La flagrancia

UV fluence (often also called the UV dose, expressed as mJ cm-2_{) is the total incident radiant energy}

on an infinitesimal sphere, or in other words, the delivered UV radiation resulting in TrOC breakdown [335]. It is often linked to TrOC removal rates in photolytic and advanced oxidation processes. The fluence F (mJ cm-2_{) is the product of the average incident UV irradiation of the UV}

lamp 𝐼 (mW cm-2_{) and the irradiation time 𝑡 (s), and can be calculated as in Equation 5 [162].}

𝐹 = 𝐼 ∗ 𝑡 (Eq. 5)

The average incident UV radiation of the UV lamp is essentially the photon flux, corrected for the absorbance of the solution, which is denoted in Equation 6, where 𝐴 is the solution absorbance at 253.7 nm (determined using a VWR UV-1600PC spectrophotometer, and given in Table D.2 in Appendix D), 𝐿 is the effective path length (cm) and 𝐼0 is the photon flux (mW cm-²) [335].

The photon flux 𝐼0 is usually determined by KI/KIO3 actinometry2 and is expressed in units of

einstein s-1 _cm-2 _{or mW cm}-_{² [292]. In collimated beam setups, where the UV source is located}

above a petri dish and the UV beams are assumed to be parallel, 𝐼0 is calculated by dividing the

photon irradiation (amount based, in einstein s-1 _{or mW) over the area of the irradiated petri dish}

(in cm²). The setup used in this study was not a collimated beam, because the UV lamp was mounted inside the reactor vessel. As a consequence, UV light was radiated inside the solution from different angles, and was affected by reactor geometry, reflection and attenuation of probes (pH and temperature), sampling port, N2 flushing port and stirring turbine3. For this reason, the

photon irradiation 𝐼0 in this study was determined on an amount basis (in einstein s-1) using

KI/KIO3 actinometry [292], and was found to be 3.52 * 10-6 einstein s-1, or 1658 mW (using a

471528 J einstein-1 _{conversion for monochromatic light at 253.7 nm).}

The effective path length 𝐿 can be determined by H2O2 photolysis, which follows first order

kinetics when the absorbance is low [336]. H2O2 photolysis was performed in the reactor by

dissolving 5 mM H2O2 in mili-Q water. The H2O2 concentration was measured over time during

photolysis, using a VWR UV-1600PC UV-VIS spectrophotometer at 253.7 nm. H2O2 photolysis is

shown in Figure D.1 in Appendix D. From the first order H2O2 removal rate constant, the effective

path length can be calculated with Equation 7 [336, 337]. 𝐿 = _{−2.303∗𝜀}𝑘𝐻2𝑂2∗𝑉

𝐻2𝑂2∗𝐼0∗𝛷_𝐻2𝑂2 (Eq. 7)

With 𝑘𝐻2𝑂2 the H2O2 photolysis rate constant (determined to be -3.1325 * 10-4 s-1), 𝑉 the volume

(1.5 L), 𝜀𝐻2𝑂2 the molar extinction coefficient of H2O2 (determined in this study to be 19.65 M-1

cm-1_{, which is in agreement with a previously reported value of 19.6 M}-1 _cm-1 _{[170]), 𝐼}

0 the photon

irradiation on an amount basis (3.52 einstein s-1_{) and 𝛷}

𝐻2𝑂2 the quantum yield for H2O2 photolysis

(1 mol einstein-1_{). The effective path length was found to be 2.95 cm, which seems a reasonable}

2 _{Actinometry is a means of measuring the photon flux, i.e. the amount of photons that the UV lamp emits. It is done}

using a chemical of which the quantum yield is known (quantum yield is the amount of molecules decomposed divided by the amount of photons absorbed), and the reaction products can be easily analyzed. In KI/KIO3 actinometry,

iodine ions absorb photons yielding iodine atoms (I•_{). Iodate scavenges the bulk electrons, forming I3-, which is}

measured spectrophotometrically.

3 _{The standard method for fluence determination involves collimated beam irradiation. The fluence90 values in this}

work were calculated from experiments involving a non-collimated beam setup, and therefore should not be interpreted as exactly accurate. However, the impact from using a non-collimated setup on the (relative) results as reported in this work is expected to be minimal.

estimate considering the reactor geometry and lamp position inside the reactor. It has to be noted that this path length is averaged over the entire solution volume (or reactor geometry) for non- collimated UV radiation. This average path length was used in Equation 6 to correct for solution absorbance when calculating the UV fluence.

In Equation 6, 𝐼0 is the incident photon flux (in units of einstein s-1 cm-2) for collimated beam,

however since the UV irradiance in our system was not a collimated beam, a more correct terminology to use would be the apparent incident intensity (averaged over the entire solution volume) 𝐼𝑠. In non-collimated beam setups, 𝐼𝑠 can be calculated from the photolysis of H2O2, in

units of einstein s-1 _cm-2_{, as given in Equation 8 [170].}

𝐼_𝑠 = 𝑘𝐻2𝑂2

−2.303∗𝜀_𝐻2𝑂2∗𝛷_𝐻2𝑂2∗1000 (Eq. 8)

𝐼_𝑠 (in non-collimated beam) can be used as a substitute for 𝐼0 in collimated beam [170], and was

used in Equation 6 to determine the average incident UV intensity in our experiments. 𝐼𝑠 was found

to be 6.92 * 10-9 _{einstein s}-1 _cm-2 _{or 3.26 mW cm}-2_.

TrOC breakdown curves were plotted as ln (_𝐶𝐶

0) in function of the UV fluence corrected for the

solution absorbance calculated as shown in Equations 5 and 6, from which pseudo first order reaction rate constants (𝑘′𝑇𝑟𝑂𝐶 in units of cm2 mJ-1) were determined. Subsequently, the UV fluence

necessary to achieve 90% TrOC removal was calculated as Equation 9. 𝑓𝑙𝑢𝑒𝑛𝑐𝑒90 (𝑚𝐽 𝑐𝑚−2) =

ln_100%10% 𝑘′_{𝑇𝑟𝑂𝐶} =

−2.303

𝑘′𝑇𝑟𝑂𝐶 (Eq. 9)

3. Results and discussion