Multimodal Comparison of Cold Atmospheric Plasma Sources for Disinfection

Abstract

While atmospheric pressure plasma sources are emerging as potentially innovative instruments in medicine, some aspects of the interaction between plasma and biological substrates remain unclear. The high diversity in both sources and applications in the literature, and the lack of a systematic testing protocol, has resulted in a wide variety of devices that cannot be efficiently compared with one another. In this work, an integrated benchmark involving physical, chemical, and biological diagnostics is proposed. The setup is designed to be stable and fixed, while remaining adaptable to different sources. Three different sources, for a total of five configurations, are compared, demonstrating the possibility of obtaining multimodal data. Comparing the biological effects in terms of E. coli abatement between direct and indirect treatments allowed for the exclusion of short-timescale species and phenomena to have a key role in the abatement. The chemical characterisation describes the equilibrium of reactive oxygen and nitrogen species in treated samples, whose presence in the water has been found to be coherent with the plasma operating gas and the nitrogen vibrational temperatures. Nitrate, nitrite and peroxide are excluded from having an autonomous role in the inactivation biochemistry, suggesting the presence of a synergistic effect.

1. Introduction

Following the rapid development of plasma medicine in recent years, a plethora of different cold atmospheric plasma sources have been developed in laboratories across the world [1,2,3,4,5]. Most of these sources have been designed for a specific purpose and tested within their framework of interest, achieving, in some cases, outstanding results up to reaching the clinical trial phase [6]. Among the most studied applications, many experiments have been carried out on disinfection and decontamination. Although the bactericidal properties of plasma treatments are commonly recognised, there is still open debate about the actual processes involved. Plasma is a complex system, involving free charges, excited atoms, reactive species, and radiation. Moreover, biological systems are even more complex. A complete description of all the biochemical processes involved, with the understanding of the triggered biochemical pathways, is still far from being available.

Among the other components, a fundamental role in the plasma–cells interaction is expected to be played by reactive oxygen and nitrogen species (RONS) [7,8,9,10]. They are typically the results of multi-step chemical chains. Primary species like hydroxyl radical, atomic oxygen, or atomic nitrogen are directly obtained from operating gases by electron-induced dissociation and then react with each other to produce secondary RONS [11]. In the presence of a liquid surface, interfacial phenomena and solubility drive the absorption of species into water, with the consequent development of higher-order chemistry [12].

ROS, which include hydrogen peroxide (H₂O₂) and ozone (O₃), are characterised by a strong redox potential. Maintaining a redox equilibrium between the cytoplasm and the environment is fundamental for biological systems, and most of them possess defence mechanisms to counteract imbalances [13]. However, when these mechanisms are not sufficient, the so-called oxidative stress can lead to protein damage and compromise the organism homeostasis. Similarly, RNS, such as nitrous acid (HNO₂), nitric acid (HNO₃) and their unprotonated forms (NO₂⁻ and NO₃⁻), can lead to cell damage via nitrosative stress. The cell membrane is known to have a key role in maintaining equilibrium, but its integrity can be harmed by the plasma treatments [14]. The whole picture of the different biochemical pathways is still mostly obscure, with thousands of proteins involved whose exact function in still unknown [13]. This poses important obstacles to a complete theoretical understanding of the chemistry, suggesting a more suitable phenomenological modelling.

A full characterisation of a plasma source in action requires investigation of the plasma state, the gas phase, the liquid phase and the biological effects. Moreover, all these points of view should be combined in a single setup. In fact, it has been observed that the same plasma source can behave differently when operated under different environmental conditions [15], meaning that each setup is unique and diagnostic data cannot be easily compared. On the other side, for each diagnostic (or set of diagnostics) multiple operating parameters and sources should be characterised under a fixed data collection protocol. The gathering of all the collected data, for the different measured quantities and the different sources, constitute the behavioural map which can suggest a phenomenological model. There are a few studies in the literature that perform a systematic comparison of plasma sources for biological and medical applications, as an example comparing the RONS in water [16,17,18]. From the point of view of bacteria inactivation, Park et al. [19] published a study on the comparison of DBD and nanosecond pulsed plasma for the inactivation of multidrug resistance bacteria, and Mann et al. [20] compared a plasma jet and a surface DBD, proposing also a standardized technique for source comparison. Similar works are available also in the contexts of wound healing [21], of cancer treatments [22] or of agricultural applications [23,24]. The broader the characterisation, the easier it is to track the main processes involved, allowing the phenomenological model of the plasma–tissue interaction to be improved. Having a model would mean being able to define a direction for optimising the sources, to tailor devices for specific functions, and to perform real-time diagnostic during treatments.

This work reports the results of a comparison of different plasma sources using a shared protocol in a fixed common setup, tracking the action of these devices from plasma generation to biological action. The investigated processes concern the bactericidal properties of direct plasma treatments, using E. coli as a model organism. The preliminary results here reported demonstrate the potential of the setup and the protocol in producing reliable and comparable data. The results of optical emission spectroscopy are found to be compatible with the production of chemical species in the sample, and peroxides and nitroxides are proven not to play an autonomous role in the triggering of biological effects, providing a first set of conditions on which to build a model.

2. Setup

The setup (Figure 1) is designed to unify all diagnostics in a single treatment protocol. Above a grounded metallic table, a $35\mathrm{mm}$ diameter glass Petri dish is positioned. A set of spacers is fixed on the table such that the dish is always placed back in the exact same position. A metallic flange is fixed in place above the table with a set of custom supports, and used to sustain the different plasma sources. Custom 3D-printed adapters are designed for each source, such that each source tip is positioned at the same distance from the bottom of the Petri dish.

The treated sample consists of 6 mL of ultrapure water (VWR Puranity TU 3UV/UF+); the choice of ultrapure water reflects the aim of simplifying as much as possible the chemistry involved, and avoid complex reactions and species which can be generated in standard culture media when interacting with the plasma; such a large sample volume, with respect to the microlitre scales usually reported in the literature [8], was chosen to prevent liquid overheating and evaporation while better simulating real-world applications. The distance between the source nozzle and the water surface is fixed at $6\mathrm{mm}$. The treatment duration was fixed at 10 min for all samples.

The bacteria abatement capabilities of the different sources have been evaluated using Escherichia coli (K-12, MG1655) as a model organism. Samples of bacteria, selected in their exponential phase after overnight incubation and reactivation, and centrifuged to remove the growth medium (Luria–Bertani broth), are resuspended in ultrapure water immediately before the treatment. The initial concentration is estimated to be around 10⁶–10⁷ CFU/mL. The treated solutions are then sampled just after the treatment, serially diluted, and inoculated onto agar plates. After overnight incubation, the colonies are counted and the colony-forming units (CFU) estimated. The results are reported in terms of logarithmic abatement:

$$\mathrm{logAbat}:={log}_{10}{\displaystyle \frac{{\mathrm{CFU}}_{\mathrm{control}}}{{\mathrm{CFU}}_{\mathrm{treated}}}}$$

Multiple verification tests have been carried out to confirm that the various steps of the protocol do not affect cell viability. No significant CFU variations have been observed even when performing the entire protocol, including exposure to the sources and gas flows, if the plasma is not switched on.

All measurements reported in this work are the results of at least three independent experiments, and the uncertainties in the final values are estimated by statistically combining the standard deviation of the replicates with the nominal uncertainty of each method.

These results will be labelled in the following as direct treatments. Another set of measurements, labelled as indirect treatments, has been carried out following the same protocol previously described, but inoculating the bacteria in the water just after, and not just before, the plasma treatment. Mimicking the direct protocol, ultrapure water has been treated for 10 min. Immediately after, the bacteria sample is injected in the water and left in the setup for 10 min (without plasma). The protocol is then resumed with serial dilutions and inoculation in Agar dishes.

By exploring the causal chain which connects the plasma to the bacteria, the treated water has been characterised from a physiochemical point of view. A fast measurement of the activation of the treated water is performed by testing the oxidation-reduction potential (ORP) immediately after the treatment with a dedicated probe (Metrohm, Herisau, Switzerland). The measurement of pH proves to be more challenging because, due to the low conductivity of the solutions (below $100\mu \mathrm{S}/\mathrm{cm}$), the use of standard probes results in unreliable data [25]. A rough estimation of the pH, however, is provided in Section 4, based on indicator strips and observations on the chemistry.

The presence of long-lived reactive oxygen and nitrogen species in the treated water is analysed using a VIS spectrophotometer (Spectroquant Prove 100, Merck, Darmstadt, Germany). In particular, the concentrations of H₂O₂, NO₂⁻ and NO₃⁻ in the treated water have been measured. Just after the treatment, a sample was mixed with the appropriate reagents in the Spectroquant test cells (Merck Spectroquant 1.14731 for H₂O₂, 1.14563 for NO₃⁻, 1.00609 for NO₂⁻) and, after the suggested reaction time, it was measured in the spectrophotometer. For NO₃⁻ and H₂O₂, it was necessary to dilute the sample 1:2 in deionised water before measuring the chemicals, since the quantification kits require $10\mathrm{mL}$ samples.

To better understand the long-timescale chemistry, the short-timescale chemistry and dynamics were also investigated. A real-time indication of the plasma properties during the treatment is obtained by optical emission spectroscopy, using a high-resolution spectrometer (Avantes, Apeldoorn, the Netherlands) and an optical fibre. Focusing on the 360–385 nm range, the orientation of the optical fibre and the integration time have been optimised for each measurement to maximize the collected signal. A fit of the second positive system of N₂ [26,27] is performed using MassiveOES routines [28,29,30], obtaining the rotational and vibrational temperatures of N₂ molecules.

3. Plasma Sources

Three different plasma sources relevant for biomedical applications have been selected for this work: a micropulsed source, which can operate with either helium or argon; a radiofrequency source, which similarly can operate with helium or argon; and a DBD-like source, operating in air. For each source, after a preliminary sampling of chemicals production in different power supply configurations, a set of electrical parameters has been selected such that they all produce similar amounts of reactive species. Considering the three sources and their different working gases, a total of five configurations are investigated.

The first source (Figure 2a) is a micropulsed plasma jet, previously reported in the literature as Plasma Coagulation Controller (PCC) [31,32]. It consists of a tungsten electrode, covered with a closed capillary that acts as dielectric, and excited with a positive microsecond pulse ($0.45\mu \mathrm{s}$ FWHM) at $7.4{\mathrm{kV}}_{\mathrm{peak}}$, repeated at a frequency of $5\mathrm{kHz}$. The power absorbed by the full system, including the plasma and the transformation circuit, is measured to be $8.8\pm 0.3\mathrm{W}$. A glass nozzle is inserted coaxially with the electrode (internal diameter at the tip $3\mathrm{mm}$), and between the capillary and the nozzle a noble gas flows at $1\mathrm{L}/\min$. An external ground ring acts as the grounding reference. This configuration results in the formation of a plasma plume, generated inside the nozzle and extending into the air. The characteristics of the plume differ for the two gases: while the helium plume is more diffuse and uniform, the argon plume appears filamentary.

The second source (Figure 2b) is similarly a plasma jet, but operating in the radiofrequency regime. The central electrode, made of tungsten, is covered in alumina except for the last $10\mathrm{mm}$, which are exposed. A glass nozzle is positioned coaxially with the electrode (internal diameter at the tip $2.7\mathrm{mm}$), and either helium or argon is flown at $1\mathrm{L}/\min$. The electrode is excited with a radiofrequency signal, $8.8\mathrm{MHz}$ at $2.5{\mathrm{kV}}_{\mathrm{ptp}}$, modulated with a $10\%$ duty cycle at $2.0\mathrm{kHz}$ to limit gas heating. The averaged power absorbed by the plasma is measured to be $6\pm 1\mathrm{W}$. Similar to the previous source, the plume is generated inside the nozzle and expands into the surrounding atmosphere. With helium, the plume appears to be diffuse and uniform but relatively short, not exceeding a few millimetres in length. With argon, instead, the plume is longer and it is characterised by a single central channel surrounded by a diffuse, weaker glow.

The last source, called MibJet (Figure 2c), is an original prototype developed in our laboratories, designed to operate directly in air without the need of an external gas supply. It consists of a tube of alumina, with internal diameter $19\mathrm{mm}$, in which air flows at $0.5\mathrm{L}/\min$. Inside the alumina tube a grounded stainless steel cylindrical grid is placed; outside, a band of copper tape is excited with positive pulses $26\mu \mathrm{s}$ wide and $5.5{\mathrm{kV}}_{\mathrm{peak}}$ high, repeated with a $1\mathrm{kHz}$ frequency. The average power absorbed by the plasma, measured using Lissajous figures method [33], is estimated to be $3.4\pm 0.3\mathrm{W}$. Plasma is ignited like a surface DBD in the holes of the inner grid, and the air flow collects the species which are delivered to the substrate. Contrary to the previous two sources, in this case the plasma is not directly in contact with the substrate, which instead is treated with the afterglow. However, it is considered of interest to analyse the performance of this source, which does not require an external gas supply and instead works with ambient air.

For all the sources, the negligibility of the thermal load delivered to the samples has been verified: in all the experimental conditions considered, the variation in temperature of the treated water was measured to be below $0.5°\mathrm{C}$ and the evaporation of the sample below $1\%$ of the total volume.

4. Results and Discussion

The biological effects, reported in the first column of

Table 1. Measured logarithmic abatement of E. coli for the five configurations, in direct and indirect treatments.

	Abatement [logAbat]
	Direct	Indirect
Micropulsed/He	>6	>6
Micropulsed/Ar	$3.5\pm 0.2$	$3.2\pm 0.4$
Radiofrequency/He	<0.5	<0.5
Radiofrequency/Ar	<0.5	<0.5
MibJet	$1.3\pm 0.2$	$2.9\pm 0.5$

with the label direct, appear to be very different for the five configurations. The micropulsed source when operating in helium is very effective, inducing an abatement larger than 6 orders of magnitude, which corresponds to the upper detection limit of the method. The same source, when operated in argon, shows a significantly reduced effectiveness, reaching $3.5\pm 0.2$ logarithmic abatement. In contrast, the radiofrequency source seems not to be effective at all, whether using helium or argon. Finally, the MibJet source is observed to reduce the bacterial load by slightly more than one order of magnitude.

The differences in the biological performance are evident. It is therefore worthwhile to further analyse the factors involved in the interaction, to track down their possible responsibility in determining such a huge gap. The phenomena having a role in plasma–biological interaction can be categorised in two classes: immediate, with time scales spanning in the range ns-$\mathsf{\mu }$s-ms, and long-term, with time scales from seconds and higher.

The first class includes UV radiation, electric fields, local heating of the substrate, and short-lived reactive chemical species. UV radiation, besides its importance in the chemistry of the gas phase and in the interaction with air molecules to create radicals, is generally not strong enough to penetrate the water layer and induce significant damage to the cell membrane [34]. On the other hand, electric fields can play a role: especially when the jet reaches the water, on the tip of the plasma, local high fields are reached, and it is proven that high electric fields can induce lethal electroporation [35,36]. A priori, heat can also play a role in the bactericidal processes. Even if the bulk water has been observed not to heat, in fact, local temperature peaks can be reached in the contact point between plasma jets and the water surface. The sensitivity to temperature changes is highly dependent not only on the bacterial species considered, but also on the specific strain and substrain. The bacteria used in this work are resistant up to $50°\mathrm{C}$, while they die in several minutes above $54°\mathrm{C}$ [37]. Finally, short-lived species like O₂⁻, atomics ones like O and H, and molecular radicals as OH and NO, are known to be highly cytotoxic [38]. In this class, we are referring here to species directly generated by the plasma or in the gas–liquid interaction, and not to the tertiary species which can be generated in further reactions between long-lived species, which instead belong to the long-timescale chemistry. The second class, in fact, collects all the reactive oxygen and nitrogen species which accumulate in water, as for examples nitrates, nitrites and peroxides. Bacteria are exposed for many minutes to these species, and to the environment—in terms, for example, of pH—that they create.

A fast and reliable way to split the effects of the two classes is to perform indirect treatments. With indirect treatments, immediate effects are excluded, being the bacteria still not present in the water, and only long-term effects can have a role. The logarithmic abatements obtained, reported in the last column of Table 1 with the label indirect, turn out to be compatible with the values of direct treatments. Again, the radiofrequency source is not efficient at all while the micropulsed one outperforms when operated in helium, reaching more than 6 logarithmic reduction. An exception is represented by the MibJet, whose result for indirect treatments is measured higher in abatement with respect to the direct ones. This further confirms the role of long-living species, considering that while in direct treatment the sample is exposed to rising concentrations, in the indirect one the source is exposed from the beginning to the final maximum concentrations, resulting for a given treatment time in a larger integrated effect.

These results identify the long-timescale chemistry as the main responsible for defining the plasma–biological interaction, and it is particularly interesting in the framework of designing efficient and ad hoc plasma sources. It is important to emphasize that this result is limited to the sources considered here and should not be generalised. Other plasma sources, in different setups and operating conditions, proved to be effective only in direct treatments [39]. In the reported setup, however, due to the geometry of the sources, the electric field applied to the substrate is expected to be relatively modest, and the water is expected to substantially screen the UV flux, leaving to the chemistry a leading role.

Having proven the fact that the interaction between plasma and bacteria is mainly chemical, it becomes of fundamental interest to measure the concentration of the chemical species which are formed in the treated samples. Figure 3 reports the concentration measured by spectrophotometry in water for the different sources. The immediate outcome is that, while the biological effects span more than six orders of magnitude, the chemical concentrations are measured to be all on the same order. Moreover, while the micropulsed helium source outperforms in the biological treatments, this uniqueness is not reflected in the chemical characterisation.

Both the micropulsed and the radiofrequency sources, when operating in argon, show an improved production of hydrogen peroxide, which is present in concentrations substantially higher than in the helium-operating ones. The increased presence of peroxide in argon plasma, however, does not correspond to an improvement in the abatement power. Moreover, in previous studies, higher concentrations of hydrogen peroxide are reported to be required to induce bacterial inactivation: the H₂O₂ minimal concentration to achieve 1 logAbat in 10 min treatment is reported to be around $40\mathrm{mg}/\mathrm{L}$ [40], more than double with respect to the $18.7\pm 1.1\mathrm{mg}/\mathrm{L}$ measured in the micropulsed helium source. Hydrogen peroxide can be therefore excluded from having, individually, the main role in bacterial abatement.

A similar conclusion can be drawn regarding nitroxides: the MibJet, in fact, is measured to be very efficient in the production of nitrates, and even for nitrites the obtained values are slightly above those of the micropulsed helium source ($68\pm 3$ vs. $13.1\pm 0.6$ for NO₃⁻, $14\pm 2$ vs. $10\pm 2$ for NO₂⁻). Again, this does not correspond to an higher abatement power. Moreover, literature studies reported lower inactivation rates with higher NO₂⁻ and NO₃⁻ concentrations (as an example, less than 1 log-reduction for $54\mathrm{mg}/\mathrm{L}$ of NO₂⁻ and $41\mathrm{mg}/\mathrm{L}$ of NO₃⁻ in [14]). Similarly to peroxides, therefore, nitrates and nitrites cannot be individually responsible for the biological effects.

The results of the chemical characterisation are further confirmed by emission spectroscopy data, with the interpolations of rotational and vibrational temperatures of nitrogen reported in Figure 4. In argon plasma, with heavier atoms and colder electrons with respect to helium one [41,42], vibro-rotational relaxation is faster, leading to lower vibrational temperatures (below $2000\mathrm{K}$) and slightly higher rotational ones (between 500 and $1000\mathrm{K}$). In helium plasmas, oppositely, rotational temperatures remains below $500\mathrm{K}$ while vibrational temperatures can reach more than $3000\mathrm{K}$. In defining the equilibrium between oxygen and nitrogen species, a key role is played by the reaction N₂($\nu$) + O → NO + N, whose rate depends on the vibrational excitation of nitrogen molecules. At higher vibrational temperatures as in helium, therefore, the reaction is boosted, increasing the production of NO and, through HNO₂ and HNO₃, of nitrates and nitrites in water [43,44]. Moreover, the reaction consumes part of the available atomic oxygen, competing with the chemistry of oxygen species and resulting in a reduction of peroxides. Together with other reactions, the equilibrium is expected to move towards nitrogen chemistry at higher vibrational temperature: this prediction matches the results from the chemical characterisations, which reported a shift of the chemistry from peroxides towards nitrogen species when switching from argon to helium. For the air jet, since nitrogen is directly involved in plasma formation, both rotational and vibrational states are highly excited, resulting in a predictable increase in nitrates and nitrites.

The chemical measurements described above can also be used to estimate the pH of the solution. Test strips have been preliminary used, sensing a neutral pH in all the solutions. Performing a second characterisation of the chemical composition of the treated samples one hour after the treatment, concentrations compatible with those measured immediately after the treatment are found. It is known that, in acidic environments, the balance between nitrates and nitrites is expected to drift over time [11]. The absence of such drift provides further confirmation that the treated solutions are almost neutral. It should be noted that most of the characterisations reported in the literature measured an acidification of the samples after the treatments, while a neutral pH is seldom found [45], confirming again the uniqueness of each experimental setup and the complexity of the involved chemistry.

To further improve the characterisation of the treated samples, the oxidation-reduction potential was measured. Starting from an initial value of $320\pm 10\mathrm{mV}$, all the sources showed an increase in potential, reaching $439\pm 4\mathrm{mV}$ for the micropulsed helium source. A similar value was measured in the micropulsed argon source ($429\pm 2\mathrm{mV}$), while values lower than $400\mathrm{mV}$ were detected in the radiofrequency source. Notably, the MibJet-treated samples, with their high nitroxide content, led to a high value ($577\pm 11\mathrm{mV}$), far surpassing the other sources. It is worth mentioning that NO₃⁻ is a strongly oxidising agent. It is therefore not surprising that the ORP measurements follow in trend the nitrate concentration.

Previous studies have attempted to mimic the effect of plasma treatments using chemical solutions. The synthesis of mock plasma-treated solutions typically focuses on reproducing the pH and the concentrations of NO₂⁻, NO₃⁻, and H₂O₂; however, the results are inconclusive, reporting biological effects both different from [46,47] and comparable to [14,48] those of the corresponding plasma-treated water. Most of the studies, however, report significant acidity, while the solutions considered in this work are neutral.

Having excluded an autonomous role of the analysed chemicals, the remaining hypothesis involves a synergistic effect among them and the involvement of additional species not measured here. As an example, one further species to investigate is the peroxynitrite ion (ONOO⁻), originating from the combined presence of nitrites and peroxides [49] and hypothesised to play a possible role in cold atmospheric plasmas [46] even when pH is close to neutral [50]. Another possible species involved in the chemistry could be ozone. However, in all the treated samples, a presence of NO₂⁻ larger than $10\mathrm{mg}/\mathrm{L}$ has been measured, and therefore any eventual presence of ozone would be scavenged by the reaction ${\mathrm{O}}_3+\mathrm{N}{\mathrm{O}}_2^{-}\to \mathrm{N}{\mathrm{O}}_3^{-}+{\mathrm{O}}_2$. For the concentrations and pH measured in this work, the half-life of ozone is expected to be on the order of milliseconds [51], thus falling into the category of short-living species.

5. Conclusions

Plasma medicine is a constantly growing field, already reaching clinical practice with some of its most studied technologies. However, the chain of components and processes involved in the interaction between plasma, atmosphere, liquids, and cells remains obscure in many aspects. The presence of numerous sources in various laboratories used for widely disparate applications represents an obstacle in defining a common standard which suits the research groups involved. An extensive campaign of source characterisation, following strict protocols and unified operating conditions, is fundamental to further understand the mechanisms involved, and therefore to design optimised sources that maximize the desired effects while preventing collateral damages.

This work presents a proof of concept for such a campaign. Multiple sources with multiple diagnostics were tested on a fixed common setup. The protocol was defined in detail, allowing for its reproduction in different laboratories. The obtained measurements range from physical to chemical and biological analyses.

The three different sources, in a total of five configurations, span six orders of magnitude in terms of biological effects. Connecting the physiochemical characterisation of the treated samples with the response of treated cells still remains a challenge. Indirect treatments, performed in the same setup as direct ones, led to excluding short-timescale phenomena from playing a key role in the considered treatments, confirming the fundamental importance of long-lived RONS. The chemical characterisation of the treated samples reported different equilibriums between oxygen and nitrogen species. These results well reflect the differences in nitrogen rotational and vibrational temperatures measured via emission spectroscopy, which can therefore be used as a real-time probe of the ongoing treatment. The analysed sources do not induce any significant change in pH, and in particular the samples treated with the micropulsed helium source reach more than 6 logAbat in a neutral solution, highlighting that acidification is not fundamental as reported in other setups [52]. Finally, the results from the chemical characterisation prove that the biological effects cannot be connected to an individual species: the samples with highest concentration of specific RONS does not coincide with the samples with the highest bacterial abatement, suggesting the presence of an optimal mix of nitrates, nitrites, peroxides (and eventually other species not measured here) for an efficient bactericidal effect. These comparisons between different sources and diagnostics have been possible thanks to the common setup in which all diagnostics are gathered together and the sources are adapted, confirming the need to define a complete and common protocol to obtain meaningful data.

Further steps will involve enriching the diagnostic methods, aiming to add more detail to the source characterisation while maintaining the consistency of the setup and protocol, and broadening the range of available sources, gathering more evidence toward a full understanding of the entire process.