Inactivation of bacterial endospores on surfaces by plasma processed air

In case of biological hazards and pandemics, personal protective equipment of rescue forces is currently manually decontaminated with harmful disinfectants, primarily peracetic acid. To overcome current drawbacks regarding supply, handling and disposal of chemicals, the use of plasma processed air (PPA) represents a promising alternative for surface decontamination on site. In this study, the sporicidal efficiency of a portable plasma system, designed for field applications, was evaluated.


Introduction
Biological agents like pathogenic bacteria, viruses or toxins represent serious threats for human health. In case of biological hazards and pandemics, rescue forces such as firefighters, civil defence and disaster control or medical staff, must be protected from possible contamination, on the one hand by wearing suitable personal protective equipment and on the other hand by efficient, but usercompatible decontamination measures of their protective equipment. In biological worst-case conditions, the disinfection of personal protective equipment suspected to be contaminated with hazardous biologically agents should be highly effective within short time, easy to use and compatible with the environmental conditions (Lemmer et al. 2012). Peracetic acid (PAA) is used as a cold sterilization agent in many areas and it is currently mostly used as sporicidal agent for the disinfection of personal protective equipment of rescue forces. Although it represents a potent disinfectant with sporicidal activity, there are several drawbacks, including the necessity of rescue teams to transport hazardous chemicals, the required proper disposal of the chemicals as well as the need for elaborate scrubbing of the personal protective equipment with brushes.
The use of cold atmospheric plasma (CAP) as sporicidal agent for the surface disinfection of protective suits could be a suitable approach to overcome these detriments. Plasma is referred to the fourth state of matter, a partially ionized gas composed of various reactive species including electrons, ions, free radicals, atoms and molecules in the ground or excited state as well as electromagnetic radiation (Bourke et al. 2017). Cold atmospheric plasma is not in thermal equilibrium since the temperature of its electrons is much higher than the temperature of ions and neutrals (M€ uller et al. 2018). The microbicidal activity of cold plasma has long been known and was demonstrated in various research studies over the last two decades. For disinfection and sterilization purposes it is most commonly created by dielectric barrier discharge (DBD) and atmospheric pressure plasma jets (Scholtz et al. 2015;Bourke et al. 2017). Various parameters are known to affect the sterilization efficiency, including the reactor geometry as well as the operating conditions like the gas pressure, the type of gas mixture, the gas flow and the frequency and power of plasma excitation. Plasma generated from air comprises reactive oxygen species (ROS) and reactive nitrogen species (RNS), such as atomic oxygen (O), singlet oxygen ( 1 O 2 ), superoxide anion (O 2 À ), ozone (O 3 ), hydroxyl radicals (•OH), atomic nitrogen N, excited nitrogen N 2 (A), nitric oxide (NO) or NO 2 (Scholtz et al. 2015). These oxygen-and nitrogen-based reactive species have strong oxidative effects on microbial structures like lipids, proteins and DNA (Laroussi 2005). The microbicidal effect of CAP depends on various factors, especially the mixture of generated reactive species and the operational mode. Regarding the operational mode, it is decisive whether the surface to be disinfected is in direct contact to the plasma glow or indirectly treated by remote mode. The sporicidal action of atmospheric plasma in direct application mode is believed to be dominated by spore erosion due to a surface action of radicals and internal oxidative irreversible damage, subsequent to the diffusion of the radicals deeply inside the spores. UV photons causing DNA lesions can also be dominating the spore inactivation, depending on the operating conditions (Boudam et al. 2006). Direct plasma treatment of surfaces within the glow discharge is known to be highly efficient but its suitability for the disinfection of large areas with complex geometries appears to be limited (M€ uller et al. 2018). If exposed remotely, the surface to be disinfected is spatially separated from the plasma volume or in an adjacent chamber (Laroussi 2005). Although usually requiring longer treatment times, remote plasma offers a more gentle treatment of complex targets (M€ uller et al. 2018). Charged particles then play a minor role for disinfection since they recombine before arriving at the substrate, which is likewise the case for short-lived neutral reactive species (Laroussi 2005;Misra et al. 2011). In a few studies, the surface to be disinfected was placed in the flowing afterglow of the plasma source (Boudam et al. 2006;Moisan et al. 2014;M€ uller et al. 2018). The sporicidal effect of plasma processed air (PPA) which is flushed into a separated chamber of a larger volume has rarely been studied. In order to disinfect three-dimensional structures (e.g. protective suits) the entire surface must be exposed to the reactive species in order to enable an even and fast process. This could be achieved by use of plasma nozzles where the afterglow stream is directed into decontamination chambers, for example, tents. The use of such a reactive gas mixture appears to be a suitable alternative in case of biological hazards in order to enable an on-site decontamination of surfaces without chemicals. In this study, the sporicidal efficiency of a newly designed portable plasma system based on a DBD and operated with ambient air was investigated. Since spores represent the most resistant form of micro-organisms, they are commonly used to assess the efficiency of sterilization processes. This is also a prerequisite for a later approval of this technology by the relevant authorities (e.g. Robert Koch Institute). Bacillus spores were used to simulate worst-case conditions and served as surrogates for any potentially relevant pathogenic bacteria, for example, Bacillus anthracis, which is of high relevance in the area of civil security. Plasma processed air from a plasma nozzle was introduced in a closed treatment chamber with a volume of approximately 300 l. The inactivation of bacterial endospores dried on different surfaces was assessed under various conditions. The complexity of the surface structure, the way the endospores are distributed on the surface, the impact of an organic matrix as well as the combination of PPA exposure with mechanical wiping was assessed. To the best of our knowledge, the sporicidal action of PPA has not been investigated in a comparable way yet.

Materials and methods
Plasma source and treatment conditions A portable plasma system developed by the company Plasmatreat GmbH (Germany) was used for all test trials. Plasma processed air was generated with a CD40 plasma nozzle based on a DBD equipped with an air cooling system. Ambient air at a pressure of 6 bar was used as process gas (7Á5 l min À1 ). The input power was approximately 200 W and the discharge frequency was 13 500 Hz. In previous trials, it has been found that the introduction of water strongly improves the microbicidal action of PPA. Therefore, if not stated otherwise, the plasma afterglow was humidified by introducing water (1 ml min À1 ) via a carrier gas stream (0Á5 l air min À1 ) and an evaporator (150°C). The water vapour was directed into the afterglow at the outlet of the plasma nozzle approximately 10 cm from the plasma discharge. A HPLC-pump Gynkotek M 300 (Gynkotek, Munich, Germany) was used for water injection. The CD40 plasma nozzle was centrally placed on the bottom of a Glovebox made from acrylic glass (depth: 0Á5 m; height: 0Á6 m; width: 1 m). The Glovebox was equipped with a valve to avoid overpressure, ozone-resistant gloves PIEC16750Y106/10 (Piercan, Bondy Cedex, France) and a positing system to introduce and remove samples. The experimental setup is schematically shown in Fig. 1.

Test strains, spore and sample preparation
Two different endospore-forming bacteria belonging to the genus Bacillus were used as test micro-organisms. B. subtilis DSM 4181 and B. atrophaeus DSM 675 were obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ). The two strains are recommended for the microbiological validation of aseptic filling machines operated by hydrogen peroxide and PAA according to the Industry Association for Food Processing Machinery and Packaging Machinery (VDMA 2016). Preparation of spore suspensions was done as described previously by Muranyi et al. (2007) using manganese sulphate for induction of sporulation. After harvesting, spore suspensions were stored at 5°C.
PET-films and Tychem â F protective suits made from Polyethylene (Tyvek with coated polymers) (DuPont, Neu-Isenburg, Germany) were used as substrates for surface disinfection trials. Pre-cut parts of 4 9 4 cm were wiped with ethanol and four pieces (samples) were placed together in quadratic Petri dishes respectively. Inoculation was done by either spraying or spot inoculation. Spray inoculation provides a more homogeneous distribution of spores on the respective surface, thereby avoiding the formation of agglomerates. In dependency of the endospore density of the applied suspension, the sprayed volume and other process parameters, it has been shown that monolayers of spores can be obtained (Muranyi et al. 2007). Spot inoculation on the other hand leads to the formation of spore agglomerates and multilayers, which may retard the inactivation. Furthermore, the application of small spots was conducted according to a method of the German Federal Office for Civil Protection. The distribution of endospores on PET samples was examined microscopically.
For this purpose, spores were washed two times with deionized water and finally diluted to approximately 5 9 10 7 CFU per ml either in deionized water or in an aqueous solution of 1-10 g l À1 bovine serum albumin (BSA) (Biomol, Hamburg, Germany). BSA was applied to investigate the sporicidal efficiency of the plasma gas when endospores are embedded in an organic matrix. In case of spraying, 5-7 µl of a prepared spore suspension was sprayed on an area of 2 9 2 cm with a two-substance nozzle (Schlick, Untersiemau, Germany) using nitrogen (2 bar) as process gas. The applied method was previously described by Muranyi et al. (2007). In case of spot inoculation, 5 9 2 µl of the prepared spore suspension was deposited on the sample surface respectively (Lemmer et al. 2012). Inoculated samples were immediately dried in a laminar flow safety cabinet for 10 min. Endospores did not germinate under these conditions, which was checked by phase-contrast microscopy.

Exposure of test micro-organisms to plasma processed air
Prior to each inactivation trial, the treatment chamber (Glovebox) was flushed with PPA for 20 min. The ozone concentration inside the chamber was measured with a GM-6000-RTI Ozomat (Anseros, Germany). After 20 min, the ozone concentration reached a maximum of 1Á8 AE 0Á2 g m À3 when 1 ml of water min À1 was injected into the plasma afterglow and 2Á1 AE 0Á2 g m À3 without humidification. The humidity and temperature inside the chamber was measured with a Basetech BTTH-1014 Thermo-/Hygrometer (Conrad Electronics, Hirschau, Germany). The humidity inside the treatment chamber increased from 25 AE 5% to 90 AE 10% RH when 1 ml of water min À1 was pumped into the plasma afterglow. The temperature increased from 21 AE 1 to 25 AE 1°C. After the plasma nozzle had been turned off, the ozone concentration inside the chamber decreased by about 50% (at 90% RH and 25°C) within 15 min. For the inactivation trials, four individual samples placed in one quadratic Petri dish were introduced into the Glovebox through a positing system and placed on lifting plates centrally located in the treatment chamber. In order to assure an indirect treatment, the turbulent PPA stream from the outlet of the plasma nozzle was not directed towards the sample surface (Fig. 1). The samples were treated inside the glovebox for different times up to a maximum of 10 min. Longer treatment times were not considered since they are not of practical relevance for the decontamination of personnel equipment of rescue forces. As long as not otherwise stated, the plasma system was in operation throughout the whole treatment duration. The impact of the parameters on the sporicidal efficiency were studied under the conditions indicated in Table 1.
In case of additional mechanical wiping, endospores (B. atrophaeus) were dispersed in 10 g l À1 BSA, spot inoculated (5 9 2 µl) on the Tychem F protective suit and finally dried. The endospore concentration of the inoculation suspension was adjusted to 10 9 CFU per ml, resulting in endospore counts of 10 7 CFU per sample. While being exposed to the PPA for either 1 or 10 min in total, each sample surface was individually wiped with a swab with bristles (Master Amp Buccal Swab Brush; Epicentre Technologies, Madison, WI, USA) for 1 min. For this purpose, 10 µl of either sterile deionized water or plasma processed water (PPW) were used to resolve the dried protein matrix with embedded spores during wiping. PPW was obtained by collecting condensate in a flask at the nozzle outlet. The resulting pH value of the condensate was~1. PPW was used immediately. Four individual replicate samples were treated respectively and the swab as well as the sample surface (Tychem F) were regarded as individual samples. The number of colonyforming units present on the swabs and the corresponding Tychem F samples was therefore determined separately. Untreated reference samples (four replicate samples) were manually wiped as well in order to determine the distribution of spores after wiping. For this purpose, the number of CFU on the Tychem F samples and on the swabs after wiping was likewise separately determined. The trials including mechanical wiping were performed twice in analogous manner.

Sample handling and enumeration of colony-forming units
Directly after exposure to PPA in the glovebox, the samples were transferred to 50 ml of sterile 1/4 strength ringer solution supplemented with 0Á1% Tween 80 as detergent (Sigma Aldrich, St. Louis, MO, USA ) in stomacher bags. All samples were manually rubbed for 1 min to detach spores from the sample surface and to suspend them homogeneously. The number of colony-forming units of the sample suspensions was determined by the pour plate technique using tryptic soy agar. 1 ml of undiluted or serially (10-fold) diluted sample suspension in 1/4 strength ringer solution was transferred to Petri dishes in triplicates (3 9 1 ml) respectively. After solidification, all plates were incubated at 30°C for 48 h before the number of colony-forming units was counted manually. Colony counts were calculated as CFU per sample on the basis of mean values from the three replicate plates and the respective dilution factor. Results are presented as mean values (n = 4 or n = 8) with standard deviations and respective log 10 reductions (log 10 (N o N À1 )). The errors of the log 10 -reductions were derived from Gaussian error propagation. Where indicated, significant differences of mean values were assessed by t-testing using SIGMAPLOT 12.5 (Systat Software Inc., San Jose, CA, USA) on a significance level of P < 0Á05.

Results
Endospores are known to be among the most resistant micro-organisms towards various chemical and physical stressors, including CAP (Bourke et al. 2017). Therefore, endospores from the two different bacterial species B. subtilis (DSM 4181) and B. atrophaeus (DSM 675) were used as surrogates for pathogenic micro-organisms. Both strains are recommended by the VDMA for microbiological validations of aseptic filling machines operated with PAA or hydrogen peroxide (VDMA 2016). Preliminary tests have shown that these two strains are slightly more resistant to PPA than B. pumilus DSM 492 and B. thuringiensis DSM 350 (data not shown). Latter is usually used as a surrogate strain for Bacillus anthracis. When endospores were either spot inoculated or sprayed onto PET films and exposed to PPA without humidification of the plasma afterglow, the sporicidal effect was negligible in both cases (Table 2). Subsequently, different flow rates of water were applied in order to investigate the impact of water which is delivered to the evaporator connected to plasma nozzle (Table 3). For this purpose, a rather low spore density was adjusted by spray inoculation on smooth PET films in order to avoid any interfering matrix effects coming from agglomerate formation or complex surface microstructures as far as possible. A high sporicidal effect of PPA (reduction of the spore count by more than 3 log 10 within 2 min) was obtained with flow rates of 0Á5 or 1 ml water min À1 . The humidity increased to more than 90% RH in both cases. The injection of a sufficient amount of water vapour into the plasma afterglow therefore proved to be crucial in order to obtain a high sporicidal effect. Based on these results, a flow rate of 1 ml water min À1 was used for all further trials. It was furthermore investigated, whether a significant sporicidal effect can also be obtained when the plasma nozzle is switched off after flushing the treatment chamber with humidified PPA. For this purpose, B. atrophaeus endospores were spray inoculated on PET films and treated in the chamber after the plasma nozzle and the vaporizer (1 ml water min À1 ) had been running for 20 min. A maximum reduction by only 0Á63 log 10 was found after a treatment for 10 min whereas a reduction by more than 4Á25 log 10 was reached within 5 min when the nozzle and the vaporizer were operating during the treatment (Table 4). These findings showed that the plasma nozzle as well as the vaporizer necessarily need to be in operation during sample treatment in the chamber in order to obtain a high sporicidal effect.
In a next step, the inactivation efficiency of PPA on surfaces with different complexity of the microstructure was determined. In this context, the impact of the distribution of spores on the respective surface was assessed as well. For this purpose, the spores were either spray inoculated (Table 4) or spot inoculated (Table 5) and then dried on either PET films or Tychem F samples. Spores were either evenly distributed without interfering matrix effects coming from agglomerates or organic loads (spraying) or present spore multilayers and clusters (spot inoculation) which was likewise confirmed microscopically. Table 4 shows the inactivation of endospores on PET films and the Tychem F protective suit when samples were spray inoculated and exposed to PPA in the treatment chamber for different times. On PET films, the detection limit was reached after 5 min which corresponded to a reduction in the initial load by 4Á24 log 10 (B. atrophaeus) and 4Á25 log 10 (B. subtilis) respectively. The endospore reduction on the Tychem F protective suit was considerably lower. In this case, reductions in the initial load by 3Á69 log 10 (B. subtilis) and 3Á54 log 10 (B. atrophaeus) were reached after PPA exposure for 10 min. The presence of spores in agglomerates distinctively affected the inactivation efficiency compared to spray inoculation on both surfaces (Table 5). After the maximum treatment time of 10 min, reductions in the initial endospore load by 2Á72 log 10 (B. subtilis) and 2Á47 log 10 (B. atrophaeus) were found on the Tychem F samples. The inactivation efficiency on PET film was again higher, resulting in count reductions by 4Á04 log 10 (B. subtilis) and 4Á41 log 10 (B. atrophaeus) after 10 min. In addition, a lower flow rate of 0Á5 ml water min À1 which is vapourized and injected into the afterglow stream was applied under these experimental conditions ( Table 6). The determined colony counts on the Tychem F surface or the PET films were not significantly different to those obtained at a flow rate of 1 ml water min À1 , which confirmed that the sporicidal effect is not affected by the water supply within a flow range between 0Á5 and 1 ml min À1 .  The impact of an organic matrix, in which the endospores are embedded, was investigated using BSA. Endospores were suspended in BSA solution and subsequently spot inoculated and dried on the sample surfaces. The protein matrix should serve as a model for practically relevant organic contaminations on protective suits like, for example, blood, vomit or dirt. A concentration of 10 g l À1 almost completely prevented the sporicidal action of PPA for both species within 10 min ( Table 7). The BSA concentration was then stepwise reduced in order to find the threshold where a significant spore inactivation can be obtained. It was found that only at 1 g l À1 , a pronounced sporicidal effect was obtained. B. atrophaeus was reduced by 1Á46 log10 after 10 min on Tychem F samples (Table 7). These results showed that even comparably low concentrations of protein may inhibit the sporicidal effect of PPA.
Since the protein matrix severely affected the sporicidal effect, PPA exposure was also tested in combination with mechanical wiping using swabs with bristles. In this case, a high inoculation density was chosen (10 7 CFU per sample), together with the highest protein concentration applied (10 g l À1 ) and spot inoculation on the more complex surface of the Tychem F protective suit. These conditions ought to represent a worst case scenario. After manual wiping of untreated reference samples for 1 min, it was found that the distribution of endospores in the swab or on the Tychem F surface was rather equal. Approximately 50% of the spores initially dried in the protein matrix were found in the swabs, while about 50% of the endospores were found on the Tychem F surface after wiping. Exposure of samples to PPA for 10 min without mechanical wiping did not cause any sporicidal effect at all (data not shown). However, when the organic matrix was resolved mechanically by manual wiping with a swab and 10 µl of liquid (water or PPW) for 1 min, the inactivation of spores on the Tychem F samples was significantly increased (Fig. 2). Bacillus atrophaeus endospores were reduced by 4Á59 log 10 within a treatment time of 10 min when 10 µl of water were used. In case of wiping with PPW, the spore count was reduced to the detection limit corresponding to a reduction by more than 5Á64 log 10 . When samples were exposed to PPA for only 1 min, the reduction of the initial loads was 1Á31 log 10 (water) and 3Á53 log 10 (PPW). The spore reduction was therefore distinctively higher when PPW was applied instead of water. In contrast to the Tychem F surface, the inactivation of endospores present on the swabs was considerably lower. Reductions by 0Á84 log 10 (wiping with water) and 3Á26 log 10 (wiping with PPW) were found after 10 min treatment. When the treatment time was only 1 min, spore reductions by 0Á66 log 10 (water) and 0Á67 log 10 (PPW) were determined on the swabs (Fig. 2).

Discussion
Sporicidal effect of plasma processed air The sporicidal action of different CAP systems has been demonstrated with different setups (Boudam et al. 2006;Bourke et al. 2017;Hertwig et al. 2018). It is well accepted that the highest sporicidal effect can be achieved when spores are directly exposed to the plasma discharge (Moisan et al. 2014). This is due to the incidence of UV photons as well as presence of highly reactive short-lived species, including accelerated ions and electrons, but also uncharged particles, such as excited atoms, molecules and radicals (Moisan et al. 2001). In the present study, significant inactivation of endospores has also been found with a remote operational setup where the target is not directly in the plasma discharge but spatially separated from the plasma source (Fig. 1). UV-irradiation was therefore not involved in the spore inactivation. Similar findings have already been reported in previous studies (Moisan et al. 2014;M€ uller et al. 2018) but it is yet difficult to attribute the endospore inactivation of remote plasma applications to certain reactive species (M€ uller et al. 2018). The plasma afterglow contains relatively few charged particles, being essentially comprised of neutral atoms, radicals and molecules, some of which are in an excited state (Herrmann et al. 1999;Moisan et al. 2001). When the gas exits the discharge volume, ions and electrons are rapidly lost by recombination (Herrmann et al. 1999) so only comparably stable species like, for example, ozone are likely to reach the target. An ozone concentration of about 2 g m À3 close to the sample surface has been proven in the present study. The sporicidal effect of reactive components which are selfquenching, with a relatively short half-life, might depend on the respective time of flight to the target (Misra et al. 2011). Since the plasma discharge in the present study was at least 1 m away from the target surface, highly reactive components most likely did not reach the target. Using a circulating plasma afterglow generated by a surface microdischarge with humid air, M€ uller et al. (2018) reported about the inactivation of B. atrophaeus endospores by remote plasma in different treatment volumes of 0Á54, 1Á8 and 2Á6 l by 4Á4 log 10 within 10, 20 and 30 min respectively. Their approach is comparable to our study, apart from the plasma source and the volume of the treatment chamber being more than 100-fold larger in the present study. Moisan et al. (2013) reported about the sporicidal effect of a flowing afterglow from a reduced-pressure N 2 -O 2 discharge, where reductions in the initial load of B. atrophaeus, B. pumilus and Geobacillus stearothermophilus by approximately 3-4 log 10 were reached within 15 min in a treatment chamber with a volume of 5Á5 l (Moisan et al. 2013). Furthermore, Schnabel et al. (2019) found inactivation rates of B. atrophaeus endospores up to 2Á74 log 10 within 5 min in a treatment chamber of 4Á5 l which was flushed with PPA derived from a microwave discharge. Similar spore reduction rates were obtained in the present study but the treatment chamber volume of 300 l was considerably larger compared to those mentioned in the above-cited studies. This shows that the DBD-plasma nozzle exhibits a high potential for the disinfection of surfaces even when operated in remote application mode. In addition, our findings indicate that primarily waterbased reactive species are responsible for the sporicidal effect. Ozone alone cannot be the cause for the sporicidal action. When no water vapour was injected into the plasma afterglow, no inactivation of spores was found within 10 min although the resulting ozone concentration in the treatment chamber was equal. A rather slow sporicidal effect of ozone has been demonstrated previously. Mahfoudh et al. (2010) reported that dry ozone at a concentration of about 8 g m À3 causes a comparably slow inactivation of B. atrophaeus endospores. Two phases in the survival curves were found, exhibiting D-values of 130 and 360 min respectively. At 82% RH, significantly shorter D-values of 11 and 128 min were determined  Aydogan and Gurol (2006) who likewise found that the inactivation of endospores depends on the respective humidity. A 3 log 10 reduction of B. subtilis endospores dried on different surfaces was consequently found within 4 h at a concentration of 3 g m À3 of gaseous ozone under high humidity (90% RH) while at 70% RH, the spore reduction was <2 log 10 within 4 h (Aydogan and Gurol 2006). According to these previous findings, an increased humidity appears to increase the sporicidal effect of ozone. However, the sole combination of ozone with a high humidity cannot be seen as the cause for the spore inactivation found in the present study, since significant endospore reductions were only found when the plasma nozzle and the vapourizer were in operation during the treatment. In contrast, almost no inactivation of B. atrophaeus was found when the treatment chamber was initially flushed with PPA for 20 min before the plasma nozzle and the vaporizer were switched off and the samples subsequently placed in the treatment chamber. Therefore, the combination of ozone and high humidity (>90% RH), being also present when the nozzle was not in operation during the treatment, was not the cause for the spore inactivation in this study.
In case of high humidity, H 3 O + , OH -, •OH radicals or H 2 O 2 are generated in nonthermal gas plasmas (Scholtz et al. 2015). Hydroxyl radicals are known to be highly reactive ROS species formed in air plasmas (H€ ahnel et al. 2010). Our previous work (Muranyi et al. 2008) has shown that highly oxidizing hydroxyl radicals derived from elevated humidity in a DBD are likely to be involved in the destruction of spore structures. Patil et al. (2014) showed that B. atrophaeus endospores can be reduced by more than 6 log 10 within 1 min inside a sealed package with a DBD. A preferably high humidity proved to be a critical factor for the pronounced sporicidal effect, contributing to the formation of reactive species other than ozone (Patil et al. 2014). Using surface micro-discharge plasma, Jeon et al. (2014) reported about an increased inactivation of G. stearothermophilus endospores with increasing humidity. Their findings were attributed to a higher generation of water related species under high humidity such as hydrogen peroxide and hydroxyl radicals. The sporicidal effect of ozone was in contrast negligible but may have contributed to chemical reactions that formed water-related sporicidal species such as •OH and H 2 O 2 (Jeon et al. 2014). The same authors furthermore assumed that condensation of plasma activated water on the sample surface may contribute to the inactivation of spores (Jeon et al. 2014). This effect could also have been relevant in the present study since the humidity in the *Inoculated surface.  The air humidity had a strong influence on the reduction rate. Higher humidity of the process gas led to higher inactivation rates and this was attributed to the formation and impact of hydroxyl radicals (H€ ahnel et al. 2010). However, since the gasphase lifetime of hydroxyl radicals is reported to be very short, it is unlikely that hydroxyl radicals created in the discharge reach the target in a remote operation mode when the distance to the treatment site is more than 1 cm (Plimpton et al. 2013). But even though hydroxyl radicals represent a short-lived species, their contribution to the sporicidal effect cannot be ruled out, since at high humidity, they might be produced by the reaction of the longlived species such as ozone with water molecules close to the sample surface (Jeon et al. 2014). The generation of hydroxyl radicals via secondary chemical processes at the point of delivery was observed by Plimpton et al. (2013).
The authors reported about hydroxyl radical detection at a remote treatment site more than a meter away from the plasma discharge (DBD). The production of hydroxyl radicals was mainly attributed to secondary reactions of stable (ozone) and semi-stable (ozonide) species present in the device effluent (Plimpton et al. 2013). Since ozonide has been shown to exhibit a relatively long lifetime of several seconds (Sehested et al. 1982), its reaction with water may result in secondary hydroxyl radicals close to the sample surface. Advanced oxidative processes between ozone and hydrogen peroxide may also contribute to the presence of hydroxyl radicals (Plimpton et al. 2013). However, based on the presented results as well as the above cited findings, further studies are still needed in order to clarify which water related species are exactly responsible for the sporicidal effect of PPA and in which way they may be interrelated with other components.

Impact of the surface topography on the sporicidal effect of PPA
The sporicidal effect of PPA was in general lower on the Tychem F surface compared to the PET film even though the inoculation procedure was equal and the wettability was comparable. The complexity of the Tychem F surface microstructure therefore had significant impact on the spore inactivation. The Tychem F suit is a laminate containing Tyvek â , a spunbonded material made from highdensity polyethylene fibers. Light microscopic investigations had shown that the Tychem surface exhibits crevices and irregularities where endospores might be protected. The surface topography has been shown to have general   implications on the inactivation efficiency of atmospheric plasma. In case of complex food surface structures which can be found, for instance, on fresh produce like lettuce, the microbicidal action is commonly affected by cavities or crevices which provide shelter for micro-organisms (Bourke et al. 2017). Butscher et al. (2016) reported about the impact of the surface structure of the substrate to be disinfected. The inactivation of G. stearothermophilus by a DBD was more efficient on polypropylene than on wheat grains due to their rough surface and the deep ventral furrow. While on polypropylene granules, 10 min of treatment yielded spore reductions by about 5 log 10 , the maximum spore inactivation on wheat grains was 3 log 10 units after 60 min of treatment (Butscher Inactivation of Bacillus atrophaeus DSM 675 endospores dried in a protein matrix (bovine serum albumin, 10 g l À1 ) on a Tychem F protective suit by plasma processed air (PPA). Samples were either exposed to PPA for 1 min (a) or for 10 min (b) whereat each sample was individually wiped for 1 min with a swab using either 10 µl of water (H 2 O) or plasma-activated water (PPW) to resolve the protein matrix.  Hertwig et al. (2015) reported about a significantly lower sporicidal (B. subtilis) action of CAP on peppercorns compared to flat glass or spherical glass beads. This was explained by the more complex surface structure of peppercorns which is characterized by cracks, grooves and pits, which might cause shadow effects for the different generated components of the plasma (Hertwig et al. 2015). Regarding the use of PPA for the disinfection of surfaces like, for example, the protective equipment of rescue forces, the topography (smoothness) of the respective devices or protective suits needs to be considered and the treatment parameters adjusted accordingly.

Impact of agglomerates and organic loads
It is known that with the exception of gamma and ebeam irradiation, more or less all sterilization techniques are to some degree affected by two major aspects, which limit the accessibility of micro-organisms. The presence of micro-organisms in agglomerates or multiple layers as well as their presence in organic matrices like proteins or polysaccharides (Moisan et al. 2014). The results of this study are in accordance with previous findings and show that matrix effects due to microbes present in agglomerates as well as embedment in organic loads are absolutely decisive for the achievable inactivation of endospores by PPA. When spot inoculation was applied, endospores were partly present in agglomerates, which significantly affected the sporicidal efficiency of PPA compared to spray inoculation, where only monolayers were present. The inactivation effect was negligible when endospores were embedded in a protein matrix, which showed that matrix effects due to organic loads are presumably even more critical than agglomeration of spores. Accordingly, the penetration depth of the reactive species seemed to be rather limited. A high fraction of endospores which are not directly accessible are likely to survive longer treatment periods. M€ uller et al. (2018) likewise reported that the presence of spore clusters can affect the sporicidal efficiency of remote plasma due to shielding effects. Deng et al. (2005) previously investigated the impact of the microbial load (B. subtilis endospores) on substrates (polycarbonate membranes) treated with atmosphericpressure glow discharges. They found that increasing inoculation densities over four orders of magnitude led to significantly higher D-values. This was attributed to stacked endospores forming multilayered structures when more than 6Á0 9 10 7 spores were present on a filter of 0Á78 cm 2 . The top layers of endospores (even if inactivated) might form a physical barrier to shield those beneath them from gas plasma penetration and hence contribute to increased survival (Deng et al. 2005).
According to Moisan et al. (2014), there are two phases in the spore inactivation kinetics when layers of spores are present. The first phase corresponds to isolated spores or those on top of a stack while spores underneath a stack are inactivated in a second phase. The inactivation of B. atrophaeus endospores with a flowing late afterglow of a N 2 -O 2 microwave discharge (200 W of microwave power at 2450 MHz) in a 50 l treatment chamber exhibited two different D-values. The first phase resulted in a 5 log 10 reduction after 15 min (D 1 = 3Á1 min) which is comparable to the spore inactivation obtained in the present study. The second phase between 15 and 60 min indicated a D-value of 33 min (Moisan et al. 2013;Moisan et al. 2014). It was therefore assumed, that matrix effects due to spores within a stack, aggregated, located in cavities or crevices, or covered by some bio-product/debris primarily have impact on the inactivation efficiency.
In contrast, phenotypic variations (e.g. more resistant subpopulations) are likely to play a minor role (Moisan et al. 2013). Furthermore, the microbicidal efficiency of cold atmospheric gas plasma is also known to be affected when bacteria are protected in biofilms (Scholtz et al. 2015;Bourke et al. 2017). In order to overcome such shielding effects, the combination of PPA exposure with mechanical disintegration of matrices may synergistically improve the sporicidal effect.

Impact of mechanical wiping and PPW during PPA exposure
When high loads of endospores were dried in a protein matrix and treated with PPA in combination with mechanical wiping using either small amounts of deionized water or PPW (condensate), a high inactivation of endospores on the Tychem F surface by up to more than 5 log 10 was achieved. In contrast, there was no inactivation at all without mechanical wiping. It is reasonable to assume, that the previously embedded spores were distributed on the sample surface after resolving the matrix during wiping and therefore became exposed to PPA. Using 10 µl of PPW (condensate) instead of water to resolve the matrix further improved the sporicidal effect significantly (Fig. 2). Plasma processed water (PPW) is a known disinfectant comprising a complex mixture of various chemical species belonging to ROS and RNS. The use of oxygen, nitrogen and water as parent molecules in PPW production will result in various primary species produced in the plasma discharge which continue to react to form more stable secondary species (Thirumdas et al. 2018). Being characterized by a typically high oxidation-reduction potential, increased conductivity and low pH due to the formation of nitric and nitrous acid, significant antimicrobial activity of PPW was found in many . The presence of nitrogen oxides (NOx) and corresponding acids, hydrogen peroxide (H 2 O 2 ), ozone (O 3 ) and peroxynitride (ONOO À ) are assumed to be responsible for the microbicidal activity (Scholtz et al. 2015). The sporicidal effect of PPW has not been studied in detail to date. In the present study, PPW (collected condensate) alone did not provide any sporicidal effect within 10 min, neither when endospores were dried on the Tychem F surface and submerged in PPW nor when spores were directly suspended in PPW (data not shown). Similar findings have already been published previously. In fact, Sun et al. (2012) found a strong sporicidal (B. subtilis) effect of PPW using a Cold Atmospheric-Pressure Air Plasma Microjet but only when spores were already present in the water during PPW generation. They assumed that short-lived species were most likely responsible. However, when B. subtilis endospores were treated in previously generated PPW, Sun et al. (2012) did not observe any inactivation. Long lived species therefore did not provide a sporicidal effect within the applied treatment time of 20 min (Sun et al. 2012). Oehmigen et al.
(2010) likewise did not prove any sporicidal effect when B. atrophaeus was treated for up to 30 min in physiological saline which was activated by a DBD. The same accounts for Schnabel et al. (2019) who did not find a sporicidal effect of PPW derived from a microwave discharge at a frequency of 2Á45 GHz and an input power in the range of 1Á1 kW. Nevertheless, as stated above, in the present study it was found that the endospore inactivation on the Tychem F samples was more pronounced when PPW was used instead of water during mechanical wiping, which points to a synergistic effect between PPW and PPA. Previous studies already speculated that the acidity and the plasma derived reactive species are interconnected. A lower pH may enhance the penetration of reactive species to penetrate cell walls while the presence of reactive species may reduce the resistance to an acidic environment (Oehmigen et al. 2010;Sun et al. 2012). A synergistic effect of PPA and PPW for the inactivation of various vegetative bacteria as well as B. atrophaeus endospores was indeed recently reported by Schnabel et al. (2019) when PPA and PPW were applied serially. However, more detailed studies are necessary to confirm a synergistic sporicidal effect of PPW combined with PPA during mechanical wiping. Overall, it can be concluded that the developed portable plasma system is suitable to reduce microbial loads on surfaces which are placed in a treatment chamber of large volume (here: 300 l) and flushed with a humidified afterglow of the DBD plasma nozzle. The reactive species in PPA with respect to ROS and RNS remain to be identified in future studies as well as the exact composition of plasma-activated water (PPW). PPW may serve as an additional on-site generated disinfectant, which synergistically improves the sporicidal action of PPA.