2. Methods 2.3 CIMS Experiments utilizing the bisulfate/sulfate buffer (IO3-5, IO8, SW3-5, SW8, and CL1) sometimes exhibited cyclical CIMS signal changes for Br 2 (m/z 285, 287, 291), IBr (m/z 333, 335) with no attributable cause. These signal changes occurred seemingly at random and to varying extents. In Fig. S1a, Experiment IO4 (pH = 1.7, includes H 2 O 2) demonstrates the most extreme example of this behaviour that almost appears to affect the analysis. First at t =-3, the Br 2 rises briefly before falling. Then at t=2, the Br 2 signal begins to resemble a sine wave. All data beyond t=2 is not considered for this specific experiment. In Fig S1b, the effect during Experiment SW5 (pH = 1.7, includes H 2 O 2) is more muted, beginning at approximately t =-6 for IBr and Br 2. As represented by these figures, this behaviour being farther away from our periods of integration is typical of the remaining experiments. Because these signal changes occurred outside of the experimental periods analyzed (i.e., before irradiation, and after O 3 had been active for one hour), they are therefore not believed to affect our results and their interpretation. 3. Results 3.1 Dark reaction production of I 2 In cases without OH precursors at pH < 2, significant photochemical I 2 production still occurs (integrated production of 14 ± 10 nmol for IO8, and 6.0 ± 2.0 nmol for SW8), while Br 2 and Cl 2 concentrations remain below limits of detection (consistent with Abbatt et al., (2010), in which no Br 2 was observed without an OH-precursor) (Table 2, main text). This production likely stems from the mechanisms outlined by Kim et al. (2016) (R13-14, R10-R12), discussed in the Sect. 1. As discussed in Sect. 3.1, H 2 O 2 or NO 2 can react directly with I-, thereby reducing the available [I-] for photochemical OH oxidation when pH < 2. When H 2 O 2 was the oxidant, integrated I 2 production amounts were found to be ≤ 0.82 nmol (IO4, IO5, and SW5), likely due to this initial dark depletion. When instead NO 2 is used (as in IO3 and SW3),