One batch of sow manure and one batch of dairy cattle manure were collected from two commercial farms. To obtain three manure samples per animal type with different amounts of potentially ozone reactive compounds, two separation techniques were applied to both manures.

One part of each manure was separated on-site with a commercial farm-scale screw press (SB Engineering, Aalestrup, Denmark) using a 250 µm filter.

One part of the sow manure was separated in a commercial farm-scale unit (AL-2 Teknik, Hovborg, Denmark) using polymer flocculation and filtration. One part of the cattle manure was separated in 0.6-L batches in the laboratory using polymer flocculation and filtration. Cationic, high molecular weight, linear polyacrylamide was used in both treatments. The optimal polymer charge density and dosage were determined based on floc size, dewaterability, volume separation, liquid turbidity, and solid dry matter content [19]. For the sow and cattle manures, 0.45 mL of 50% Superfloc c-2260 emulsion (40 mol% charge density) and 1.7 mL of 50% Superfloc c-2240 (20 mol% charge density) emulsion (Cytec, NJ, USA) were used per liter of manure, respectively. The sow manure was sieved on a 250 µm roller belt filter, and pressure was applied using a drum roller. The cattle manure was sieved through a 200µm filter, and no pressure was applied.

Six samples were therefore obtained: one raw sow manure (s1), one screw-pressed liquid from the sow manure (s2), one flocculated liquid from the sow manure (s3), one raw cattle manure (c1), one screw-pressed liquid from the cattle manure (c2), and one flocculated liquid from the cattle manure (c3).

Vessels (5 L, Ø 170 mm) were filled with 2.5 L batches of manure, and the manure was continuously stirred. Two different stirring methods were used. Tests proved that they resulted in equal effects on redox potential and H₂S emissions. The sow manure was stirred with an overhead stirrer (RZR 2041; Heidolph, Schwabach, Germany) and an impeller (BR 13; Heidolph) at 180 rpm. The cattle manure was stirred manually by spinning the gas inlet diffuser at ∼85 rpm.

An ozone-enriched oxygen stream was generated by passing pure oxygen through an electrical discharge ozone generator (Ozonia LAB2B; Degrémont Technologies–Triogen, East Kilbride, Scotland). The ozone generator was fed oxygen from a gas cylinder at a flow of 2 L/min at a pressure of 1.5 bar. The generator was set to its maximum output level, resulting in a gas containing 1.5–2.2% O₃ and 97.8–98.5% O₂. The ozone concentration was measured before each experiment with an ozone analyzer (UV-100; Eco Sensors, Fresno, CA, USA), using pre-dilution in order to be within the analytical range. The ozone-enriched oxygen stream was led through Teflon tubes, and injected at the bottom of the manure-filled vessel through a stainless steel solvent filter (A-230A; Upchurch Scientific, Oak Harbor, WA, USA, pore size 20 µm) used as a diffuser for dispersion of the added gas. The O₂/O₃ mixture was added continuously at a constant rate, for the various experiments at 0.075–0.18 L/min. Depending on outlet ozone concentration measured before each experiment, the addition rate was 2.5–5.5 mg O₃/min. The maximum ozone addition was 350 mg O₃/L manure and 17,000 mg O₂/L manure. To apply different ozone doses, the treatment time of the 2.5-L sample was varied between 10 and 160 min. The ozone was fully dissolved and consumed in the manure for all treatments. This was ensured through continuous measurement of the ozone level above the surface of the manure.

Additionally, two control treatments were performed on another pig manure by adding N₂ and O₂. For the N₂ treatment, pure nitrogen was added at 0.06 L/min for 80 min, and for the O₂ treatment pure oxygen was added at 0.07 L/min.

The analyses performed on the non-ozonated manure samples were: pH, particle sizes, total dry matter, total chemical oxygen demand (COD), volatile fatty acids, total NH₃/NH₄⁺ concentration and total H₂S/HS⁻/S^2- concentration. Particle size range was analyzed using laser diffraction (Master Sizer 2000; Malvern Instruments Ltd, Worcestershire, UK). pH was measured using a pH electrode (InPro450IVP/PT100 SG; Mettler-Toledo, Zurich, Switzerland). Total dry matter was determined gravimetrically as the weight loss upon heating the sample to 105 °C (APHA, 1992). The total COD concentration was determined by performing destruction, color reaction, and spectrometric quantification using the Spectroquant Kit 114555 (Merck KGaA, Darmstadt, Germany). The total NH₃/NH₄⁺ concentration were measured by performing color reaction and spectrometric quantification using the Spectroquant Kit 100683 (Merck KGaA). The volatile fatty acids, including butanoic acid, were measured on a gas chromatograph [20] The total H₂S/HS⁻/S²⁻ concentration was measured by precipitating with zinc, capture of the H₂S gas, color reaction and spectrometric quantification using the method described by Eriksen et al. [21].

The continuous analyses performed during treatment were: manure pH, manure redox, gaseous H₂S concentration, and gaseous ozone concentration. The pH and redox electrodes were submerged 5 cm below the surface of the stirred manure. To measure the pH and redox of the manure continuously, the pH electrode and an oxidation reduction potential (ORP) electrode (Pt4805-DXK-S8/120; Mettler-Toledo/Thornton, Switzerland) was used. The inlet tubes for the gaseous ozone and H₂S concentration analyses were placed 10cm above the manure surface and 10 cm below the vessel top. Ozone and H₂S emissions were measured continuously using the ozone analyzer and a gold film detector (Jerome 631x; Arizone Instruments LLC, Chandler, AZ, USA), respectively. The detection limits of the apparatus used for the H₂S emissions analyses were >0.001 ppm and <45 ppm. When lower and higher levels were detected, the values of 0.001 ppm and 45 ppm, respectively, were logged.

In total, 29 experiments with additions of ozone enriched oxygen were performed. The number of treatment replicates for the individual manure samples s1, s2, s3, c1, c2, and c3 were 5, 4, 4, 6, 6, and 4, respectively. Measurement of H₂S was performed on 25 treatments. The number of treatment replicates for the individual manure samples s1, s2, s3, c1, c2, and c3 were 4, 4, 4, 5, 5, and 3, respectively.

Mean curves of “ozone amount versus redox potential” and “redox potential versus gaseous H₂S concentration” were plotted using a moving average trendline with the number of points being equal to the number of replicates.

The amount of ozone required to minimize H₂S emissions was calculated as the intercept of the linear regression for two of the observed redox phases: the redox rise and the final stable redox phase (Figure 1). The ozone amounts were compared by applying Student's t-test (2-sided, α = 0.05).

Six animal slurry batches in three particle size ranges (Table 1) were oxidized using ozone enriched oxygen in order to remove malodorous H₂S emissions. The two full slurries (s1 and c1) were different in the composition. Separation caused the particle ranges to change, as did also the content of particulate compounds. The content of soluble compounds such as H₂S/HS⁻ were also affected, as these can be adsorbed to the negative particles, acting as particle counterions, adsorbed to their cationic polymer or dissolved in the water retained in the solid fraction. Hence, the batches contained different amounts of potentially ozone reactive compounds. Despite the differences between the batches, the oxidation profiles exhibited similar rises in redox potential (Figure 2).

A relatively rapid rise in redox potential was observed approximately from −275 mV to −10 mV (Figure 2). Although the specific redox potential profiles varied, all exhibited as a minimum a rapid rise from −200 mV to −80 mV. Hence, similar reactions occur in the manure samples. The amount of ozone enriched oxygen added during this rise differed between the six manure batches. The redox potential increase was followed by a phase with relatively constant redox potential, which was observed to continue until the experimental maximum additions, i.e., up to 350 mg O₃/L manure.

The gaseous concentration of H₂S above the manure changed as a function of addition of the ozone enriched oxygen along with the change in redox potential (Figure 3). Reductions in concentration were also observed upon addition of pure oxygen, whereas N₂ addition did not cause reductions in the H₂S emissions.

The measured H₂S emissions upon the treatment with ozone enriched oxygen remained constant at approximately 10 ppm from −400 mV until −200 mV; however the measurements were limited by the upper detection limit of the analysis equipment of 45 ppm. The H₂S concentration in general decreased from this level to <0.05 ppm at around −80 mV (Figure 3). This is equivalent to a decrease in H₂S emissions of more than 99%.

Previous studies have indicated that the H₂S emissions were depleted between −208 mV and 0 mV [9,17,18]. This is largely in agreement with the present findings of reductions in H₂S emissions around −80 mV upon ozone addition.

The decrease in H₂S emissions upon addition of ozone enriched oxygen (Figure 3) occurred simultaneously with the observed redox potential rise (Figure 2), which for all samples was observed from −200 mV to −80 mV. The ozone amount that had been added when this change took place differed between the six manure samples. For all samples, however, the amount of ozone needed for reducing gaseous H₂S concentration (corresponding to a rise in redox potential from −200 mV to −80 mV) was low compared with the total amount of ozone added when the minimum H₂S emissions were obtained (25–65%).

The observed decrease in the gaseous concentration of H₂S above the manure is ascribed to removal of H₂S in the manure. This could be through oxidation by ozone, but it may also occur by oxidation by molecular oxygen, microbial oxidation, acid-base reaction and volatilization during gas addition.

The total ozone requirement for minimizing the H₂S emissions (Figure 4) increased as the total H₂S/HS⁻ concentrations in the non-ozonated manure (Table 1) was increased (R² = 0.89) (Figure 5(a)). Hence, ozone's initial reaction may be a reaction with reduced dissolved sulphur.

The very initial reactivity of ozone towards H₂S and HS⁻ compared to the reactivity towards other dissolved species can be assessed based on existing kinetic data from the literature [13,22]. The time for reduction of the ozone concentration to 1/e (Eulers number) times the initial concentration, i.e., the lifetime of ozone τ_O3, with respect to dissolved species, can be estimated if the initial concentration of the dissolved species is known: (1) d [ X ] / dt = k 2 nd [ X ] [ O 3 ] = k ′ [ O 3 ] ; k ′ [ O 3 ] = k 2 nd [ X ]where k' is the pseudo-first-order rate constant, [X] denotes the concentration of the compound that reacts with ozone, and k^2nd denotes the second order rate constant for the reaction of ozone with compound X. Because the initial concentration of ozone is very low, the concentrations of other reactants in the manure can be assumed constant in a short initial time interval, and a pseudo-first-order expression for ozone's lifetime can be applied: (2) τ o 3 = 1 / k ′ = 1 / ( k 2 nd [ X ]

The calculated lifetimes are only approximations of the very initial reactivity, i.e., until the reactions leads to reductions in concentrations of the species reactive towards O₃. Despite this, it is however possible to assess the relative importance of various compounds for consumption of ozone.

The calculated initial ozone lifetime with respect to manure relevant species are the lowest for HS⁻ (Table 2). This supports the observed increase in the total ozone requirement at increasing H₂S/HS⁻ concentration in the non-ozonated manures (Figure 5(a)). Consequently, it is assessed that only when the concentration of H₂S/HS⁻ is lowered several orders of magnitude will ozone be available for degradation of other compounds. Non-dissociated phenols are, despite the low concentration, the largest competitor with HS⁻ for reaction with ozone, which in particular would be the case if the pH was higher. The calculated ozone lifetimes, however, indicate that it is very likely, that the added ozone oxidizes HS⁻ in the manure.

The calculated lifetime of ozone upon reaction with carboxylic acids, 4-methylphenol, other phenols and anions of phenols are high compared to the lifetime upon reaction with HS^– (Table 2). Hence, the main ozone consumption is not indicated to be caused by dissolved organic compounds. No relationship between ozone requirement and dry matter content or total chemical oxygen demand (COD) was observed (Figure 5(a), Table 1, Figure 4). Neither were dry matter and total COD entirely responsible for the ozone requirements within the manure types. This is in agreement with previous studies, in which only weak correlations were found between dry matter content and the oxidation level [23,24] and corroborates that ozone predominantly is consumed by H₂S/HS⁻ as long as it is present.

If ozonation was the only reaction responsible for the H₂S removal and if competing ozone reactions are assumed to be insignificant, the initial amount of the reduced sulphide (Table 1) should approximately be equal to the amount of ozone required to minimize H₂S emissions (Figure 4). The molar ratio of ozone required for H₂S elimination relative to the initial amount of total reduced sulphide were calculated to be in the range of 0.1 to 1.1 for the six batches of manures. Assuming a 1:1 stoichiometry of the reaction and no secondary reactions of importance, only in one batch (c3; molar ratio = 1.1) is the amount of ozone added sufficient to explain the removal of the reduced sulphide. Hence, the reduced sulphide is also removed by other mechanisms than ozonation.

Chemical or microbial oxidation by co-added oxygen (the ozone-enriched oxygen stream contained 1.5–2.2% O₃ and 97.8–98.5% O₂) may cause the additional decrease in the H₂S emissions [25–28]. Based on a previous study [28] it is estimated that <3% of the added O₂ is consumed by the slurry. Oxygen oxidation is observed in the control experiment in this study upon addition of pure oxygen (Figure 3). However, reaching the low H₂S emission level required 0.2–0.7 L gas per L manure upon addition of the O₃ and O₂ gas, while it required approximately 1.4 L gas per L manure upon addition of pure O₂. That is the approximately 2% O₃ in the added gas causes the reaction rate per gas volume to increase 2–7 times. Hence, chemical or biological oxidation by O₂ also contributes to reducing H₂S concentration, but the reaction rate with O₂ is much lower than O₃.

The total experimental treatment time was 10 to 160 min. The significant decrease of 99.5% in gaseous H₂S concentration, however, occurred within 3–6 min. The short experimental duration and the peak in the removal rate are atypical for microbial oxidations. Hence, microbial oxidation is not likely to have caused the decrease in gaseous H₂S concentration. Hence, the observed oxidation by O₂ were due to chemical oxidation.

An increase in pH and hence a decrease of dissolved H₂S relative to HS^– would also lead to lower gaseous emissions of H₂S. A continuous pH increase, due to the emission of e.g., H₂S and CO₂, was in fact observed during addition of the ozone enriched oxygen (Figure 6), with the average pH increase being 1 pH unit per 230 mg O₃ added per L manure. However, the pH increase observed simultaneously with the decrease in the H₂S emissions (–200 mV to −80 mV) was only 0.01–0.05 pH units. Based on calculations of the H₂S/HS^–/S^2– distribution from pK_a-values and observed pH, it is estimated that the pH increase could only result in a reduction of the emissions by ∼0–7%. Hence, the pH increase cannot account for the observed decrease of the H₂S emissions.

Volatilization of the H₂S when adding gas (flushing) could cause the additional decrease in the gaseous concentration of H₂S above the manure. This is observed not to be the case, as the control treatment with N₂ addition did not change the H₂S emission (Figure 3). This can also be estimated from the volume of added O₂ gas, the Henry's law constant, pK_a of H₂S and pH. In all cases, the relative amount of total reduced sulphide removed by gas flushing is estimated to be <5%. Hence, the flushing cannot account for additional removal of H₂S.

In conclusion, it is ozone and oxygen as oxidizing agents that causes the minimization of H₂S emissions. This is well aligned with the observed correlation between increase in redox potential and decrease in H₂S emissions.

It has previously been difficult to accurately calculate the exact amount of oxidizing agent required to minimize H₂S emissions from manures [1,9,14,15]. The present study indicates some useful methods for determining the optimal amount of oxidation agent needed.

The amount of ozone required for H₂S removal was observed to correlate with the H₂S/HS⁻ content of the non-ozonated manure. Thus, as an alternative to monitoring the actual H₂S emissions continuously, the H₂S/HS⁻ concentration of the raw slurry may be analyzed.

The decrease in H₂S emissions was observed to occur simultaneously with a rise in redox potential from ‐200 mV to ‐80 mV (Figures 2 and 3). Hence, in batch oxidation operations, it should be possible to apply the redox potential at termination of the rapidly changing redox phase and initiation of the stable redox phase (approximately −10 mV) in order to adjust the dose of oxidizing agent. This can be calculated as the intercept of the linear regression for the two redox potential phases (Figure 1).

For the six manure batches in this study, the required amount of ozone is estimated to be 6–24 mg O₃/L manure (Figure 4). In previous studies of manure, amounts of 250–3,000 mg O₃/L have been added to eliminate emissions of odorous compounds by ozonation [1,8,11,12]. These studies focused on compounds other than H₂S, but the present study demonstrates that if the main objective of oxidation is to eliminate H₂S emissions instantaneously, much lower doses can be used when applied with proper process control.

Monitoring redox continuously to assess the appropriate dose of oxidizing agent has been demonstrated to be applicable to two manures in three particle size ranges (Table 1), which covered different manure characteristics: both cattle and pig manure, and both high and low dry matter contents, total COD contents, redox potentials, H₂S/HS⁻ concentrations, and H₂S− to-HS⁻ ratios, and different particle size ranges. Hence, the control method is robust with respect to different manure characteristics. It is therefore applicable independently of farm operation and the continuously varying manure characteristics. This indicates that similar relations may be expected in other media than manure, such as drinking water and wastewater [16,29]. However, it is possible that these media contains oxidizable compounds competing with H₂S/HS⁻ in terms of O₃ and O₂ reactivity, resulting in less distinct changes in redox. Analogous relations between redox potential and reduction in H₂S emissions may also be expected with respect to oxidation using other oxidizing agents than ozone and oxygen, such as aeration, hydrogen peroxide treatment, permanganate treatment, Fenton's process, photo-oxidation, and anodic oxidation [2,10,24,30–32].