Synergistic Effect of Atmospheric-pressure Plasma and TiO2 Photocatalysis on Inactivation of Escherichia coli Cells in Aqueous Media

Atmospheric-pressure plasma and TiO2 photocatalysis have been widely investigated separately for the management and reduction of microorganisms in aqueous solutions. In this paper, the two methods were combined in order to achieve a more profound understanding of their interactions in disinfection of water contaminated by Escherichia coli. Under water discharges carried out by microplasma jet arrays can result in a rapid inactivation of E. coli cells. The inactivation efficiency is largely dependent on the feed gases used, the plasma treatment time, and the discharge power. Compared to atmospheric-pressure N2, He and air microplasma arrays, O2 microplasma had the highest activity against E. coli cells in aqueous solution, and showed >99.9% bacterial inactivation efficiency within 4 min. Addition of TiO2 photocatalytic film to the plasma discharge reactor significantly enhanced the inactivation efficiency of the O2 microplasma system, decreasing the time required to achieve 99.9% killing of E. coli cells to 1 min. This may be attributed to the enhancement of ROS generation due to high catalytic activity and stability of the TiO2 photocatalyst in the combined plasma-TiO2 systems. Present work demonstrated the synergistic effect of the two agents, which can be correlated in order to maximize treatment efficiency.

Non-equilibrium low temperature plasma inactivation of microorganisms has attracted increased attention in recent years due to its efficacy against a wide range of bacteria and fungi 1,2 . Among plasma-generated effects, highly-reactive chemical species, in particular reactive oxygen (ROS) and nitrogen species (RNS), are generally considered to be responsible for the observed antimicrobial efficacy 3 . To date, most of the studies have focused on the activity of species generated in the gas phase, however, the lifetime of chemical species generated in the gas phase is relatively short, as they are quenched rapidly during their frequent collisions with molecules or other species. High reactivity and short lifetime also means that only a fraction of the species generated during plasma treatment are able to penetrate the gas− liquid interface to reach the intended target 4 , potentially limiting the efficiency of gas phase plasma inactivation against microorganisms in a moist environment or in bulk liquids.
Killing efficiency can be enhanced by circulating bacterial suspension to facilitate the exposure of microorganisms to the species generated at the gas− liquid interface 5 , however this may not be feasible in some treatment systems. Alternatively, sufficiently high concentrations of reactive species can be attained by generating .
Highly-reactive plasma environment may provide an alternative mean for the enhancement of the TiO 2 catalyst efficiency. By adjusting the processing parameters and treatment conditions, plasma can be modulated to preferentially generate metastables, ions, radicals, or (V)UV photons. As such, the high voltage discharge zone can serve as an effective source of (V)UV photons for TiO 2 photocatalysis, generating a sufficient number of oxidative species to effectively inactivate Bacillus subtilis spores 25 . Furthermore, plasma-generated strong electric field may prevent recombination of electrons with holes on the surface of TiO 2 , thus improving the quantum efficiency of photocatalysis 23 . Moreover, ultrasound and shock waves generated in electrical discharge process can clean the TiO 2 surface and enhance the mass transport of the reactants to the solid surface, thereby facilitating participation of more active catalyst sites in the heterogeneous catalytic reactions 26 .
The formation of strong electric fields, thermal effects, UV radiation, ultrasound and shock waves depend strongly upon the discharge parameters, including gas composition. Therefore, in this study, we investigated the antibacterial efficiency of plasma/TiO 2 photocatalysis system (Fig. 2), where microplasma were generated in four types of gases, namely N 2 , He, air and O 2 , known to produce vastly different biological effects. The effect of treatment system geometry and energy of the discharge on the antibacterial efficacy was investigated. The inactivation efficiency of Escherichia coli cells, evaluated via Colony-Forming Units (CFU) counts on petri dishes, was correlated with the concentrations of key reactive oxygen species (ROS), i.e. hydrogen peroxide (H 2 O 2 ), ozone (O 3 ), hydroxyl radicals (·OH), to understand the mechanisms that may govern the observed antibacterial efficacy.

Results
Chemical reaction rate model and inactivation efficiency. Plasma-generated species, such as ·O, ·OH, H 2 O 2 or O 3 , play an important role in microorganism inactivation and cell death 25 . Different empirical kinetic models can be used to elucidate the observed bacterial inactivation process. Among them, pseudo-first-order kinetics has been previously shown to produce best fit to the experimentally obtained data for the inactivation of microorganisms by advanced oxidation processes (AOPs). Given that the kinetic rate is widely applied to evaluate different disinfection processes, the process of E. coli disinfection in this study was fitted with CW model expressed by Eq. 1 23,27,28 : where C 0 and C t are the respective concentration of live E. coli cells at the start (t = 0) and the end of treatment (t), respectively, and k is the kinetic rate.
Effect of gas chemistry and treatment time on bacterial inactivation. The sensitivity of E. coli cells to different types of plasma treatment is presented in Fig. 3. The efficacy of the treatment showed strong dependence on both the treatment time and the type of gas being used. Under the same experimental conditions (4.5 kV, 2 SLM flow rate, 10 6 CFU/ml), the highest inactivation efficiency was observed for O 2 microplasma, followed by air microplasma, He microplasma and N 2 microplasma in the order of decreasing efficacy. Increasing exposure time significantly enhanced the decontamination effect of the treatment, with 4 min and 5 min being sufficient to achieve > 99% bacterial inactivation using O 2 and air plasma, respectively. The experimentally obtained inactivation data were then evaluated using a kinetic model. To calculate the pseudo-first-order rate constant associated with the disinfection kinetics, a linear regression was performed on the data plotted in Fig. 3(e). Figure 3(e) shows the ln(C t /C 0 ) values measured as a function of N 2 , He, air, or O 2 plasma treatment time and fitted theoretically by using Eq. 1. As expected, the ln(C t /C 0 ) values decreased Top view of the microplasma device (a) and cross-sectional view of single-discharge system without TiO 2 (SD) (b), SD system with a circular TiO 2 film (CTD) (c), and SD system with a plate-like TiO 2 film (PTD) (d).
with increasing the treatment time. The calculated theoretical kinetic rates of k(N 2 ) = 0.009 s −1 , k(He) = 0.022 s −1 , k(air) = 0.052 s −1 and k(O 2 ) = 0.076 s −1 were in good agreement with the measured values (R 2 > 0.995 for all feed gases). This indicates that the plasma-generated ROS may play a crucial role in killing the microorganisms, with the efficacy of the treatment directly linked to the density of oxidants generated inside the microplasma. Among these four types of microplasma arrays, the O 2 microplasma array was the most efficient in killing E. coli cells in water, which can be attributed to its relatively high density of ROS.
Effect of TiO 2 photocatalysis on the antibacterial performance of O 2 microplasma. Identified as the most promising type of plasma for bacterial inactivation, the O 2 discharge was then selected to investigate the effect of TiO 2 photocatalysis on the antibacterial performance of the plasma in water system. Figure 4 shows the comparison between the ln(C t /C 0 ) values measured in E. coli inactivation experiments performed with the O 2 plasma-SD, CTD, and PTD systems and fitted theoretically by Eq. 1. Kinetic rates (k) and correlation coefficients (R 2 ) derived from the fittings are summarized in Table 1. The theoretical values are in agreement with the measured values, with correlation coefficients as high as 0.995. Increasing the energy of discharge (V P ) positively contributed to the enhancement of inactivation efficiency of the treatment across all three discharge systems (SD, CTD and PTD).
Furthermore, integration of TiO 2 film into the treatment system greatly improved its ability to inactivate E. coli cells independent of input voltages. CTD configuration was particularly effective, showing consistently higher inactivation rate and shorter treatment times. E. coli cells with the concentration of 10 6 CFU/mL were completely inactivated by O 2 microplasma in the CTD system at V P = 6.0 kV within the treatment time of 1.0 min. Such increase in activity was attributed to a higher concentration of ROS generated in aqueous solution by the TiO 2 film photocatalyst and the O 2 microplasma.   Table 2 shows changes in ROS (H 2 O 2 , O 3 and ·OH) concentration in aqueous solutions treated with N 2 , He, air and O 2 plasma as a function of treatment time. Longer treatment times resulted in a significant increase in the concentration of all ROS. The concentration of H 2 O 2 in the distilled water attained levels of 1.463, 2.036, 5.644 and 14.013 mg/L after 5 min treatment with N 2 , He, air and O 2 plasma discharges, respectively, at 4.5 kV. This was attributed to the high electron density, energy of the plasma and long lifetime of the excited species that facilitate the energy transfer between the excited plasma species and water molecules, leading to H 2 O 2 formation (*e − + H 2 O → ·H + ·OH + e − , ·OH + ·OH → H 2 O 2 ) 2 . In addition, the dissolved zone (O 3 ), acting as ROS, has been considered to be very effective against a wide range of microorganisms 3 . Compared with that in the air and O 2 plasma treated solutions, the generated ozone concentration was significantly lower in solutions treated with either He or N 2 microplasma. The highest ozone concentration of approximately 0.758 mg/L was produced in the O 2 microplasma system, followed by that in the air microplasma (at 0.700 mg/L), He microplasma (at 0.080 mg/L), and N 2 microplasma (at 0.047 mg/L).

Concentration of reactive oxygen species (ROS) under different plasma conditions.
More importantly, the concentration of OH radicals, a species widely considered to be among the most important agents in the bacterial inactivation process 25,29 , showed a remarkable increase with the treatment time, attributed to electron impact dissociation of water 30 . Treatment with O 2 microplasma showed the highest ·OH concentration (3.998 mg/L) after 5 min, with slightly lower values obtained in solutions treated with air plasma (at 3.683 mg/L). On the other hand, ·OH concentration in solutions treated by He and N 2 microplasma were relatively low (at 0.363 mg/L and 0.247 mg/L, respectively). The ROS concentration in the plasma-treated solutions is well related with the rate of killing of E. coli cells in water. The higher the ROS concentration is, the higher the sterilization efficiency would be.
Addition of TiO 2 photocatalyst to O 2 plasma system resulted in a notable increase in the concentration of H 2 O 2 , O 3 and ·OH, with the results for different plasma discharge systems (SD, CTD and PTD) summarized in Table 3. Similar to the single-discharge system (without TiO 2 photocatalyst), the respective concentrations of H 2 O 2 , O 3 and ·OH increased with treatment duration. Increasing input voltage also resulted in a notable increase in the concentration of ROS in all systems. Interestingly, an increase of input voltage from 3 kV to 4.5 kV led to a multi-fold increase in the concentration of all ROS, whereas subsequent increase of input voltage to 6 kV only produced a modest increase in the ROS concentration.
In terms of system configuration, the highest concentrations of H 2 O 2 , O 3 and ·OH were produced in the CTD system, reaching the levels of 17.985, 0.989 and 8.098 mg/L, respectively. Slightly lower values of 16.909, 0.865 and 6.909 mg/L, respectively, were attained using the PTD system, with the lowest respective levels of 15.494, 0.837 and 5.570 mg/L measured in the solution treated with SD system. These results indicate that the concentration of H 2 O 2 , O 3 and ·OH in plasma-treated solution was significantly related to the TiO 2 photocatalytic activity, and the synergistic effect between the TiO 2 photocatalyst and O 2 plasma may be optimized by changing the configuration of the treatment system. Furthermore, since the sterilization efficiency under 4.5 kV discharge in CTD and PTD was similar to that at the higher input voltage, 4.5 kV was recommended as the optimum discharge voltage in terms of energy consumption.   Activity and stability of TiO 2 films in O 2 plasma system. E. coli inactivation efficiency of an original TiO 2 film was compared with that of a 'used' TiO 2 film, the latter being used for six treatments prior to this measurement. Figure 5 shows the performance of CTD, and PTD O 2 plasma systems containing either the original or used TiO 2 film when operated at 4.5 kV. In the CTD system containing the used TiO 2 film, the inactivation efficiency of E. coli was similar to that fitted with an original TiO 2 film. In the PTD system, however, the E. coli inactivation efficiency decreased when the used TiO 2 film was employed, although it was still higher than that in the SD system. The results indicate that a TiO 2 film deployed in the CTD system retained high photocatalytic activity over a longer period of use, whereas its activity was partly lost in the PTD system. This can probably be attributed to the change of TiO 2 phase film during discharge channel formation in the PTD system but not in the CTD system, which was proven with the experimental XRD data.
Structural characterization of TiO 2 films before and after O 2 plasma discharge. Figure 6 shows the XRD patterns of TiO 2 films before and after their repeated use in the O 2 plasma treatment system operated at 4.5 kV. After exposure to O 2 discharge in the CTD system, the TiO 2 film retained the same anatase phase diffraction peaks as the original sample, which also confirmed that the TiO 2 film had not been changed during plasma treatment. In the PTD system, however, the phase of TiO 2 had been altered since several peaks representing the rutile phase were observed in the diffraction pattern. In the PTD system, the plate-like TiO 2 film was attached to the stainless steel surface acting as the ground electrode, the temperature of which would increase rapidly in the discharge process, leading to the TiO 2 crystal structure transfer from anatase phase to rutile phase 19,31 . The rutile TiO 2 has been previously demonstrated to possess lower catalytic activity than the anatase counterpart 11,31 , which may account for the decreased photocatalytic activity and inactivation efficiency of the used TiO 2 in the PTD system.

Discussion
Plasma-generated ROS, such as singlet oxygen (O radical), hydrogen peroxide (H 2 O 2 ), ozone (O 3 ) and hydroxyl radical (·OH), are generally considered to play significant roles in sterilization and decontamination of surfaces  and liquids. When generated in gas phase for the treatment of pathogens in liquid phase, these species need to cross the gas− liquid interface in order to reach the intended target. This presents a considerable challenge, since the characteristic time of reaction between the O radical and pure water is τ 1 = 0.02 s, the diffusion coefficient in water is D 1 = 2 × 10 −9 m 2 s −1 and thus the estimated diffusion distance is τ = ≈ . L D 6 7 1 11 μ m 32,33 . Therefore, it is difficult for these highly reactive radicals to penetrate the gas− liquid interface (several μ m to hundreds of μ m) or diffuse into the solution within their short lifetime during plasma treatment 33,34 2 2 uv ), although the oxidation-reduction potential of H 2 O 2 (acid, 1.776 V) is much lower than the corresponding value for O 3 (2.07 V) 35,36 . By generating species in the close proximity to the intended microbial targets, i.e. by creating a discharge in the solution itself, it is possible to significantly enhance the decontamination efficacy of plasma treatment.
Despite significant progress in our understanding of the electrical discharges in aqueous solutions, many chemical and physical aspects of this process remain poorly understood. This is in part due to the complexity of the plasma-generated hydrodynamic effects, such as high liquid inhomogeneity due to complex nonlinear behavior which may affect the electric field, discontinuous liquid-vapor boundaries due to liquid vaporization, increased solution conductivity due to plasma generation of charged species, and increase in molecular density in the vicinity of the high voltage electrode. These effects will also depend on the operating parameters of the plasma, such as applied energy or gas chemical composition.
Our results show that among the N 2 , He, air, and O 2 microplasma arrays, the O 2 microplasma is the most effective in killing E. coli cells suspended in water, at least in part due to the relatively higher concentration of the ROS generated by this plasma in the treated solution. Once generated in the gas phase, these ROS species can immediately attack E. coli cells at the liquid/gas interface, or they may dissolve into the water, reaching the cells further away from the discharge. The increase in the inactivation efficiency with the treatment time is well explained by the chemical reaction rate model.
Depending on the voltage and the chemistry of the gas used to produce the plasma, as well as the duration of exposure, plasma may trigger either the physical disruption of bacteria or a cascade of biochemical reactions leading to bacterial death 3,29 . The Lissajous figures of the N 2 , He, air, and O 2 microplasma arrays indicate that the discharge power showed little dependence on the chemistry of the feed gas used. However, the discharge power was found to increase rapidly, from 12W to 50W, as the V P increased from 3.5 to 6.0 kV. Figure 7 shows the total ROS concentration in solutions treated in SD, PTD and CTD O 2 plasma systems as a function of plasma discharge power. The CTD O 2 microplasma system was found to generate the highest concentration of ROS under every discharge power, generating ∼ 27.5 mg/L of ROS at 50W, compared to ∼ 21.8 mg/L and ∼ 24.7 mg/L generated at the same discharge power using SD and PTD systems, respectively. Lower efficiency of PTD system may be attributed to the partial transition of TiO 2 from the anatase into less catalytically-active rutile phase as a result of plasma-generated heating of the ground electrode onto which the TiO 2 film was mounted.
In all systems tested, a strong positive correlation between the sterilization efficiency and the ROS concentration was demonstrated, suggesting a possible avenue for process optimization, for instance, through fast and efficient transport of plasma-generated species to the target organisms 2,35,37 . Generally, cell membranes containing unsaturated fatty acids and protein molecules can sustain significant damage from the strong oxidizing effects of ROS 6 , interfering with the ability of the cell membrane to effectively transport ions and polar compounds. Moreover, the plasma sheath can be formed in vicinity of E. coli cells located at the plasma/liquid interface, leading to charge accumulation on the bacterial surface 36 . The electrostatic force caused by charge accumulation on the outer surface of the cell membrane could overcome the tensile of the membrane and cause its rupture 29,36 .
Our study also showed the synergistic activity that arises as a result of combining two advanced oxidation processes, namely TiO 2 photocatalysis and microplasma treatment. The improvement in killing efficacy was attributed to enhanced generation of ROS, which was in line with previous studies that showed substantial increase in the relative emission intensities of OH (313 nm) and O (777 nm) radicals in distilled water (1.7 and 3.8 times, respectively) in plasma systems augmented with TiO 2 photocatalytic layer 16 . Figure 8 shows the likely reaction mechanism in the TiO 2 -microplasma photocatalysis process in the CTD system. When the plasma is generated between the high-voltage electrodes and ground electrode, large amounts of UV photons and ozone are produced in the plasma-discharge zone. These plasma-generated photons reach the surface of the TiO 2 film, inducing photocatalysis, resulting in the generation of more OH radicals, and other active species (e.g., H 2 O 2 and ·O 2 − ). In addition, ozone accepts electrons to produce ·O 3 − radicals, from which OH radicals are generated by radical chain reactions. A series of photocatalytic redox reactions are presented in Eq. 2-6. Active species (H 2 O 2 , O 3 , ·O 2 − and ·OH) produced as a result of the aforementioned reactions in the plasma-enhanced TiO 2 photocatalytic system can then be used to decontaminate polluted water, by directly killing pathogenic microorganisms, such as E. coli cells in this study, via oxidative stress, and by oxidizing the organic pollutants, such as organic dyes and phenolic derivatives, into smaller, less-hazardous molecules [15][16][17]38 . Depending on the chemistry of the pollutant or the species of the pathogen, it may be necessary to optimize the treatment system. For instance, in our study, highest efficiency was attained in the TiO 2 -microplasma combined systems in the CTD configuration fed with O 2 gas. The presence of TiO 2 provides an additional mechanism for bacterial inactivation, whereby pathogens located on TiO 2 surface are subject to direct electron/hole exchange with the surface, affecting the cell membrane, which renders them more susceptible to oxidative damage by generated ROS. Once inside the cell, ROS can induce significant damage to intracellular components of the organism, including carbohydrates, lipids, proteins and nucleic acids, affecting cellular metabolism, and, at sufficiently high concentration, leading to cell death 14,19 . Importantly, our results show that the photocatalytic activity of TiO 2 can be maintained over time, suggesting the proposed system may be used over relatively long periods of time without substantial loss in decontamination efficiency.

Conclusion
In this study, the plasma inactivation of E. coli cells suspended in water was performed by using atmospheric-pressure N 2 , He, air, and O 2 microplasma arrays. Compared to the N 2 and He microplasma arrays, the air and O 2 microplasma arrays are more effective in inactivating E. coli cells in water. Addition of TiO 2 photocatalysis to the O 2 microplasma system can lead to substantial improvements in the inactivation efficiency against E. coli, attributed primarily to the enhancement of ROS generation. System configuration, specifically the placement of TiO 2 thin film, notably affected both the ROS generation and the longevity of the photocatalyst, with the highest inactivation efficiency and stability observed in the CTD system operated with O 2 as the feed gas. The chemical reaction rate model provided good fit to the experimentally-obtained data, and may be used to quantitatively describe the efficiency of E. coli cell inactivation in single-discharge and plasma-TiO 2 combined systems.

Methods
Atmospheric-pressure microplasma arrays were used to inactivate E. coli cells in aqueous media, as shown in Fig. 2 7,39 . Briefly, the plasma device consists of 29 microplasma jet units housed in a glass cup (inner diameter: 50 mm). Each discharge tube consists of three parts: (1) an inner electrode, (2) a quartz sleeve of inner electrode, and (3) an external quartz tube. The separation between the external quartz tube and the inner quartz sleeve is approximately 80 μ m. When the gas passes through the separation distance, the plasma is produced by dielectric barrier discharge 37 . The feed gas, such as He, N 2 , air, or O 2 is supplied into the discharge tubes from the bottom of the device at the flow rate of 2.0 standard liter per minute (SLM). A stainless steel ground electrode surface (Ø 30 mm) is positioned about 10 mm above the discharge tubes in the discharge reactor (as shown in Fig. 2).
Film of TiO 2 nanotubes were prepared by an electrochemical method 40 . A 300-μ m-thick titanium foil, acting as the anode, was cleaned with acetone for 15 min and then washed with distilled water for 15 min in an ultrasonic cleaner machine. One nickel plate was also immersed in electrolyte (0.08 mol/L C 2 H 2 O 4 ·2H 2 O and 0.135 mol/L NH 4 F), acting as the cathode. The power supply was utilized to provide a DC output with the voltage of 20 V. After the 2 h electrochemical reaction, the anodized samples were calcined at 400 °C for 2 h in a muffle furnace.
Once prepared, TiO 2 films were utilized in the plasma reactors in two configurations. In the first case, a circular TiO 2 film was distributed around the inner surface of the discharge reactor cylinder (CTD), while in the second case, a plate-like TiO 2 film was attached to the stainless steel ground electrode surface in the discharge reactor (PTD), as shown in Fig. 2(b) and (c), respectively. The power supply (TCP-2000K, Najing Suman Electronic, China) capable of supplying bipolar AC output with the peak voltage (V P ) of 0-20 kV at an AC frequency of 9.0 kHz was used. The discharge power can be calculated by a Lissajous figure formed with the charges across this capacitor and the applied voltage across the discharge chamber.
A colony of E. coli was dispersed into approximately 200 ml of LB liquid medium (Tryptone: 2 g, Yeast Extract: 1 g, NaCl: 2 g, distilled water: 200 g, pH = 7.2) and was incubated on the shaking table (rotation speed: 200 rpm, 28 °C, 24 h) until a sufficient number of E. coli cells were obtained. Then, test solutions were prepared by inoculating sodium chloride solution (1% NaCl) to the concentration of 10 6 CFU/ml. Then, a 100 ml aliquot of the test solution was treated by N 2 , He, air and O 2 microplasma or these microplasma coupled with TiO 2 films, with the treatment time ranging from 1 to 5 min. Control sample received no plasma treatment. After these treatments, 100 μ l of the treated solution was inoculated on the standard petri dishes 9 cm in diameter containing approximately 15 ml of LB solid medium. The plates were then incubated for 24 h at 37 °C, at which stage colony counting was performed to estimate the survival rate of the treated bacteria.
The concentration of hydrogen peroxide in the plasma-treated water was obtained by color forming reactions and spectrophotometric measurements. When titanium oxysulfate (TiOSO 4 ) reacted with H 2 O 2 , a yellow-colored complex (pertitanic acid) was formed and UV-Vis measurement was done at 407 nm to colorimetrically determine the concentration of H 2 O 2 (TiO 2+ + H 2 O 2 → [TiO(H 2 O 2 )] 2+ ) 41 . The concentration of dissolved ozone was determined by the indigo method 42 . For hydroxyl free radical detection, the salicylic acid (SA) was used as a selective trapping reagent of ·OH and the production of ·OH radicals was quantified by High-performance liquid chromatographic (HPLC) assay 43 .