Hydrogen Peroxide Formation during Ozonation of Olefins and Phenol: Mechanistic Insights from Oxygen Isotope Signatures

Mitigation of undesired byproducts from ozonation of dissolved organic matter (DOM) such as aldehydes and ketones is currently hampered by limited knowledge of their precursors and formation pathways. Here, the stable oxygen isotope composition of H2O2 formed simultaneously with these byproducts was studied to determine if it can reveal this missing information. A newly developed procedure, which quantitatively transforms H2O2 to O2 for subsequent 18O/16O ratio analysis, was used to determine the δ18O of H2O2 generated from ozonated model compounds (olefins and phenol, pH 3–8). A constant enrichment of 18O in H2O2 with a δ18O value of ∼59‰ implies that 16O–16O bonds are cleaved preferentially in the intermediate Criegee ozonide, which is commonly formed from olefins. H2O2 from the ozonation of acrylic acid and phenol at pH 7 resulted in lower 18O enrichment (δ18O = 47–49‰). For acrylic acid, enhancement of one of the two pathways followed by a carbonyl–H2O2 equilibrium was responsible for the smaller δ18O of H2O2. During phenol ozonation at pH 7, various competing reactions leading to H2O2 via an intermediate ozone adduct are hypothesized to cause lower δ18O in H2O2. These insights provide a first step toward supporting pH-dependent H2O2 precursor elucidation in DOM.


Table of Contents
. Model compound and DMSO concentrations used in the experiments based on the apparent second-order rate constants k for the reactions with • OH and O3.Indication of pH and applied O3 dose (cO3).

Model compound C (µM) cO3 (µM)
pH kOH, app (M -1 s -1 ) kO3, app (M -1 s - a the apparent second-order rate constants for the reactions with O3 were calculated based on the known speciesspecific second-order rate constants and the respective pKa.Measurement procedure of H2O2 quantification by the singlet oxygen method.The procedure for the quantification of H2O2 by the singlet oxygen method proceeded as follows: (1) addition of 1 mL sample to a 1cm quartz cuvette and closing with a lid containing a syringe hole, (2) addition of 1 mL 50 mM chlorine in 100 mM phosphate buffer at pH 7.0 to a syringe, (3) start measurement of the counting unit (gate time 5 ms, 2000 measurement points, C8855-01, Hamamatsu Photonics K.K Electron Tube Division), (4) fast injection of the chlorine solution by the syringe to the quartz cuvette, 5 (5) recording and integrating peaks with the spline method and processed in R (LOQ = 2.5 µM).
Table S3.Ozonation of model compounds: H2O2 yields at pH 3 and 7 determined by the singlet oxygen method.The analytical procedure for the determination of δ 18 O values of H2O2 is described in Figure S1.Steps 1-4 serve to transform H2O2 to gaseous O2 while minimizing the contamination by S6 ambient O2.Steps 5-6 involve a previously established methodology for measurement of 18 O/ 16 O ratios in O2 by gas chromatography isotope ratio mass spectrometry (GC/IRMS). 10,11gure S1.Schematic representation of the experimental procedure to transform H2O2 to O2 for oxygen isotope ratio analyses.
The six procedural steps are as follows.

1.
A standard solution contained up to 120 µM of H2O2 in 100 mL aqueous solution of 10 mM phosphate buffer in 100 mL serum bottles.Samples from ozonation experiments of different target compounds were prepared in 10 mM phosphate buffer and acidified instantly at the end of an experiment with phosphoric acid.From these reactors, aliquots of up to 10 mL were removed to determine the H2O2 concentration and measure residual concentrations of olefins/phenols and selected organic products.

2.
The H2O2-containing samples (in serum bottles) were purged for 10-15 min with N2 (sufficient time for 90-110 mL of solution) using a silicone frit inserted in a butyl stopper, to remove dissolved O2.

4.
The transformation of H2O2 to O2 by HOCl was performed by maintaining the solution pH at 7 (pH was monitored before and after transformation for each experiment).
Typically, 50-200 µL of HOCl solution (~1.5-1.7 M) and the same volume of 2 M ascorbic acid were used and injected sequentially using gas-tight Hamilton syringes.
Both HOCl and ascorbic acid solutions were prepared freshly for each experiment in 12 mL headspace vials while purging of the headspace with N2.Additionally, aliquots of 5 M NaOH were injected to adjust the pH to pH 7 (NaOH was added to the same syringe as HOCl and thus injected just prior to HOCl).

5.
A headspace was created by manually replacing 3 mL of the sample solution with N2 while holding the vials upside down. 12Hereby, a small overpressure of 1.3-1.4bar was created to minimize transfer of atmospheric O2 into the vials.After creation of headspace, the vials (still in upside down position) were placed on an orbital shaker for 30 min at 200 rpm to facilitate transfer of O2 to the gas phase.

6.
Immediately after the O2 extraction into the headspace, the samples were placed on the PAL-autosampler of the GC/IRMS device and 1500 µL of each headspace was injected.
Single injections of different 21 mL vials served as replicates.To account for O2contamination, control samples (blanks in same sample matrix containing phosphate buffer, HOCl and ascorbic acid) were prepared in the same manner, but omitting H2O2 addition or generation through ozonation.Details regarding the instrumental analysis are described in Section S3.3.

S3.2 Dissolved oxygen measurements
Dissolved O2 concentrations were measured before and after transformation of H2O2 to O2 in one of the replicates, using a needle-type oxygen microsensor (NTH-PSt7, PreSens Precision Sensing GmbH, Germany), which was freshly calibrated before each measurement campaign.
O2 concentrations after purging solutions with N2 and before initiating the H2O2 transformation to O2 were typically between 1.5-5 µM (Figure S2a, were used for blank correction) and reached up to 100 µM after addition of HOCl and ascorbic acid (Figure S2b).Blank corrected O2 concentrations of the transformed samples from ozonation of acrylic acid, cinnamic acid, phenol and sorbic acid are summarised in Figure S3.

S3.3 Oxygen isotope ratio measurements by GC/IRMS
Instrumental analysis mostly followed procedures developed by Bopp et al., 2022 and procedural descriptions are reproduced here with some modifications.Gaseous and headspace samples in 21 mL crimp vials were analysed by gas chromatography (GC)/IRMS consisting of a GC coupled via a Conflo IV interface to a Delta V Plus isotope ratio mass spectrometer (Thermo Fisher Scientific, Reinach, Switzerland).All samples were placed on a Combi PAL autosampler (CTC Analytics, Zwingen, Switzerland) for isotopic analysis.The injection was performed using a 2.5 mL gas-tight headspace syringe (HD-type, gauge 23, CTC Analytics).
Prior to sample loading, the syringe was flushed with N2 for 1 min.

S3.4 Data evaluation
Evaluation of 18 O/ 16 O ratio measurements of O2 followed peak integration and blank correction procedures as described in detail previously. 10,11In brief, automatic peak detection was performed with Isodat NT 3.0 (Thermo Fisher Scientific) applying time-based background determination 1 min prior to the O2 peak.Blank correction for diffuse contamination from ambient O2 were performed with blank (in ultrapurified water) and control samples (in sample matrix), respectively, according to established procedures. 14The peak areas of blank samples Figure S4.Variation of mean amplitudes of the blanks (light blue) and reference gas (petrol) measured on the GC/IRMS over the course of 12 individual measurement days.After higher variations on measurement day 4 and 9, the gas-tight headspace syringe for injection was replaced.The filament of the GC/IRMS had to be replaced after measurement day 4. δ 18 O of blanks and reference gas, respectively: 15.45 ± 2.06 ‰ (n = 52) and 23.88 ± 0.07 ‰ (n=221).

S4.1 Conservation of H2O2 prior to conversion to O2
Effects of time, pH and organic peroxides on H2O2 concentrations: At room temperature, H2O2 very slowly (within several days) disproportionates to H2O and O2. 15 Therefore, for the timeframe of the transformation experiment (within 2-3 hours) the disproportionation is negligible.Consequently, H2O2 concentrations originating and determined from commercially available stocks are considered stable and reliable for the transformation reactions.However, H2O2 which is formed upon ozonation of model compounds might be influenced by other solution components.One example is the presence of glyoxylic acid, a frequently detected ozonation-induced byproduct (i.e.100% yield during cinnamic acid ozonation).Glyoxylic acid and H2O2 react to formic acid and CO2 with pH dependent second-order rate constants ranging from ~100 M -1 s -1 (pH 10) to < 0.3 M -1 s -1 (below pH 5). 6Control experiments at pH 3 showed that H2O2 was stable over 2 hours (typical time period required for sample processing for conversion to O2) in contrast to samples at pH 7 (Figure S5a).Consequently, all ozonated model compound solutions were immediately acidified with phosphoric acid to pH 3 to ensure stability of H2O2.Furthermore, the presence of organic peroxides can lead to H2O2 formation as they are in equilibrium with a carbonyl compound and H2O2 (Figure 1, main manuscript).However, typically these rate constants to achieve equilibrium are in the order of 10 -6 up to 10 -4 s -1 , 16 making formation of H2O2 within the time frame of the experiments negligible.Effects of purging time on H2O2 concentrations: 10-15 min purging was sufficient to reach O2 levels < 0.4 ppm for volumes between 80-110 mL (Figure S5b).Purging removes samples were directly placed on the autosampler.Therefore, all samples could be within 12 hours after the experiment.

Effects of pH on conversion efficiency:
The rate of the reaction of HOCl with H2O2 is higher at higher pH, with a maximum at pH 9.65. 5However, the pH was always £ 7, because otherwise, side reactions may occur.At pH 3, the reaction conditions might not allow a complete transformation of H2O2 to O2 with the applied HOCl dose.Thus, the H2O2 conversion to O2 was always performed at pH 7.

Effects of DMSO on H2O2 conversion efficiency:
In ozonation experiments, an .OH radical scavenger is typically added to ensure only direct reactions between ozone and the model compound.In the present study, DMSO was used as radical scavenger, which reacts with HOCl with a kHOCl-DMSO of 315 M -1 s -1 at pH 7, 20 °C. 18Thus, higher HOCl concentrations were necessary in the presence of DMSO to maintain the same turnover.The latter is illustrated in Figure S6a, which shows the transformed H2O2 after a first (black circles) and a second (red

Effects of ascorbic acid and diffusion on O2 concentrations:
O2 can react slowly with ascorbic acid (the quenching reagent) which would lead to a depletion of O2 and thus potential isotopic fractionation.However, control measurements did not indicate a relevant decrease in O2 for isotopic measurement at pH 7. Further, diffusion of O2 from ambient air into the vials is possible, which would lead to an O2 increase and influence the δ 18 O.The latter was monitored using a blank which underwent the same procedure and no substantial increase in O2 was observed over the measurement sequence (low variation for blanks in Figure S4).O2   procedure. 19The O isotope ratios can be determined as low as 3 μM in ultrapurified water and 12 μM in sample matrix, respectively, corresponding to signal amplitudes of 376 and 496 mV.
With a typical experimental matrix (10 mM phosphate buffer and 5 mM DMSO) a higher MDL is rationalised by higher blanks.Section S5.Approach for the derivation of δ 18 O in O3 δ 18 O of O3 was determined indirectly in a mass-balance approach by measurements of O isotope ratios of O2 by GC/IRMS in an identical manner as described for transformed H2O2.
Given that O3 typically coexists with residual O2 in aqueous solutions, δ 18 O of O3 (δ 18 OO3) was derived from the comparison of δ 18 O from solutions type (i) containing both O3 and O2 (δ 18 OO3+O2) with δ 18 O of solutions type (ii) where O3 was removed and only the residual O2 (δ 18 OO2) was left behind.

Figure S3 .
Figure S3.H2O2 concentrations from the ozonation of olefins and phenol vs. corresponding measured O2 concentration (measured with a needle-type oxygen microsensor (NTH-PSt7)) after transformation of H2O2 by HOCl.The line represents a 1:1 molar ratio.

9 Figure
Figure S1.(a) Relative residual H2O2 (% molar) as a function of time and effects of pH and T in solutions with initial concentrations of 500 μM glyoxylic acid, 500 μM H O of the individual GC/IRMS measurements are shown in Figure S4.The peak areas of control samples of the model compound ozonation experiments are shown in Figures S10-S13.

Figure
Figure S5.(a) Relative residual H2O2 as a function of time and effects of pH and T in solutions with initial concentrations of 500 μM glyoxylic acid, 500 μM H2O2 and cinnamic acid (1 mM at pH 7, 0.5 mM at pH 3) and 10 mM phosphate buffer.(b) Measured oxygen removal during purging with N2 over time of a 110 mL H2O2 solution.No duplicate experiments were performed.

Figure
Figure S1.(a) Relative residual H 2 O 2 (% molar) as a function of time and effects of in solutions with initial concentrations of 500 μM glyoxylic acid, 500 μM H 2 O 2 and acid (1 mM at pH 7, 0.5 mM at pH 3) and 10 mM phosphate buffer.(b) Measure removal during purging with N 2 over time.No duplicate experiments were performe Effects of sample transfer on H2O2 and O2 concentrations: The transfer of sample vials into the glovebox was performed immediately after purging solutions with N2 to remove O2 and thus it was again assumed that disproportionation of H2O2 is negligible.O2 concentrations are expected to decrease further during sample handling because of the overpressure and continuous flushing with N2 in the glovebox.

circles) addition of 1 mL
HOCl to ~50 µM H2O2 solutions as a function of increasing DMSO (NIR-PMT measurements).The doses of HOCl and NaOH to be added to achieve full conversion in the experiments was assessed prior to each H2O2 transformation experiment with a control sample and was in the range of 9-20 µL NaOH (5M) and 50-200 µL HOCl (~1.5-1.7 M) for the 21 mL crimp vials.

Figure S6 .
Figure S6.Influence of DMSO on analytical procedure for H2O2 in buffered solutions.(a) H2O2 concentration (measured as 1 O2) as a function of increasing DMSO concentration (black circles).A second HOCl dose in NIR-PMT was necessary to quantify the remaining H2O2 (red circles) and achieve a complete mass balance (dotted line).The solid line represents the nominal H2O2 concentration.(b) Injected O2 (amplitudes in mV) detected by the IRMS as a function of increasing DMSO concentration, (c) δ 18 O values as a function of increasing DMSO concentration.Amplitudes ranged between 300 and 2000 mV.Experimental conditions: 48 μM H2O2, 0-40 mM DMSO, phosphate buffer pH 7, 10 mM.S4.3 Effects of solution constituents on 18 O/ 16 O ratio measurements by GC/IRMS
the sample handling (needle injections) were corrected using matrix blanks as described inPati et al., 2016.Effects of pH on18 O/16 O ratio measurements: The importance of a consistent pH, which ensures a complete turnover of H2O2 to O2 is exemplified in FigureS7with δ 18 O values of transformed H2O2 from cinnamic acid ozonation at pH 3 with and without pH adjustment to pH 7.

Figure
Figure S8.(a) Amplitude test for O2 from transformed H2O2 standard solutions and (b) corresponding δ 18 O values as a function of the H2O2 concentrations.Black dotted line: mean δ 18 O.Experimental conditions: ultrapurified water without pH control.

Figure 11 Figure
Figure S9.(a) Amplitude test for O2 from transformed H2O2 buffered solutions and (b) corresponding δ 18 O values as a function of the H2O2 concentrations.Black dotted line: mean δ 18 O.Experimental conditions: ultrapurified water with pH control, 10 mM phosphate buffer (pH 7) and 5 mM DMSO.

Figure
Figure S6.(a) Amplitude test for O2 from transformed H2O2 standard solutions and (b) corresponding δ 18 O values as a function of the H2O2 concentrations.Black dotted line: mean δ 18 O.Experimental conditions: in 10 mM phosphate buffer (pH 7) and 5 mM DMSO.

Figure
Figure S5.(a) Amplitude test for O2 from transformed H2O2 standard solutions and (b) corresponding δ 18 O values as a function of the H2O2 concentrations.Black dotted line: mean δ 18 O.Experimental conditions: ultrapurified water without pH control.

FigureS4. 5
Figure S6.(a) Amplitude test for O2 from transformed H2O2 standard solutions and (b) corresponding δ 18 O values as a function of the H2O2 concentrations.Black dotted line: mean δ 18 O.Experimental conditions: in 10 mM phosphate buffer (pH 7) and 5 mM DMSO.

Table S2 .
Measurement details for the model compounds.
a benzaldehyde was measured instead of cinnamic acid.

Table S4
. H2O2 yields (% of consumed O3) for the reaction of phenol with ozone at pH values between 3 and 8.

Analytical approach for the determination of δ 18 O in H2O2 S3.1 Preparation of samples for 18 O/ 16 O ratio determination in H2O2
For each measurement, 1500 µL of gaseous or headspace sample was single injected into the split injector with He as carrier gas (99.999%) and a split flow of 40 mL /min.Chromatographic separation of O2, N2 13,11r was achieved by two 30 m Rt-Molsieve 5 A PLOT column (Restek from BGB Analytik; 30 m x 0.32 mm ID, 30 μm film thickness) and a PLOT column particle trap (Restek from BGB Analytik; 2.5 m x 0.32 mm ID).10,11The O2 pulses were introduced into a GC combustion III interface (Thermo Fisher Scientific) equipped with a Nafion membrane to remove water and subsequently into a Delta Plus XL isotope ratio mass spectrometer (Thermo Fisher Scientific).δ18 O values were determined from ratios of peak areas of masses 32 and 34 versus reference gas pulses of O2 introduced at the beginning of each chromatogram (99.995%, 2.98 ± 0.15 V peak height,  δ 18 O = 0.07 ‰, n =221).The δ 18 O value of the reference gas was adjusted to O2 peaks from on-column injections of ambient air (70 µL) thereby assuming a constant δ 18 O of 23.88 ‰.13

Table S5 .
Overview δ 18 O of H2O2 from the selected model compounds at different pH values (10 mM phosphate buffer).