Effects of vegetables and fruit with varying physical damage, fungal infection, and soil contamination on stability of aqueous ozone

.


Introduction
The food industry has always had to respond to numerous challenges related to safety, quality, and economical imperatives.In the recent decades, the fresh food sector has had to adapt to the growing demand of healthier, less-processed food products with minimal chemical treatments and environmental impact.Amongst methods available in the industry to respond to those concerns, the use of ozone has a particularly interesting potential, owing to its strong antimicrobial effect, lack of harmful direct by-products, absence of need for storage or handling of chemicals, and low energy consumption (Aslam et al., 2020;Brodowska et al., 2018).Indeed, ozone has been used for decades in treatment of drinkable water and wastewater due to both its effectiveness and the fact that only oxygen is formed upon its degradation, leaving no unwanted chemical residues (Ikehata & Li, 2018).Thus, considering the needs of the industry, the advantages ozone confers, and the current expertise, there is a strong interest from the fruit and vegetables industry to use ozone as a postharvest cleaning medium for its products.
Ozone (O 3 ) is a powerful oxidant; it has a higher oxidation potential (2.07 V) than other common oxidants (chlorine: 1.36 V; hydrogen peroxide: 1.78 V; hypochlorous acid: 1.49 V), which effectively makes it the strongest oxidant available in the food industry (Karaca & Velioglu, 2007).Ozone is a broad-spectrum antimicrobial agent which has been confirmed to act on a wide range of microorganisms, such as bacteria, viruses, fungi, protozoa and spores; but its efficacy is known to depend on numerous variables (microbial strain, density, ozone concentration, presence of interfering compounds, or method of application) (Guzel--Seydim et al., 2004;Khadre et al., 2001).While there is no consensus agreeing on the exact mechanism of action, it is generally accepted that ozone reacts with various cell membrane and wall constituents, and internal cell content (e.g.proteins, enzymes, nucleic acids), ultimately initiating cell lysis (Brodowska et al., 2018).Moreover, the antimicrobial activity of ozone is highly circumstantial, depending on parameters such as the targeted species, mode of application or environmental conditions (Karaca & Velioglu, 2007).As such, potential matrix effects should always be considered because the toxicity of ozone is considered to not only to be due to its high oxidation potential, but also to the generation upon its decomposition of highly reactive free radical species (Tzortzakis & Chrysargyris, 2017).This includes well-known reactive oxygen species (ROS) such as hydroxyl radical (⋅OH), hydroperoxyl (HO 2 ⋅) and superoxide (O 2 − ⋅) (Fábián, 2006).
The non-selectivity of ROS affects the overall biocidal effectiveness of ozone, as organic matter and other impurities are also oxidized before, or in parallel with the target microorganisms (Aslam et al., 2020).In aqueous medium, the half-life of ozone is known to vary from seconds to hours depending on a myriad of parameters, including temperature, pH, fluid-dynamics conditions, dissolved organic compounds, water hardness, presence of redox-active species and surfaces (Gardoni et al., 2012;Staehelin & Hoigne, 1985).This instability can be seen both as an advantage and as a liability when using ozone; and means that ozone must be produced in situ.While higher ozone doses are generally linked to higher microbial reductions, overextending ozone concentrations for fruit and vegetable treatments may also lead to loss of visual, textural, or nutritional qualities (Karaca & Velioglu, 2007).This makes the use of aqueous ozone extremely commodity-and conditions-dependant.An efficient application of ozonated water in the food industry therefore relies on the actual ozone concentration during the treatment rather than the initial concentration, the type of commodity (fruit and vegetables in particular) and water quality (Aslam et al., 2020).
The effect of the presence of impurities and other compounds influencing on the ozone decay is defined as the "ozone demand of the medium" (Karaca & Velioglu, 2007;Kim et al., 2003).But since most research have focused almost exclusively on water quality parameters under well-defined laboratory conditions, there is a knowledge gap on intrinsic ozone kinetics during "real-life" situations, where the required effective ozone concentration should be achieved in presence of food commodities and where damaged, dirty, or rotten specimens contribute with reactive organic compounds to the water medium.From an industrial perspective, this type of intrinsic kinetic modelling would be essential to proper implementation of ozone-based solutions in the food industry, as it would permit correct prediction of the required contact time between products and ozone or serve as tool to adjust operating conditions (van Leeuwen, 2015).However, most studies on the use of aqueous ozone in the food industry have focused exclusively on the biocidal effects of the process, or on the effect of ozone on different food quality parameters of the products, such as colour and content of antioxidants; these types of studies have been extensively reviewed (Aslam et al., 2020;Glowacz et al., 2015;Karaca & Velioglu, 2007;Shezi et al., 2020;Tzortzakis & Chrysargyris, 2017;Vijay Rakesh Reddy et al., 2022).On the other hand, decay of ozone in aqueous medium have mostly been described using a purely physico-chemical approach targeting reaction mechanisms, the effects of parameters like temperature, pH, and water hardness and subsequent reaction kinetics under pure analytical conditions (Ershov & Morozov, 2009;Staehelin & Hoigne, 1982).Thus, there is a lack of knowledge on the intrinsic behaviour of aqueous ozone in the food industry, even if the general kinetic scheme is usually accepted (Gardoni et al., 2012;Ignatiev et al., 2008).
It is the authors' hypothesis that a larger organic load, caused either by the fruit or vegetables themselves, soil impurities, or possible fungal infections, leads to a faster ozone depletion, due to increased ozone demand of the medium.Similarly, the type of commodity and damage would also impact the decay rate of ozone, where the higher content of organic compounds and oxidable material would translate to shorter ozone half-life in water.Ultimately, a set of kinetic data can be presented which allow fruit and vegetables producers to adapt aqueous ozonebased technologies to handle various products and their physical state.Hence, the objective of this study was to outline the effects of physical damage, fungal infections and soil contamination on the decay rate of ozone when used for surface disinfection of fruit and vegetables in a water bath.Specifically, the ozone decay kinetics has been studied during application to apples, carrots, onions, celeriac, and pears; these crops have been selected as they are common commodities produced in Denmark with an annual productions between 2.5 tons/year (celeriac) to 100,000 tons/year (carrots) (Danmarks Statistik, 2020).The results will also be useful for future optimal design and interpretation of studies of disinfection of fruits and vegetables with aqueous ozone.

Chemicals and reagents
Potassium indigotrisulfonate (ITS; > 60% purity), hydrochloric acid, phosphoric acid and monosodium phosphate were obtained from Sigma-Aldrich/Merck (Burlington, MA, USA).Sodium dihydrogen phosphate were from VWR chemicals (Radnor, PA, USA).Ozone saturated water was made by bubbling ozone through an 8 mm PTFE tube into DI water prior to use.Ozone was produced in situ by a ONY-20 ozone generator (INFUSER ApS, Copenhagen, Denmark).Using ambient air as a starting material, the generator concentrates the oxygen to high levels (≥92%), before feeding it to a dielectric barrier discharge (DBD) system, which converts the oxygen to ozone.By using oxygen purified from air, one can avoid the production of NO x gases and aqueous nitric acid (HNO 3 ), inevitable by-products of conventional ozone production.Malt extract agar substrate was from Sigma-Aldrich (Saint-Louis, MO, USA).All experiments were conducted with deionized water (EC < 18.2 MΩ).

Fruits and vegetables
Fresh organic apples, pears, baby carrots, Zittauer onions and celeriac were either purchased at a local store or directly provided by collaborating producers (Hunsballe Grønt, Denmark; and Gl Estrup Gartneri A/S, Denmark).The products were bought or obtained as close as possible to the testing day, not prewashed, and if needed, stored at 5-8 • C until use.Before every experimental run, the weight and surface area were obtained, if possible.Outer surface area was recorded for apples as per the method developed by Clayton et al. (1995); and for onion a spherical shape was assumed, and the surface area was calculated from the diameter.Similarly, cross-cut area for all commodities were assumed to be round-shaped.

Fungal cultures
Mixed fungal cultures were grown from apples, pears, and onions.No isolation and identification of individual species were made, and the different cultures were named after the parent medium, e.g.fungi mix apple.Fungal mixtures were made by having small pieces of fruits and vegetables cut and placed in malt extract agar and left to incubate at 22 • C for 3-5 days.Resulting mycelia were placed in eight large petri dishes further incubated 4-8 days at 22 • C, in order to produce enough to use for ozone treatment and dry matter determination.Description and visuals of fruits/vegetables fungi can be found in the Supplementary Material.

Soil samples
An existing collection of Danish reference soils kept air-dry at University College Copenhagen were used.The soils included exhibited a variation in pH, soil organic matter (as expressed in the carbon-content) and particle size distribution covering normal agricultural soils in Northern Europe and North America.Agricultural as well as natural soils were included.
Soil characterization was previously done by University College Copenhagen using classical approaches.was recorded as pH CaCl2 (Thomas, 1996); and soil texture was defined following a combined sieving and hydrometer method (Day, 1965).Specific surface area was determined by the authors using the ethylene glycol monoethyl ether (EGME) method (Cerato & Lutenegger, 2002).
The overview of the soil samples used in the present study can be found in the Supplementary Material.

Measurement of ozone concentration
Measurement of ozone took place using the indigotrisulfonate (ITS) discoloration method (Bader & Hoigné, 1981).The colorimetric process is standard method for ozone determination in water (APHA, 2018).This method is suitable for slow-to-moderate ozone decay rates due to the stability of the indicator and the short time required between sampling points.The presence of plant materials or soil particles do not cause interference.The absorbance (λ 600nm ) was measured in plastic cuvettes using a Shimadzu UV-1280 Spectrophotometer or a Shimadzu UV-1800 (Shimadzu USA Manufacturing Inc, Canby).The concentration of ozone was subsequently calculated by Equation ( 1): where [O 3 ] = ozone concentration (mg/L), V total = total volume (mL), V sample = sample volume (mL), A blank = Absorbance value of the blank sample, A sample = Absorbance value of the sample, b = cell pathlength (1 cm), and f = 0.42.The factor f is based on a sensitivity factor of 20,000 cm − 1 at 600 nm, as given by the standard method developed by Bader and Hoigné (1981); as well as on an absorption coefficient for aqueous ozone of 2950 M − 1 cm − 1 at 258 nm.

Water quality
Acidity (pH), O 2 -content, electric conductivity and temperature were monitored before and during all ozonation experiments (MULTI meter MU 6100H; pHenomenal electrodes 111, CO11 and OXY11; VWR International bvba, Leuven, Belgium).The instrument was calibrated at regular intervals using pH 4.00 and pH 7.00 buffer standard solutions and a 0.01 M KCl conductivity standard solution (VWR International bvba, Leuven, Belgium).

Experimental setupall ozonation experiments
The experiments were designed to cover the industrial washing process using aqueous ozone comprising 3 separate steps: 1) Saturation of water with ozone to a desired concentration level; 2) Washing fruit and vegetables including fungi and soil in ozonated water; and 3) Removal of the fruit and vegetables for further processing leaving process water behind.Measurements of ozone and water quality parameters took place during each step.
To remove any potential impurities and interfering material, preozonation of the water was done the day prior to each experiment.The ITS reagent was prepared in advance and divided in smaller beakers 9.7 mL aliquots in prevision of the ozone measurement, where one beaker corresponded to one timepoint.A recipient was filled with 0.8-15 L of pre-ozonated water, depending on the size of the commodity of interest.The temperature of the wash water was kept at 15 ± 3 • C using a cooling system and stirring was kept to a minimal level.The wash water was then ozonated for approximately 7 min, enough to reach an initial ozone concentration of 7-11 mg/L and left to rest for 3 min.
The ozone decay experiments were divided into three phases mimicking an industrial fruit and vegetable washing process (Fig. 1): 1) Initial ozone decay before washing; 2) Decay in presence of fruit/vegetables/fungi/soil; and 3) Residual decay after washing.A minimum of 5 sample points was taken for the first phase, around 10 (or enough to cover 2-3 ozone half-lives) for the second phase and if possible, another minimum of 5 sample points for monitoring the residual ozone in phase 3. The initial decay consisted of the natural aqueous ozone decay.After this phase, the studied commodity was added to the wash water until all products were completely submerged.They were left in the wash water enough time for the ozone to decay around two or three half-lives; at that point they were removed from the ozonated water.When applicable, i.e. if the ozone concentration did not drop under the lower detection limit, the residual ozone concentration was monitored for a few minutes.
The ozone concentration during the run was measured as follow: at every timepoint, an aliquot of the ozonated water was taken and immediately mixed with the corresponding ITS reagent beaker.As the ozone decolorized ITS reagent is stable for up to 4h under normal conditions, the absorbance reading for all timepoints was done in one go after each individual run.The decay rate constant, k, was calculated assuming a first-order reaction (Equation ( 2)).
The determination of ozone decay in the presence of soil and fungi mix followed the same procedure.The amount of soil per run (10 g/L) was determined during pre-trials in order to achieve an ozone decay measurable by the ITS method.To avoid having soil particles interfere with the colorimetric measurements, the ozone-quenched ITS samples were transferred to the reading cuvette by means of a syringe fitted with a 0.45 μm pore size filter.A pre-study was first conducted to ensure that the filtering step did not affect the reading.Fungal mycelium (0.25 g/L) was placed in a steel mesh tea strainer during ozonation.Ozone decay for the experimental setup was monitored with and without the tea strainer submerged when no fungi mix was added.The setup was not found to influence the ozone degradation rate.

Preparation of commodities
To emulate broken fruit/vegetables during the cleaning phase at the production site, each commodity was divided in up to five different damage levels corresponding to increased exposed cross-cut surface, starting from the intact commodity (cross-cut surface = 0 cm 2 ) (Table 1).The products were cut in a way to leave a cross-section as round as possible to ease the area calculation.Tests were done in triplicate.

Statistical analysis
Data analyses (non-linear regression, descriptive statistics) were performed in Microsoft Excel 2010 (including data analysis package).For the main part of the experiment (damage levels), Principal Component Analysis (PCA) was conducted using the software LatentiX 2.12 (LatentiX ApS, Copenhagen).The different samples (commodity x damage level) were used as objects, and the other variables considered were as follow: (1) initial ozone concentration, (2) phase 2 (during commodity submersion) k value, (3) total ozone exposure (area under the ozone concentration curve during phase 2, AuC), (4) weight, and (5) surface area (cross-cut).The data was autoscaled before calculating the PCA models.The raw data tables can be found in the Supplementary Material (Online Resource 2).
The phase 1 k value (initial decay) was used as a quality control parameter where decay rates larger than 5.78 ⋅ 10 − 4 s − 1 were rejected.This corresponds to an ozone half-life of 1200 s (20 min), which is the generally accepted lower range for distilled water at 20 • C (Khadre et al., 2001).It is assumed that higher decay rates would suggest unacceptable levels of interference from external parameters on the aqueous ozone stability.In average, the k value for the initial decay was 3112 s (52 min), owing to the colder temperature of the water during this experiment.

Multivariate analysis
A Principal Component Analysis (PCA) of data related to the ozone decay experiments was carried out to identify important relationships (Fig. 2).Sample point clusters were easily identified in the score plot when sorting the data points either according to type of commodity (Fig. 2A), or by damage level (Fig. 2B).The commodity clusters tended to be separated along the PC2 axis, and as seen in the scores plot, weight and commodity were both located along the PC2 axis (Fig. 2C).The clusters of damage levels, on the other hand, tended to be separated along the PC1 axis, where the scores plot shows that this is correlated to ozone decay rates, damage level and surface area.In general, intact products tend to be clustered together (Damage level 1, Fig. 2B), with samples with increasing ozone decay rates, k, corresponding to increased cross-cut surface areas (Fig. 2B).The area under curve parameter, AuC, which is equivalent to the total amount of ozone that was consumed in the presence of the commodities was as expected located close to the rate of ozone decay in the scores plot.Overall, the PCA showed that commodity type, weight, surface area and damage levels were the most important factors determining the ozone decay kinetics.

Initial ozone decay
Assuming a first-order reaction following Equation (2), the average k value for the first phase (initial decay) reported across all samples averaged 2.2⋅10 − 4 s − 1 with a standard deviation of 1.1⋅10 − 4 s − 1 .This value was obtained with conditions favourable to ozone stability, using triple-distilled water at temperatures between 8 and 15 • C. The temperature range may have affected the variation of the initial decay, which is consistent with previous findings.Gardoni et al. (2012) inferred from six different studies a decay constant k for a first-order reaction with respect to ozone of 2.0⋅10 − 4 to 3.0⋅10 − 4 s − 1 in the absence of compounds susceptible to reaction with ozone.The ozone half-life in undisturbed water in the experimental settings was thus about 52 min, well within the 40-60 min range as per Gardoni's findings.

Ozone decay in presence of intact and cut commodities
When considering decay of ozone in presence of intact commodities (phase 2, level 1), the values of k for both pears and apples were within or slightly over the range of natural ozone decay as described previously: (2.5 ± 0.3)⋅10 − 4 and (3.2 ± 0.5)⋅10 − 4 s − 1 (95% C.I.), respectively, indicating minimal influence on the ozone decay.On the other hand, the rate constants determined during phase 2 for intact onions and carrots were found to be (1.2 ± 0.3)⋅10 − 3 and (1.3 ± 0.1)⋅10 − 3 s − 1 , respectively.These rates correspond to a half-life of ozone of roughly 9 min, against a half-life of 35-45 min for pear and apple samples.This demonstrates that the ozone demand of the medium is highly dependent on the type of commodity present, as intact apples and pear would tend to react less with ozone due to the lower reactivity of their skin with oxidation process, as opposed to their flesh (Juhnevica-Radenkova et al., 2018).
As confirmed by the PCA, the most obvious trend with the ozone decay is the difference between the diverse commodities on their respective aqueous ozone demand when cross-cut sections were exposed (Table 1).Onions with a cross-cut surface were the most ozoneconsuming samples with an average k value of 6.1⋅10 − 3 s − 1 at damage level 5, resulting in an ozone half-life just under 2 min (112 s).On the other hand, with a higher cross-cut surface area exposed to aqueous ozone (around 270 cm 2 for apples and pears, against 220 cm 2 for onions), apples and pears gave a k value of 2.38⋅10 − 3 and 2.74⋅10 − 3 s − 1 , corresponding to half-lives of 290 and 250 s, respectively.For the three commodities onions, pears and apples, the ozone demand of the medium increased linearly with the exposed cross-cut surface area (Fig. 3), with a rate of 1.85 ⋅ 10 − 5 , 9.4 ⋅ 10 − 6 and 7.0 ⋅ 10 − 6 s − 1 /cm 2 of cross-cut surface for onions, pears and apples, respectively.The linearity of the relationship is further supported by the R-squared value for each product: 0.942 (onions), 0.952 (apples) and 0.986 (pears).Carrot samples however did not follow such a regular trend, and their k values for phase 2 (between 1.2 ⋅ 10 − 3 s − 1 and 2.2 ⋅ 10 − 3 s − 1 ) were corresponding to an ozone half-life comprised between 5 and 9 min, with seemingly no relation to the area of the cross-cut surface (R-squared of 0.317).These results are expected to be linked to the amounts of cell components that become soluble upon the tissue breakage by the cross-cutting.This is related to the phenomenon juiciness, which is the release of intracellular content during biting and chewing (Mercado et al., 2019).Carrots, which have a low juiciness, gave a low ozone decay rate, while apples, which reacted faster with ozone, have a higher juiciness (Harker et al., 1997).As carrots would mainly consume ozone through a surface reaction, this would limit ozone demand to a narrower zone of action, while reaction across the whole container for juicier commodities would trigger a larger chain reaction starting with ozone and water, cascading into complex radical formation further intensified by the presence of organic compounds from the fresh produces (Staehelin & Hoigne, 1985).
In terms of type and amounts of potential ozone-scavenging compounds in the different tested products, pears and apples would tend to present a similar profile, while onions and carrots would exhibit more different contents.As fruits, both apples and pears have high sugar and polyphenols contents (Hyson, 2011); those compounds contain several conjugated double bonds and hydroxyl groups, which are notorious targets of ozone-induced oxidation.Onions also present compounds readily oxidized by ozone: in addition to carbohydrates, they exhibit higher content of sulphur-containing products as well (Butnariu & Butu, 2015).Carrots, of course, contain high amounts of αand β-carotene, also known targets of oxidation by ozone due to their conjugated structure (Benevides et al., 2011).However, carotenes are also known for their low solubility in water (Ishimoto et al., 2019); this would also influence on the lower effect that carrots had on aqueous ozone consumption.

Effect of soil contamination on ozone decay
Since soils potentially could be attached to vegetables, the decay of ozone was also tested in slurries of soils.Each soil type showed a marked effect on the aqueous ozone decay (Table 2).As previously stated, the lowest accepted initial half-life for aqueous ozone in this experiment was set at 20 min.In average, the initial decay recorded during the commodities experiment was 2.18 ⋅ 10 − 4 s − 1 , corresponding to a half-life of 53 min.In contrast, the soil that had the strongest ozone demand (for a load of 10 g/L) exhibited an ozone half-life of a mere 51 s, while the soil with the least demand still showed a bit over 10 min ozone half-life (626 s).
Multiple Linear Regression and Partial Least Squares Regression were attempted to model the rate constant as function of soil physicochemical variables.However, the attempt was unsuccessful.As Danish soils were chosen (an overview of the soil samples parameters can be found in the Supplementary Material), all samples roughly shared the same type of soil texture: with higher content of sand compared to clay or silt in the ranges of 1-14% and 0-16% w/w respectively.Virtually all samples were classified as loamy sand or sandy loam.The resulting specific surface area exhibited limited variation (126-190 m 2 /g).As such, particle size could be considered a non-factor in this experiment.The main difference in the soils lied in their horizons: A, B or C.Those tree horizons refer to their typical depth and composition, where the surface horizon A would typically contain more organic matter as a result of accumulation (the mull/mor/plow-layer), while horizon C would usually exhibit little alteration to the composition of the parent material comprising glaciofluvial and calcareous morainic deposits from the last glacial stage in Denmark (soil age approximately 10-15.000years) (Madsen et al., 1992;Schoeneberger et al., 2012).The logical outcome of this is visible in the results: soils from horizon A mostly have the highest ozone demand, while soils from the horizon C are demanding the least.The two B horizons represent two main classes: Bhs having accumulated eluvial humic matter and precipitation of sesquioxides while Bt have an accumulation of clay particles.Although of different origin, the specific surface area is approximately the same.Ozone depletion is faster in the Bhs horizon indicating the effect of organic matter.It is important to note that amongst A horizons the differences can be drastic.For example, in the case of the Knudshoved soil, the top layer showed a lower ozone demand than the B horizon of the same location and results closer to C horizons.The origin of the organic matter could be an explanation: The Bhs comprise older and eluvial organic matter, i.e., previously dissolved organic matter that has precipitated with depth, while the A horizon comprise more recently formed and less humified organic matter from living plants.Further research is required to shed light on this.
Soil contamination of commodities will result in a marked increase in the ozone demand during the disinfection process.Contamination with organic matter rich soil material will have the most marked impact on the ozone demand, but in general the difference between different soil types is low -within a factor of 10 lowered ozone half-life.Hence, soil can influence the ozone demand during industrial surface disinfection and should be taken into consideration.

Effect of fungi on ozone decay
Ozone degraded significantly faster when in contact with 0.25 g/L fungal mycelium.From an initial decay rate constant of 1.53⋅10 − 4 s − 1 (corresponding to an ozone half-life of 4516 s), ozone degraded 3 to 5 times faster in presence of fungi (Table 3).When ozonated water was in contact with pear-, apple-and onion-derived fungi, the decay rates constants were found to be 5.80, 7.65, and 8.19 s − 1 , respectively, corresponding to ozone half-lives of 1196, 906, and 850 s, respectively.Small droplets of water and/or exudates were present on the hyphae adding some uncertainty to the experimental sampling (RSD dry matter content up to 67%).However, this did not influence the decay rates exhibiting narrow 95% confidence intervals.Considering the trace amounts of fungal contamination that is expected in production sites (Tournas, 2005), this parameter is expected to have a minor effect on the ozone decay during industrial surface disinfection.

Conclusion
Fruit and vegetable commodities proved to have different effects on the decay rate of ozone in aqueous solutions.Intact commodities have little to moderate effect on the ozone decay rates (ozone half-life > 20 min for apples and pears, approximately 10 min for carrots and onions); while the presence of damaged goods could cause a significantly increased ozone demand, reducing the ozone half-life to under a minute in some cases.The variation observed between the different products suggests that a preliminary screening would be required for any production wishing to implement an ozone-based washing step.This screening step would be important to estimate the effective rate for ozone replenishment, as failure to address this could cause an early depletion and ineffective disinfection.Moreover, any site with such a system implemented would also have to examine their production line for damaged products and/or soil impurities.Ideally, and especially in the case of tuber or root vegetable production, a pre-washing step postharvest could be necessary prior to ozone treatment.
The kinetic data would be valuable to estimate the necessary ozone input throughout the process; more data should be collected for additional commodities.From a cost-and environmental perspective, the use of ozonated water for disinfection presents promising upsides; and its biocidal efficacy has been extensively demonstrated.However, the present study highlights the need to further consider the ozone demand of the medium, especially when operating under aqueous conditions.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 1 .
Fig. 1.Overview of ozone decay experiment with three phases: (1) Pure ozonated water: Initial phase where the decay of ozone is governed by pure aqueous phase reaction kinetics; (2) Presence of commodities: Phase where rapid degradation of ozone is observed due to reactions with surface compounds, juicy leachates, fungi and soil; and (3) Effect of residuals: Final phase where residual ozone continues to react with compounds left in solution after the removal of the commodities.

Fig. 2 .
Fig. 2. Principal component analysis of data related to ozone decay experiments (Table 2 in supplementary material) (A) PCA score plot colored according to type of fruit and vegetable products.(B) PCA score plot, sorted by damage levels with numbers corresponding to increasing cross-cut area as reported in Table 1.(C) PCA loading plot associated with score plots A and B.

Table 1
Ozone degradation in presence of fruit and vegetables.