea men 617
would also be imp o ed om an economic poin o iew. 618
Rega ding he e ec o he in luen was ewa e composi ion, i was ound ha a ge ing single 619
s mul iple con aminan was ewa e ma ices signi ican ly a ec ed he en i onmen al p o ile 620
o he ea men , wi h an inc ease be ween 186-264% ac oss he LCA impac ca ego ies 621
conside ed. He e, again, hese impac s de i ed om he nega i ely a ec ed deg ada ion 622
e iciencies, which esul ed in longe eac ion imes and highe ene gy demands by he di e en 623

22
pumps in ol ed. 624
In his s udy, wo di e en eac o ypes we e scaled up and compa ed in en i onmen al e ms: 625
a s anda dised modula eac o and a e ical pla e s i ed ank eac o . The la e p esen ed 626
he ad an age ha i allowed o a ully cus omisable design, as i was no es ic ed o he 627
anode geome y, size o numbe o comme cially a ailable cell modules. None heless, he 628
e ical pla e s i ed ank eac o sco ed highe o he majo i y o LCA midpoin and endpoin 629
ca ego ies, wi h di e ences up o 54% and 49%, espec i ely, esul ing om he inc eased ene gy 630
equi emen s by he elec odes and he eci cula ion pump. Based on complemen a y analysis 631
ega ding he o e sizing e ec , i was ound ha e en i 17.7% was o e sized, he modula 632
eac o was he mos a ac i e con igu a ion. 633
Consequen ly, he mos en i onmen ally iendly con igu a ion o he elec ochemical oxida ion 634
o CBZ h ough BDD anodes co esponded o ba ch ope a ion in a s anda dised modula eac o , 635
p e e ably when he in luen was ewa e ma ix had a low con en o sca enge s, such as o he 636
ions, o ganics o pollu an s. The associa ed en i onmen al impac s a he midpoin o hese 637
condi ions a e de ailed in Table B.1. Despi e he lack o compa able e e ences on LCA applied 638
o pilo -scale eAOPs o CBZ emo al, a p elimina y compa ison o ou mos p omising 639
con igu a ion in single and mul icomponen sys ems (i.e., he ECSWL-B, ECSWH-B and MSWL-640
B scena ios in he s anda dised modula eac o ) wi h espec o o he pilo -scale ea men s in 641
e ms o GWP is p esen ed he e. Gi en he di e si y o unc ional uni s, he epo ed alues ha e 642
been ex apola ed o a common e e ence o kg CO2 eq pe g CBZ emo ed (Fig. 12). Conside ing 643
ha he a e age CBZ in ake o adul s is 600 mg/day [60] and ha app oxima ely 72% is 644
abso bed by he human body [8], he CO2 emissions associa ed wi h he emo al o he daily 645
CBZ discha ge pe pa ien ha e been co ela ed o he equi alen dis ance co e ed by an 646
a e age passenge ca o he same en i onmen al impac (conside ing ha an a e age o 107.5 647
g CO2/km was emi ed in 2020 o new passenge ca s egis e ed in Eu ope [61]). The unde lying 648
calcula ions can be ound in Table B.2, al hough i should be no ed ha he compa ison be ween 649
he s udies should no be aken unques ionably, as he e a e conside able di e ences in hei 650
LCA scopes. 651
23
652
Figu e 12: Repo ed GWP impac s (in kg CO2 eq pe g CBZ emo ed) o se e al was ewa e ea men s. GWP esul s 653
a e also linked o he equi alen dis ance (in km) a elled by an a e age passenge ca o he same emissions. 654
T ea men s including he impac s associa ed wi h he in as uc u e a e deno ed wi h *. GAC: g anula ac i a ed 655
ca bon, NF: nano il a ion, SPF: sola pho o-Fen on. 656
Pesquei a e al. (2021) conduc ed an LCA on pilo -scale sola -based ea men s, including sola 657
pho olysis and TiO2 pho oca alysis (wi h and wi hou H2O2 addi ion) and nea -neu al pho o-658
Fen on [62]. Thei LCA was based on he chemical and ene gy consump ion in he pho o eac o 659
and, a a la e s age, he impac o i s cons uc ion was also conside ed (indica ed wi h * in 660
Fig. 12). Sola pho olysis exhibi ed he lowes associa ed GWP (i.e., 5 kg CO2 eq pe g CBZ 661
emo ed excluding in as uc u e), al hough i was a gued ha he applicabili y o he p ocess 662
was hinde ed by he lowe mine alisa ion e iciencies a ained. On he o he hand, sola pho o-663
Fen on p esen ed he highes impac (i.e., 57 kg CO2 eq pe g CBZ emo ed excluding 664
in as uc u e) due o he need o acidi ica ion, neu alisa ion and i on emo al s eps. As a 665
esul , sola TiO2-P25 ea men wi hou H2O2 was p esen ed as he mos sui able al e na i e, 666
conside ing ha he ca alys should be eused a leas 5 imes [62]. In absolu e e ms, his 667
ea men esul ed in a GWP o 22 kg CO2 eq pe g CBZ emo ed, which inc eased by 15.5% when 668
conside ing in as uc u e impac s. In ou s udy, he scena io ha would ou pe o m TiO2-P25 669
pho oca alysis would be he ECSWL-B scena io, wi h 7.6 kg CO2 eq pe g CBZ emo ed. By 670
inc easing he complexi y o he was ewa e ma ix, he ECSWH-B and MSWL-B scena ios 671
showed impac s o up o 18.9 and 26.5 kg CO2 eq pe g CBZ emo ed, espec i ely. Howe e , he 672
analysis by Pesquei a e al. (2021) did no include he impac s associa ed wi h he ope a ion o 673
o he equipmen such as pumps, which a e he main sou ces o elec ici y consump ion and hus 674
24
GWP. Consequen ly, i can be a gued ha ou elec ochemical se up en ails a signi ican ly lowe 675
en i onmen al impac han sola pho o-Fen on o he h ee scena ios selec ed, while a mo e 676
comp ehensi e analysis o he elec ical ene gy consump ion by sola TiO2-P25 pho oca alysis 677
is needed. None heless, he elec ochemical ea men has he added alue o no equi ing a 678
ca alys , and he e o e, a oiding he need o op imise he euse, egene a ion, ope a ing cos s 679
and en i onmen al impac s o he ca alys ma e ial. 680
Gallego-Schmid e al. (2019) e alua ed se e al pilo -scale sola pho o-Fen on (SPF) p ocesses 681
in combina ion wi h nano il a ion (NF). Thei esul s showed ha he NF uni helped o educe 682
he en i onmen al impac o acidic and neu al SPF by 38-43% by enhancing he ea men 683
e iciency. In e ms o SPF pe o mance, neu al SPF was hampe ed by he impac associa ed 684
wi h he use o an i on complexing agen , making i less en i onmen ally iendly han 685
con en ional acid ea men [63]. The highe impac s achie ed by he neu al SPF (i.e., 167.3 kg 686
CO2 eq pe g CBZ emo ed) compa ed o hose o Pesquei a e al. (2021) could be a ibu ed o 687
he highe numbe o a ge pollu an s (and hus p ocess e iciency a ec ed), he highe numbe 688
o consumables (including eagen s and i on complexing agen s) and he inclusion o anspo 689
and disman ling in he scope. As con i med h ough his s udy, ou elec ochemical ea men 690
s ands ou as mo e sus ainable, as GWP impac s a e be ween 2.2 and 22 imes lowe . Zepon 691
Ta pani and Azapagic (2018) also in es iga ed he en i onmen al impac o SPF, ozona ion and 692
o he con en ional was ewa e ea men s, such as g anula ac i a ed ca bon (GAC) and 693
nano il a ion (NF). In e ms o CBZ emo al, hei ou ea men s showed conside ably highe 694
GWP impac s han any p e ious wo k (i.e., be ween 189.1 and 312.9 kg CO2 eq pe g CBZ 695
emo ed), p esumably also due o he la ge scope o he LCA [64]. 696
In ela ion o p e ious li e a u e on LCA applied speci ically o elec ochemical oxida ion, he 697
s udies om Cha zisymeon e al. (2013) and Li e al. (2022) a e a ailable, al hough hey we e 698
applied o oli e mill was ewa e ea men and PFAS emo al om g oundwa e , espec i ely 699
[65, 66]. In bo h s udies, i was concluded ha he en i onmen al impac o he elec ochemical 700
ea men was p ima ily de e mined by he elec ical ene gy consump ion, which was also 701
obse ed in ou s udy. Thei absolu e GWP alues eached 160 and 0.205 CO2 eq pe cubic me e 702
o ea ed was ewa e , espec i ely. Unde he s anda dised modula eac o con igu a ion, he 703
ECSWL-B, ECSWH-B and MSWL-B scena ios p esen ed GWP impac s o 7.6·10−3, 1.9·10−2 and 704
2.7·10−2 kg CO2 eq pe 1 mg CBZ emo ed pe cubic me e o ea ed was ewa e du ing one day 705
o ope a ion. Compa ison be ween he h ee assessmen s is ce ainly hampe ed by he di e en 706
a ge pollu an s, was ewa e o igins, unc ional uni s and LCA scopes conside ed, as e lec ed 707
in he di e en o de s o magni ude o he esul s ob ained. The e o e, u u e LCA s udies on 708
elec ochemical oxida ion applied o he emo al o pha maceu icals a e necessa y o 709
consolida e he en i onmen al p o ile o eAOPs. 710
5 Conclusions 711
This s udy has demons a ed he impo ance o scaling up labo a o y esul s o a mo e 712
comp ehensi e e alua ion o an elec ochemical ea men , since mos o he en i onmen al 713
impac s in a modelled scaled-up pilo ope a ion we e ound o be ela ed o he elec ical ene gy 714
25
consumed by complemen a y pumps and no he elec ochemical eac o i sel . Consequen ly, 715
op imising he ene gy equi emen s o all pieces o equipmen is c ucial o aim owa ds 716
sus ainable and ca bon neu al was ewa e ea men . In his way, he e o s made o achie e 717
SDG No. 6 o Clean Wa e and Sani a ion a e no jeopa dised by inc easing he le els o CO2 and 718
o he g eenhouse gases in he a mosphe e. 719
F om an en i onmen al poin o iew, his wo k has shown ha he mos p omising eAOP o he 720
emo al o CBZ is ca ied ou in a s anda dised modula eac o ope a ed in ba ch mode, 721
p e e ably when he complexi y o he in luen was ewa e is as low as possible. Unde hese 722
condi ions, ou eAOP has been shown o ou pe o m p e iously epo ed AOPs, such as ozona ion 723
and sola Pho o-Fen on, in e ms o GWP (i.e., anging om 10% o 96% less kg CO2 eq pe g CBZ 724
emo ed). Howe e , u he LCA s udies on simila eAOPs a e equi ed o con i m hei sui abili y 725
o u u e applica ions, especially i he scope o he LCA can be ex ended o a ull plan 726
ope a ion. 727
Finally, i should be no ed ha his echno-en i onmen al analysis is based on s eady-s a e 728
modelling and would he e o e bene i om alida ion s udies. To co obo a e he obus ness 729
and e ec i eness o he eAOP o eal was ewa e ea men , expe imen s should be eplica ed 730
on a la ge scale, and dynamic modelling aspec s, such as possible al e a ions o p ocess 731
a iables (e.g., low a e, cu en densi y and was ewa e composi ion) and de e io a ion o 732
equipmen o e ime (e.g., ouling o elec odes) should be e alua ed. In addi ion, a oxici y 733
assessmen o he ea ed e luen is ecommended o ensu e sa e and iable ope a ion. 734
Acknowledgemen s 735
This esea ch ecei ed unding om he Eu opean Union’s EU F amewo k P og amme o 736
Resea ch and Inno a ion H2020 unde G an Ag eemen No 861369 (MSCA-ETN Inno EOX), om 737
he KU Leu en Indus ial Resea ch Council unde g an numbe C24E/19/040 (SO4ELECTRIC), 738
and om he HP-Nanobio p ojec (PID2019-111163RB-I00), g an ed by Spanish Minis y o 739
Science and Inno a ion. S. Es é ez hanks he Spanish Minis y o Science, Inno a ion and 740
Uni e si ies o inancial suppo (G an e e ence PRE2020-092074). 741
Compe ing in e es s 742
The au ho s decla e no compe ing in e es s. 743
Supplemen a y Ma e ial 744
The Supplemen a y Ma e ial includes addi ional esul s ega ding he scale-up modelling and 745
en i onmen al analyses as well as a li e a u e e iew. 746
32
229–234. doi:10.1016/j.jclep o.2013.05.013. 965
[66] G. Li, J. Dunlap, Y. Wang, Q. Huang, K. Li, En i onmen al Li e Cycle Assessmen (LCA) o 966
T ea ing PFASs wi h Ion Exchange and Elec ochemical Oxida ion Technology, ACS ES&T Wa e 967
2 (9) (2022) 1555–1564. doi:10.1021/acses wa e .2c00196. 968