Enhanced Direct Photolysis of Organic Micropollutants by Far-UVC Light at 222 nm from KrCl* Excilamps

Krypton chloride (KrCl*) excilamps emitting at far-UVC 222 nm represent a promising technology for microbial disinfection and advanced oxidation of organic micropollutants (OMPs) in water treatment. However, direct photolysis rates and photochemical properties at 222 nm are largely unknown for common OMPs. In this study, we evaluated photolysis for 46 OMPs by a KrCl* excilamp and compared it with a low-pressure mercury UV lamp. Generally, OMP photolysis was greatly enhanced at 222 nm with fluence rate-normalized rate constants of 0.2–21.6 cm2·μEinstein–1, regardless of whether they feature higher or lower absorbance at 222 nm than at 254 nm. The photolysis rate constants and quantum yields were 10–100 and 1.1–47 times higher, respectively, than those at 254 nm for most OMPs. The enhanced photolysis at 222 nm was mainly caused by strong light absorbance for non-nitrogenous, aniline-like, and triazine OMPs, while notably higher quantum yield (4–47 times of that at 254 nm) occurred for nitrogenous OMPs. At 222 nm, humic acid can inhibit OMP photolysis by light screening and potentially by quenching intermediates, while nitrate/nitrite may contribute more than others to screen light. Overall, KrCl* excilamps are promising in achieving effective OMP photolysis and merit further research.

were purchased from Sigma-Aldrich. Sodium phosphate dibasic heptahydrate (98-102%) was purchased from EMD Millipore. Information on vendor, purity, and abbreviation of tested organic micropollutants (OMP) are listed in Table S1. All chemicals were used as received. All aqueous solutions were prepared using ultrapure Milli-Q water.

Text S2. Determination of effective path length and average fluence rate of the KrCl* excilamp and LPUV lamp setups
The effective path length of the KrCl* excilamp ( Figure S1) was determined by dividing the solution volume (20 mL) by the surface area (r = 2.5 cm; A = πr 2 = 19.63 cm 2 ). Hence, d was 1.0 cm for the KrCl* excilamp. For the LPUV lamp setup with a cylindrical unit ( Figure S1), the top circular shape was assumed to be a square shape, so the effective path length was estimated by assuming the same area between these two shapes. The effective path length was estimated to be the side length of the square as 3.545 cm, obtained from d = √πr 2 (r, the radius of the circle 2 cm).
Our previous study 1 using model compounds and based on the volume-based fluence rate has obtained a similar path length and verified this assumption.
The average fluence rate of the KrCl* excilamp and the LPUV lamp photoreaction setups, respectively, was determined by iodide-iodate actinometry based on reaction S1 2 : The Φλ of triiodide ion (I 3 -) was reported as around 0.94 at 222 nm and 0.72 at 254 nm in a previous study, 3 where Φλ is the quantum yield of I 3 at λ nm, d (m) is the effective path length, and t (s) is the reaction time. The effective path lengths for the KrCl* excilamp and the LPUV lamp setups were 1.0 cm and 3.545 cm, respectively. The same setup for OMP photolysis was used for the actinometry test, so the fluence rates already accounted for reflection factor (RF), divergence factor (DF), and petri factor (PF).

Text S3. Calculation of quantum yield and fluence rate-normalized photolysis rate constant in buffered deionized water
The kinetics of direct photolysis of OMP follows equation S2: 4 where C (M) is OMP concentration, t (s) is photolysis time, Eavg(λ) (Einstein·m -2 ·s -1 ) is the averaged fluence rate at λ nm, d (cm) is effective path length, Φλ (dimensionless) is the quantum yield at λ nm, ελ (M -1 ·cm -1 ) is the molar absorption coefficient of the OMP at λ nm, and aλ (cm -1 ) S5 is the absorbance by background water matrix at λ nm. In this study, absorbance by background water matrix is minimal compared with OMP (i.e., ε λ C ≫ a λ ), so equation S2 can be simplified to: The direct photolysis of OMP also followed pseudo-first-order kinetics as shown in equations S4 and S5: where k(λ) (s -1 ) is the time-based pseudo-first-order rate constant at λ nm. One constraint for this modification is that the total light absorbance (i.e., F = 1-10 -ε λ Cd ) by OMP is less than 0.1. 5 However, for some OMPs featuring high absorbance at 222 nm or 254 nm, F is greater than 0.1, so k changes with OMP concentration along photolysis (equation S5). To obtain accurate values of k for this group of OMPs, the change of (1-10 -ε λ Cd )/Cd was controlled to be less than 10% by taking samples with small changes of C. Therefore, photolysis samples were taken before 50% removal for three naphthyl OMPs (i.e., 2-NAP, NAP, and 2-NAPA) and four triazines pesticides (i.e., ATR, CYA, PRO, and SIM) at 222 nm, before 50% removal for BZQ and ACE at 254 nm, before 30% removal for nine sulfonamides (i.e., SFA, SMX, SFZ, SMZ, SDA, SMR, SMT, SCP, and SDM), three fluoroquinolones (i.e., FLU, ENR, and CIP), and two dyes (i.e., MO and CV) at 254 nm, and before 20% removal for 9-ACA at 254 nm.
Quantum yield of OMP photolysis was reported to be dependent on the initial OMP concentration. 6 In other words, quantum yield was higher at lower concentrations of OMP due to the weaker "self-quenching" process. We aimed to run experiments at OMP concentrations close S6 to those in real wastewater or surface water samples (commonly in nM levels 7,8 ). This low OMP concentration can also maintain a small number (< 0.1) for F = 1-10 -ε λ Cd to qualify the constraint of first-order kinetics. However, due to the need to meet the instrument detection limit for analysis, 2 µM was the lowest initial concentration allowed for all OMPs for photolysis experiments. Hence, the quantum yield (Φλ) was calculated for OMP at 2 μM by the equation S6: where C is the initial concentration 2 μM.
In the literature, 5 Because of the different path lengths between KrCl* excilamp (1.0 cm) and LPUV lamp (3.545 cm) in the experiments, kE(254) was recalculated for LPUV lamp at effective path length of 1.0 cm using equation S8, so that it can be compared with results of the KrCl* excilamp. S7 where C (M) is the initial OMP concentration (2 μM Assuming that the photolysis in the presence of humic acid also follows pseudo-first-order kinetics, we can obtain the following equation: Equations S7 and S9 can be combined into equation S10: S8 Based on equation S10, kE(222) is dependent on the concentration of BPA along the photolysis experiment. To test the assumption for pseudo-first-order kinetics, the combined water factor and light absorption � was calculated for BPA concentrations from 2 μM to 0.01 μM. Figure S5 shows that the change of � is within 3% compared with the initial condition with 2 μM. Therefore, it is valid to use initial BPA concentration of 2 μM to calculate the modeled kE(222) with the consideration of light screening in this study.

Text S5. Analytical methods
The  Table S2. Two dyes (CV and MO) were analyzed by colorimetric methods at 590 nm and 464 nm, respectively. The concentrations of total organic carbon (TOC) of prepared humic acid and fulvic acid solutions were measured using a Shimadzu TOC analyzer (TOC-ASI-L; Kyoto, Japan).

Text S6. Discussion on the quantum yield of sulfadimethoxine
As an exception, sulfadimethoxine (SDM) with a six-membered ring substituted by two methoxy groups showed a relatively low increase in quantum yield. Two SO2 extrusion pathways were proposed in previous studies. [9][10][11] The major difference between these two pathways is the formation of an intermediate with or without a secondary amine group between the two rings ( Figure S4, pathway for sulfachloropyridazine (SCP) as an example). For SMT, the pathway with intermediate 4-(2-imino-4,6-dimethylpyrimidin-1(2H)-yl)aniline (without the amine) is dominant, which was previously determined by 1 H-NMR spectra. 11 However, a quantum chemical study reported that the formation of an intermediate with the amine group was prevailing for SCP. 9 Similarly, this pathway was proposed for SDM. 10 To explore why the impact of 222 nm on sixmembered sulfonamides was different between SDM and others, further studies are warranted using high-resolution mass spectrometry and NMR spectroscopy to elucidate the photolysis mechanism.

Text S7. Comparison of the roles of humic acid between 222 nm and 254 nm
A previous study at 254 nm reported that 1-5 mg·L -1 HA (Sigma Aldrich; potential carbon content 40%-56% 12-14 ) improved BPA removal. 15 This discrepancy in the effects of HA at 222 nm and 254 nm is potentially due to the different roles of direct and indirect photolysis of BPA at these two wavelengths. DOM can not only screen light and quench intermediates to inhibit the direct photolysis of BPA but also sensitize the indirect photolysis of BPA. 16 At 222 nm, BPA followed a strong direct photolysis (0.667 cm 2 ·μEinstein -1 ), and indirect photolysis of BPA sensitized by HA is potentially minor compared with the direct photolysis. Therefore, HA mainly inhibits the direct photolysis of BPA at 222 nm through screening and quenching, as discussed in the main S10 text. However, the direct photolysis at 254 nm is relatively weak (0.016 cm 2 ·μEinstein -1 ; 42 times lower than that at 222 nm), so the indirect photolysis caused by HA as a sensitizer may outperform the inhibition by HA on direct photolysis, and the net effect of HA can facilitate BPA photolysis.
It should be acknowledged that these hypotheses are only based on the results for BPA and HA.
Further studies are indeed required to assess the roles as light screener, quencher, and sensitizer played by different DOMs at 222 nm for different OMPs.    0.091 0.0021 a Molar absorption coefficient: M -1 ·cm -1 is for nitrate, nitrite, and iodide; L·g C -1 ·cm -1 is for humic acids and fulvic acids. b Environmentally relevant concentrations are 1-4 mg C·L -1 for humic acids or fulvic acids 17 , 2-10 mg N·L -1 for nitrate 18