Antimicrobial Applications of Ambient–Air Plasmas

Home , Plasma medicine

Matthew John Pavlovich

A dissertation submitted in partial satisfaction of the

requirements for the degree of

Doctor of Philosophy

Chemical Engineering

in the

Graduate Division

of the

University of California, Berkeley

Committee in charge:

Professor Douglas S. Clark, Co-Chair

Professor David B. Graves, Co-Chair

Professor Kevin Healy

Spring 2014

Abstract

Antimicrobial Applications of Ambient-Air Plasmas

Matthew John Pavlovich

Doctor of Philosophy in Chemical Engineering

University of California, Berkeley

Professors Douglas S. Clark and David B. Graves, Co-Chairs

The emerging field of plasma biotechology studies the applications of the plasma phase of matter to biological systems. “Ambient-condition” plasmas created at or near room temperature and atmospheric pressure are especially promising for biomedical applications because of their convenience, safety to patients, and compatibility with existing medical technology. Plasmas can be created from many different gases; plasma made from air contains a number of reactive oxygen and nitrogen species, or RONS, involved in various biological processes, including immune activity, signaling, and gene expression. Therefore, ambient- condition air plasma is of particular interest for biological applications.

To understand and predict the effects of treating biological systems with ambient-air plasma, it is necessary to characterize and measure the chemical species that these plasmas produce. Understanding both gaseous chemistry and the chemistry in plasma-treated aqueous solution is important because many biological systems exist in aqueous media. Existing literature about ambient-air plasma hypothesizes the critical role of reactive oxygen and nitrogen species; a major aim of this dissertation is to better quantify RONS by produced ambient-air plasma and understand how RONS chemistry changes in response to different plasma processing conditions. Measurements imply that both gaseous and aqueous chemistry are highly sensitive to operating conditions. In particular, chemical species in air treated by plasma exist in either a low-power ozone-dominated mode or a high-power nitrogen oxide-dominated mode, with an unstable transition region at intermediate discharge power and treatment time. Ozone (O3) and nitrogen oxides (NO and NO2, or NOx) are mutually exclusive in this system and that the transition region corresponds to the transition from ozone- to nitrogen oxides-mode. Aqueous chemistry agrees well with to air plasma chemistry, and a similar transition in liquid-phase composition from ozone mode to nitrogen oxides mode occurs as the discharge power increases.

One prominent example of plasma biotechnology is the use of plasma-derived reactive species as a novel disinfectant. Ambient-air plasma is an attractive means of disinfection because it is non-thermal, expends a small amount of power, and requires only air and electricity to operate. Both solid surfaces and liquid volumes can be effectively and efficiently decontaminated by the reactive oxygen and nitrogen species that plasma generates. Dry surfaces are decontaminated most effectively by the plasma operating in NOx mode and less effectively

1 in ozone mode, with the weakest antibacterial effects in the transition region, and neutral reactive species are more influential in surface disinfection than charged particles. Aqueous bacterial inactivation correlates well with ozone concentration, suggesting that ozone is the dominant species for bacterial inactivation under the condition of a low-power discharge. Alternatively, air plasma operating in the higher-power, nitrogen oxides-rich mode can create a persistently antibacterial solution. Finally, when near-UV (UVA) treatment follows plasma treatment of bacterial suspension, the antimicrobial effect exceeds the effect predicted from the two treatments alone, and addition of nitrite to aqueous solution, followed by photolysis of nitrite by UVA photons, is hypothesized as the primary mechanism of synergy.

The results presented in this dissertation underscore the dynamic nature of air plasma chemistry and the importance of careful chemical characterization of plasma devices intended for biological applications. The complexity of atmospheric pressure plasma devices, and their sensitivity to subtle differences in design and operation, can lead to different results with different mechanisms.

Table of Contents

Chapter 1: Introduction to Ambient-Condition Plasmas and Plasma Biotechnology...... 1 1.1 Abstract ...... 1 1.2 Ambient-Condition Plasmas ...... 1 1.3 Plasma Biotechnology ...... 3 1.4 Plasma Disinfection ...... 4 1.5 Ambient-Air Plasma Chemistry ...... 5 1.6 Comparison to Similar Techniques and Chemistries ...... 7 1.7 Conclusion and Outline...... 8

Chapter 2: Methods and Materials ...... 10 2.1 Abstract ...... 10 2.2 The Dielectric Barrier Discharge (DBD) Device ...... 10 2.3 Plasma Generation and Characterization ...... 13 2.4 Gaseous Chemistry Measurements ...... 13 2.5 Aqueous Chemistry Measurements ...... 14 2.6 Antibacterial Measurements ...... 15

Chapter 3: Effect of Discharge Parameters and Surface Characteristics on Ambient-Gas Plasma Disinfection ...... 16 3.1 Abstract ...... 16 3.2 Introduction ...... 16 3.3 Experimental Section ...... 16 3.4 Results ...... 17 3.5 Discussion ...... 23 3.6 Conclusion ...... 26

Chapter 4: Quantification of Air Plasma Chemistry for Surface Disinfection ...... 28 4.1 Abstract ...... 28 4.2 Introduction ...... 28 4.3 Experimental Section ...... 29 4.4 Results ...... 29 4.5 Discussion ...... 37 4.6 Conclusion ...... 42

Chapter 5: Ozone Correlates with Antibacterial Effects from Indirect Air Dielectric Barrier Discharge Treatment of Water ...... 44 5.1 Abstract ...... 44 5.2 Introduction ...... 44 5.3 Experimental Section ...... 45 5.4 Results ...... 46 5.5 Discussion ...... 56 5.6 Conclusion ...... 59

Chapter 6: Antimicrobial Synergy Between Ambient-Gas Plasma and UVA Treatment of Aqueous Solution ...... 61 6.1 Abstract ...... 61 6.2 Introduction ...... 61 6.3 Experimental Section ...... 63 6.4 Results ...... 64 6.5 Discussion ...... 72 6.6 Conclusion ...... 75

Chapter 7: Conclusion...... 78 7.1 Abstract ...... 78 7.2 Gas-Phase Chemistry and Surface Disinfection ...... 78 7.3 Liquid-Phase Chemistry and Aqueous Disinfection ...... 78 7.4 Outlook ...... 79

Appendix A: Long-term antibacterial efficacy of air plasma-activated water ...... 80 A.1 Abstract ...... 80 A.2 Introduction ...... 80 A.3 Experimental Section ...... 80 A.4 Results and Discussion ...... 81 A.5 Conclusion ...... 84

Appendix B: Air Spark-like Plasma for Frugal Antimicrobial NOx Generation ...... 85 B.1 Abstract ...... 85 B.1.1 Frugal engineering ...... 85 B.1.2 Antimicrobial air plasma and plasma-activated water ...... 87

B.2 Experimental Section ...... 88

B.2.1 NOx Box with spark-like discharge ...... 88 B.2.2 High-voltage power supplies ...... 89 B.2.3 FTIR spectroscopy ...... 91 B.2.4 Aqueous chemistry measurements and antibacterial evaluation ...... 92 B.3 Results and Discussion ...... 92 B.4 Frugal Plasma...... 99 B.5 Conclusion ...... 100

Appendix C: Disinfection in South Africa and an Outlook for Plasma Disinfection ...... 102 C.1 Introduction ...... 102 C.2 Technical Background ...... 102 C.3 Locations Visited ...... 102 C.4 Findings...... 108 C.4.1 Disinfection Protocols ...... 108 C.4.1 Access to Resources ...... 112 C.5 Outlook for Frugal Plasma Disinfection ...... 112 C.6 Summary ...... 113

References ...... 115

iii

Acknowledgments

Here’s where I get to thank everyone who has made this Ph.D. a success! Every acknowledgments page should, of course, start with effusive thanks to one’s advisers. Rest assured that without the academic prowess, encouragement, and enthusiasm of Douglas Clark and David Graves, this dissertation would not have been possible. Many thanks to Dr. Yukinori Sakiyama, a former staff researcher in the Graves lab, who was a fantastic mentor over the first three years of my Ph.D. Almost all of my experiments were aided in some way by one of the undergrads I had the pleasure to mentor: Zhi Chen, Phillip Tu, Alex Lill, and Connor Galleher, you guys were awesome.

More accolades are due to my collaborators and co-authors, especially Professor Zdenko Machala (Comenius University in Bratislava, Slovakia), Dr. Hung-wen Chang (National Taiwan University), Toshisato Ono (Tokyo Tech), and Dr. Matt Traylor (UC Berkeley). Other Berkeley faculty who have been helpful include Professors Susan Amrose (Blum Center), Kevin Healy (Bioengineering), Michael Lieberman (EECS), and Kara Nelson (Environmental Engineering). Really special thanks to Professor Jane Dai (Deakin University) who graciously hosted my trip to Australia. And to all the Clark and Graves group members past and present, thank you for being there to eat lunch with and to bounce ideas off of.

Anyone familiar with my graduate career might have seen it take some unusual turns. Learning about development engineering isn’t something included in most engineering Ph.D.’s, but it’s been one of the most meaningful things I’ve done here. Though we received a little funding from the Blum Center early on, it wasn’t until a few years later that the “frugal” angle really took off. Attending the Clinton Global Initiative University conference gave some unexpected inspiration from the former President. During our subsequent exploits, we received a travel grant from the Development Impact Lab and USAID to study disinfection in Cape Town, South Africa. Our experience there was possibly the single biggest highlight of my entire time as a graduate student. One of the reasons I’ve been so excited about the frugal engineering effort is that it’s brought my lab research tantalizingly close to an application. Now Plasmachine remains in the capable hands of the Graves lab, who I’m confident will do a great job with it.

And there are a whole host of family, friends, and funding sources who need to be included here. Exactly how many people, directly or indirectly, made a huge impact is tough to guess.

A great place to start is with the Department of Energy’s Plasma Science Center, which provided a majority of my project’s financial support. Later funding came from the Department of Energy for a project with our collaborators at the University of Illinois. (Thanks to Dr. Sung- Jin Park and Dr. Gary Eden at UIUC for joining us on that project.) And the STEM fellowship from the Department of Homeland Security supported me for the first three years of my Ph.D. During my fellowship, I interned at Lawrence Livermore National Lab for a summer, where I learned so much about how to do biology research, thanks to Lisa Vergez and Dr. Paul Jackson. One unexpected benefit of that internship was meeting my girlfriend Stephanie, for whom I have run out of adjectives to describe my appreciation, but who has been there for me every step of

iv the way. Of course, my parents, other family, friends, roommates, and assorted colleagues have done their part too and have my gratitude. Nearly everyone I have worked with and interacted with over the past five years has been pleasant, kind, and helpful, and I have all of you to thank for a fantastic graduate school career.

Chapter 1: Introduction to Ambient-Condition Plasmas and Plasma Biotechnology

1.1 Abstract

Ambient-condition plasmas are introduced along with a summary of the emerging field of plasma biotechnology, which aims to use the plasma phase of matter as a novel bio-active agent. An overview of the current biological applications of plasma is presented. Major chemical species thought to be involved with plasma biological effects are discussed, with a particular emphasis on reactive species implicated in plasma disinfection.

1.2 Ambient-Condition Plasmas

Plasma is a reactive, gas-like phase of matter. It is similar to gas in that it has low viscosity, density, and molecular order, but plasmas typically contain much more reactive chemistry than gases. Sometimes referred to in popular culture as an “ionized gas,” a plasma contains not only charged species like ions and electrons, but also reactive neutrals (including radicals and energetically excited species) and photons, among other reactive species. Historically, plasmas have been divided into two major categories: thermal plasmas, including those found in astrophysical phenomena and those considered for nuclear fusion reactors, are characterized by equal gas and electron energies and therefore temperatures. Non-thermal plasmas are characterized by electrons with much greater energy than ions and neutral species, so these electrons are considered “hot” (i.e., 10,000 K or more) compared to the “cold” ions and neutrals (which often exist at or near room temperature). Because these different components of the same phase exist at different temperatures, the plasma is said to be “non-equilibrium.”

Industrially, the most important application of non-thermal plasmas is the low-pressure processing of materials, often for microelectronics and semiconductor fabrication. A textbook by Lieberman and Lichtenberg gives a detailed description of the discharge physics of low-pressure, non-thermal plasmas in the context of etching, deposition, implantation, and other surface modifications. [1] Compared to igniting a plasma discharge at atmospheric pressure, creating a plasma at low pressure (typically < 1 Torr) is relatively easy because atomic collisions happen so infrequently that reactive species have a relatively long lifetime.

However, a number of methods exist to create stable discharges at ambient conditions: at or near atmospheric pressure and room temperature. In 1857, Siemens developed an atmospheric-pressure electrical discharge for the production of ozone. [2,3] By the early 20th century, the so-called Siemens generator was a well-known method for generating ozone, and contemporary studies focused on optimizing such devices for ozone production. [4] Devices like the Siemens generator became known as “dielectric barrier discharges” because they forced current through glass (an electrical insulator) to create a discharge.

Other devices for igniting stable plasmas at ambient conditions have been developed more recently and include the gliding arc, corona, electrospray, plasma jet, and various alternative configurations of the DBD. [5–9] Figure 1.1 illustrates representative examples of

1 these types of discharges. Fridman et al. published a detailed review of the discharge physics of atmospheric-pressure plasma discharges and some examples of strategies used to create ambient- condition plasmas. [10] In another review, Tendero et al. describe sources and applications of atmospheric-pressure plasma with a particular focus on spectroscopic characterization. [11]

Figure 1.1. Various ambient-condition plasma discharges. Clockwise from top left: corona discharge, electrospray, tubular DBD, helium jet. All photographs courtesy of Steve Graves except tubular DBD (http://web.science.mq.edu.au/~rmildren/UVLamp2.jpg).

One method of operating a DBD is as a surface micro-discharge (SMD), sometimes called a “remote,” “surface,” or “indirect” DBD. SMDs are advantageous sources of reactive species for several reasons: treated materials are electrically isolated from high-voltage electrodes; the discharges are nonthermal, increasing adjacent gas temperature by only a few degrees; devices can be scaled simply by changing electrode size and input power; and the discharges can operate in ambient air without requiring a noble gas admixture. [12–14] Kogelschatz and Eliasson studied and modeled the discharge characteristics and chemistry of air DBDs, mostly focusing on the “silent discharge” to produce ozone. [13,15,16] The experiments

2 and results described in this dissertation primarily focus on plasma produced by dielectric barrier discharge.

Film deposition and polymerization is one commonly studied application of ambient- condition plasmas. Polymerization and deposition of polyethylene was among the first applications of ambient plasma discharges reported in the literature. [17] Other examples of ambient-plasma surface modification have included deposition of dielectric silicon dioxide films [18], altering polymer surface energy and hydrophobicity [19], creating surface-sensitive signal transducers [20], and modifying the wettability and shrinkage characteristics of textiles. [21] Recently, plasma modification of surfaces has expanded to include treatment of biomaterials, tissues, and living organisms.

1.3 Plasma Biotechnology

Non-thermal, atmospheric-pressure plasmas are of increasing interest for biological and medical applications in the emerging field of plasma biotechnology, also called plasma health care or plasma medicine. Stoffels [22,23] and Laroussi [24,25] were among the first to study the effects of plasmas on living organisms including bacterial and mammalian cells. Reactive species formed in plasma are often relevant to biology, and ambient-condition plasmas represent a novel method for delivering biologically-active species to living organisms. In a recent review, Graves summarizes the biological significance of plasma-derived reactive species. [26]

The use of plasma to inactivate bacteria and other microorganisms, sometimes called plasma disinfection, plasma sterilization, or plasma decontamination, is one of the most definitively established applications of plasma biotechnology. The next section summarizes the current understanding of using plasma to decontaminate surfaces and liquid volumes. In addition to inert surfaces and liquids, plasma has been proposed as a means of decontaminate wounds, facilitating faster healing. Recent studies have shown plasma to be an effective treatment for treating human skin and wounds, including in the cases of ulcers [27], lesions arising from Hailey-Hailey disease [28], normal skin flora in hair follicles [29], biofilms on skin [30], and wounds infected with Pseudomonas and Staphylococcus bacteria. [31] A computational simulation suggests that consistent application of plasma to an infected wound could decrease the time required for the wound to heal by days or weeks. [32] Because mammalian cells have mechanisms to prevent or repair biochemical damage that bacterial calls lack, it has been hypothesized that plasma treatment has the potential to be a selective skin antiseptic, removing pathogens while not harming human skin tissue. [33]

Cancer treatment by plasma is another area of interest in plasma medicine. Non-thermal plasmas have been shown to induce apoptosis in mammalian cancer cells [34–36] and to degrade adhesion proteins associated with cancer cells. [37] Some evidence suggests that plasma induces DNA damage in mammalian cells. [38] Even for cancer cells resistant to chemotherapeutics, plasma has been shown to restore sensitivity to those chemotherapeutic agents, suggesting that plasma could be a valuable adjunct to conventional cancer treatment. [39] Reactive oxygen species in particular are implicated in plasma treatment of cancer [40], owing to the Warburg effect, by which cancer cells are especially vulnerable to reactive oxygen attack due to their

3 reliance on aerobic metabolism. [41,42] Because of the differential metabolism in cancer versus healthy cells, it is believed that plasma treatment might be selective for toxicity to cancer cells over healthy cells, and histological observations of plasma-treated healthy mammalian cells has shown no long-term damage. [27]

Further areas of interest in plasma medicine include the use of plasma for blood coagulation [43]; pesticide, toxin, and dye destruction [44–46]; and tooth bleaching, among other dental applications. [47] The use of plasma to treat the surfaces of biomaterials is a large and established field closely related to plasma medicine. Example applications include rendering implant surfaces antibacterial [48], improving cell adhesion and migration properties [49,50], increasing biocompatibility [51], and immobilizing biomolecules. [52] A review by Chu gives an extensive discussion of the means and motivation for plasma treatment of biomaterial surfaces. [53]

Comprehensive reviews summarizing these and other potential applications of plasma medicine have recently been published by Fridman [54], Kong [14], and von Woedtke [55].

1.4 Plasma Disinfection

Disinfection of bacteria and other microorganisms has been one of the most widely studied applications within the emerging fields of plasma biotechnology and plasma medicine. Plasma created from air at or near atmospheric pressure and room temperature has been shown to be effective in disinfecting both aqueous volumes and solid surfaces. [56–60] In 1996, Laroussi showed that a helium DBD disinfected liquid medium with E. coli growing in it. [56] Two years later, Kelly-Wintenberg and coworkers reported using air DBD at atmospheric pressure to disinfect a polypropylene surface contaminated with E. coli. [61]

Since then, many different configurations of atmospheric pressure plasmas have been studied for disinfection of a variety of surfaces and liquid media, killing or inactivating a variety of microorganisms (including viruses, fungi, and Gram-positive and Gram-negative bacteria). For example, ambient-condition plasma disinfection has been shown to be effective against microorganisms normally difficult to inactivate, including MRSA, biofilms, viruses and bacterial spores. [62–65] Summarizing this work, Ehlbeck et al. recently published a review of ambient- condition plasma disinfection including a representative list of disinfection results under various experimental conditions. [59]

The application of ambient-gas plasma disinfection in healthcare is particularly promising, including in hospital and emergency-medicine settings. Nosocomial, or hospital- acquired, infections account for more than 100,000 deaths per year in the United States alone. [66] These infections affect 5-10% of patients and increase the average cost of stay by more than $12,000 per infected patient [67]; thus, there is a pressing need for safe and broadly effective strategies for hospital infection control. Atmospheric pressure plasmas have been suggested as a means of infection control for several situations relevant to healthcare and other settings, including hand-hygiene prophylaxis [12], and killing of drug-resistant bacterial strains. [68]

Air plasma disinfection technology also has the potential advantage of operating with only air and electricity. The plasma power supplies can be portable, and a preliminary discussion of portable, handheld plasma devices is given in Appendix B. In situations where supply chains are disrupted or inconvenient, air plasma disinfection might provide crucially important disinfection and antisepsis. On-demand, point-of-use disinfectants could be critical in developing world situations such as refugee camps or after natural disasters, for certain military applications, or for wilderness medicine applications.

It has been shown that treating an aqueous solution with plasma under certain conditions can create “plasma-activated water,” which can be applied to a contaminated surface or liquid volume and used as a disinfectant. [69] Further, it has been reported by several groups that this acidic solution is effective in killing bacteria in suspension. [7,70–72] If the pH rises above about 3–4, the antibacterial effectiveness is known to drop significantly. [73] Recent results show that plasma-activated water remains antimicrobial for one week or more following its initial exposure to plasma. [8,74] Treating aqueous solutions under slightly different processing conditions can lead to disinfection of the aqueous medium by ozone; in industrial waste-water treatment, such discharges are used to inactivate various classes of water-related pathogens, as well as to improve taste, decompose contaminants, and remove odor. [75]

Despite an increasing selection of plasma devices and operating conditions that have antimicrobial effects, several fundamental questions about plasma disinfection remain unanswered. Previous studies have shown the antimicrobial effect of air DBD on E. coli and other microorganisms on several different surfaces. [9,57,76] Some studies have made important comparisons of conditions affecting the disinfection capability, including discharge mode on the same substrate [9], type of microorganism on the same substrate [77], and different substrates with the same device. [57] However, relatively few studies have directly related the nature of the discharge and surface to the antimicrobial effect. Therefore, it remains mostly unknown how discharge parameters affect plasma antimicrobial activity and whether the type of surface, the nature of the discharge, the electrical characteristics of the plasma, and/or the treatment time most strongly influences plasma-based disinfection. Furthermore, studies using similar methods of generating plasma have not always produced similar results.

1.5 Ambient-Air Plasma Chemistry

Ambient-air plasma is understood to inactivate bacteria and other microorganisms by producing reactive oxygen and nitrogen species, and one important research objective in ambient-air plasma disinfection is identifying which reactive species are most responsible for the inactivation effect. [26] Chemical species formed in the air plasma region itself are thought to include atomic nitrogen (N) and oxygen (O), excited states including singlet-delta oxygen * + (O2(a¹Δg)) and vibrationally excited nitrogen (N2(v) ), ions including oxygen molecular ion (O2 ) + nitrogen molecular ion (N2 ), and high-energy photons in the UVB–UVA range. [15,16,46,78] Other neutral and ionic species are likely present in the plasma, including association and dissociation products of the species listed above. Optical emission spectroscopy (OES) is the most common technique for spectroscopically identifying chemical species present in the plasma region. [79]

The chemical mechanism of disinfection is somewhat understood for aqueous systems and less well characterized for non-aqueous systems. The list of RONS thought to be involved can differ greatly depending on the study, the plasma source, and the operating parameters. Of particular importance for disinfection is the overlap between plasma-generated reactive species and the chemistry produced by the innate immune system to clear pathogens, which depends heavily on RONS like nitric oxide (NO), nitrogen dioxide (NO2), hydroxyl radical (OH), and hydrogen peroxide (H2O2). [80,81] In at least a few studies, the ability of neutrophils to produce ozone or a related oxidant was considered. [82–85]

Fourier-transform infrared (FTIR) spectroscopy has been used in some studies to identify reactive species produced by dielectric barrier discharges; measurements are typically made in a post-discharge gas region rather than in the plasma itself, such that observed reactive species have diffused away from the discharge region. Commonly identified species include ozone (O3), nitrous oxide (N2O), nitric acid (HNO3), and nitrogen dioxide, with ozone associated with lower- power discharges and nitrogen dioxide with higher power. [16,86,87] Numerical models of air plasma chemistry have focused on calculating the time evolution of reactive species, and the number of reactions and species involved in such models highlights the complexity of ambient air plasma chemistry. Sakiyama’s model, for example, considers an SMD operating at relatively low power and includes 624 reactions and 53 chemical species, including ions, electrons, radicals, and other reactive neutrals. Under the conditions in this model, the most prominent reactive species in the post-discharge “afterglow” region include ozone, nitrous oxide, dinitrogen pentoxide (N2O5), and nitric acid. [88,89]

- It has been established that nitric acid/nitrate (HNO3/NO3 ), nitrous acid/nitrite - (HNO2/NO2 ), and hydrogen peroxide are among the reactive species present in plasma-treated aqueous systems. [46,70,74,86] Nitric acid is a very strong acid and is nearly fully ionized to - form NO3 in aqueous solutions that are not buffered. Nitrous acid has a pKa of about 3, so it may or may not be fully ionized into nitrite, depending on the pH, as established, for example, by the concentration of HNO3.

Recently, peroxynitrous acid/peroxynitrite (HONOO/ONOO-) has been identified as a potentially important species in air-plasma treated aqueous chemistry. A well-established method - to generate peroxynitrite is to mix aqueous solutions of acidified H2O2 with NO2 . [90] Both of these species are known to be created in air plasma, so it seems likely that ONOO- (or ONOOH) is also generated in water exposed to air plasma. [5,91] At room temperature and non-basic pH, aqueous peroxynitrite has a relatively short lifetime, on the order of seconds. However, at -80 C and strongly basic pH, it can be maintained for up to one year in storage. [90] Other, shorter- - lived aqueous-phase species are known or thought to include hydroxyl radical, superoxide (O2 ), nitric oxide, and nitrogen dioxide. [46] These highly reactive species probably do not persist under typical conditions. OH radicals, for example, are expected to react within microseconds or even shorter times, depending on what else is present to react with it. Detection of these species is more challenging but may be accomplished, for example, by radical-sensitive techniques like electron paramagnetic resonance (EPR). [92]

Some recent studies have attempted to “recreate” the antimicrobial effect of plasma- treated water by preparing chemical mixtures of these reactive oxygen and nitrogen species. Naïtali and coworkers found that an acidic solution of nitrate, nitrite, and hydrogen peroxide produced nearly the same antimicrobial effect as plasma-activated water. [93] However, in a similar study, Oehmigen et al. found that a solution containing the same components accounted for only a fraction of the plasma’s antimicrobial activity. [86] These results demonstrate the need for further study to determine the antimicrobial role for these reactive species.

Aqueous solutions of nitrite ions at pH below 4–5 are known to be antimicrobial even outside the context of plasma disinfection. So-called “acidified nitrite” has received a great deal of attention in the biomedical literature due to the generation of nitric oxide from the decomposition of nitrous acid. Nitrite ions are thought to play a significant biological role as an intermediate in the path from dietary nitrate to nitric oxide. Lundberg et al. review the multiple ways the coupled nitrate, nitrite, and nitric oxide pathways are involved in physiology, underlying promising therapies for conditions including heart attack, stroke, hypertension, and gastric ulceration. [94] Among a long list of applications, including acidified nitrite creams for use as topical NO-donating wound healing agents [95], concentrated acidified nitrite solutions were shown to be very effective in surface disinfection, even for C. difficile spores. [96] Kono et al. note the synergistic antimicrobial effects of nitrite, H2O2, and low pH. [97]

Ozone is another reactive species present in ambient-gas plasma discharges under certain conditions. It is well known that under conditions of relatively low discharge power sometimes called the “silent discharge” regime, ozone is relatively abundant. [15,16] Ozone, a powerful oxidant, is a potentially significant reactive species especially in the context of plasma treatment of aqueous solutions because of the widely established use of ozone in large-scale water treatment. Ozone presents a different and complementary means of treating aqueous systems because it does not acidify water but has a relatively short half-life on the order of minutes, depending on solution pH. [98]

Mass transfer from the gas to the aqueous phase is another important characteristic for determining aqueous-phase chemistry. NO, NO2, O3, and H2O2 will dissolve directly from the gas phase into adjacent water, although their aqueous solubilities differ markedly. [99] Therefore, the need to enhance mass transfer efficiency, such as through bubbling or shaking, is more pronounced when treating liquids with certain plasma-generated species. In particular, some nitrogen oxides like dinitrogen tetroxide (N2O4) exhibit much greater Henry’s coefficients than ozone. Coupled air plasma and chemical kinetic models and corresponding gas phase and water chemistry studies are beginning to appear in the literature. [6,88,100,101] The complex chemistry of air plasmas and the associated plasma-water chemistry are areas of active research and further investigation will be needed to fully understand these systems.

1.6 Comparison to Similar Techniques and Chemistries

In terms of reactive species chemistry and envisioned applications, plasma biotechnology is comparable to a number of other phenomena and chemical pathways. Photodynamic therapy (PDT) is a technique that can use, for example, a highly conjugated porphyrin derivative

7 photosentitizer, light at a specific wavelength, and molecular oxygen to achieve biological effects, most commonly inactivating microorganisms. Like ambient-air plasma, PDT generates reactive oxygen species (ROS) including superoxide and singlet-delta oxygen. [102] PDT has been proven effective in decontaminating localized infections in vivo [103,104], inactivating antibiotic-resistant bacterial species [105], and curing skin tuberculosis (lupus vulgaris). [106]

Another established antimicrobial method that uses ROS is solar disinfection, or SODIS. SODIS uses photons from sunlight at UV wavelengths to decontaminate water: UVC (280-100 nm) and UVB (315-280 nm) directly damage pathogen DNA, while UVA (400-315 nm) forms ROS including hydroxyl radical and superoxide. [107] One study showed that aerobic respiration involving oxidation enzymes is a critical biochemical pathway targeted during SODIS disinfection. [108] ATP production, as well as the transmembrane proton gradient, are also thought to be affected during SODIS. [109] Hydrogen peroxide has been used as an adjunct in SODIS to accelerate the antibacterial effect. [110]

Solar disinfection is one of several techniques classified as an advanced oxidation process (AOP), a method of disinfecting water or aqueous solution by production of ROS. Other examples of AOP mostly use UV light to form ROS in water. UVC is a well-known technique for decontaminating water and can be derived from lamps [111] or LEDs. [112] UVA-producing LEDs also exhibit antimicrobial activity against contaminated water, and ROS including hydroxyl radical and hydrogen peroxide are hypothesized as reactive intermediates. [113] Finally, photocatalytic oxidation using titanium dioxide (TiO2) and UVA light has been shown to disinfect surfaces [114,115], and titanium dioxide has been used as an adjunct to achieve a synergistic sterilization effect with an atmospheric-pressure plasma jet. [116]

1.7 Conclusion and Outline

The potential of ambient-air plasma to achieve novel biological effects has clearly been demonstrated in various studies over the last twenty years. Plasma follows a number of other technologies that use reactive oxygen and nitrogen chemistry in a biological context, and a major advantage of plasma is the ease and convenience of creating reactive species. While the importance of reactive oxygen and nitrogen has been stressed in earlier studies, both the knowledge of what reactive species are produced under particular plasma processing conditions, and an understanding of how those reactive species produce biological effects, are still lacking. A more complete characterization of both plasma chemistry and the relationship between plasma chemistry and its associated biological effects is necessary before plasma biotechnology can be implemented clinically. Therefore, the experiments in this dissertation are intended to study how plasma chemistry responds to plasma discharge parameters and how that chemistry determines biological effects. Disinfection is considered as a “model” biological effect, though characterizing plasma chemistry is essential to better understanding all of the various applications of plasma biotechnology.

The next chapter, Chapter 2, describes the dielectric barrier discharge device used to generate the plasma in this study. Chapter 2 also introduces the assays used to measure gaseous chemistry and aqueous chemistry, as well as protocols for culturing E. coli and quantifying

8 antibacterial effects. Chapter 3 focuses on disinfection of dry surfaces and determining which plasma processing parameters and surface characteristics affect disinfection efficacy. In Chapter 4, gas-phase reactive species are measured and quantified as functions of treatment time and discharge power density, and gaseous chemistry is related to surface disinfection. The two chapters that follow describe aqueous chemistry and aqueous inactivation. Chapter 5 describes aqueous chemistry varying with treatment time and power density, and it shows one set of conditions where ozone correlates strongly to antimicrobial effects in liquids. Chapter 6 demonstrates the use of UVA photons in conjunction with plasma treatment for rapid disinfection. Finally, Chapter 7 concludes the work and discusses open questions and future directions for the field.

Two appendices describe related studies: Appendix A shows post-treatment chemical and antimicrobial effects of plasma-treated solution, and Appendix B describes a class of “frugal” plasma devices intended for sanitization applications in resource-constrained settings. Lastly, Appendix C contains a report about disinfection in the developing world and how plasma technology might contribute to solving the problem.

Chapter 2: Methods and Materials

2.1 Abstract

In this chapter, experimental methods, equipment, and materials used throughout the dissertation are introduced. Various devices for producing atmospheric-pressure plasma are described and diagrammed. Assays for measuring gaseous and aqueous chemistry are described along with protocols for culturing bacteria and quantifying bacterial inactivation. General methods used in multiple chapters of this dissertation are described here, while specific methods pertaining to particular experiments are mentioned in the following chapters.

2.2 The Dielectric Barrier Discharge (DBD) Device

A dielectric barrier discharge is an example of a low-temperature plasma. Both low- pressure and atmospheric-pressure DBDs are possible, though only the atmospheric-pressure discharge is considered here. In a DBD, a high-voltage powered electrode and a ground electrode are separated by a dielectric layer and a gas gap. Charge accumulates on the surface of the dielectric layer until the breakdown potential of the surrounding gaseous medium is reached. Then, transient filaments form between the dielectric layer and the grounded electrode, which is where ionization occurs and the plasma itself is formed. In classical plasma processing, both the powered electrode and grounded electrode are fixed metal plates or cylinder surfaces. Kogelschatz summarizes the history, physics, and principal applications of the DBD in a comprehensive review. [13]

Two variations on the DBD were used in these experiments. For both configurations, the powered electrode was a cylindrical copper block (47 mm in diameter) covered by a thin quartz plate (1 mm in thickness). The first configuration is the “indirect” DBD, also called the surface micro-discharge (SMD). Similar to the device described in a series of reports by Morfill et al., the SMD is a configuration of a DBD where charged particles are confined to a plasma generation region near a grounded metal electrode attached to a dielectric layer. [12,63,87] Only reactive neutrals interact with treated surfaces, and diffusion of reactive species from the plasma region to the post-discharge “afterglow” is an important process. Figure 2.1 shows the SMD device used in these experiments. A stainless-steel woven wire mesh was attached to the quartz plate and used as a ground electrode. The wire diameter was 0.5 mm and the mesh density was 8 × 8 meshes per cm2. In some experiments, the SMD device was placed on top of a cylindrical acrylic enclosure measuring 46 mm in height and 56 mm in diameter. In others, the SMD was placed into a Petri dish, 15 mm in height and 100 mm in diameter.

Figure 2.1. Schematic of the air SMD device. The device is shown with a glass vial containing aqueous solution.

The second configuration of the DBD is the “direct” DBD or “floating electrode” DBD. In this configuration, filaments form between the dielectric and the treated surface, and the treated surface itself acts as a ground electrode. In addition to reactive neutrals, treatment by direct DBD involves charged particles and electric fields. The floating electrode configuration is similar to one used by Fridman et al. [9,35] Figure 2.2 shows schematics and inset photographs of the direct and indirect DBD. In direct mode, the treated surface was connected to ground by placing the surface on top of a grounded steel plate and placed at 1 mm from the quartz plate. A discharge was ignited between the quartz plate and the treated surface. Figure 2.3 shows a high- resolution photograph of the SMD device used in this study.

Figure 2.2. Schematics and photographs of the DBD device operating in the two different modes: (a) direct, or floating-electrode; and (b) indirect, or surface micro-discharge.

Figure 2.3. Photograph of the SMD device. The orange color (top-middle) comes from the powered copper electrode, and the horizontal and vertical lines are the grounded stainless steel mesh. The purple-blue glow is the plasma itself (the discharge region of ionization). The color is characteristic of air plasma and arises from photoemission from excited oxygen and nitrogen states. Photo courtesy of Steve Graves.

2.3 Plasma Generation and Characterization

A function generator (Protek, Model 9301) and high voltage amplifier (Trek, 10/40A) were used to generate a high-voltage alternating current measuring 2.6 to 10 kV at 1 to 10 kHz. The voltage waveform was monitored by an oscilloscope (Tektronix, TDS 2022B). The power consumption was measured by the Lissajous method using a 100 nF capacitor inserted between the grounded electrode and ground. [78] Power was measured every 60 s to monitor for fluctuations, and power usually fluctuated by less than 15% for a fixed frequency and voltage. Power density was calculated as total power divided by the electrode area; absolute power ranged from 0.30 W to 10.0 W.

2.4 Gaseous Chemistry Measurements

Fourier transform infrared (FTIR) spectroscopy was performed as an in-situ measurement of the gaseous chemical composition in the SMD device described above. The reactor was placed inside an infrared spectrometer (DIGILAB Excalibur Series, MOD. FTS 3000) so that the infrared beam passed through infrared-transparent windows (ZnSe polished disc, International Crystal Laboratories), located approximately 2 cm below the plasma discharge, in the walls of the reactor. Absorption FTIR spectra were recorded over the duration of plasma ignition, up to 20 min. The wavenumber resolution of the measurements was 2 cm-1, and between 20 and 200 scans were typically averaged to create each spectrum. The time resolution of the measurements was between 30 s and 3 min. The following vibrational bands were used as diagnostics for the -1 -1 -1 presence of reactive species: N2O at 2235 cm ; NO at 1900 cm ; NO2 at 1630 and 2916 cm ; -1 -1 HNO3 at 1325 and 1718 cm ; and O3 at 1055 cm .

The absolute concentrations of N2O, NO, NO2, HNO3, and O3 were estimated by fitting the measured FTIR spectra to spectra simulated using parameters in the HITRAN database (http://hitran.iao.ru/). To generate simulated standard spectra, the following parameters were used: a wavenumber range of 500 to 4000 cm-1, a wavenumber computational step of 0.1 cm-1, an optical path of 0.05 m (corresponding to the diameter of the acrylic reactor), an assumed Gaussian apparatus function, and an apparatus resolution of 2 cm-1. The species concentration was adjusted to find the best fit with the experimentally measured spectra. Unless otherwise noted, the relationship between absorbance and concentration remained linear over the range tested, though non-linearity in gas-phase FTIR spectroscopy at high gaseous concentrations is a known phenomenon that has been discussed in the literature. [117]

Gaseous ozone was measured with UV absorption at 254 nm. A UVC light source (UVP PenRay) was situated perpendicular to UV-transparent fused silica windows (Edmund Optics), and light that passed through the windows was collected with a fiber optic cable attached to a spectrometer (OceanOptics 2000). Ozone concentration was calculated by Beer’s law using an absorption cross section of 1.14 x 10-21 m2. [118]

2.5 Aqueous Chemistry Measurements

To measure aqueous-phase chemistry, 150 µl of either physiological saline solution (0.9% sodium chloride solution, Sigma) or phosphate-buffered saline solution (PBS; 10 mM sodium phosphate, 0.9% NaCl, pH 7.4; Sigma) was inserted into glass vials measuring 15 mm in diameter and 45 mm in height, or 7950 mm3 in volume. After insertion of the aqueous solution, the remaining air space in the vial measured 14 mm in diameter and approximately 42 mm in height, or approximately 6500 mm3 in volume. Solutions were exposed to plasma at power densities varying between 0.02 and 0.4 W cm-2 for exposure times ranging from 30 to 300 s. In general, when power density was varied, exposure time was held constant, and when exposure time was varied, power density was held constant. Exposure time was measured starting from the ignition of plasma. It is likely that the distributions of plasma generated species did not reach steady state.

Following plasma treatment, the vials were immediately capped and vortexed for 5 seconds to ensure thorough mixing. The aqueous-phase chemistry was then analyzed via specific assays for nitrate, nitrite, hydrogen peroxide, and ozone concentration. pH was measured using a probe (Hanna Instruments, HI 3221). To quantify aqueous nitrate and nitrite, UV absorbance scans were performed in the 200 to 400 nm wavelength range. Linear regressions were performed against nitrate and nitrite standards to quantify the concentration of nitrate and nitrite in plasma-treated samples. Concentration standards for nitrate and nitrite were generated with sodium nitrate and sodium nitrite and prepared for pH values between 2 and 7 to obtain extinction coefficients for each wavelength. The nitrate absorbance peak is near 306 nm, and the nitrite absorbance peak is near 360 nm. Hydrogen peroxide concentration was measured using a biochemistry analyzer (YSI, 2700 Select).

Dissolved ozone concentration was measured by the indigo method as described by Bader and Hoigné [119,120]. The indigo trisulfonate reagent (Sigma) was calibrated as suggested by Gordon and Bubnis [121] and the “apparent” molar absorptivity ε was 18,700 M-1 cm-1, corresponding to a sensitivity coefficient f of 0.389 L mg-1 cm-1. Indigo Reagent II was used to assay for aqueous ozone, and the ratio of sample to reagent was decreased as recommended for high anticipated concentrations of ozone. [120] To correlate the change in absorbance to a concentration of ozone, the “Spectrophotometric, gravimetric method” as described by Gordon and Bubnis was used. [121] There is some debate in the literature about the ability of indigo dye measurements to unambiguously detect ozone if other oxidants such as superoxide might be present. [82] However, in the experiments that follow, the fact that high concentrations of gas phase ozone was measured by UV and FTIR absorption below the active plasma zone strongly suggests that it is indeed dissolved O3 that is detected by this method in these experiments.

2.6 Antibacterial Measurements

Escherichia coli K12 were cultured in lysogeny broth (LB) growth medium to an optical density at 600 nm (OD600) of 1.0, corresponding to a bacterial concentration of approximately 3 × 108 colony-forming units per ml (cfu ml-1). The culture was incubated for 18 hours at 37º C and shaken at 200 rpm.

For surface disinfection experiments, 100-µl aliquots of the culture were withdrawn and pelleted by centrifugation at 5000 rpm for 10 minutes. The pellets were spread evenly across test surfaces using a plastic inoculating loop and allowed to dry completely in a sterile environment. All surfaces were disinfected by an application of 70% ethanol prior to seeding with E. coli. Once the surfaces were dry, they were exposed to plasma as described in the following chapters. Following exposure, treated surfaces were immediately exposed to surrounding air and immersed in 10 ml PBS to quench further reactions by residual plasma-generated species. The antimicrobial activity was determined using a modified version of Sattar’s quantitative assay for virucidal activity. [122] The surfaces were vortexed for 10 minutes to ensure complete elution of the bacterial inoculum into the PBS. Then, the eluent was diluted in PBS by successive tenfold serial dilutions.

For aqueous disinfection experiments, cultures were diluted by a factor of 100 and suspended in either physiological saline solution or PBS. 150 l of each suspension was transferred into a glass vial prior to treatment with plasma. Vials were placed inside the SMD device described above and in the following chapters and exposed to plasma. Following plasma treatment, the vials were immediately capped and vortexed for 5 seconds. Then, the suspension was withdrawn from the vial and serially diluted in PBS. During the dilution, the concentrations of aqueous-phase plasma-generated species were diluted by a factor of 105.

After plasma treatment and serial dilution, the bacterial dilutions were plated on LB agar and incubated overnight at 37º C, after which colonies were counted to determine the number of viable cells. The antibacterial effect of treated solutions was quantified by the log reduction, log(N0/N), where N0 is the number of viable cells prior to exposure, and N is the number of viable cells after exposure. N0 fluctuated by approximately a factor of 5; possible explanations for this fluctuation include slight day-to-day differences in the initial cell culture density as well as variation in the efficiency of recovering the bacteria from untreated samples. The maximum log reduction detectable using this approach is approximately 4–6, depending on experimental conditions. In some cases, the actual log reductions could have been higher than the reported values.

Chapter 3: Effect of Discharge Parameters and Surface Characteristics on Ambient-Gas Plasma Disinfection

Originally published in Plasma Processes and Polymers, DOI 10.1002/ppap.201200073, courtesy of Wiley-VCH Verlag GmbH

3.1 Abstract

Ambient-gas plasma, or plasma created from air at ambient conditions, has been studied as a means of disinfecting surfaces. Possible applications of ambient-gas plasma disinfection include infection control in hospital, emergency medicine, and community settings. However, the parameters that determine the effectiveness of ambient-gas plasma disinfection, and the optimal conditions for its use, are not well understood. This chapter reports results from a series of surface sterilization experiments that evaluate the practical suitability of ambient-gas plasma as a disinfectant, and the characteristics that are responsible for its antimicrobial effect against the model bacterium E. coli are determined. The results reported in this chapter suggest that neutral reactive species are more influential in surface sterilization than charged particles and electric fields under the conditions considered here. In addition, bacteria on non-biological surfaces (i. e., metal, rubber, and silicon) are considerably more susceptible to plasma disinfection than bacteria on pig skin. Interestingly, air plasma performed at least comparably to the conventional antiseptics ethanol and dilute chlorhexidine, indicating that ambient-gas plasma treatment could be a promising strategy for both clinical and community infection control.

3.2 Introduction

In this chapter, results are presented with the goal of better understanding the disinfection capability of air DBDs by more closely examining which parameters determine the antimicrobial effectiveness against bacteria in dry films on various surfaces. Surface disinfection results are reported using ambient-air plasma, or air DBD generated at room temperature and atmospheric pressure. Two modes of operation, direct and indirect mode; a range of applied AC waveform voltages and frequencies; several different contaminated substrates; and a range of treatment times are considered. In addition, the antiseptic effectiveness of plasma is compared to that of conventional chemical disinfectants. Antimicrobial effectiveness for each condition is calculated using E. coli as a model organism.

3.3 Experimental Section

Section 2.2 shows the experimental setup for direct and indirect mode. For both configurations, the treated surface was set at 1 mm from the electrode. In indirect mode, no visible discharge was observed between the mesh and the treated surface, while in direct mode, a discharge was ignited between the quartz plate and the treated surface. For both direct and indirect mode, the discharge area was limited to approximately 2.5 cm2 by using a Teflon cover, and for indirect mode, the mesh was cut to an area of approximately 2.5 cm2, so that the

16 discharge size in both modes was same as the area of the treated surface. Although it is possible that the discharge on the quartz disc was slightly nonhomogeneous, the entire discharge area is considered in all cases to standardize the measurements. Treated surfaces were placed into a Petri dish (100 mm in diameter and 15 mm in height) during the experiments to avoid possible disturbance by external air flow.

Three test surfaces were used to simulate surfaces relevant to a healthcare setting: stainless steel, to simulate surgical instruments; silicone rubber, to simulate devices such as catheters and IV lines; and pig skin, to simulate human skin. All surfaces were disinfected by an application of 70% ethanol prior to seeding with E. coli. Surfaces were intentionally contaminated with bacteria as described in Section 2.6. Pig skin was purchased with coarse hair clipped and the fat layer removed by the supplier (Pel-Freez Biologicals). Excess hair was shaved, and samples were washed thoroughly with Nanopure water prior to use. The samples were kept frozen at -20° C until use and allowed to thaw completely in a sterile environment prior to seeding with E. coli. Each sample of pig skin was tested for the presence of residual antibiotics as suggested by Bush and coworkers [123]; no residual antibiotic activity was found. The typical thickness of the pig skin was 4 mm ± 1 mm. For stainless steel, silicone rubber, and pig skin surfaces, the test surface was a disc measuring approximately 2.5 cm2 in area as mentioned above. Some additional experiments were performed on silicon surfaces; for these experiments, the test surface was a square measuring approximately 2.5 cm2 in area cut from a silicon wafer.

Once the surfaces were dry, they were exposed to either plasma or another disinfectant. For the plasma disinfection experiments, the surfaces were exposed to ambient-gas plasma generated by the air DBD device described above. The following variables were altered independently: exposure time (t), frequency (f), and voltage amplitude (V) of the applied AC waveform; mode of discharge (direct or indirect); and type of surface. Plasma was generated and characterized as described in Section 2.3. To compare plasma disinfection to other disinfectants, some surfaces were treated with 100 µl 70% ethanol (v/v in Nanopure water) or 100 µl chlorhexidine digluconate (2%, 1%, 0.5%, and 0.2% w/v in Nanopure water; Sigma).

Following exposure, treated surfaces were diluted, plated on LB agar, cultured overnight, and observed for bacterial colony growth, as described in Section 2.6. The maximum log reduction was approximately 5–5.5.

3.4 Results

The effects of the discharge mode, the discharge parameters, and the type of surface on the antimicrobial activity were evaluated by performing surface sterilization experiments and independently varying each variable. In Figures 3.1–3.6, error bars represent the standard deviation about the mean, and each data point represents the average of between four and eight samples. First, the antimicrobial effects of direct and indirect mode were compared. The applied frequency and voltage were adjusted in both modes to achieve an average power density in each discharge of approximately 0.75 W cm-2. For direct mode, f = 10 kHz and V = 6 kV; for indirect mode, f = 10 kHz and V = 4.9 kV. Figure 3.1 shows the antimicrobial effect of direct-mode and

17 indirect-mode plasma as a function of exposure time. Data are shown for a stainless steel surface; similar trends were observed for all surfaces.

Figure 3.1. Time-dependent bacterial inactivation in direct mode (black bars) and indirect mode (white bars). Discharge power was fixed at 0.75 W cm-2.

Figure 3.2 shows the effect of discharge frequency and voltage on antimicrobial activity. Data are shown for indirect mode on a stainless steel surface; similar trends were observed for all surfaces tested. Two conditions, f = 10 kHz and V = 4.9 kV, and f = 4 kHz and V = 6.5 kV, were selected. The average power density was 0.75 W cm-2 in both cases.

Figure 3.2. Time-dependent bacterial inactivation for indirect mode at low-frequency and high- voltage (black bars: 4 kHz and 6.5 kV) compared to high-frequency and low-voltage (white bars: 10 kHz and 4.9 kV). Discharge power density was fixed at 0.75 W cm-2.

The effect of power density on antimicrobial activity is shown in Figure 3.3. Data are shown for indirect mode on a stainless steel surface; similar trends were observed for all surfaces tested. Applied frequency was held constant at f = 10 kHz; voltage was varied to change power density. An applied voltage of V = 4.4 kV resulted in power density 0.54 W cm-2; V = 4.9 kV resulted in power density 0.75 W cm-2, and V = 5.4 kV resulted in power density 0.99 W cm-2.

Figure 3.3. Time-dependent bacterial inactivation in indirect mode with power density 0.54 W cm-2 (white bars), 0.75 W cm-2 (gray bars), and 0.99 W cm-2 (black bars).

Figure 3.4 shows the average survival curve for both modes on stainless steel surfaces. The survival curve of E. coli on stainless steel surfaces was determined to be a “single-slope” survival curve. The “D-value,” or the time required to achieve 1 log reduction, was calculated to be about 60 s when disinfecting a stainless steel surface.

Figure 3.4. Survival curve on a stainless steel surface, averaging all direct-mode and indirect- mode data. Under the experimental conditions tested here, E. coli exhibited single-phase survival kinetics with a “D-value” of approximately 60 s.

In addition to the parameters mentioned above, the dependence of plasma disinfection on the type of surface was investigated. Figure 3.5 compares disinfection results for four surfaces treated in indirect mode for 60 s and 300 s. The power consumption was approximately 0.75 W cm-2.

Figure 3.5. Bacterial inactivation on several test surfaces. Surfaces were exposed to indirect- mode plasma for 60 s (black bars) and 300 s (white bars). Discharge power density was fixed at 0.75 W cm-2.

Finally, the antimicrobial effect of plasma on pig skin was compared to the antimicrobial effect of 70% ethanol and various concentrations of chlorhexidine digluconate, both commonly used and studied skin antiseptics. [124,125] Figure 3.6 demonstrates the antiseptic ability of indirect-mode plasma, ethanol, and chlorhexidine for 1- and 5-minute treatments.

Figure 3.6. Comparison of plasma disinfection to other disinfectants. Surfaces were exposed for 60 s (black bars) and 300 s (white bars). Plasma: indirect-mode plasma; EtOH: 70% ethanol in water; CHX: chlorhexidine digluconate in water. 2% chlorhexidine showed complete disinfection within the detection limit.

3.5 Discussion

Antimicrobial activity increased with increasing exposure time, but no significant difference was found between the antimicrobial activity of direct mode and indirect mode at any exposure time. Increased temperature was not responsible for the inactivation in either mode; the temperature of the treated surface increased by about 5° C in direct mode and less than 1° C in indirect mode. Neutral species are present in both modes, but charged particles interact with the surface in direct mode only. This difference suggests that the contribution of reactive neutral species is more important than that of charged particles (i.e., electrons and ions) under the conditions considered here. The results can be explained by the short duration time of filamentary discharges reaching the treated surface. Under typical conditions, the electric field on the treated surface (E) is estimated to be around 100 kV cm-1 and the lifetime of each filament (τc) is estimated to be 10 ns. [126] At 10 kHz, the ratio of fluence of reactive neutral species (Fn) to charged particles (Fc) per cycle is given as

(3-1) where nn and nc are density of neutral and charged particles, respectively. T is the gas temperature (300 K), m is molecular weight, k is the Boltzmann constant, and µ is the mobility of electrons (~0.03 m2/Vs) and ions (~0.001 m2/Vs). τn is the period of the applied voltage waveform (= 100 µs). A numerical simulation [88] of air DBD indicates that the peak density of charged particles is lower than the peak density of reactive neutral species, suggesting that in direct mode, the time-averaged dose of neutral species on treated surfaces is much higher than that of charged particles. Therefore, despite the additional presence of charged particles in direct mode, no significant difference was observed in the inactivation efficacy between direct and indirect mode.

However, indirect mode was found to be more reproducible in terms of both bacterial killing and power consumption than direct mode. The difference in reproducibility was attributed to the electrical impedance of the system. Because the treated surface and the discharge volume are part of the electrical circuit for direct mode, a slight difference of the gap and electrical properties of the treated surface can alter the impedance of the system. Therefore, in direct mode, various discharge properties can be significantly modulated, including power consumption, duration, and intensity of each filament, although the applied peak voltage is fixed. However, for indirect mode, the electrical circuit is closed between the powered electrode, dielectric material, and ground electrode. The discharge properties are independent of the gap distance and electrical properties of the treated surface. The stability and reproducibility of the indirect discharge would be a significant advantage for practical applications. Moreover, when pig skin was treated with direct-mode plasma for longer than 60 s, significant burning of the skin was observed (data not shown). The observed burning can be explained by filaments tending to congregate at the part of the skin closest to the electrode due to the skin’s topical nonuniformity. In addition, it is possible that such a nonhomogeneous discharge decreased the antimicrobial effect under this condition due to uneven treatment of the surface.

Previous results are inconsistent regarding the role of the discharge mode (direct or indirect) in determining antimicrobial effectiveness. Fridman et al. report “complete sterilization” of typical resident skin flora on an agar surface after 15 seconds of treatment with a direct-mode air DBD. They note a less-pronounced “disinfection” of the surface after 2 minutes of treatment operating the same device in indirect mode. [9] However, Machala et al. report no significant difference between direct-mode and indirect-mode treatment when sterilizing Salmonella typhimurium on an agar surface using a plasma needle. [127] The results described here agree with those of Machala and coworkers, supporting the hypothesis that reactive neutral species are the most important contributors to the antibacterial effect of the plasma under these conditions. However, this is not a general result: ion and neutral chemistry likely play different roles depending on the device configuration and applied waveform. For example, Vandamme et al. describe a configuration where liquid is plasma-treated in direct mode then applied to cells; reactive neutrals, but not electrical fields or currents, are found to be important under their

24 conditions [40]. Although their study considered the effects of plasma on mammalian cancer cells, not bacteria, it is possible that a similar configuration could be of interest to the disinfection applications of ambient-gas plasma.

Plasma operating at the high-frequency, low-voltage condition had the same antimicrobial activity as plasma operating at the low-frequency, high-voltage condition when the discharge power density was the same. In addition, increasing or decreasing the power density by 30% did not alter the antimicrobial effect of the plasma. In all of these cases, antimicrobial activity increased with increasing exposure time, but antimicrobial activity was not found to vary significantly with applied frequency, voltage, or power consumption. Therefore, the data shown in Figures 3.2 and 3.3 suggest that density of reactive neutral species responsible for inactivating E. coli is not strongly modulated when frequency, voltage, and power is changed under conditions considered here. However, this consequence is not necessarily general because the range of parameters tested was relatively narrow.

Finally, the single-phase D-value of 60 s demonstrated in Figure 3.4 suggests first-order inactivation kinetics between reactive neutral species and bacteria on a steel surface. Log reduction was found to be a linear function of time with a single slope; similar survival curves and D-values are reported, for example, in a review paper by Laroussi. [58] However, the results described here do not indicate exactly which species react with bacteria and what the relevant reactions are. To this end, more studies of gas-phase chemistry are necessary, particularly relating to the kinetics of putative antimicrobial species. Chapter 4 describes in more detail the gaseous chemistry from air treated by ambient-condition SMD.

As demonstrated in Figure 3.5, at 60 s, the antimicrobial effect was approximately 1 log reduction on all surfaces. However, for a 300 s exposure, the antimicrobial efficacy was significantly less on pig skin than on various non-biological surfaces. This difference may develop after longer exposure times because of the skin’s irregular topography; once some base level of disinfection has occurred at a short exposure time, it is difficult for reactive species to contact and inactivate additional E. coli. These “multi-phase” inactivation kinetics suggest that, at least for long exposure times, reactive species flux is not the limiting factor in pig skin antisepsis. Skin is known to be a relatively difficult substrate to disinfect. Previously proposed reasons for the difficulty in disinfecting skin include roughness, porosity, and surface heterogeneity (especially the presence of hair follicles), and the presence of multiple stratum corneum layers. [128]

To test the hypothesis that surface roughness affects antimicrobial activity, atomic-force microscopy (AFM) measurements were performed on some of the surfaces and compared the measured surface roughness to known values of skin surface roughness. The root-mean-square (RMS) roughness of silicon was 0.29 nm and the RMS roughness of silicone rubber was 16.53 nm. Even though rubber is rougher than silicon by a factor of more than 50, the antimicrobial effect on rubber was nearly identical to the antimicrobial effect on silicon, suggesting that the sub-micrometer scale surface roughness between silicon and rubber is not an important factor in disinfection efficacy.

In contrast, the RMS roughness of human skin has been estimated to fall in the range of 60 to 150 µm, although it is known to vary according to the age and the location on the body. [129] In at least one study, the surface roughness of pig skin was estimated to be close to the estimated range for human skin. [130] Hence, it is possible that surface roughness larger than ~10 µm has a significant impact on the inactivation of E. coli. For instance, the rough surface of skin could act as a shield from the plasma-generated reactive species if the size of E. coli cells (~1 µm) is smaller than the roughness. To better understand the antiseptic capability of plasma, further experiments should be performed to determine the effects of porosity and surface nonuniformity on plasma’s ability to disinfect the surface.

Previous studies considering the antimicrobial effectiveness of air DBD plasma have reached different conclusions regarding the time needed to achieve a given log reduction on a given surface. Kelly-Wintenberg et al. report at least a 6-log reduction of both E. coli and Staphylococcus aureus after 30 seconds of direct DBD treatment disinfecting a polypropylene surface. [61] In contrast, Leipold et al. showed that a 340-second treatment was required to achieve a 5-log reduction of Listeria innocua on a stainless steel surface. [131] This study found between a 4- and 5-log reduction of E. coli on stainless steel after a direct DBD treatment time of 300 seconds, in close agreement with the results of Leipold and coworkers.

A similar range of results exists for indirect DBD treatment. Ben Gadri et al. report a 5- log reduction of E. coli on polypropylene after 24 seconds of treatment. [57] Schwabedissen et al. achieved between a 4- and 5-log reduction of B. subtilis on a paper surface after 10 minutes of exposure with their “PlasmaLabel” device, though the authors note their device is not optimized in terms of electrode geometry. [132] This study found an intermediate result: 5 minutes of exposure to indirect DBD plasma results in a 4- to 5-log reduction of E. coli on stainless steel, silicone rubber, and silicon surfaces. It is important to note that the various DBD devices all have different geometries; differences in reactor construction (such as allowing treated air to recirculate, as in the device of ben Gadri and coworkers) might account for substantial increases in effectiveness.

Different sources report widely varying effectiveness of chlorhexidine as an antiseptic, but the results shown are in general agreement with previously published results. [133] No antiseptic showed a significantly increased effect after 5 minutes of treatment compared with 1 minute of treatment, suggesting that the effectiveness of skin antisepsis is not a strong function of treatment time for these conditions. Although higher concentrations of chlorhexidine were more effective than plasma at treating skin, indirect-mode plasma under these conditions had a similar antiseptic effect to 70% ethanol and lower (≤0.5%) concentrations of chlorhexidine. Plasma antisepsis may be an effective alternative to chlorhexidine in situations where chlorhexidine is an inappropriate antiseptic (such as in the treatment of open wounds) or unavailable (such as in the developing world). For example, plasma could be a good alternative antiseptic in the rare case of chlorhexidine allergy [134] or to avoid the problem of tooth staining when using chlorhexidine as a disinfectant mouthwash. [135]

3.6 Conclusion

Ambient-air plasma, or plasma created from air at room temperature and atmospheric pressure, can disinfect surfaces contaminated with E. coli. The nature of the surface plays an important role in the effectiveness of plasma as a disinfectant: 5 minutes of exposure to indirect- mode plasma achieved a 4-to-6-log reduction in bacterial load on stainless steel, silicone rubber, and silicon surfaces but less than a 2-log reduction in load on pig skin. In contrast, small fluctuations of the plasma parameters—the frequency and voltage of the applied waveform and the power density in the plasma—appear to have little to no effect in determining the antimicrobial effectiveness of the plasma, at least within the range where the device used here created a sustained discharge. Furthermore, direct-mode and indirect-mode discharges showed similar disinfection capability for the same power density, treatment time, and surface type. In both modes, neutral species diffuse out of the plasma, through the gas phase, and onto the treated surface; direct mode adds the additional effect of charged particles interacting with the treated surface. Therefore, these results suggest that electrical effects and charged species play a less important role than neutral species in air DBD disinfection and that reactive neutral diffusion away from the discharge region is a dominant mechanism of inactivation under the conditions described here. More experiments are needed to determine which plasma-generated species are most responsible for plasma’s antimicrobial activity, and correlation between plasma chemistry and biological effect remains an important research objective for the field of plasma biotechnology. The next three chapters aim to establish connections between the presence of certain reactive species and antibacterial effects.

In addition, although indirect-mode and direct-mode air DBD are equally effective disinfectants, indirect mode has several practical advantages over direct mode. The power consumption of plasma in indirect mode is more stable and reproducible. Moreover, significant burning of pig skin occurred when it was treated with direct-mode plasma for longer than 60 s. Therefore, indirect-mode discharges represent a more promising method of ambient-gas plasma disinfection for practical healthcare settings, especially when sensitive or nonuniform surfaces are involved.

Chapter 4: Quantification of Air Plasma Chemistry for Surface Disinfection

4.1 Abstract

Atmospheric-pressure air plasmas, created by a variety of discharges, are promising sources of reactive species for the emerging field of plasma biotechnology because of their convenience and ability to operate at ambient conditions. One biological application of ambient air plasma is microbial disinfection, and the ability of air plasmas to decontaminate both solid surfaces and liquid volumes has been thoroughly established in the literature. However, the mechanism of disinfection and which reactive species most strongly correlate with antimicrobial effects are still not well understood. Gas-phase measurements of plasma chemistry are quantified via infrared spectroscopy in confined volumes, focusing on air plasma generated via surface micro-discharge (SMD). The measurements described in this chapter extend earlier findings that gaseous chemistry is highly sensitive to operating conditions. The gaseous concentrations of ozone (O3) and nitrogen oxides (NO and NO2, or NOx) are quantified throughout the established “regimes” for SMD air plasma chemistry: the low-power, ozone-dominated mode; the high- power, nitrogen oxides-dominated mode; and the intermediate, unstable transition region. The results presented here are in good agreement with previously published experimental studies of aqueous chemistry and parameterized models of gaseous chemistry. The major result of this chapter is the correlation of bacterial inactivation on dry surfaces with gaseous chemistry across these time and power regimes. Bacterial decontamination is most effective in “NOx mode” and less effective in “ozone mode,” with the weakest antibacterial effects in the transition region. The results described here underscore the dynamic nature of air plasma chemistry and the importance of careful chemical characterization of plasma devices intended for biological applications.

4.2 Introduction

The chemical mechanism for disinfection of dry surfaces by atmospheric-pressure air plasma is remains relatively poorly understood. It is accepted that reactive oxygen and nitrogen species (RONS) are responsible for surface disinfection, and it has been shown that ozone generation by air SMD in confined volumes is particularly dynamic. Under conditions of moderate discharge power density, around 0.1–0.3 W cm-2, ozone concentration in a confined air volume treated by SMD first rises then falls, implying a transition from an ozone-rich regime to a nitrogen oxides-rich regime at constant power. [87] A simple parameterized model developed in conjunction with the same study supports the finding of a mode transition, suggesting that vibrationally excited nitrogen and nitric oxide are important species in the ozone-quenching, mode transition process. [87]

Although the characterization of ozone in the mode transition has improved the understanding of ambient-air plasma chemistry, neither the gas-phase production of NOx by ambient air surface micro-discharge nor the relationship between reactive species concentration and antibacterial activity has been thoroughly quantified. In this chapter, the concentrations of nitrous oxide, nitric oxide, nitrogen dioxide, and ozone are shown as they evolve over time and vary with discharge power density. In addition, the variation of surface antibacterial activity

28 according to the gaseous chemistry is shown. Finally, results are compared to those of previous studies involving SMD air plasmas. The “ozone mode” to “nitrogen oxides mode” transition is both a critical characteristic of gaseous chemistry and a determinant of disinfection efficacy; this mode transition is observed in a variety of SMD devices but may occur at different power densities or treatment times depending on device configuration.

4.3 Experimental Section

Figure 4.1 shows the experimental configuration. The device shown here is a modification to the device shown in Figure 2.2, with the addition of optically transparent windows to measure gaseous chemistry. The device was operated in the surface micro-discharge configuration, or “indirect mode,” as described in Section 2.2. Plasma was generated and characterized as described in Section 2.3.

Figure 4.1. Schematic of the SMD device, showing optically transparent windows.

Fourier-transform infrared spectroscopy (FTIR) and UV absorption measurements were performed in the region just below the plasma zone to qualitatively identify some of the various plasma-generated species in the reactor, as described in Section 2.4. Surface disinfection was measured as described in Section 2.6. The maximum log reduction was approximately 4.

4.4 Results

Gaseous reactive species concentrations were measured via FTIR as functions of time for eight different power densities. Figure 4.2 shows the temporal evolution of a representative fixed-power FTIR spectrum over 20 min of plasma treatment. Prominent peaks correspond to nitrous oxide (N2O), nitric oxide (NO), nitrogen dioxide (NO2), and ozone (O3). By repeating similar experiments at differing power densities, the two-dimensional variations in the concentrations of N2O, NO, NO2, and O3 were determined as functions of time and power density. Nitric acid (HNO3) was observed as well, typically in the range of 0–300 ppm (data not shown), but the concentration of HNO3 fluctuated with time and power density. It is possible that the gaseous concentration of nitric acid is particularly sensitive to ambient water vapor concentration (i.e., humidity), and no attempt was made to normalize the air conditions or otherwise control for variations in humidity. No correlation between nitric acid concentration fluctuation and bacterial deactivation rates was detected.

Figure 4.2. Representative FTIR spectra showing evolution of reactive species over time. Spectra were acquired at 2 min (dotted line), 8 min (dashed line) and 16 min (solid line) after ignition of plasma. Power density was fixed at 0.30 W cm-2.

Figure 4.3 shows contour plots illustrating the concentrations of N2O, NO, NO2, and O3 varying with both time and power density. In order to generate contour plots, concentrations were interpolated between measured time and power density points. The concentration of nitrous oxide increases monotonically with both time and power density, indicating that at least some nitrous oxide is likely present when running an SMD device under any operating condition. Both nitric oxide and nitrogen dioxide increase with increasing time and power, but only above a -2 -2 certain power density threshold, about 0.10 W cm for NO and 0.20 W cm for NO2. In contrast, ozone is observed only below a power density threshold of about 0.40 W cm-2. Between 0.05 W cm-2 and 0.40 W cm-2, ozone concentration first increases then decreases with time, reaching a local temporal maximum at some point during the first 10 minutes of operation. Finally, below 0.05 W cm-2, ozone concentration increases with time.

Figure 4.3. Contour plots of reactive species concentrations as functions of time and plasma power density, measured by FTIR absorption. First plot: nitrous oxide; second plot: nitric oxide; third plot: nitrogen dioxide; fourth plot: ozone. Concentrations are displayed in parts per million (ppm) by volume. Note that the scale for the ozone contour plot is different from the scales in the other three plots.

The discharge antibacterial properties were measured, as shown in Figure 4.4. Antibacterial effect, as measured by log reductions, increases over time for any given discharge power density. However, the bacterial inactivation after 20 min is lowest at intermediate power density, intermediate at low power density, and highest at high power density.

Figure 4.4. Bacterial inactivation in log reductions as a function of time and discharge power density.

In addition to the measurements with FTIR, the gaseous concentration of ozone was measured via UV absorption at 254 nm. Measurements of ozone with UV absorption were conducted in the ozone-rich operating conditions below about 0.20 W cm-2. Figure 4.5 shows the time and power variation of the ozone concentration as measured by UV absorption.

Figure 4.5. Ozone concentration as a function of time and discharge power density, measured by UV absorption. The data shown in this graph are analogous to those displayed in the fourth plot of Figure 4.3. Concentration is displayed in parts per million (ppm) by volume.

The absolute gas phase ozone concentration estimated by FTIR, at the highest ozone concentration measured here, is about a factor of 2.2 less than the estimate using UV absorption, though the two methods give similar qualitative profiles of ozone concentration over time and power density. Other studies have examined disagreements between infrared and UV measurements of gaseous ozone and found that exact agreement is rare, though discrepancies are usually on the order of 10% or less. [118,136] Separate measurements of liquid phase ozone concentration using indigo dye after thoroughly mixing with water and using published values of the Henry's law coefficient are in good agreement with the UV absorption measurements, so it was determined that the gas phase UV absorption measurement is more accurate. Figure 4.6 shows the gas–liquid equilibrium data for ozone dissolved in water using both FTIR- and UV- derived values for gaseous concentration. Fitting a regression line to each set of data gives an -1 -1 estimate of the Henry’s coefficient for each: kH = 0.028 mol kg bar for the FTIR data and kH = 0.014 mol kg-1 bar-1 for the UV data. The most commonly reported Henry’s coefficient for the -1 -1 ozone–water system is kH = 0.013 mol kg bar . [137] Although the cause of the apparent inaccuracies in absolute ozone concentration with the FTIR instrument used here and the HITRAN database remains uncertain, the FTIR-derived values have been corrected to reflect the self-consistent UV absorption/indigo dye measurements.

Figure 4.6. Gas–liquid equilibrium data for ozone dissolved in water, using gaseous ozone concentrations measured from (a) FTIR and (b) UV absorption.

4.5 Discussion

These results show two distinct “regimes” of plasma-generated chemical species when operating the SMD in ambient air, which is referred to in previous studies and in Chapter 5 as “ozone mode” and “nitrogen oxides mode.” In particular, in Chapter 5, operation at power density < 0.20 W cm-2 is denoted as ozone-rich “low-power” mode, and > 0.25 W cm-2 as nitrogen-oxides rich “high-power” mode. [138] The presence of ozone at lower power densities and nitrogen oxides at higher power densities is in good agreement with earlier results and is consistent with the observations of Kogelschatz in a flowing DBD system. [15,16,138,139] Shimizu et al. also observed concentration maxima when studying ozone production via SMD as a function of time and power density, but these authors used an SMD device with different dimensions and other experimental details, so agreement with the results reported here is not expected to be quantitative. [87] However, the present results are in good qualitative agreement with both Shimizu’s experimental results and a parameterized model included in his study, which predicted mutual exclusivity between ozone and NOx in an enclosed system treated by air SMD. [87]

With better quantification of the gaseous chemistry, and by considering the temporal evolution of chemical species over a greater range of plasma treatment times, a comprehensive definition of the operating regimes is provided here. Operating at low power density and for short treatment times tends to produce gaseous chemistry dominated by ozone, as shown in Figures 4.3 and 4.5. At least some ozone tends to be present immediately after plasma ignition at any power density, though it may be quenched within the first few seconds of operation at the highest accessible power densities (> 0.40 W cm-2). For power densities < 0.40 W cm-2, ozone exists as a persistent species in the post-discharge chemistry for some measurable length of time and increases in concentration for at least 30 s before decaying or being quenched. At the lowest power densities (< 0.05 W cm-2), ozone persists for at least 20 min, and no decay or quenching behavior is observed. When ozone concentration is increasing in the post-discharge region, it is likely that ozone is being created in the plasma itself. “Ozone mode” is therefore defined as the set of conditions where ozone concentration increases over time; that is, the set of conditions where ozone is actively being produced in the plasma.

In contrast with ozone, and again in agreement with previous results, nitrogen oxides dominate the gaseous chemistry at high power density and relatively long treatment time. NOx (NO and NO2) are not produced during the first 30 s of plasma operation under any of the conditions tested here. However, at sufficiently long treatment times under sufficiently high power densities—that is, sufficiently high energy input into the system—the NOx concentration increases over time. Once NOx appear in the system, as long as the plasma remains ignited, no decay or quenching of NOx is observed, suggesting that these species represent a stable end-point for gaseous chemistry under sufficiently high power density (> 0.20 W cm-2).The operation time needed to reach a stable nitrogen oxides mode increases with decreasing power density, requiring only 30 s above 0.60 W cm-2 but 1200 s at 0.20 W cm-2. As with ozone mode, the “nitrogen oxides mode” is defined as the set of conditions in which NOx concentrations are increasing over time.

Finally, at intermediate treatment times and discharge power densities, the gas-phase chemistry exists in a “transition region.” The transition region corresponds to transition from the ozone-dominated mode to the nitrogen oxides-dominated mode. It consists of the conditions where ozone concentration is decreasing over time but non-zero, and the NOx concentration is zero. Figure 4.7 summarizes the ozone mode, nitrogen oxides mode, and transition region.

Figure 4.7. Summary of the modes of operation when treating air by SMD in a confined volume. In “ozone mode” (blue), ozone concentration increases with treatment time; in “nitrogen oxides mode” (red), NOx concentration increases with treatment time; and in the “transition region” (green), ozone concentration decreases with treatment time. The power density and time axis scales are identical to those used in Figure 4.3.

The transition from ozone mode to nitrogen oxides mode does not indicate an immediate replacement of ozone by nitrogen oxides in the post-discharge region of the reactor. Instead, while operating in the transition region, ozone concentration gradually decreases over time. In all conditions where NO and NO2 were produced, the ozone concentration reached zero before NOx concentration began to increase, suggesting that the transition region may correspond to quenching of ozone by nitrogen oxides. Kogelschatz and Eliasson summarized the reaction processes associated with ozone quenching by nitrogen oxides, noting that “[s]ince these reactions involving NO and NO2 have large rate constants, relatively low nitrogen oxide

39 concentrations can seriously interfere with ozone generation.” [15,16] As Eliasson and Kogelschatz describe, NO and NO2 can act “catalytically” in the presence of ozone and atomic oxygen to quench ozone-producing reactions, such that a very small concentration of NO and/or NO2 can consume the ozone present in the system and quench further ozone production. [15] Therefore, while operating in the transition region, nitrogen oxides likely are generated in the plasma region, diffuse into the gas phase of the reactor, and react with gaseous ozone. Once all the ozone has been consumed, nitrogen oxides persist in the gas phase rather than being created and destroyed in ozone-quenching reactions, and the discharge operation enters the nitrogen oxides mode.

The transition region, with the lowest total concentration of gaseous reactive species, also corresponds to the least antimicrobial activity, as shown in Figure 4.4. High concentrations of both ozone and NOx decontaminate surfaces effectively, indicating that reactive species diffuse to the contaminated surface and react with and inactivate the E. coli bacteria. However, in the transition region, it is hypothesized that reactive species preferentially react with other reactive species in the gas phase rather than reacting to inactivate bacteria. Regardless of power density, there is no bacterial inactivation for the first 2–3 min of plasma treatment. The diffusion length for reactive species away from the plasma is L = 4 cm, and using an approximate gas-phase diffusivity of D = 0.1 cm2 s-1, an estimate of the characteristic diffusion time is L2/D = 160 s, or 2.67 min, assuming no convective transport within the reactor.

Therefore, the following mechanism for bacterial inactivation inside the SMD reactor is proposed. First, reactive species are created in the plasma region adjacent to the electrodes, then the reactive species diffuse away from the plasma region. Diffusion to the bottom of the reactor takes approximately 2–3 min, after which reactive species react with the bacteria to inactivate them. As the plasma remains active, additional reactive species are produced, which either diffuse to the bottom of the reactor or are quenched by reaction with other gas-phase species. Such quenching reactions are particularly pronounced in the transition region, where ozone is gradually replaced by nitrogen oxides. A parameterized model developed to describe the mode transition incorporates these reactions and agrees qualitatively with what was observed in this study. [87]

Both ozone and NOx were observed to disinfect surfaces, though NOx appeared to be a more effective disinfectant because bacterial inactivation was more rapid in operating regimes where NOx dominated the gaseous chemistry. Ozone is commonly used as a disinfectant for wastewater [75,140,141], though it has not been used as frequently for surface decontamination. Mahfoud and coworkers studied the inactivation of bacteria on surfaces using dry ozone and note that “there is no mention in the literature of the possibility of inactivating dried vegetative bacteria with dry ozone: the ozonation process is reported to be efficient on vegetative bacteria only in humid media, water, agar or with airborne bacteria.” Although their study used different operating conditions and bacterial species than the ones used here, these experiments find similar results of a biphasic survival curve when using 4000 ppm ozone to inactivate non-sporulating bacteria on dry surfaces. Furthermore, Mahfoud’s results suggest a strong dependence of surface type on inactivation efficacy; the bacterial log reduction varied over four orders of magnitude simply by changing the nature of the dry surface when inactivating the same bacterium under otherwise identical operating conditions. [142] In Shimizu’s study, 2000 ppm ozone produced by

SMD achieved a 5-log reduction in E. coli after only 30 s of treatment, though bacteria were grown and treated directly on agar, a relatively easy substrate to disinfect. [87]

NOx are known to be toxic, and nitric oxide (NO) is implicated in a number of immunological and therapeutic pathways including immune signaling, gene expression, vascularization in ischemia-reperfusion injury, and direct damage to pathogens. [80,94,143–145] Surface disinfection by NO2 is a relatively novel disinfection strategy; one study suggested that Venezuelan equine encephalomyelitis (VEE) virus was susceptible to low concentrations of NO2, but the same study found that Bacillus subtilis was not significantly inactivated by NO2; another report concluded that NO, but not NO2, was bactericidal against a variety of pathogens. [146,147] However, the concentrations of NO2 used in those studies were many orders of magnitude below the concentrations observed here. A recent United States patent describes extremely rapid bacterial disinfection using NO2 at concentrations equal to or greater than those measured here. [148]

The experiments described in Chapter 3 used the same SMD design but with two modifications to the operating parameters. First, power density varied in a range from 0.50–1.00 W cm-2, power densities high enough for the device to enter nitrogen oxides mode after only a few seconds of operation. Under those conditions, inactivation did not vary significantly with discharge power density. [60] Although gaseous chemistry was not measured in the experiments in Chapter 3, it is likely that the gas-phase chemistry did not change with increasing power past a sufficiently high power density, so a higher discharge power density did not necessarily lead to an increase in disinfection rate under those conditions. Second, the experiments in Chapter 3 used a diffusion gap of only 1 mm instead of the 4 cm described here. Using the same characteristic diffusion time scale described above, an estimate of the diffusion time for reactive species in under the conditions of Chapter 3 is L2/D = 0.1 s. Inactivation was much more rapid under those configurations, reaching a 4-log reduction in bacterial count within 5 min, indicating a 4-fold increase in the rate of inactivation compared to the results described here. [60] However, based on the diffusion time scale alone, a 1600-fold increase in inactivation rate would be expected. Therefore, disinfection under the conditions employed in this chapter is not purely diffusion-controlled, and reactions between gaseous RONS and bacteria might be the kinetically limiting step. [149] Free convection within the reactor may also play a role in transporting reactive species to the contaminated surface. More studies of the inactivation and gas-phase reaction kinetics are necessary to better understand the mechanism of disinfection.

Other plasma disinfection systems have used the SMD or similar configurations for the production of reactive species. Maisch et al. used a wire-mesh surface micro-discharge reactor operating at 0.02 W cm-2 to inactivate Candida albicans yeast. Analysis of the gaseous chemistry in their reactor showed approximately 500 ppm ozone and 3 ppm or less NO and NO2 after 60 s of plasma treatment. [64] Similarly, Klämpfl and coworkers used an SMD device for rapid and efficient disinfection of a variety of spore-forming and non-spore-forming bacteria on agar; their study reported a power density of 0.035 W cm-2 and 500 ppm gaseous ozone after 60 s of treatment. [65] Although inactivation rates vary considerably across devices, substrates, and types of microorganism inactivated, both results show similar chemistry to the chemistry generated by the device used here, where operation at low power density for short treatment times results in gaseous chemistry dominated by ozone. Jeon and coworkers used the same

41 device to generate ozone from various mixtures of ozone and nitrogen or ozone and argon. Their study measured 1500–3000 ppm of ozone in synthetic air (20% O2 and 80% N2) after 120 s of plasma operation with a slight decline in concentration over time, consistent with the behavior in the transition region, when the device used here is operated at 0.15–0.30 W cm-2 for 120 s. Their study also measured E. coli inactivation on agar with changing gas composition. Interestingly, although the most ozone was generated when creating a plasma from pure oxygen, inactivation of E. coli did not vary significantly with gas composition in their study. [150] In contrast, Zimmermann et al. used a nearly identical device at higher power density, 0.50 W cm-2, to treat adenoviruses in solution. Under their conditions, only 10–20 ppm of ozone was observed after 5 min of operation. [63] NOx concentrations were not quantified, but a spectral peak assigned to NO was present, indicating their device was operating in nitrogen oxides mode.

Different constructions of the SMD besides wire-mesh devices have been reported. Oehmigen et al. describe a configuration using concentric ring-shaped electrodes, which they refer to as a “surface DBD.” [73,86] Their device consumed 0.25 W cm-2 of power and ran for up to 12 minutes, a set of conditions that corresponds to the transition region for the device used here, as shown in Figure 4.7. Although gaseous composition was not quantified in their study, their published gas-phase FTIR spectrum showed a strong ozone peak, but not a nitric oxide or a nitrogen dioxide peak. Interestingly, the treated aqueous phase in their study contained nitrite and nitrate (which form when NOx dissolve in water) rather than ozone. Unpublished experiments under conditions similar to those used here have shown that sufficiently large volumes of liquid water in the confined SMD chamber can drastically alter the time evolution of the gas phase species. It is therefore likely that the relatively high solubility of some nitrogen oxides in water favored dissolution of NOx in the aqueous phase rather than gas-phase quenching reactions of ozone. [99] In another report, Schwabedissen and coworkers used a pair of aluminum electrodes inside and outside a polyethylene envelope. Their study shows between 800 and 2000 ppm of ozone after 7 minutes of operation with power density between 0.02 and 0.05 W cm-2. [132] Their study notes marked differences in both power consumption and ozone generation depending on the shape of the electrodes. Different device configuration, and different electrical input parameters, can alter the relationship between input power and gaseous chemistry; in particular, the “map” of where an SMD device produces ozone or nitrogen oxides can differ from device to device and should be carefully characterized when, for example, using the device for biological applications.

4.6 Conclusion

Gaseous species important to ambient-air plasma disinfection were quantitatively measured by infrared spectroscopy inside a surface micro-discharge (SMD) reactor. The measurements made here support other measurements of aqueous chemistry, including ones described in Chapter 5, and semi-empirical models of gaseous chemistry. Immediately after plasma ignition, ozone is produced for at least a few seconds and up to twenty minutes, depending on discharge power density. At high enough power density, above approximately 0.10 W cm-2, the gaseous chemistry will eventually transition from an ozone-producing mode to a nitrogen oxides-producing mode; ozone concentration reaches a temporal maximum before decreasing to zero, after which NOx (nitric oxide or NO, and nitrogen dioxide or NO2)

42 concentration starts to increase. The “transition region” likely corresponds to the gas-phase quenching of ozone by nitrogen oxides. The mode transition is also apparent in aqueous chemistry when water is treated by plasma under the same conditions described in this chapter. The mode transition in the aqueous phase, as well as the associated antibacterial effects, are described in detail in Chapter 5.

Antibacterial effects were measured using E. coli as a model bacterium, and inactivation correlates reasonably well with gas-phase reactive species concentration. A high concentration of NOx is most closely associated with surface disinfection, while ozone appears to be a slightly less effective surface disinfection agent. The transition region is associated with the smallest antimicrobial effect, likely because reactive species are consumed in gas-phase reactions rather than diffusing to the contaminated surface. Diffusion path length has an effect on how long disinfection takes, but an estimate of the characteristic diffusion time scales suggests that disinfection is not entirely a diffusion-controlled process, and the reactions of plasma species with bacteria may control the overall kinetics of the antimicrobial activity.

Finally, these results are comparable to other studies using similar “indirect DBD” or SMD devices. In particular, other SMD devices configured similarly to the one described here display similar chemical characteristics, with ozone production associated with low power density and short exposure time, and NOx production associated with higher power densities and longer exposure time, for a number of different devices. However, different constructions of the SMD or other approaches to “indirect DBDs” can produce differences in the detailed chemistry despite operating under similar conditions. A potential advantage for SMD-based plasma disinfection is its ability to “tune” the chemical species produced based on operating conditions, but it should be emphasized that careful control of those operating conditions is necessary for reproducible results.

Chapter 5: Ozone Correlates with Antibacterial Effects from Indirect Air Dielectric Barrier Discharge Treatment of Water

Originally published in Journal of Physics D: Applied Physics, DOI 10.1088/0022- 3727/46/14/145202, courtesy of IOP Publishing

5.1 Abstract

Ambient-condition air plasma produced by indirect dielectric barrier discharges can rapidly disinfect aqueous solutions contaminated with bacteria and other microorganisms. In this chapter, key chemical species in plasma-treated aqueous solutions and the associated antimicrobial effects are measured for varying discharge power densities, exposure times, and buffer components in the aqueous medium. The aqueous chemistry corresponded to air plasma chemistry, and a transition in composition from ozone mode to nitrogen oxides mode was observed as the discharge power density increased. The inactivation of E. coli correlates well with the aqueous-phase ozone concentration, suggesting that ozone is the dominant species for bacterial inactivation under these conditions. Published values of ozone-water antibacterial inactivation kinetics as a function of the product of ozone concentration and contact time are consistent with these results. In contrast to earlier studies of plasma-treated water disinfection, ozone-dependent bacterial inactivation does not require acidification of the aqueous medium and the bacterial inactivation rates are far higher. Furthermore, the antimicrobial effect depends strongly on gas-liquid mixing following plasma treatment, apparently because of the low solubility of ozone and the slow rate of mass transfer from the gas phase to the liquid. Without thorough mixing of the ozone-containing gas and bacteria-laden water, the antimicrobial effect was not observed. However, it should be recognized that the complexity of atmospheric pressure plasma devices, and their sensitivity to subtle differences in design and operation, can lead to different results with different mechanisms.

5.2 Introduction

The contribution of ozone to aqueous-phase chemistry and the aqueous-phase antimicrobial effect has apparently not been considered in the context of the surface micro- discharge (SMD), also described as the “indirect” dielectric barrier discharge (DBD), employed in this study. Experiments published to date showing the antibacterial effects of plasma-activated water generally involved relatively long bacterial incubation times following plasma-water treatment. That is, bacteria suspended in plasma-activated water were allowed to remain in the treated water for times ranging from 15 minutes to 3 hours before testing the extent of bacterial killing by solution quenching, followed by serial dilutions and bacterial plating. [46,74,86] In contrast, as shown in this chapter, if ozone is the active aqueous antibacterial agent, much shorter contact times between the bacteria and the ozone-containing water result in extensive bacterial inactivation.

Although the present chapter focuses on aqueous disinfection, it is worth noting that ozone has been used for many different biomedical applications for many years. Currently,

44 widely reported clinical application areas of either gaseous ozone or ozone-treated liquids include dentistry [151–153]; treatment of herniated disks [154]; oxidative pre-conditioning for ischemia-reperfusion injury [155]; wound healing and disinfection [156,157]; and various forms of blood treatment [158], among others. Bocci et al. discuss the biomedical applications of ozone in the context of the recent realization that ozone is one of several biologically important and therapeutically useful gaseous species that include hyperbaric oxygen (O2) nitric oxide (NO), carbon monoxide (CO), carbon dioxide (CO2), xenon (Xe), hydrogen sulfide (H2S) and hydrogen (H2). [159] There are also reports of endogenous formation of ozone, for example by antibodies or amino acids [85,160], although the unambiguous detection of ozone in this context is a topic of active research. [82,161,162]

A recent report showed that the distribution of gas-phase reactive species in confined indirect air DBD, similar to the device used for the present chapter, is surprisingly dynamic and transient. [87] Chapter 4 describes gaseous chemistry in air treated by DBD in more detail and shows transient chemical regimes based on treatment time and power density. For instance, at relatively low power (~0.05 W cm-2), gas-phase ozone concentration is measured to reach more than 1000 ppm after 1 minute of treatment and continues to increase for the treatment duration. By contrast, when input power is increased to around 0.2 W cm-2, ozone density in the gas-phase starts to decrease in a few tens of seconds at a constant power density, showing a peak ozone density.

A semi-empirical model suggests that the mode transition is initiated by vibrationally excited states of nitrogen leading to higher concentrations of nitrogen oxides that quench ozone. [87] The observed transition is significantly different from the case in classical ozone reactors in that the transition takes place over time at a constant power. This dynamic behavior also illustrates that relatively small changes in atmospheric-pressure plasma device design and operation can result in significant differences in plasma chemistry with corresponding differences in biochemical effects. Therefore, it is incorrect to assume that every atmospheric pressure-based plasma disinfection system operates via the same mechanism or involves the same biochemical reactions.

In this chapter, the dynamics of indirect DBD air plasma-generated species are further examined, focusing on aqueous-phase species created when the plasma-generated species enter the liquid phase. The power density, exposure time, acidification of the aqueous phase, and mixing of the treated solution are investigated. The aims of this chapter are to quantify the effects of those parameters on the aqueous-phase chemistry and antimicrobial effect for relatively short (i.e., less than 5-minute) treatment times. A clear correlation between liquid composition and disinfection efficacy is demonstrated under the conditions examined. In addition, aqueous ozone concentration shows an unexpectedly strong dependence on mixing of the aqueous phase with the gas above the liquid surface following treatment, suggesting an important role for mass transfer of ozone. Finally, the pH of the system is considered to determine whether the inactivation is acid-dependent under these conditions. E. coli was used as a model organism to determine the antimicrobial effect.

5.3 Experimental Section

Figure 2.1 shows the experimental setup. The device was operated in the surface micro- discharge configuration, or “indirect mode,” as described in Section 2.2. Plasma was generated and characterized as described in Section 2.3. Fourier-transform infrared spectroscopy (FTIR) absorption measurements were performed in the region just below the plasma zone to qualitatively identify some of the various plasma-generated species in the reactor, as described in Section 2.4. [88] The FTIR absorption measurements were done without the glass vial and liquid medium. It is possible that the vial with liquid medium changes distributions of plasma- generated species. However, preliminary measurements suggested that those influences are minimal under conditions considered here. Aqueous chemistry was measured as described in Section 2.5, and aqueous disinfection was measured as described in Section 2.6. The maximum log reduction was approximately 5.5.

5.4 Results

The effects of the discharge power density, exposure time (i.e., the duration of time the discharge was sustained), and type of solution (i.e., buffered or non-buffered) on the gas-phase chemistry, aqueous-phase chemistry, and antimicrobial effect were investigated. The goal of this chapter was to determine which reactive oxygen and nitrogen species were present and which species appeared most responsible for the observed antimicrobial effect. In Figures 5.1–5.5, vertical error bars represent the standard deviation of the measurement about the mean, and each data point represents the average of between 3 and 6 samples. For measurements with varying power density, horizontal error bars represent the standard deviation of the power density about the mean power density. Where no error bar is present, the error interval is smaller than the size of the marker on the graph.

Aqueous-phase chemistry was measured in plasma-treated liquids. Figure 5.1 shows the concentration of aqueous-phase hydrogen peroxide, nitrite, and nitrate as a function of power density after exposure to plasma for a fixed exposure time of 300 s. Figure 5.2 shows the concentration of ozone in the aqueous phase for varying discharge power density at a constant exposure time (5.2a) and for varying exposure time at a constant power density (5.2b). In Figures 5.1 and 5.2, data are shown for PBS; similar trends were observed for non-buffered saline.

Figure 5.1. Aqueous chemistry in plasma-treated PBS varying discharge power density. (a) Hydrogen peroxide. (b) Nitrite (black) and nitrate (white).

Figure 5.2. Ozone chemistry in plasma-treated PBS. (a) Varying discharge power density at a fixed exposure time of 300 s. (b) Varying exposure time at fixed power densities of 0.05 W cm-2 (black) and 0.30 W cm-2 (white).

The plasma’s antimicrobial effect was also measured. Figure 5.3 shows the effect of varying discharge power density at a constant exposure time (5.3a) and of varying exposure time at a constant power density (5.3b) on the antimicrobial effect. Figure 5.4 demonstrates the effect of vortexing the treated solutions on the plasma’s antimicrobial effect with the discharge power density fixed at 0.05 W cm-2. Data are shown for PBS; similar trends were observed for non- buffered saline. Finally, Figure 5.5 compares liquid-phase chemistry and antimicrobial effect as a function of power density for buffered and non-buffered saline solutions. The time that bacterial suspensions were exposed to plasma was fixed at 300 s for all data shown in Figure 5.5, although Figures 5.2(b) and 5.3(b) suggest that the liquid phase ozone concentration and bacterial inactivation rates do not increase after 120 seconds of plasma exposure. Treated samples were vortexed following exposure in all of the antimicrobial experiments except those described by the “not vortexed” curve in Figure 5.5. The temperature of the treated solutions increased by a maximum of 3°C at the highest power density and longest exposure time, suggesting that thermal effects do not account for the plasma’s antibacterial effect.

Figure 5.3. Antimicrobial effect in plasma-treated PBS. (a) Varying discharge power density at a fixed exposure time of 300 s. (b) Varying exposure time at fixed power densities of 0.05 W cm-2 (black) and 0.30 W cm-2 (white).

Figure 5.4. Antimicrobial effect in plasma-treated PBS when the aqueous phase is vortexed for 5 seconds (black) or not vortexed (white).

Figure 5.5. Comparison of chemistry and antimicrobial effect in buffered (black) and non- buffered (white) aqueous solutions. (a) pH endpoint. (b) Ozone concentration. (c) Antimicrobial effect.

5.5 Discussion

The quantitative FTIR gas phase measurements shown in Chapter 4 indicate that ozone is dominant at lower power (0.05 W cm-2), and that nitrogen oxides become dominant at higher power (0.30 W cm-2). At low power density, ozone concentration in the gas phase increases continuously for at least twenty minutes after a discharge is ignited; however, at high power density, gas-phase ozone is present for the first minute of treatment but is completely quenched afterwards. As noted before, the transient presence of ozone in the gas phase is consistent with a previous report by Shimizu et al. [87] As noted above, a semi-empirical model suggests that the transition from ozone mode to nitrogen oxide mode is initiated by vibrationally excited states of nitrogen reacting to form nitrogen oxides that quench ozone.

The aqueous-phase measurements shown here indicated that aqueous-phase chemistry follows the same mode transition as in the gas-phase as described in detail in Chapter 4. The

56 definitions of operating regimes used in this chapter are consistent with the ones described in Chapter 4 but represent a simplification for the fixed treatment time of 300 s. In aqueous-phase at low power densities (< 0.20 W cm-2), the concentration of dissolved ozone is relatively high, and the concentrations of nitrogen oxides and hydrogen peroxide are relatively low. The opposite trends prevailed at power densities greater than about 0.25 W cm-2: low ozone concentration but high concentrations of nitrogen oxides and hydrogen peroxide. At a treatment time of 300 s and intermediate power densities between about 0.20 W cm-2 and 0.25 W cm-2, a “transition” regime was evident in which the chemistry is highly sensitive to small changes in power density as shown in Figures 5.1 and 5.2.

Of the reactive species measured, ozone correlated most strongly to the plasma’s antimicrobial effect. The data in Figures 5.2 and 5.3 show similar profiles of aqueous-phase ozone concentration and log reduction as functions of both exposure time and discharge power density. Ozone is commonly used in industrial and municipal water treatment as a disinfectant for drinking water, and its microbial inactivation kinetics have been widely studied, though published results differ by several orders of magnitude. [163] Some of these differences have been attributed to differences in temperature, pH, contamination or “ozone demand” in the water, and initial density of bacteria, among other factors. [140,141,164]

Many previous studies have treated clean water with ozone and mixed this ozonated water with contaminated water rather than directly exposing contaminated water to ozone. It is possible that having bacteria present in the treated water in this system increases the ozone demand enough to account for the observed apparent lower effectiveness. Finch and coworkers noticed that “cellular debris” from inactivated bacteria can “shield” remaining bacteria from ozone. [140] Increasing the ozone demand in treated water has been shown to increase the necessary ozone concentration for inactivation by a factor of 10 or more. [163] Buffer solutions prepared in sterile, purified water and lacking any residual ozone-reactive compounds have been described in previous studies as “ozone demand-free buffer”. When ozone demand-free buffer was ozonated and mixed with bacterial suspensions near room temperature, an ozone dose, or Ct value, of between 0.001 and 0.05 mg ozone min-1 l-1 was required to achieve a 3.5-log inactivation of E. coli. [141,165] However, when laboratory wastewater was used, Farooq and Akhlake calculated a required Ct value of 0.30 mg ozone min-1 l-1. [166]

Many previous studies of E. coli inactivation by ozone in the aqueous phase used fed- batch or continuous-flow systems, while this system most closely resembles a simple batch reactor. Reactor configuration is also important in determining the inactivation efficacy, implying an important role for mass transfer in this system. As noted by Hunt and Mariñas, incomplete mixing and deviations from ideal reactor conditions could lead to reductions in the antimicrobial effect. [141] To evaluate the importance of mass-transfer effects, the experiments were repeated without vortexing the treated vials after plasma treatment. Figure 5.4 shows the comparison of the antimicrobial activity with and without vortexing. When the treated samples were thoroughly mixed following treatment, the log reduction increased linearly with treatment time up to 120 seconds. However, without mixing, the antimicrobial effect was nearly zero for all treatment times. These results strongly suggest that mass transfer effects are important and that vigorously mixing the system overcomes barriers to mass transfer. The data shown in Figure

5.4 imply that most or all of the inactivation effect happens during the vortex duration of 5 seconds. The treated solutions were diluted by a factor of 105 immediately after vortexing.

Though some ozone remained in the solution while the bacteria were plated, the ozone concentration was 5 orders of magnitude smaller during plating than it was during vortexing. Therefore, any duration of time following the vortexing and dilution likely does not contribute significantly to the antimicrobial effect. Using 5 seconds as the exposure time, the ozone dose calculated for this system is 0.29 mg ozone min-1 l-1 to achieve at least a 5.5-log reduction of E. coli. This result agrees most closely with the findings of Farooq and Akhlaque [166], suggesting that the ozone demand in this system most closely resembles that of laboratory wastewater. Considering both the previous results of ozone disinfection and the results presented here, it appears crucial that aqueous disinfection systems employing ozone include some method of improving the mass transfer of ozone into the aqueous phase. Such strategies might include the use of a counter-current sparged column, a stirred-tank reactor, or a bubbler, as suggested by Glaze. [75]

Figure 5.5 illustrates that the antimicrobial effect under these conditions is not pH- dependent. Prior reports in the plasma-activated water literature have described an important role for acidification. The rates of inactivation reported here are much faster than the rates observed in previous experiments with plasma-activated water. [74] Incubation times in these previous studies typically ranged from 15 minutes to 3 hours, whereas the contact time between bacteria and the plasma-generated species in the present work was only 5 seconds. Oehmigen et al. and Naïtali et al. described similar antimicrobial effects in plasma-treated water: following a 20- minute exposure, approximately 6 logs of inactivation were observed in plasma-treated non- buffered water, but less than 1 log of inactivation was observed in treated neutral buffer. [73,93] Previous studies have proposed several hypotheses for why acidification might be necessary, including nitrite/nitrous acid equilibrium [74] and superoxide/perhydroxyl equilibrium [167].

However, in the present experiments in which it appears that ozone is the dominant antibacterial component, antimicrobial effects are observed even when the treated solutions do not appreciably acidify. Figure 5.5(a) shows the pH endpoint after 300 seconds of plasma treatment as a function of discharge power density for both a buffered and non-buffered aqueous system. Treated physiological saline acidifies to below pH 3, while the pH of treated PBS decreases only slightly. Figure 5.5(b) shows that the concentration of ozone in the aqueous phase is similar for buffered and non-buffered aqueous solutions, and Figure 5.5(c) shows similar inactivation profiles for buffered and non-buffered solutions.

Although hydrogen peroxide, nitrate, and nitrite were present in addition to ozone, significant antimicrobial activity cannot be attributed to those species. Hydrogen peroxide is known to be antimicrobial in high doses, but concentrations of mM to hundreds of mM are typically used. [168,169] Watts and coworkers observed a 6-log reduction of E. coli after treatment with 3 mM hydrogen peroxide, but a treatment time of 120 minutes was required. In contrast, sub-mM concentrations of hydrogen peroxide had no antibacterial effect in their experiments. [168] The maximum hydrogen peroxide concentration measured here was less than 0.1 mM. Nitrate and nitrite have been shown to be indirectly antimicrobial at low pH by producing nitric oxide through the “acidified nitrite” pathway. [94,143,170] However, the

58 strongest antimicrobial effect was observed when nitrate and nitrite concentrations are lowest and found that inactivation did not depend on pH. Therefore, under these conditions, neither hydrogen peroxide nor acidified nitrite were responsible for the antimicrobial effect.

The results described in Appendix A demonstrate a set of conditions, using the same device with a similar configuration, that likely involved a different inactivation mechanism than the ozone-dependent inactivation described here. [74] Although ozone was not measured previously, it is unlikely that ozone was involved in the previously observed antibacterial effect. Under the conditions of Appendix A, water was treated with a discharge power density of approximately 0.30 W cm-2, for which no ozone is present in the aqueous phase. In addition, the antimicrobial effect of plasma-treated water persisted for at least seven days after treatment, and that a relatively long plasma-water exposure time of 20 minutes was required. The half-life of ozone in aqueous solution is relatively short, on the order of minutes, so it is not possible that ozone would persist for several days and account for a long-lasting antimicrobial effect. [98]

From the measured liquid phase concentration of ozone, and using a Henry’s law constant of 0.013 mol kg-1 bar-1 [137], the equivalent gas phase ozone concentration can be estimated, assuming that vortexing achieves vapor-liquid equilibrium. Under these circumstances, the estimated gas-phase ozone concentration falls between 5000 and 6000 ppm after 5 minutes of plasma treatment at power density < 0.20 W cm-2. This estimate agrees well with the measured gas-phase ozone concentration (from UV absorption measurements; see Figure 4.4) of around 5000 ppm after 5 minutes of treatment. 5000–6000 ppm is a substantial concentration of ozone, greatly exceeding the United States Occupational Safety and Health Administration (OSHA) recommended exposure limit of 0.1 ppm and underscoring the need for controlled, directed applications of reactive species in plasma disinfection devices. [171]

Interestingly, the liquid-phase ozone concentration reaches a maximum after 2 minutes of plasma treatment, which was enough time to inactivate all of the bacteria in the sample to within the detection limit. A simple estimate of the time needed for ozone to diffuse into the vial of depth L = 4 cm with an estimated gas-phase diffusivity D = 0.1 cm2 s-1 is L2/D = 160 s. Therefore, the hypothesized primary effect of plasma treatment in the low-power regime is to create ozone in the air space above the aqueous phase in the vial. After about 2 minutes of plasma treatment, enough ozone has diffused into the air space of the vial to inactivate all the E. coli. When the vial is capped and vortexed, the gas-phase ozone transfers into the well-mixed aqueous phase and rapidly reacts to inactivate the bacteria.

The scenario above neglects any effects of other plasma-generated species and is therefore probably incomplete. The most important result is that given the measured ozone concentrations in the gas and liquid phases, the measured rate of bacterial inactivation is completely consistent with published values of the product of ozone concentration and contact time. Ozone could be an important component in indirect air DBD devices such as the one used here and should not be neglected in future considerations of plasma biomedicine using discharges in air.

5.6 Conclusion

The results presented in the previous chapter demonstrate that ambient-condition air plasma operating as a confined, indirect DBD will create relatively high concentrations of gas- phase ozone at low powers (below about 0.2 W cm-2). This chapter demonstrates that, if this ozone-rich gas is thoroughly mixed into bacteria-laden water for even a short time of 5 seconds, suspended E. coli is completely inactivated to within the limits of detection. By contrast, high- power treatment above about 0.25 W cm-2 produces less than a 1-log reduction in viable E. coli. Complete inactivation requires about 2 minutes of plasma operation, and this corresponds to the characteristic diffusion time for the plasma-generated species to diffuse into the gas space above the liquid in the vial containing the bacteria in water. The antimicrobial effect correlates well with aqueous-phase ozone concentration, but not with pH or concentration of hydrogen peroxide, nitrite, or nitrate, suggesting an ozone-dominant mechanism of inactivation under these conditions.

Further support for this hypothesis comes from the measured liquid phase ozone concentration, which is consistent with the estimated equilibrium ozone concentration in the gas phase following vigorous mixing of the phases. In addition, the product of ozone liquid phase concentration and contact time, and the measured rates of bacterial inactivation, are well within the range of published values characteristic of ozone inactivation of E. coli suspended in water. The rates of bacterial inactivation with ozone are much faster than rates previously reported for “plasma-activated water” in which nitrogen oxides and low pH correlate with bacterial inactivation. The antimicrobial effect under these conditions appears to be mass-transfer limited. When the treated samples were not thoroughly mixed by vortexing following plasma exposure, no significant inactivation was observed.

That ozone-mediated inactivation is not pH-dependent is in contrast to many of the prior studies of plasma-treated aqueous systems where acidification was necessary to achieve an antimicrobial effect. Previous experiments demonstrated an antimicrobial effect that persisted for at least seven days after water treatment but required relatively long exposure times and acidification of the aqueous medium. Using the same device, the conditions discussed here result in an antimicrobial effect that requires much shorter exposure times and is independent of the pH of the aqueous medium. Therefore, aqueous-phase chemistry and bacterial inactivation can be “tunable” based on discharge power density, demonstrating the complexity and versatility of ambient-gas plasma chemistry and its biological interaction.

Finally, ozone has not generally been emphasized as an important reactive oxygen species in atmospheric pressure plasma biomedical applications to date. The present results suggest that ozone can be an important, if not dominant, factor in microbial inactivation. However, the present results do not imply that ozone is always the most important ROS associated with air plasmas. It should be recognized that the complexity of atmospheric pressure plasma devices, and their sensitivity to subtle differences in design and operation, can lead to different results with different mechanisms.

Chapter 6: Antimicrobial Synergy Between Ambient-Gas Plasma and UVA Treatment of Aqueous Solution

Originally published in Plasma Processes and Polymers, DOI 10.1002/ppap.201300065, courtesy of Wiley-VCH Verlag GmbH

6.1 Abstract

This chapter describes the photochemical and antimicrobial synergy between chemical species generated by ambient-condition air plasma and ultraviolet photons at near-UV (UVA) wavelengths. The interaction of UVA photons, produced by a UVA-emitting LED, with plasma- generated aqueous-phase species is investigated under different discharge power densities, corresponding to different regimes of plasma chemistry. Higher power treatments simultaneously - - generate species such as hydrogen peroxide (H2O2), nitrite (NO2 ), and nitrate (NO3 ) in aqueous solution. When relatively high-power plasma treatment was followed by UVA treatment of an aqueous suspension of E. coli, the antimicrobial effect significantly exceeded the effect predicted from the two treatments alone. The activity of various plasma-created species was tested by creating “artificial” aqueous solutions containing nitrite, nitrate, or hydrogen peroxide, added at the concentrations observed from plasma activation of water. Exposing the individual plasma- associated species to UVA photons indicated that the antimicrobial effect is associated with nitrite but not with nitrate or hydrogen peroxide. The presence of the antioxidant ascorbate during UVA treatment effectively prevented the enhancement of the antimicrobial effect, suggesting an antioxidant-mediated mechanism of bacterial inactivation. Addition of nitrite to aqueous solution, followed by photolysis of nitrite by UVA photons with wavelengths near 360 nm, is hypothesized as the primary mechanism of the synergy between air plasma and UVA treatments, and hydroxyl radical is thought to be most directly responsible for the bactericidal effect. The full effect of plasma-activated water acting synergistically with UVA is reproduced by UVA treatment of water containing both nitrite and hydrogen peroxide added at the concentrations created by the air plasma.

6.2 Introduction

Ambient-air plasmas generated by SMD are known to contain a dynamic distribution of reactive species in the gas phase depending on the plasma operating conditions, with discharge power density (expressed in power per unit electrode area, typically W cm-2) appearing to be a critical parameter. [15,16] Recent reports, along with the information presented in Chapter 5, have shown that for disinfection of aqueous solution, the aqueous chemistry and disinfection capability associated with ambient-air plasma correlate well with the established gas-phase chemistry. [87,138] As shown in Chapter 4 along with other studies, under relatively high power density (>> 0.1 W cm-2) and for long treatment times, various nitrogen oxides, along with hydrogen peroxide, are among the principal reactive species produced. In contrast, at relatively low power density (<< 0.1 W cm-2) and/or for short treatment times, ozone appears to be the dominant reactive species. At intermediate power density and treatment times, the plasma chemistry undergoes a “mode transition” from low-power or “ozone” mode to high-power or

“nitrogen oxide” mode. The mode transition is particularly pronounced when the discharge zone is spatially confined, minimizing mixing with adjacent air. [172] The critical power density for mode transition is not constant from device to device and can vary with geometry and electrode design as well as with the operating temperature and humidity. [172]

To summarize the results presented in Chapter 5, the difference in reactive species chemistry between high- and low-power mode has a profound effect on the efficacy and speed of disinfection. In high-power mode, where nitric acid, nitrous acid, and hydrogen peroxide are the most abundant aqueous species, disinfection is relatively slow, requiring bacteria to be incubated in plasma-treated solution for minutes to hours. Bacterial inactivation under these conditions requires acidification of the surrounding medium, as little to no antimicrobial effect is observed in plasma-treated neutral buffer solutions. [73,74,86,93] As shown in Appendix A, water treated by plasma under these conditions, sometimes referred to as “plasma-activated water,” retains some of its antimicrobial effect for at least seven days after treatment, over which time the concentration of some reactive species gradually declines, but the pH remains constant. [8,74] Inactivation in this “persistently” antimicrobial plasma-activated water is hypothesized to involve both nitric acid and cytotoxic chemical species produced by reactions among plasma- generated species. In particular, nitrite is depleted over the course of plasma-activated water’s lifetime, and it is possible that nitrite acts as a reservoir for the production of the antimicrobial species nitric oxide (NO) via the “acidified nitrite” pathway. [94,170]

In low-power mode, with ozone acting as the primary antimicrobial agent, the disinfection characteristics of plasma-treated solution are markedly different. Inactivation of bacteria in water, described by the product of ozone concentration (C) and incubation time (t), commonly referred to as the “Ct” product, can be much more rapid when treated by ozone compared with nitrogen oxides. For example, the data presented in Chapter 5 indicate that ozone aqueous concentrations of 3–4 mg l-1 acting for 5 seconds results in a 4-log reduction for E. coli. [138] Acidification of the solution is not necessary for ozone disinfection, as low-power plasma treatment will disinfect even buffered solutions. [138] However, ozone treatment of aqueous solution requires vigorous mixing or some enhancement of mass transfer of gaseous ozone into the aqueous phase, apparently because ozone is relatively insoluble in water compared to many nitrogen oxides. [137] Finally, ozone-containing water treated with low-power plasma does not exhibit a persistent antimicrobial effect days after its treatment because ozone decays in water with a half-life of only a few to a few tens of minutes. [98]

Nitrite is one of the primary species produced in high-power plasma treatment of water, as discussed previously. It has limited direct antimicrobial action, but it can react under a variety of conditions to produce more reactive species. In addition to serving as a reservoir for nitric oxide production in the nitrate-nitrite-nitric oxide pathway, nitrite participates in a number of biologically significant reactions. According to a recent review by Lundberg et al., nitrite can be incorporated into fatty acids or reduced to nitric oxide and other biologically active nitrogen oxides by heme, and it is under active investigation for various therapeutic roles, for example in ischemia-reperfusion injury. [173] In another review, Graves describes the biological role of various plasma-associated RONS including nitrite and nitric oxide. [26] Finally, nitrite is known to undergo photolysis when exposed to UV photons at wavelengths near 360 nm to create nitric oxide (NO) and hydroxyl radicals (OH). [174–178] Both nitric oxide and hydroxyl have known

62 innate immune activity; they can react directly with pathogens and with other RONS intermediates, and are active in immune signaling, among many other biochemical processes. [80]

UV radiation, at several different wavelengths across the UV spectrum, has been extensively studied for water and surface disinfection. UVC light generated by a mercury lamp has a strong bactericidal effect because the photoabsorption cross section of DNA peaks around 254 nm and the wavelength coincides with a mercury emission peak. However, in some cases, UVC disinfection is limited by the shallow penetration depth. [113,114] Recently, Hamamoto et al. and Mori et al. have reported inactivation of E. coli in aqueous solution using high-intensity UVA LEDs (~ 1 W cm-2) with emission maxima near 360 nm [113,179]. In addition to external sources of UV light, ambient-condition air plasma discharges emit some UV radiation. Laroussi and Leipold reported that UV emission from air plasma discharges occurs mostly at UVA and UVB wavelengths, though typically at intensities much lower than the UVA LED used in this study; no significant emission was observed in the UVC spectrum. [180] Other disinfection strategies, including solar disinfection (SODIS) and photocatalytic oxidation using titanium dioxide (TiO2) and UVA light are discussed in Section 1.6.

This chapter describes the use of a high-intensity UVA LED in combination with ambient-air plasma treatment to disinfect aqueous suspensions of E. coli. While inactivation was achieved by UVA treatment and plasma treatment alone, a synergistic interaction between the two treatments produced an antimicrobial effect greater than expected when both treatments were combined. The synergistic effects can be reproduced partially by adding nitrite to water at the concentration measured following plasma exposure. Combining nitrite and hydrogen peroxide at the appropriate plasma-generated concentrations fully reproduced the observed synergy. UVA photolysis of nitrite is known to form nitric oxide and hydroxyl radicals. The antimicrobial effect achieved by plasma/UVA and nitrite/hydrogen peroxide/UVA treatment is significantly reduced in the presence of the antioxidant ascorbate, implicating oxidizers produced by the photolysis of nitrite as the primary active antimicrobial species.

6.3 Experimental Section

Figure 2.1 shows the experimental setup. The device was operated in the surface micro- discharge configuration, or “indirect mode,” as described in Section 2.2. Plasma was generated and characterized as described in Section 2.3. Power density was calculated as total power divided by the electrode area of 17.4 cm2. Absolute power was fixed at either 0.80 W or 5.0 W, corresponding to power densities of 0.05 W cm-2 (i.e., low-power) or 0.30 W cm-2 (i.e., high- power).

For UVA experiments, a UVA-emitting LED (Nichia, NC4U133A(T)) was situated approximately 5 mm below the bottom of a glass vial with same dimensions as above. The intensity of UVA light 5 mm above the LED was measured with a photodetector (Ophir, Nova II). Applied voltage and current were measured with an external voltmeter (Protek 506), and light intensity was adjusted by manipulating the applied voltage. All experiments were conducted using an applied voltage of approximately 15.87 V, corresponding to a photon

63 intensity of approximately 810 mW. Over the course of these experiments, applied voltage fluctuated by less than 0.03 V, corresponding to a fluctuation of photon intensity of less than 25 mW. The emission maximum was observed at 369 nm.

For some experiments, chemical compounds were added to bacterial suspensions, including sodium nitrate (NaNO3, Sigma), sodium nitrite (NaNO2, Sigma), hydrogen peroxide (H2O2, Fisher), and sodium ascorbate (C6H7NaO6, Alfa Aesar) at concentrations indicated below. 150 µl of each suspension was transferred into a glass vial prior to treatment with plasma or UVA. For plasma treatment, vials were placed inside the SMD device shown in Figure 2.1 and exposed to plasma. For UVA treatment, vials were held in place 5 mm above the UVA-LED and exposed to UVA light. Samples were exposed to plasma only, UVA only, plasma then UVA in series, or UVA then plasma in series.

Bacteria were cultured and aqueous disinfection was measured as described in Section 2.6. The maximum log reduction was approximately 5. .

6.4 Results

The antibacterial effects of UVA treatment, plasma treatment, and both treatments in combination were measured to determine whether UVA and plasma produced a synergistic antimicrobial effect. Figures 6.1–6.4 show E. coli inactivation profiles as functions of UVA exposure time with various combinations of plasma treatment and additional chemical species present in the bacterial suspension. In Figures 6.1–6.4, each point represents the mean log reduction of at least 3 trials, and error bars represent the standard deviation about the mean. Where no error bar is present, the error interval is smaller than the size of the marker on the graph.

First, the inactivation efficacy of the UVA device used here was compared to that of other high-intensity UVA systems. Figure 6.1 shows results for UVA-only treatment (i.e., no plasma treatment) as a function of both exposure time and photon fluence. Other results for the inactivation of E. coli using only a high-intensity UVA-emitting LED include, for example, Hamamoto et al. [113] and Mori et al. [179].

Figure 6.2. Bacterial inactivation as a function of exposure time and photon fluence from UVA- LED treatment.

Next, plasma and UVA treatment were combined and different combinations of the two treatments were analyzed for interactive effects. The “expected additive” effect as a function of UVA treatment was defined as the antimicrobial effect from UVA treatment by itself (i.e., the data shown in Figure 6.1) plus the antimicrobial effect from a fixed duration of plasma treatment by itself. In high-power or nitrogen oxide-rich mode (0.30 W cm-2), 5 minutes of plasma treatment resulted in 0.20 log reductions, and in low-power or ozone-rich mode (0.05 W cm-2), 30 seconds of plasma treatment resulted in 0.50 log reductions. Therefore, the expected additive inactivation curves were calculated by adding 0.20 log reductions (for high-power mode) or 0.50 log reductions (for low-power mode) to the UVA-only inactivation curve shown in Figure 6.1.

In this chapter, plasma-only inactivation kinetics were not measured in detail, but the results described above are in good agreement with previous measurements of plasma inactivation in solution. In particular, the result of 0.50 log reductions after 30 seconds of low- power plasma treatment is consistent with a previous report of the effects of low power “ozone” mode on bacteria in solution, discussed in further detail in Chapter 5. [74,138] In the experimental configuration used, about 2 minutes are required for ozone to diffuse into the vial containing the water and bacteria. Therefore, after 30 seconds the O3 concentration in the water 65

(after vortexing to mix the gas with the liquid in the vial) should be about 1–2 mg l-1, corresponding to ~0.5–1 log bacterial reduction. [138]

Figure 6.2 compares the combined antimicrobial effect of UVA treatment coupled with plasma treatment to the expected additive effect. As shown in Figure 6.2(a), when high-power plasma treatment followed UVA treatment, the antimicrobial effect matched the expected additive effect, and no interaction between the two treatments was evident. However, when UVA treatment followed high-power plasma treatment, the antimicrobial effect was significantly enhanced relative to the expected additive effect, showing clear evidence of the synergistic effect between UVA photons and plasma-generated species. In contrast, when low-power plasma was used instead of high-power plasma, this enhancement was not observed, as shown in Figure 6.2(b). The antimicrobial effect of both combined treatments, low-power plasma followed by UVA and UVA followed by low-power plasma, matched the expected additive antimicrobial effect.

Figure 6.2(a). Antimicrobial effect from combined UVA and high-power (5 minute treatment at 0.30 W cm-2) plasma treatment. Open circles ( ): expected additive effect; closed squares ( ): UV followed by 5 minutes of plasma treatment; closed triangles ( ): UV following 5 minutes of plasma treatment. A clear antimicrobial synergy is shown: UVA exposure following plasma- activation of water results in much greater antimicrobial action than adding their individual effects.

Figure 6.2(b). Antimicrobial effect from combined UVA and low-power (30 s treatment at 0.05 W cm-2) plasma treatment. Open circles ( ): expected additive effect; closed squares ( ): UV followed by 5 minutes of plasma treatment; closed triangles ( ): UV following 5 minutes of plasma treatment.

To determine which reactive species are responsible for the synergy between high-power plasma treatment and UVA treatment, a variation of the experiment described above was performed. Instead of treating with plasma, variations of “artificial” plasma-activated water were created. In these experiments, individual species or combinations of species were added to the bacterial suspension prior to exposure to the UVA-LED. This experiment was performed with 5 mM nitrite, 5 mM nitrate, and/or 100 µM hydrogen peroxide. The concentrations used here approximate the aqueous-phase concentrations of the species measured after 5 minutes of plasma treatment in the high-power nitrogen-oxide mode as described in Chapter 5. [74,138] The mixture of all three species in the same solution was defined as “artificial plasma-treated PBS”. There are likely other short-lived species present in actual plasma treatment that were not measured or considered here. Incubating E. coli in neutral-pH buffered suspensions with nitrite, nitrate, and hydrogen peroxide for 5 minutes, with no UVA treatment, produced no antimicrobial effect (data not shown).

Figure 6.3 shows the antimicrobial effect as a function of UVA exposure time with these species added to the bacterial suspension. No plasma treatment was performed as noted above. Figure 6.3(a) shows the effects of individually adding single plasma-associated species. Under these conditions, neither adding nitrate nor adding hydrogen peroxide enhanced the log reduction compared to UVA treatment without added nitrate or hydrogen peroxide. Adding nitrite to the suspension before UVA exposure enhanced the antimicrobial effect but not to the same extent as actual plasma treatment. However, as shown in Figure 6.3(b), artificial plasma-treated solution showed the same synergistic antimicrobial effect as solution treated with air plasma before UVA exposure. Interestingly, a solution with added nitrite and hydrogen peroxide, but not nitrate, also reproduced the synergistic effect observed with plasma treatment, suggesting that it is the combination of nitrite and hydrogen peroxide that is responsible for the plasma-UVA synergy.

Figure 6.3(a). Antimicrobial effect from UVA treatment with single added species. Gray circles ( ): UVA only; gray squares ( ): UVA following 5 minutes high-power plasma. Black circles ( ): 5 mM nitrite; black squares ( ): 5 mM nitrate; black triangles ( ): 100 µM hydrogen peroxide. Individual additions of the three known components of plasma-activated water show only nitrite plays a partial role in accounting for plasma-UVA synergy.

Figure 6.3(b). Antimicrobial effect from UVA treatment with multiple added species. Gray circles ( ): UVA only; gray squares ( ): UVA following 5 minutes high-power plasma. Closed inverted triangles ( ): 5 mM nitrite plus 100 µM hydrogen peroxide; closed diamonds ( ): 5 mM nitrite plus 100 µM hydrogen peroxide plus 5 mM nitrate. UVA treatment of the combination of nitrite and hydrogen peroxide at the concentrations observed after plasma treatment reproduces the plasma-UVA synergy.

Finally, the effects of increasing the concentration of nitrite and of adding an antioxidant to the suspension prior to UVA treatment were studied, shown in Figure 6.4. Figure 6.4(a) compares the antimicrobial effect of adding 5 mM and 50 mM nitrite; increasing the nitrite concentration tenfold resulted in an increase of approximately 1 log reduction. Figure 6.4(b) shows the antimicrobial effect as a function of UVA treatment time with 5 mM nitrite, with and without 5 mM ascorbate. When ascorbate was added to the bacterial suspension with nitrite, no enhancement to the antimicrobial effect was observed, and the overall effect was comparable to UVA treatment alone without addition of nitrite or ascorbate. Figure 6.4(c) shows that ascorbate has a similar effect in suppressing the synergistic effect when added to solution before plasma and UVA treatment. When ascorbate was added to a bacterial suspension prior to exposure to plasma then UVA, the antibacterial effect was again comparable to treatment with UVA only.

Figure 6.4(a). Antimicrobial effect from UVA treatment with added nitrite. Gray circles ( ): UVA only; gray squares ( ): UVA following 5 minutes high-power plasma. Dotted triangles ( , · · ·): 5 mM nitrite; dashed triangles ( , - - -): 50 mM nitrite. Adding 5 mM nitrite accounts for a significant fraction of the observed plasma-UVA synergy; 50 mM nitrite added reproduces the observed synergy, but this concentration is much higher than what is observed from plasma treatment of water for the present experiments.

Figure 6.4(b). Antimicrobial effect from UVA treatment with added nitrite and ascorbate. Gray circles ( ): UVA only; gray squares ( ): UVA following 5 minutes high-power plasma. Closed triangles ( ): 5 mM nitrite; open triangles ( ): 5 mM nitrate plus 5 mM ascorbate. Ascorbate abrogates the synergistic effects of added nitrite.

Figure 6.4(c). Antimicrobial effect from UVA treatment with plasma treatment and ascorbate. Gray circles ( ): UVA only. Closed squares ( ): UVA following 5 minutes plasma; open squares ( ): UVA following 5 minutes plasma plus 5 mM ascorbate. Ascorbate abrogates most of the synergistic effects of plasma activated water.

6.5 Discussion

Most studies of the interaction of antibacterial treatments involve measuring doses, usually in terms of the minimum inhibitory concentration, or MIC. [181,182] If the MIC of the combined treatments is larger than expected from the two treatments alone, then the two treatments interact antagonistically; if it is smaller than expected, then the two treatments interact synergistically. However, it is difficult to assign an inhibitory “concentration” of ambient-air plasma treatment, so here interactions were evaluated by comparing the log reduction achieved by two treatments in combination to the sum of the log reductions achieved by each treatment independently; this sum was called the “expected additive” effect. If the effect of the combined treatment exceeded the expected additive effect, the two treatments operated synergistically. In contrast, if the combined effect was less than the expected additive effect, then the two treatments operated antagonistically. Although this measure of synergy or antagonism is less common than the usual approach described above for describing the interaction of antibacterial treatments, there is precedent for comparing log reductions to determine whether synergy exists

72 between irradiation and chlorination to kill pathogenic E. coli [183] and between nitrite and hydrogen peroxide treatment of E. coli in the presence of lactate. [97]

Treating bacterial suspensions with high-power plasma first, followed by UVA treatment, was the only experimental procedure that produced an enhancement of the antimicrobial effect above the expected additive value. Because the effect of the combined plasma/UVA treatment was much greater than the effects of the plasma treatment and UVA treatment added together, plasma and UVA treatment of aqueous bacterial suspension interact synergistically. However, a synergistic interaction only occurred in high-power mode when the bacterial suspension was treated with plasma before UVA. Chapter 5 shows that the abundant aqueous species associated with high-power plasma treatment of buffered aqueous solution are nitrite, nitrate, and hydrogen peroxide, while low-power plasma treatment tends to produce primarily aqueous ozone. [138] Therefore, the synergistic effect between the plasma and UVA treatment seems to involve the interaction between UVA photons and nitrite, nitrate, and/or hydrogen peroxide.

The photolysis of nitrite in the presence of photons with wavelengths near 360 nm has been widely studied. [174] The absorbance maximum of nitrite in aqueous solution has been reported to be in the range of 350–365 nm, and the measured photolysis products of nitrite in the presence of water are hydroxide (OH-), nitric oxide (NO), and hydroxyl radical (OH). [175–178] In the absence of other reactants, nitric oxide and hydroxyl will readily recombine to nitrous acid (HNO2). Nitrous acid then dissociates and neutralizes hydroxide to create nitrite and water, resulting in no net nitrite hydrolysis. [176,184] However, where other reactants are present, the radicals nitric oxide and/or hydroxyl will readily react; for instance, both nitric oxide and hydroxyl can react with microorganisms or other RONS intermediates to produce antimicrobial effects. [80] In contrast, nitrate and hydrogen peroxide, which exhibit absorbance maxima near 308 nm and below 200 nm, do not strongly absorb in the UVA range. [178,185] The LED photon source emits used here most strongly in the UVA range, with an emission maximum at 369 nm. Based on the known photochemistry of nitrite at UVA wavelengths, and the antimicrobial activity of its photolysis products, it was hypothesized that nitrite was the most important species for synergizing with UVA treatment.

To test the hypothesis of nitrite photolysis accounting for the plasma/UVA synergy, bacterial suspensions with and without added nitrite were exposed to UVA photons and compared the antimicrobial effect. The same experiment was repeated with added nitrate and hydrogen peroxide, as well as with all three species added to the same suspension. As shown in Figure 6.3(b), adding nitrite, nitrate, and hydrogen peroxide in combination, then treating with UVA, produced the same antimicrobial effect as treating with plasma before treating with UVA, suggesting that one or more of those three species account for the synergistic effect between plasma and UVA treatment. Adding only nitrite, shown in Figure 6.4(a), enhanced the antimicrobial effect but not to the same extent as adding all three species or treating with plasma. Therefore, while nitrite accounted for some of the antimicrobial synergy between plasma and UVA, nitrite alone could not explain the enhancement of the UVA inactivation following plasma treatment. A possible explanation is that some secondary reaction occurs, such as between nitrite photolysis products and hydrogen peroxide; for example, nitrite and hydrogen peroxide have been shown to form the potent oxidant peroxynitrous acid (HOONO) under physiological conditions. [97] This hypothesis was tested by adding both nitrite and hydrogen peroxide, but not

73 nitrate, to the bacterial suspension before exposure to UVA. Figure 6.3(b) shows that adding nitrite and hydrogen peroxide in the absence of nitrate produced the same effect as adding all three components. Therefore, nitrate does not significantly affect the antimicrobial effect under these conditions.

The combination of nitrite and hydrogen peroxide to produce enhanced antimicrobial effects has been described previously. Heaselgrave and coworkers describe enhancing the antimicrobial effect of acidified nitrite through the addition of hydrogen peroxide. [186] Although their study considered acidified nitrite and the experiments described in this chapter were conducted in neutral buffer, the enhanced antimicrobial effect may act via a nitric oxide intermediate in both cases: NO is one of the photolysis products of nitrite and also thought to be - one of the primary antimicrobial agents in the acidified nitrite pathway. Either NO or NO2 can be oxidized to peroxynitrite (ONOO-), a strong oxidant central to innate immune chemistry. [80] Jiang and Yuan also attribute the synergistic antimicrobial effect between nitrite and hydrogen peroxide to the formation of peroxynitrite, showing that the two species in combination can inactivate bacteria in biofilms that form in water treatment systems. [187] Spence et al. argue a slightly different point, that hydrogen peroxide may enhance the antimicrobial effect of peroxynitrite, showing that catalase (an enzyme that disproportionates hydrogen peroxide to water and oxygen) confers resistance to peroxynitrite treatment in Neisseria gonorrhea. [188]

All of the above studies establish a connection between hydrogen peroxide and reactive nitrogen species for enhanced antimicrobial effects. While the studies of Heaselgrave and Jiang use considerably higher concentrations of hydrogen peroxide (~1 M) than used here (100 µM), Spence and coworkers used a similar concentration (200 µM), indicating that a relatively dilute concentration of hydrogen peroxide may be enough to achieve a significant enhancement to antimicrobial effects produced by reactive nitrogen species in solution. More recently, Machala et al. showed that ambient-condition air plasmas can create peroxynitrite in aqueous solution, and under some conditions, antibacterial effects correlate well with the amount of peroxynitrite present. [5] Although Machala’s results indicated greater concentrations of peroxynitrite in acidic solution than in neutral buffer, some peroxynitrite was produced in neutral solution, and peroxynitrite formation appeared to correlate at least partially with the concentration of nitrite and hydrogen peroxide.

To confirm that the nitrite photolysis products were the species most responsible for the enhanced antimicrobial effect, nitrite and ascorbate were added to the bacterial suspension prior to UVA exposure. Figure 6.4(a) demonstrates that, by increasing the concentration of nitrite by a factor of 10 from 5 mM to 50 mM, the inactivation was enhanced by approximately 1 log reduction. This result implies that under the conditions studied here, the enhancement to inactivation above the base antimicrobial effect of UVA alone is first-order in nitrite photochemical products.

Finally, the effect of scavenging the products of nitrite photolysis and preventing those photolysis products from inactivating bacteria was investigated. Ascorbate, or vitamin C, is an antioxidant that is thought to have an important role in preventing human disease. [189] It can scavenge various oxidizing species including hydroxyl and nitric oxide, and in at least one study, ascorbate was found to react preferentially with hydroxyl over other oxidants. When ascorbate

74 was used in combination with nitrite photolysis, Opländer et al. observed that ascorbate scavenged “NO-consuming reaction partners” including hydroxyl without reacting with NO itself to produce a pure source of nitric oxide. [190] The observed “selectivity” of ascorbate to react with hydroxyl might be explained by the short half-life of hydroxyl in aqueous solution, which is several orders of magnitude smaller than that of other oxidants, including nitric oxide, indicating that hydroxyl reacts extremely quickly in solution relative to other oxidizing species. [191] Because hydroxyl is highly cytotoxic, it was hypothesized that the addition of ascorbate to scavenge hydroxyl would reduce the antibacterial effect. As shown in Figure 6.4(b), UVA treatment with and without the addition of both equimolar nitrite and ascorbate produced approximately the same antimicrobial effect, supporting the hypothesis that hydroxyl is an important contributor to the inactivation of E. coli. Figure 6.4(c) supports this conclusion as well, showing that when ascorbate was added to the suspension prior to plasma treatment, the enhancement to the antimicrobial effect from UVA treatment was abrogated, and approximately the same log reductions were observed as with treating with UVA only.

The in situ generation of hydroxyl and nitric oxide, and the ability to control their relative abundances via antioxidants like ascorbate, has many possible biomedical applications beyond disinfection. One emerging biomedical application of atmospheric-pressure plasma is the treatment of open wounds. [31] Numerical simulations of plasma-assisted wound healing suggest that the diffusion and transport of plasma RONS into infected tissue is an important step in determining the rate of wound healing. [32] Using UVA in combination with plasma treatment could accelerate the wound healing process by delivering more reactive species to infected sites. Once plasma-generated species have diffused into the fluid covering a wound, subsequent treatment by UVA could enhance the plasma’s bactericidal effect in a manner similar to the effect observed in aqueous solution.

More generally, the interaction between plasma reactive species and photons is a potentially powerful strategy for localized delivery of specialized RONS. For example, nitric oxide is biologically significant in so many contexts that it is beyond the scope of this dissertation to describe them all, but NO has been implicated in the physiology of the human vasculature, endothelium, nervous system, and immune system, among others, as described in a comprehensive review by Moncada. [143] The application of NO via chemical donors for various therapeutic purposes has also been extensively studied. [144] Further experiments are necessary to explore other plasma/photon interactions for the creation of nitric oxide or other biologically active RONS.

6.6 Conclusion

This chapter demonstrates the synergistic interaction between UVA and ambient-air plasma treatment to produce antimicrobial effects in aqueous suspension. The synergy was only observed under particular plasma operating conditions. When an aqueous suspension of E. coli was treated with high-power air plasma for 5 minutes followed by UVA for 60 seconds, the antimicrobial effect was at least a 4.5 log reduction in bacterial load. Under the same conditions, the expected additive effect of plasma treatment alone plus UVA treatment alone was a 2-log reduction in load. In contrast, this chapter has shown that under other conditions, such as treating

75 with UVA first or using low-power plasma, the combined antimicrobial effect matched the expected additive effect.

When nitrite and hydrogen peroxide were added to the bacterial suspension instead of pre-treating with high-power plasma, the same bacterial inactivation was observed as when the suspension was treated with plasma. The further addition of nitrate had no effect on the antimicrobial activity. Adding only nitrite produced a lesser synergistic effect, and the antimicrobial activity increased with increasing concentration of nitrite. However, when the antioxidant ascorbate was added, no enhancement of the antimicrobial effect was observed above that of UVA treatment alone.

Based on known photochemistry and the strong absorbance of nitrite at 369 nm, UVA photolysis of nitrite is a likely explanation for the observed synergy between UVA and plasma treatment. However, other reactions besides UVA-driven nitrite photolysis could potentially account for the synergistic antimicrobial effect between UVA photons and plasma-created reactive species. For example, UVA could photo-activate biomolecules on the E. coli outer membrane, which then react with nitrite and hydrogen peroxide. Alternatively, UVA photons could penetrate into E. coli and create reactive species inside the cell, altering the cellular response to nearby nitrite and hydrogen peroxide. Further experiments are necessary to unequivocally establish the precise photochemical and biochemical pathways involved with synergistic UVA and plasma inactivation of bacteria.

Assuming that UVA photolysis of nitrite is a major contributor to the observed synergy, the following mechanism is proposed for the synergistic interaction between high-power plasma and UVA treatments of E. coli suspended in aqueous solution. First, plasma operating in high- power mode creates gaseous hydrogen peroxide (H2O2) and nitrogen oxides, including nitrogen dioxide (NO2) and nitric acid (HNO3). When these species diffuse into water, they form aqueous - - hydrogen peroxide, nitrite (NO2 ) and nitrate (NO3 ) anions. If the aqueous solution is buffered, as it was here, nitrite and nitrate have little or no inherent antimicrobial activity alone. However, when aqueous nitrite is exposed to UVA light in the range of 360 nm, it photolyzes to form the antimicrobial species nitric oxide (NO) and hydroxyl (OH). Under these conditions, hydroxyl is likely responsible for the majority of the antimicrobial effect. The photolysis of nitrite in the 5 mM concentration produced by plasma treatment cannot account for the entire antimicrobial effect, so it is possible that “side reactions” of hydroxyl or nitric oxide produce further bactericidal species. The most likely secondary reaction is hydrogen peroxide reacting with nitrite or nitric oxide to form the strong oxidant and antimicrobial agent peroxynitrite (ONOO-). The proposed reaction pathways are shown in Figure 6.5.

Figure 6.5. Summary of reaction pathways in the antimicrobial synergy between air plasma species and UVA photons. (1) Air plasma forms RONS including nitrogen dioxide (NO2) and nitric oxide (NO). NO2 and NO, through a series of reactions, dissolve in water and hydrolyze to - yield nitrite (NO2 ). (2) In aqueous solution, nitrite is photolyzed at 369 nm to yield nitric oxide (NO), hydroxyl (OH), and hydroxide (OH-). [174–178] (3) When the 369 nm LED is turned off, in the absence of other reactants, NO and OH recombine into nitrous acid (HNO2). Nitrous acid dissociates at neutral pH to nitrite and hydrogen ion (H+), and H+ and OH- recombine to water (H2O). [97,176] (4) In the presence of E. coli, OH damages biomolecules to inactivate the bacteria. (5) In the presence of both E. coli and ascorbate (a strong antioxidant), [5] OH preferentially reacts with ascorbate, diminishing the antimicrobial effect. (6) Nitrite alone does not account for the full synergy between plasma and UVA, but the combination of nitrite and hydrogen peroxide (H2O2) does. Hydrogen peroxide is also produced in the plasma and dissolves into the aqueous phase, where it may react with NO to form another antimicrobial compound like peroxynitrite (ONOO-). [80,186,187]

The interaction of UVA photons with plasma-generated chemical species has the potential to increase the speed and efficacy of both ambient-plasma disinfection and UV-based disinfection methods. Furthermore, these results suggest the possibility of a wider application of ambient-condition plasma chemistry coupled with photochemistry to produce unique chemical and biological effects.

Chapter 7: Conclusion

7.1 Abstract

This work has studied the production of biologically active chemistry produced by ambient-condition air plasmas and has aimed to relate that chemistry to antimicrobial effects. Both gas-phase and liquid-phase chemistry were studied and related to inactivation of E. coli on surfaces and in aqueous suspension. This chapter summarizes those results and describes an outlook for future study in the field.

7.2 Gas-Phase Chemistry and Surface Disinfection

Chapters 3 and 4 describe the gaseous chemistry created when air is treated by surface micro-discharge and the related antimicrobial effects. Plasma processing parameters such as waveform voltage and frequency do not appear to significantly alter the plasma’s antibacterial effect. In addition, both “direct” (i.e., floating-electrode) and “indirect” (i.e., surface micro- discharge) mode result in similar inactivation of bacteria. Treatment time and discharge power density, on the other hand, are significantly correlated to both chemistry and antimicrobial effect.

Gaseous air plasma chemistry displays two distinct “regimes” that agree with earlier models of ambient plasma discharges: an ozone-dominated regime at low power densities and/or short treatment times, and a nitrogen oxides-dominated regime (NOx, or NO and NO2) at high power densities and long treatment times. An unstable “transition region” exists at intermediate power densities and treatment times, and it corresponds to a gradual replacement of ozone by nitrogen oxides in the gas phase. Appendix B extends the study to chemistry produced by a different method, spark discharge, which operates exclusively in the nitrogen oxides-rich mode. NOx-rich chemistry is responsible for the most efficient inactivation of bacteria on dry surfaces.

7.3 Liquid-Phase Chemistry and Aqueous Disinfection

Chapter 5 and Appendix A show the relationship between plasma chemistry, solution chemistry, and antibacterial effects in solution. Aqueous chemistry corresponds, at least qualitatively, with gaseous chemistry, and the same chemical regimes are present in the aqueous phase as in the gaseous phase. When NOx dissolve in the aqueous phase, they produce nitrite - - (NO2 ) and nitrate (NO3 ), and acidify non-buffered neutral solution in the process. NOx are more soluble in the aqueous phase than ozone, so dissolving ozone in water tends to require more agitation or careful mixing than dissolving NOx. In following the post-exposure concentration profiles of aqueous chemistry, ozone persists on the order of minutes, nitrate remains stable for at least one week, and nitrite persists on the order of hours in acidic medium and for at least one week in neutral buffer.

When treating solution with NOx-rich chemistry, antibacterial effects are highly sensitive to solution pH. In neutral buffer, treating a solution with plasma does not acidify the solution,

78 and the antibacterial effect is greatly diminished. However, treating a non-buffered solution with plasma results in a persistently antimicrobial solution that retains some antibacterial activity for at least one week following exposure. In contrast, when using ozone as the primary antibacterial agent, the bactericidal effect is not related to the solution’s buffer chemistry, and ozone decontaminates both water and neutral buffer after a few minutes of treatment.

Chapter 6 demonstrates the interaction between plasma-derived species and UVA photons. UVA apparently photolyzes nitrite to yield more reactive species that are responsible for rapid antibacterial effects. The plasma-UVA synergy likely proceeds via an oxidant-mediated mechanism because the addition of an antioxidant to solution abrogates the observed synergy. Like treating with ozone, treating with plasma and UVA in combination is not sensitive to solution pH or buffer chemistry.

7.4 Outlook

The work presented in this dissertation contributes to both the basic understanding of chemistry produced by ambient-condition plasmas and some potential applications of plasma disinfection. Although important new aspects of the relationship between plasma processing conditions, chemical effects, and biological effects have been elucidated in this work, many fundamental questions remain about the mechanisms responsible for the observed biological effects produced by plasma-generated species. Since plasma disinfection inevitably involves reactive species and bacterial biomolecules, and at least some of these reactions are similar or even identical to other biological systems (e.g. mammalian biochemistry), it is anticipated that the results presented here will be relevant to the wider field of plasma medicine. Further work will be necessary to better understand the nature of those reactions.

In addition, plasma interaction with photons is a promising field of study only briefly described here. Both plasma and energetic photons represent methods of creating reactive species from air, and the combination of the two could be a strategy to create novel chemistry on demand and at the point of use. Finally, the results described here are relevant not only to plasma disinfection and plasma medicine but to the greater field of plasma biotechnology. As plasma biomedicine devices continue to mature in the laboratory and move toward clinical and commercial applications, it is essential to characterize such devices chemically to ensure reproducible and consistent operation.

Appendix A: Long-term antibacterial efficacy of air plasma-activated water

Originally published in Journal of Physics D: Applied Physics, DOI 10.1088/0022- 3727/44/47/472001, courtesy of IOP Publishing

A.1 Abstract

Indirect air dielectric barrier discharge (DBD) in close proximity to water creates an acidified, nitrogen-oxide containing solution known as plasma-activated water (PAW), which remains antibacterial for several days. Suspensions of E. coli were exposed to PAW for either 15 minutes or 3 hours over a 7-day period after PAW generation. Both exposure times yielded initial antibacterial activity corresponding to a ~5-log reduction in cell viability, which decreased at differing rates over 7 days to negligible activity and a 2.4-log reduction for 15-min and 3-hr exposures, respectively. The solution remained at pH ~2.7 for this period and initially included hydrogen peroxide, nitrate, and nitrite anions. The solution composition varied significantly over this time, with hydrogen peroxide and nitrite diminishing within a few days, during which the antibacterial efficacy of 15-min exposures decreased significantly, while that of 3-hr exposures produced a 5-log reduction or more. These results highlight the complexity of PAW solutions where multiple chemical components exert varying biological effects on differing time scales.

A.2 Introduction

It was hypothesized that long-lived secondary products, such as H2O2, nitrite, and nitrate, would likely be responsible for the extended biological effects of plasma-activated water (PAW) after plasma treatment. This chapter reports measurements of the correlation between the manner in which PAW is created via indirect air DBD, the solution composition, storage conditions, duration of bacterial exposure, and antibacterial effectiveness for a period of up to 7 days after creation.

A.3 Experimental Section

Figure A.1 shows the experimental setup. 10 ml of distilled water (Invitrogen, UltraPure Distilled Water) or PBS were added to a glass container of 70 mm diameter, covered by the electrodes, and treated for 20 minutes to create PAW. The gap between the ground electrode and the solution surface was 42 mm. The device was operated in the surface micro-discharge configuration, or “indirect mode,” as described in Section 2.2. Plasma was generated and characterized as described in Section 2.3. The power consumption was 5 W (0.29 W cm-2). Aqueous chemistry was measured as described in Section 2.5.

Figure A.1. Experimental setup for PAW/PAPBS generation by the indirect air DBD. Inset shows a visible image of the discharge.

E. coli K12 were as described in Section 2.6. Aliquots of the culture (100 µl) were pelleted by centrifugation at 5000 rpm for 10 minutes. The pellet was resuspended in either 1 ml of distilled water, PBS, PAW, or plasma-activated PBS (PAPBS) to determine each solution’s antimicrobial effect. [72] The suspensions were incubated for either 15 minutes or three hours. In the latter case, suspensions were vortexed every 20 minutes to ensure thorough mixing. At the end of the 15-min or 3-hr incubation, the suspensions were immediately diluted in PBS. Aqueous disinfection was measured as described in Section 5.6. The maximum log reduction detectable using this approach was approximately 5.5. In some cases discussed below, the actual log reductions could have been higher than the reported values.

A.4 Results and Discussion

The antimicrobial actions of PAW were examined over periods of several days along with the composition and antimicrobial activity over the same time scale. The solution pH, composition, and antimicrobial effects were also measured just after plasma exposure, and these were found to be similar to results reported by Oehmigen et al. [73] Oehmigen et al. used normal saline and PBS-buffered saline solutions whereas these experiments began with distilled water. No significant changes in antibacterial efficacy were observed in comparisons of PAW generated from distilled water or saline solutions. Additionally, no inactivation was observed when cells were incubated with untreated distilled water (data not shown). Thus, ionic strength effects were not significant in this work.

Figure A.2 shows the antibacterial efficacy when cells were exposed to PAW for 3 hours (A.2a) and 15 minutes (A.2b). For all solutions, the pH was constant over the experiment at 2.7 ± 0.2 for PAW and 7.0 ± 0.1 for PAPBS. The pH of 2.7 is close to the reported pKa 2.8-3.2 for 81 nitrous acid. [192] Little bacterial inactivation was observed when E. coli was treated with all PAPBS solutions, while a ~5-log or greater reduction in bacterial viability was measured in both 15-min and 3-hr exposures with PAW directly after plasma treatment. Chapters 5 and 6 describe the treatment of E. coli suspended in solution and show similar results. When using plasma in a higher-power mode (0.30 W cm-2) and relatively long exposure times, the antibacterial effect is greatly diminished treating buffered solution compared to water.

The antibacterial efficacy of 3-hr exposures remained at or above 5 logs for 2 days then decreased to 2.4 logs after 7 days, while the efficacy of 15-min exposures dropped from 5.6 logs to 2.4 logs in 30 minutes, remained at about 1 log at the end of day 1 and day 2, then yielded no antibacterial activity after 7 days. The antibacterial assay detection limit is within the error bars of the measured value for the first 2 days of the 3-hr exposures; thus, the exact rate at which the antimicrobial efficacy declines over this time is unclear. As noted above, it is possible that the initial reduction following 3-hr incubation is greater than 5 logs.

Figure A.2. Time-dependence of antibacterial activity of PAW incubated with E. coli for (a) 3 hours, (b) 15 minutes, and (c) of PAPBS incubated for 3 hours after indirect air DBD treatment.

Ultraviolet spectra, resulting nitrate and nitrite concentrations, and hydrogen peroxide concentrations are shown in Figure A.3. UV absorbance data were acquired for each plasma treated solution to quantify nitrite and nitrate concentrations as described above. The absorbance scans of PAPBS, and thus the concentrations of nitrite and nitrate, were effectively constant over 7 days; however, the UV absorbance of PAW varied greatly (A.3a). Nitrite levels in PAW decreased from 1.2 mM to 2 M within two days, while the nitrate concentration increased quickly from 1.2 mM to 3 mM then increased more slowly over the following 4 days (A.3f). PAPBS exhibited an initial concentration of approximately 200 M H2O2 that did not vary significantly over the observation period (A.3e). PAW solutions yielded an initial concentration of 100 M, which dropped below the detection limit (~ 1-5 M) after 2 days.

Figure A.3. UV absorbance of (a) PAW and (c) PABS over 7 days after indirect air DBD treatment with representative spectra of the nitrite peak (red), nitrate peak (green), and the peak centered at 262 nm (blue) for (b) PAW and (c) PAPBS. Concentration of (e) H2O2 in PAW (filled squares) and PAPBS (open squares) and the calculated concentration of (f) nitrite in PAW (filled squares) and PAPBS (open squares), of nitrate in PAW (filled circles) and PAPBS (open circles) as a function of time after indirect air DBD treatment.

H2O2 is known to be involved in the antimicrobial properties some of plasma treated solutions. [7,93] Previous investigations seeking to define the antimicrobial contributions of species in PAW have attributed different potencies to H2O2. Burlica et al. found that approximately 3 mM H2O2 contributed a 2-log reduction in bacterial colony formation [7], while Naitali et al. found that 10 μM acidified H2O2 yielded 0.4-log reduction. [93] Under these conditions, the initial hydrogen peroxide concentration was an intermediate concentration of approximately 100 μM, which decreased to below the detection limit (~1-5 μM) over 2 days.

As noted above, the antibacterial activity of acidified nitrite is well known and is presumably due to the decomposition of nitrous acid to generate nitric oxide. [94] Several groups have studied the contribution of acidified nitrite to the antimicrobial capacity of PAW. Naitali et al. attributed a ~3-log reduction to acidified nitrite and a 5-log reduction to the synergistic effect of acidified nitrite with H2O2 and nitrate. [93] This synergy presumably arises from the formation of reactive species such as peroxynitrous acid, peroxynitrite, and related nitrogen oxide products. The initial concentration of 1.2 mM nitrite in PAW decreased quickly over 2 days to 2 M, during which the antibacterial efficacy of PAW was at or above 5 logs in 3-hr exposures, but dropped significantly in 15-min exposures. It should be noted that nitrogen oxide solution chemistry is far from fully determined. The complexity of this chemistry is evident from a relatively recent review on peroxynitrite by Goldstein et al. [193]

Kamgang-Youbi et al. investigated the antibacterial activity of water after treatment with a gliding arc. [71] These researchers found a substantial reduction in viable H. alvei cells upon exposure to PAW, and this activity persisted for at least 24 hours. Additionally, Oehmigen et al. found that plasma treated saline solutions aged for 30 minutes were still bactericidal when exposed to cells for 15 minutes. [73] However, these authors found that the antibacterial efficacy was greatly reduced after 30 minutes. Here, similar behavior was observed in which 15-min exposures yielded a drop in efficacy from 5.6 to 2.4 logs following a 30-minute delay between creation of PAW and incubation with bacteria.

A peak was observed at 262 nm in the PAW samples but not PAPBS (Figure A.3a). A Gaussian function fit well to this peak and the calculated peak height increased linearly over the experiment. The peak at 262 nm was only present when PAW samples were stored in sealed tubes and did not appear in samples stored in open tubes, suggesting the involvement of a gas phase species diffusing either into or out of the PAW solution. No difference was observed, however, in the antibacterial capabilities of PAW samples stored in open or closed tubes (data not shown). Many organic nitro molecules exhibit UV absorbance maxima around 260 nm. [194] Thus, the nitration of an organic contaminant could account for this peak. A similar peak was observed previously in PAW. [195]

A.5 Conclusion

In summary, the antimicrobial actions of PAW were monitored over periods of several days. The H2O2 and nitrite levels correlated with bacterial log reductions from 15-min exposures with cells, but not from 3-hr exposures, which yielded higher and sustained antibacterial efficacy that persisted for at least 7 days. These results highlight the complexity of PAW solutions where multiple chemical components exert varying biological effects on differing time scales. Further study will more completely elucidate the nature and time-dependent post-treatment concentration profile of other species in PAW.

Appendix B: Air Spark-like Plasma for Frugal Antimicrobial NOx Generation

B.1 Abstract

The generation of nitrogen oxides is demonstrated and their antimicrobial efficacy is analyzed using atmospheric air spark-like plasmas in simple, inexpensive devices. Spark-like discharges in air in a 1 L confined volume are shown to generate NOx at an initial rate of about 16 2.2 x 10 NOx molecules/J dissipated in the plasma. An inexpensive power supply dissipating about 12 W in this confined volume generates on the order of 3000 ppm NOx in ten minutes. Around 90% of the NOx is in the form of NO2 after several minutes of operation in the confined volume, suggesting that NO2 is the dominant antimicrobial component. The strong antimicrobial action of the NOx mixture after several minutes of plasma operation is demonstrated by measuring rates of E. coli disinfection on surfaces and in water exposed to the NOx mixture. The spark-like discharge systems generating these species can operate with inexpensive power supplies, simple automotive spark plugs, and relatively small sources of electricity. This application is referred to as frugal plasma, suggesting this is an example of frugal innovation. Power could be provided by compact solar panel systems to recharge modest-sized batteries that would in turn power the plasma devices. Some possible applications of frugal plasma generation of NOx (perhaps followed by dissolution in water) include disinfection of surfaces, skin or wound antisepsis, and sterilization of medical instruments at or near room temperature. This is especially promising for circumstances in which conventional sterilization, disinfectant, or antiseptic supplies are not available, such as in emergency conditions, refugee camps, or isolated, low-resource settings in general. The chapter concludes with a vision and several cautions for how this idea might develop in the future.

B.1 Introduction

B.1.1 Frugal engineering

The primary focus of this chapter is to describe a class of simple, inexpensive plasma devices operating in air, sometimes using small amounts of water, which can be exploited for applications such as biocides, disinfectants, sterilization, and antisepsis, among others. First, this chapter presents measurements of gas- and aqueous-phase species created in air spark-like plasmas powered with simple, low-cost power supplies. The key antimicrobial species created in the air spark-like plasmas are shown to be primarily nitrogen dioxide (NO2), as well as nitric oxide (NO). The literature on NO and NO2 is extensive and certainly cannot be comprehensively summarized here. [196] Briefly, NO is one of the physiologically important classes of molecules referred to as “reactive nitrogen species” (RNS). [26,145] NO2 is an important air pollutant and, - along with NO, plays a key role in atmospheric chemistry. [197] The nitrite anion (NO2 ), especially under acidic conditions in aqueous solution, has received considerable attention in the last several decades for its direct and indirect physiological roles, connection to NO in the body, therapeutic actions, and use as a disinfectant, among other reasons. [94,198] It is known that NOx generation is favored by higher temperature air plasmas, so their generation was explored using

85 air spark-like discharges. [199] Furthermore, these species have been characterized in disinfecting adjacent surfaces and in forming a strongly antimicrobial aqueous solution.

The importance of NO generated in air plasma for wound sterilization and healing and related plasma therapeutic applications has been noted by others, but the potentially key role of NO2 seems not to have been strongly emphasized in the previous air plasma literature. [200,201] Kim et al. reported the use of a microwave plasma operating at atmospheric pressure in flowing O2/N2 mixtures to create NOx, mostly in the form of NO2. [202] These authors also summarize results from various papers reporting that the production rate of NOx molecules from spark 16 17 discharges typically ranges from the order of 10 – 10 NOx molecules generated/J dissipated in the spark. This figure is referred to later in this chapter to show that this is close to the chemistry measured here.

Although it has apparently not been presented extensively in the scientific literature, recent commercial applications of gaseous NO2 (generated by vaporizing liquid NO2) as a room temperature sterilization gas has clear relevance to the ideas and results presented in this chapter. [203] The concentrations apparently used in the commercial applications of NO2 as a sterilization agent are also similar to those measured and reported here.

The simplicity and relatively low cost of the class of plasma device presented here imply that such devices can be thought of as an example of “frugal technology.” This novel perspective might open up new applications for atmospheric air plasmas. The idea of making a technology simpler to render it inexpensive and potentially more robust while still meeting important social and environmental needs has been referred to as “frugal innovation” or “frugal engineering.” [204–207] In some ways, frugal innovation is a manifestation and extension of earlier 20th century ideas about “appropriate technology.” [208] The frugal innovation concept has historical precedents that date back at least to the 18th century in Europe and North America. [205] Rao summarizes in a recent review the disruptive potential of frugal innovation in corporate settings. [204] In the arena of non-corporate developments, the “Jaipur foot” is often cited. [209] This product is a robust and inexpensive prosthetic foot made from easily available materials in India. Indeed, India and other parts of South Asia are the regions most commonly associated with the application of frugal innovation, despite its possible origins elsewhere. For example, one guiding principle in frugal innovation and engineering is the Indian/Pakistani concept of Jugaad, defined as “a creative or innovative idea providing a quick, alternative way of solving or fixing a problem.” [210] Similar interests have been observed and expressed for medical devices and technology. [211–213] Ballati et al. and Nocera address current and future energy needs in the context of frugal innovation. [214,215] Water treatment using various technologies, including low cost methods that qualify as “frugal,” was reviewed by Gadgil. [216]

Whitesides observes that there is also a science component to the development of frugal technology. [217] This important point is addressed in greater detail below, because one purpose of this chapter is to help clarify the science behind atmospheric air plasma technology used for frugal innovation. More generally, reliable, scientifically based measurements and/or calculations are needed to validate and understand frugal technology. Scientific understanding is essential in order to ensure the technology works safely and reliably in the way that is intended and that its operation is reasonably well understood and properly controlled under realistic

86 conditions. For example, in the context of frugal air plasma, identifying the key kinetics of the chemical species created in the plasma, as well as a proper understanding of the electrical power supply and chemical species delivery and disposal systems, are all needed for proper and safe use of the technology. A deeper understanding of the science behind the technology may also suggest additional novel applications, and several examples are described later in the chapter in the context of frugal air plasmas. Of course, any viable and successful technology must meet not only minimal technical specifications but must also satisfy numerous other constraints, including those related to cultural and local economic issues, among others. Ideally, the technology must also be scalable so that significant numbers of people are served. These issues are not addressed here, but they will be an essential part of any successful application in practice.

This chapter suggests that atmospheric pressure air plasma technology fits naturally into the context of frugal innovation. The chapter describes an example of a simple, low-cost, and robust device that utilizes electrical discharges in ambient air to create various potentially useful chemical species such nitrogen oxides, which in some cases are mixed into water. After briefly summarizing the state of understanding of antimicrobial air plasma and plasma-activated water, the devices, power supplies, and electrode designs are described, along with their basic discharge physics and chemistry. The effects of several low-cost, high-voltage power supplies are compared with those of a commercial laboratory power supply. The energy efficiency of the devices is highlighted, as well as the simplicity, robustness, and low cost of the devices and materials used. FTIR and UV absorption spectroscopy are used to characterize gas phase composition from the plasma devices, and liquid phase UV-Vis absorption measurements are - - made for nitrate (NO3 ), and nitrite (NO2 ). Next, these species are characterized for inactivating E. coli bacteria either dried on metal surfaces or dissolved in water. The crucially important issue of “frugal plasma” device safety and reliability is addressed as well. The chapter ends with some envisioned applications and a vision of how this field might develop in the future.

B.1.2 Antimicrobial air plasma and plasma-activated water

The recent literature on atmospheric plasma applications in surface and water disinfection is growing rapidly and the fact that air plasmas generate antimicrobial conditions and chemical species is now well known. [218,219] Air plasma using dielectric barrier discharges (DBDs), gliding arcs, and similar devices have been shown by various groups to generate strongly antimicrobial components, capable of rapidly disinfecting adjacent surfaces and liquids. [5,60,65,220] If operated near water, these discharges can create what has been termed “plasma- activated water.” Chapters 1 and 5, and Appendix A, provide more detail about the aqueous chemistry of water treated by air plasma.

For air DBD plasmas, it is known that lower gas temperatures (< 350K) and lower -2 electrical power deposition (< 0.5 W cm ), as well as relatively dry air, favor the formation of O3 over nitrogen oxides and nitric acid. [13,16,87] Generating O3 from air plasma (or pure O2) via dielectric barrier or corona discharges has been exploited commercially for many years, and there is a correspondingly large literature, especially for water disinfection. [199,221] O3 dissolved in water is one of the strongest antimicrobial agents known. Under conditions in which O3 is formed in air DBD plasmas and dissolved in water, it can in some cases be responsible for nearly all of the rapid antimicrobial effects in plasma-activated water. [138] Recently, Kim et al.

87 reported a breakthrough in the design of ozone-generating atmospheric pressure micro-plasma devices. [222] Some of the ideas in this chapter regarding frugal plasma could certainly be applied to novel ozone-generating plasma devices, but this topic is left to future study.

B.2 Experimental Section

B.2.1 NOx Box with spark-like discharge

An acrylic chamber measuring 10x10x10 cm (or 1 L in volume) with an automobile spark plug (Bosch Platinum 4417, costing approximately $6) mounted on the top, and two infrared-transparent windows on parallel sides, was used to produce and measure the chemical species produced by the spark-like plasma, consisting primarily of nitrogen oxides (NOx). A schematic of the chamber is presented in Figure B.1. This chamber is called the “NOx Box”. The spark plug was powered in turn by each of three different power supplies described below. The conventional spark plug was slightly modified to increase the inter-electrode gap distance to 5 mm. This increase in spark length significantly increased the rate of NOx generation, presumably by increasing the gas temperature in the spark. A photograph of a typical spark-like discharge is shown in Figure B.2. The relatively high-temperature plasma created in the repetitive spark-like discharge resulted in significant production of NOx that diffused throughout the box volume over periods of minutes to tens of minutes. Metal discs with E. coli bacteria dried on the surface, or vials containing bacterial suspensions in water, were placed inside the box and treated for specific times. Corresponding sets of the time evolution of gaseous NOx concentrations were measured via in-situ, real-time FTIR analysis.

Figure B.1. Sketch of the 1-liter volume “NOx Box” showing the acrylic body, grounded aluminum top, conventional automotive spark plug, and power supply. FTIR absorption was used to measure gas phase species evolving during operation and afterglow of the spark-like plasma. Three different power supplies were tested and compared.

Figure B.2. Photo of the spark-like discharge created using a spark plug altered to allow the spark to form between the center powered electrode and the grounded posts. The spark length is about 5 mm (photo courtesy of Steve Graves).

B.2.2 High-voltage power supplies

Three types of high-voltage power supplies were used in this study. The voltage was measured in all configurations by a high-voltage probe (Tektronix P6015A) and a 200 MHz digital oscilloscope (Tektronix 2024C). First, a commercial high voltage power supply, consisting of a Trek model 10/40A high voltage amplifier combined with a Protek 31 MHz synthesized function generator 9301, was used. The cost of this commercial power supply setup was approximately $12,000. This power supply generates a well-defined and stable sinusoidal AC waveform with 6 kV peak-to-peak amplitude and 5 kHz frequency. Typical operation of the NOx Box spark plug with the laboratory power supply using a sinusoidal AC waveform resulted in a power input to the spark of about 35 W. This was calculated by multiplying the directly measured electrode voltage with the indirectly measured current (as a voltage drop across a 1.25 Ω resistor) resulting in a net power efficiency based on power drawn from the grid of around 5%.

In addition, two “frugal” power supplies were tested that cost much less than the power supply described above: a commercial transformer designed to power neon signs and a homemade flyback transformer. The schematic for the neon sign power supply, and for the homemade flyback transformer power supply, coupled with a commercial spark plug coil and connected to the NOx Box, is illustrated in Figure B.3. Both of the “frugal” power supplies are illustrated here operating from a 12V battery; operating from a battery requires a DC/AC inverter (costing around $25) for the neon sign transformer.

Figure B.3. Top: sketch of the NOx Box using the neon sign transformer to power the spark plug. Bottom: corresponding sketch using the flyback transformer and ignition coil to power the spark plug. The “frugal” power supplies are shown operating from a 12V battery.

The term “spark-like” plasma is used because a spark discharge is generally considered to be a fully transient discharge. [223] In the discharges created here, however, there appears to be significant quasi-glow like character. This can be seen most clearly in the voltage characteristic of the pulsed DC flyback circuit, illustrated in Figure B.4, in which a relatively stable DC current occurs over a large part of each discharge cycle after a well-defined spark breakdown. This spark-to-glow transition has been studied previously [224,225], and the DC atmospheric- pressure air glow discharges have been characterized [79,226,227]. Preliminary tests have shown that it is the quasi-glow part of the discharge cycle that leads to the most significant gas heating and therefore the highest NOx production.

Figure B.4. Voltage and current characteristic for the flyback transformer/coil circuit powering the spark plug, showing the point of spark breakdown with excessive current pulse at about 4000 V. The long period of relatively stable DC voltage occurs at 900 V with a stable current of about 40 mA corresponding to the glow discharge mode.

The neon sign transformer (CPI - 10035DM, CPI Advanced, Chino, CA, costing approximately $80) and the accompanying DC/AC inverter ($25) used to allow it to be powered using a 12V battery cost a total of about $105. It is possible that less expensive versions, including those that can operate directly from a 12V battery, could also be used. This transformer operates with two counter-phase (“push-pull”) quasi-AC outputs providing up to 10 kV at approximately 23.5 kHz frequency, with a maximum current of about 35 mA. The neon sign transformer’s power input was about 16 W, and the power dissipated in the spark was about 12 W. If the inverter efficiency is included, a net power efficiency of about 60% was observed for this system. For the neon sign transformer, one of the two powered leads was connected to the center plug electrode, which is powered during the conventional operation of the spark plug, and the other powered lead was connected to ground. Another version of the NOx Box uses two spark plugs on the top; in this case, both spark plugs are powered by the two powered leads of the neon sign transformer, with a common ground.

The flyback transformer circuit (constructed for a cost of about $15) was combined with a commercial electronic ignition coil (MSD blaster 3, costing $15) for a total of around $30. As with the neon sign transformer, this flyback circuit/coil combination choice is probably not optimized for producing NOx, and other versions could work as well, perhaps at even lower cost. The flyback transformer/coil circuit provided a pulsed high voltage DC with an amplitude of 30 kV (although breakdown occurred at lower voltages – typically about 6 kV, as seen in Figure B.4), and the pulse frequency was in the range of 50-600 Hz. The typical plasma power was about 6 W with the flyback transformer in the pulsed DC mode, and the net power efficiency was about 50%.

B.2.3 FTIR spectroscopy

Fourier transform infrared (FTIR) spectroscopy was performed as an in-situ gas product diagnostic in the NOx Box, as described in Section 2.4. For NO, the relationship between absorbance and concentration remained linear over the range tested. For NO2, significant nonlinearity was observed in the absorbance/concentration relationship above concentrations of 1000 ppm when using the band at 1630 cm-1. The phenomenon of non-linearity in gas-phase FTIR spectroscopy at high gaseous concentrations is known and has been discussed in the literature. [117] Therefore, the gaseous NO2 concentrations determined via FTIR were confirmed with visible absorption at 404 nm. NO2 concentrations determined by visible absorption showed good agreement with concentrations determined by the 2916 cm-1 vibrational band in FTIR, so -1 all NO2 concentrations reported were determined by fitting FTIR absorbance at 2916 cm .

B.2.4 Aqueous chemistry measurements and antibacterial evaluation

Aqueous chemistry was measured as described in Section 2.5. Surface and aqueous disinfection were measured as described in Section 2.6. The maximum log reduction was approximately 4 for surface disinfection and 5 for aqueous disinfection. In some of the cases reported below, the actual log reduction likely exceeded this measurable limit of 4 or 5 log reductions.

B.3 Results and Discussion

Nitrogen oxides were generated by all three types of spark-like discharges in the closed 1-L volume of the NOx Box. NOx production was measured over the duration of the discharge operation, typically 1-20 min, and during the post-discharge period, up to 1 hour. The gas-phase concentrations of nitric oxide (NO) and nitrogen dioxide (NO2) were measured as functions of time by FTIR. A typical FTIR spectrum is shown in Figure B.5, showing the presence of primarily NO2, NO, N2O4, and HNO3.

Figure B.5. FTIR spectrum showing the temporal evolution of NO, NO2, N2O4, and HNO3 in the line-of-sight of the NOx Box powered by the neon sign transformer (12 W).

Figure B.6 shows plots of the spatially averaged NO and NO2 concentrations vs. elapsed time of spark-like discharge operation, using the laboratory power supply, the neon sign transformer power supply, and the homemade flyback transformer. Each of these power supplies delivers different amounts of power into the spark, so Figure B.7 plots the NOx concentrations normalized by power dissipated in the spark. The voltage and current waveforms from the different power supplies have different effects on the kinetics of NOx production, and further study is necessary to deconvolute the gas heating and natural convection phenomena corresponding to each case.

Figure B.6. Temporal evolution of gas-phase (a) NO and (b) NO2 concentrations in the NOx Box powered by the laboratory power supply (35 W), the neon sign transformer (12 W), and the flyback transformer (6 W).

Figure B.7. Power-normalized temporal evolution of gas-phase NO (a) and (b) NO2 concentrations in the NOx Box powered by the three power supplies under the conditions of Figure B.6. The flyback transformer produces the most NO and NO2 per unit power dissipated in the spark-like discharge.

The time derivative of NOx concentration in the volume is proportional to the net rate of NOx generation. The initial (t=0) rate of generation can be compared to previous estimates of

NOx production rates in terms of NOx/J dissipated, as noted in the Introduction. Using the NO2 values in Figure B.7, an estimate of the initial generation rate using the flyback transformer is about 267 ppm/W/5 minutes, or equivalently about 0.89 ppm/J. One part per million corresponds to about 2.5 x 1016 molecules in 1 L, assuming room temperature and atmospheric pressure in the 16 box. Therefore, 0.89 ppm/J corresponds to about 2.2 x 10 NOx molecules created/J dissipated. 16 2.2 x 10 NOx /J is in the range reported in the literature, as noted previously. The decrease in net rate of generation that is observed as time increases is likely related to various gas phase and perhaps surface loss processes, the detailed analysis of which will require further study.

An important point from Figure B.6 is that the gas phase composition is dominated by NO2 by more than an order of magnitude over NO at longer times. However, it appears that within the first minute or two, NO concentration may be comparable to NO2 concentration. NO is known to react with O2 to form NO2 and that NO2 will react with itself to dimerize into N2O4; these reactions are sensitive to temperature and pressure The plots in Figure B.6 can be rationalized in part by taking these reactions into account. However, analyzing the details of the NOx chemistry in the volume is beyond the scope of this study. The problem is complicated by spatial gradients in the gas phase: gas mixing due to convective diffusion will certainly be important in the first few minutes, so this is a topic that will also require considerable further study.

An important question for various potential applications is whether the results shown here can be scaled to larger volumes with more spark plugs. The most straightforward modification was to add a second spark plug to the existing box and power the two plugs with the two leads of the neon sign transformer. Figure B.8 summarizes the results.

Figure B.8. Addition of a second spark plug to the NOx Box: (a) NO and (b) NO2 concentrations vs. time using the neon sign transformer and turning the plasma off after 10 minutes. The power dissipated doubled from 16 to 32 W with the addition of the second spark plug (powered by the two ‘hot’ leads, cf. Figure B.3). The NO concentration increased by a factor of approximately 1.5, and the NO2 concentration increased by a factor of approximately 1.8. NO concentration falls faster in the afterglow than NO2; significant concentrations of both species remained in the box after 20 minutes with no plasma.

As can be seen in Figure B.8, adding a second spark plug doubled the transformer’s power consumption, and the peak NOx concentrations increased by a factor of 1.5 to 1.8. This result shows that it should be possible to increase the rate of production of NOx sufficiently to achieve rapid rates of disinfection or even sterilization in volumes significantly larger than 1 L with the same power as used here, around 30W. Reaching 2500 ppm NOx in a box five times larger should take about five times as long as what is observed here, around 15 minutes for a 5 L volume, rather than 3 minutes in the 1 L volume.

The effect of treating water with the spark-like plasma generated NOx was investigated - - by measuring aqueous concentrations of nitrite (NO2 ) and nitrate (NO3 ). Water treatment relied on dissolution and diffusion of nitrogen oxides into the aqueous phase; solutions were not mixed to promote aqueous-phase mass transfer, and the relatively small water volume of 0.15 ml minimized transport limitations. Figure B.9 shows the pH and the concentrations of aqueous nitrate and nitrite as a function of water exposure time in the NOx Box.

The antibacterial effects in solution and on surfaces were determined by exposing E. coli either suspended in physiological saline or dried on stainless steel discs to gaseous NO and NO2. Bacterial log reductions for both aqueous suspensions and dry surfaces are shown as functions of time in Figure B.10. Despite the different concentrations of reactive species and power consumptions when using different power supplies, the NOx Box achieved disinfection to

96 beyond the detection limit of the assays conducted here (i.e., 5 logs in water and 4 logs on surfaces) within 5 min of operation with all three power supplies tested.

Figure B.9. Temporal evolution of aqueous (a) nitrite and (b) nitrate concentrations and (c) pH in vials containing 150 μl of initially deionized water and placed in the NOx Box, powered by the three power supplies under the conditions of Figure B.6. Note the differences in scale between nitrite and nitrate concentrations: nitrate dominates in the water. Note also that the measured pH corresponds approximately to the nitrate concentration assuming full ionization of nitric acid.

Figure B.10. Temporal evolution of E. coli log reductions (a) in aqueous saline solution and (b) on stainless steel surfaces (right) in the NOx box powered by the three power supplies under the conditions of Figure B.6.

Although it is beyond the scope of this chapter to make a detailed comparison with previous plasma-based bacterial inactivation results, the present results are consistent with rates and degrees of bacterial inactivation reported previously. [60,69,73,74,86] Recent reports of using NO2 as a commercial room-temperature gaseous sterilizer for various surfaces (Noxilizer®) note rapid rates of bacterial spore inactivation (one log reductions in less than one minute) at NO2 concentration of about 2500-3000 ppm. [148,203] E. coli is generally much easier to inactivate than spore-forming bacteria, and maximum disinfection was achieved after the gas phase concentration of NO2 reached around 2000 ppm, suggesting that the gas mixtures created here were perhaps somewhat less potent. However, more detailed studies of the antimicrobial action of the gases created in the NOx Box are needed in order to draw definitive conclusions.

B.4 Frugal Plasma

This chapter now considers the question of how such devices might be used and in what ways they can be thought of as a kind of frugal innovation. It is clear from the results presented in this chapter that it is possible to create antimicrobial concentrations of NO and especially NO2 using relatively small amounts of electrical power obtained from 12V batteries, with inexpensive power supplies. Part of the motivation to develop and explore this idea, in part, came from the recent development and installation of portable solar energy panels coupled with 12V batteries for use in rural maternity clinics in the developing world. The We Care Solar (www.wecaresolar.org) team has been a leader in this kind of installation, especially in rural African maternity clinics. Quoting from their website:

We Care Solar designs portable, cost-effective Solar Suitcases that power critical lighting, mobile communication devices and medical devices in low resource areas without reliable electricity.

By equipping off-grid medical clinics with solar power for medical and surgical lighting, cell phones and essential medical devices, We Care Solar facilitates timely and appropriate emergency care, reducing maternal and infant morbidity and mortality, and improving the quality of care in Africa, Haiti and other regions.

This installation is capable of charging sealed 12V lead-acid batteries ranging in size from 12 to 140 Ah. A 36W (12V x 3 A) power load from the smallest battery would be able to power a neon sign transformer / 2 spark plug system for four hours. This is far more than what would be needed to fill a 5–10 L container with 2000 ppm NOx that might be used, for example, for medical instrument sterilization.

There are many other potential applications for the kind of plasma devices presented here. Plasma-activated water could be used for surface disinfection or even skin or wound antisepsis. Water treated by plasma contains high concentrations of nitrates and nitrites, and these compounds have been used for food preservation, especially for meats and fish, for many centuries. Nitrite has been extensively explored over the last 10-20 years, for various therapeutic applications, as has been summarized in recent reviews. [94,198]. Lundberg et al. note that acidified nitrite has been shown to “…have potent antibacterial activity against a range of pathogens, including Salmonella, Yersina and Shigella species, H. pylori, and Pseudomonas aerguinosa.” [94]

A key advantage of an air plasma-based system to generate these species is that one needs only electricity and air (and perhaps water). Many situations can be envisioned in which supplies are difficult or impossible to obtain, including natural disasters and various other serious social disturbances such as armed conflicts. A portable frugal plasma system could be lifesaving in this context, assuming small amounts of electrical power are available, which might be provided by solar charging of batteries or other sources.

It must be acknowledged, of course, that safety issues are important when using plasma devices that create large concentrations of antimicrobial compounds that are also potentially highly toxic. The toxic compounds must be kept away from vulnerable individuals. In the case of NO2, for example, chronic exposures can lead to various adverse health effects; the US EPA limits the 24 hour exposure limit for NO2 in outside air to 0.053 ppm. [228] Care must therefore be taken to ensure that risks for toxicity are minimized in whatever application is chosen for frugal plasma devices.

B.5 Conclusion

This chapter has shown that inexpensive, frugal plasma devices based on spark-like air plasmas can efficiently produce strongly antimicrobial gas mixtures at rates that make them interesting for a variety of potential applications. The spark-like air plasma creates NOx mixtures

100 that consist of 90% NO2 with 10% NO. Frugal power supplies based on commercial devices such as a neon sign transformer or a flyback transformer coupled with an automotive ignition coil will power automotive spark plugs to create these NOx mixtures. It appears that around 2000 ppm NO2 is sufficient for rapid surface or water disinfection. The scaling from a smaller volume with one or two spark plugs to a larger one is not necessarily linear, but a 5 l volume with two plugs (dissipating about 15 W per plug) is estimated to reach 2500 ppm NOx in about 15 minutes. This power consumption is well within the capabilities of a fully charged 12V battery of 14 Ah capacity, the smallest battery used in the field by We Care Solar. Therefore, the power loads required are quite reasonable for surface disinfection of medical instruments that might be used in a medical clinic in rural Africa. Many practical issues must be addressed before this frugal plasma technology is ready for field implementation, including extensive antimicrobial testing on realistic disinfection targets (such as medical instruments); tests of longevity of parts such as the spark plugs and other electrical components; tests of power supply safety and reliability operating on a daily basis for many months; and developing procedures for proper disposal of the toxic gases after instrument disinfection. Nevertheless, the present results show that frugal spark- like plasma devices operating in ambient air appear feasible and additional applications will likely emerge as researchers begin to consider other ways to use them. Appendix C further discusses the situation of disinfection protocols in sub-Saharan Africa and describes how plasma disinfection might be able to be used advantageously.

101

Appendix C: Disinfection in South Africa and an Outlook for Plasma Disinfection

C.1 Introduction

In fall 2013, we applied for and received a DIL Explore grant to study disinfection in the developing world. In April 2014, we traveled to Cape Town, South Africa to discuss with clinicians and other medical professionals the issues surrounding disinfection of medical instruments in low-resource settings and whether plasma technology would be a suitable sanitization solution in those settings. We visited three community health clinics located in the townships, high-poverty outlying residential areas of Cape Town; a suburban hospital; and an NGO, the Desmond Tutu HIV Foundation. This report gives a brief overview of “frugal plasma” technology, describes our findings, and concludes by discussing where plasma disinfection might be best used in the developing world.

C.2 Technical Background

Plasma, the “fourth state of matter,” is a high-energy phase similar to gas but generally more chemically reactive. One emerging application of plasma is in “plasma medicine” or “plasma biotechnology,” the use of plasma in a biological or medical context. Chapter 1 gives more information about the field of plasma biotechnology. Topics studied within plasma medicine have included disinfection, wound healing, cancer treatment, dermatology, toxin abatement, and dentistry, among others. We believe that plasma-based disinfection could be an ideal technology to use in the developing world: plasma creates a disinfectant on demand and at the point of use, requiring no raw materials besides electricity and air. Here, we focus on disinfection of medical instruments for clinics and hospitals, though plasma may also be useful to disinfect water for hand hygiene or laundry, textiles, or even food. Traditional devices used to produce plasma are expensive and require specialized components, though recent research in our lab has focused on developing low-cost and portable plasma sources, denoted “frugal plasma.” Appendix B describes in greater detail the devices proposed for frugal plasma disinfection.

Before implementing or field testing frugal plasma disinfection for low-resource healthcare, it would be necessary to better understand current disinfection practices in those settings: what instruments are used and need to be sanitized? How are they currently disinfected, and are current protocols deficient in any way? What resources are available, especially in terms of electrical power and capital for buying equipment? The goal of the trip was to answer questions like these to better inform the design of new prototypes, and, ideally, create a frugal plasma disinfecting device suitable for field testing.

C.3 Locations Visited

1) Desmond and Leah Tutu Research Clinic at Masiphumelele Community Health Center (Pokela Road, Masiphumelele, Cape Town)

102

Contact: Dr. Katherine Gill, Co-Investigator/Medical Officer (+27 (0)21 785 5486, [email protected])

103

This clinic serves the township of Masiphumelele (population around 40,000), which has an HIV and/or TB infection rate of about 40%. The clinic focuses on maternal health care and the needs of children under 5 such as vaccines. The clinic is government funded. The DTHF has been working in Masiphumelele for more than a decade. The research clinic was built in 2003 as an annex to the primary health care clinic. In addition to offering antiretroviral treatment, the clinic conducts research into TB and innovative HIV prevention methods such as microbicides. President Obama visited the HIV Foundation in his recent South Africa trip. The HIV Foundation also sponsors a nearby youth center, which also provides medical care, but focuses more on outreach and education to youth, providing a place for kids to do homework, eat, and learn job skills.

2) Victoria Hospital, Wynberg (Alphen Hill Road, Wynberg 7800, Cape Town) Contact: Nikki Fuller, Head of Anesthetics (+27 (0)82 395 9905)

104

Victoria Hospital is district/provincially aided hospital, located in Wynberg, with approximately 300 beds. Wynberg is a southern suburb of the City of Cape Town (population around 15,000). The hospital offers a variety of services ranging from surgery to x-rays. Victoria is classified as a “district” hospital, the basic tier of hospital in South Africa, providing primary and emergency care but not more specialized services, which would be provided at a larger and more advanced “regional” or “tertiary” hospital.

3) Emavundleni Research Centre, Crossroads (Sonwabile Road, New Cross Roads, Cape Town) Contact: Dr. Danielle Crida (+27 (0)82 395 9905, [email protected])

Designed to accommodate hundreds of trial participants, the center was built in 2007 to house the Desmond Tutu HIV Foundation’s first large-scale vaccine trials. With a small laboratory and a recently expanded pharmacy, it is ideally situated to conduct a variety of HIV prevention studies and psycho-social research on HIV issues. The clinic treats approximately 20 patients per day, 10 involving the use of surgical instruments.

4) Guguletu Community Health Clinic (NY3, Gugulethu, Cape Town) Contact: Dr. Katy Murie, Family Physician ([email protected])

105

106

107

Gugulethu (population 98,000) is a township 15 km from Cape Town, South Africa. This clinic is owned and entirely funded by the Western Cape Department of Health. It sees approximately 28,000 patients per month, with much heavier traffic on the weekdays, and around 3,000 patients per month in the emergency room. The clinic provides both primary care and maternity care.

C.4 Findings

C.4.1 Disinfection Protocols

Common instruments requiring disinfection at all the locations visited include scalpels, forceps, metal “kidney” trays, and speculums. Instruments are disinfected in groups or trays with dimensions not exceeding 12 x 24 x 6 inches. Most individual instruments are flat and shorter than 12 inches in length. A typical disinfection protocol for metal instruments is as follows:

1. Scrub or wash the instruments to remove solids.

2. Soak the instruments in various biocides, typically quaternary ammonium or peroxide/peracetic acid-based, for approximately 20-30 minutes (Victoria Hospital also uses a proteolytic enzymatic cleaner).

108

3. Sterilize the instruments in a machine, either a steam/boiler device (at Masiphumelele), a UV disinfector (Emavundleni, costing R12,000 or approximately 1,100 USD), or an autoclave (Victoria and Gugulethu; Victoria also has a “flash” autoclave in the operating theater in case instruments fall on the ground), for approximately 30 minutes.

109

4. Put the instruments in a sanitary container, such as autoclaved packaging or a metal dish rinsed with ethanol, and let them cool.

110

Clinicians and technicians described this procedure as generally sufficient, especially when all the equipment is working properly, but not perfect. Specific complaints and noted deficiencies included the following:

 At Masiphumelele, there is no method for indicating sterilization (i.e., that the steam cycle has completed) or maintaining sterility following the disinfection protocol.  Emavundleni faces similar issues as Masiphumelele: there is no method to verify the correct operation of the UV disinfector; furthermore, technicians expressed skepticism that the disinfector actually made the instruments more sanitary.

111

 At both locations, when the disinfection device is not working properly or is backlogged due to high demand, sometimes the only disinfection process is chemical. The quality of that process is judged empirically: no patients have gotten infected, so it appears that the process is working.  Although both Gugulethu and Victoria Hospital have autoclaves, reliability and maintenance of these autoclaves is a concern. A doctor at Victoria Hospital estimates the hospital has spent upward of 15,000 USD over the last two years on autoclave repair costs. The Gugulethu clinic has two autoclaves, but the larger unit in the trauma ward is currently broken with no plans to repair it, so the entire clinic relies on a smaller benchtop unit in the maternity ward.  Disinfection throughput and backlog is sometimes a problem. All the locations we visited have finite sets of instruments that can become depleted if there are a large volume of procedures in a given day. The Gugulethu clinic in particular sees such a large number of patients that instrument shortages are common.  Sending instruments “down the road” to a clinic with a working autoclave or disinfector is a common practice but greatly increases the time required.

For nonmetal instruments such as baby bottles or other plastics, a lower setting on the autoclave is used, where available. Facilities lacking a working autoclave plastic instruments to a tertiary hospital (Groote Schuur, for greater Cape Town) for ethylene oxide sterilization; this process takes about two or three days.

C.4.1 Access to Resources

Resource and training access is generally good. Most of the people responsible for disinfection either are nurses or have formal technical training in how to perform disinfection procedures. With respect to physical resources and consumables:

 Access to clean water is not typically an issue; clinicians at all the locations we visited were comfortable with drinking the tap water. Sporadic water shortages affected many of the locations but infrequently. At Masiphumelele, the HIV Foundation has a backup water filter, but the attached government clinic does not.  Access to electrical power is mostly reliable with sporadic outages. Occasional loss of power is seen as routine and not a serious problem; even closing the clinic due to a lack of power is seen as acceptable if not ideal. All of the facilities have backup generators, though at Gugulethu, the power provided by the generator is sometimes insufficient to power the whole clinic, and at Emavundleni, the UV disinfector is not powered by the backup generator.  It was noted that within greater Cape Town, utilities are the best and most reliable in perhaps all of Africa. In more rural parts of South Africa, and certainly elsewhere in sub- Saharan Africa, access to power and water is not nearly as reliable.  Consumables like gloves, needles, etc. are provided by the government and are kept well- stocked. No re-use of consumables was reported.

C.5 Outlook for Frugal Plasma Disinfection

112

While medical instrument disinfection is possibly less of an unsolved problem in the Cape Town area than anywhere else in sub-Saharan Africa, several opportunities exist for frugal plasma technology. In particular:

 Current disinfection protocols only apply to metal instruments. Metal instruments represent the bulk of what needs to be disinfected, though there is also call to disinfect other materials including surgical drapes (measuring 60 x 60 cm), baby bottles and other plastics, linens, among others. Clinicians found it promising that plasma can disinfect any surface, not only metal. Simultaneous plasma disinfection of instruments and material to pack them in represents an advantage over current procedures.  Plasma uses very little power for disinfection, approximately 15 W using current lab prototypes. It was expected that the low power draw would be attractive to low-resource clinics. However, all of the facilities we visited reported having more than sufficient access to power under most situations, except under outages, where generators are usually adequate. The Gugulethu clinic was the only facility that reported having limited power from generators during a power outage.  The cost advantage of plasma disinfection over other techniques is important. Currently, we are aiming for a cost of 100 USD or less per disinfector unit, an order of magnitude less expensive than UV disinfectors and a factor of at least 50 less expensive than autoclaves.  If an indicator could be developed to provide a visual confirmation that disinfection was successful, the technology would be adopted much more readily. Nurses and technicians complained that the only way they could track the progress of disinfection is by measuring time and trusting the apparatus is working correctly.  Plasma disinfection does produce some toxic byproducts. Before field testing, it will be necessary to determine a way to mitigate the toxicity. However, clinicians are already comfortable with the idea of gas sterilization (via ethylene oxide) and didn’t dismiss plasma outright simply because of the potential for toxicity.  There are some concerns over approvals and regulatory matters. Some technicians were worried that plasma disinfection had not been proven to make instruments safe for use with patients and did not want to be seen as “experimenting” on their patients with unproven technology.  Opportunity may exist for the focus of the project to be shifted from strictly low-cost disinfection to looking at a reasonable replacement for an autoclave. Clinicians were interested in the possibility of a device similar to an autoclave that was both cheaper and more robust. By increasing the target price, we could (for example) include a catalytic converter to convert potentially toxic nitrogen oxide compounds back to air.  Of the facilities we visited, the Gugulethu clinic seemed the most open to working with us again in the future. Dr. Murie was enthusiastic about our future progress. The Desmond Tutu HIV Foundation was extremely helpful in organizing the trip, and our experience underscores the necessity to work with an NGO or other organization “on the ground” in the developing world.

C.6 Summary

113

More sustainable, convenient, and low-cost disinfection continues to be a goal worth pursuing for developing economies, and plasma disinfection has potential to help solve the problem. A major advantage of plasma is that it can disinfect all types of surfaces, and plasma disinfectors can be constructed for far less money than autoclaves. But before plasma disinfection is implemented in the field, several issues need to be addressed first, primarily including abatement of potentially toxic gases, making sure devices can scale to accommodate greater quantities of instruments, and incorporating an indicator to show the progress of the disinfection process.

114

References

[1] M. A. Lieberman, A. J. Lichtenberg, Principles of Plasma Discharges and Materials Processing, John Wiley And Sons, Hoboken, NJ, 2005.

[2] W. Siemens, Poggendorffs Ann. Phys. Chem. 1857, 102.

[3] U. Kogelschatz, B. Eliasson, W. Egli, Pure Appl. Chem. 1999, 71, 1819–1828.

[4] A. Gray, Phys. Rev. (Series I) 1904, XIX, 347–368.

[5] Z. Machala, B. Tarabova, K. Hensel, E. Spetlikova, L. Sikurova, P. Lukes, Plasma Process. Polym. 2013, 10, 649–659.

[6] C. a J. van Gils, S. Hofmann, B. K. H. L. Boekema, R. Brandenburg, P. J. Bruggeman, J. Phys. D. Appl. Phys. 2013, 46, 175203.

[7] R. Burlica, R. G. Grim, K.-Y. Shih, D. Balkwill, B. R. Locke, Plasma Process. Polym. 2010, 7, 640–649.

[8] J. Julák, V. Scholtz, S. Kotúčová, O. Janoušková, Phys. Medica 2012, 28, 230–239.

[9] G. Fridman, A. D. Brooks, M. Balasubramanian, A. Fridman, A. Gutsol, V. N. Vasilets, H. Ayan, G. Friedman, Plasma Process. Polym. 2007, 4, 370–375.

[10] A. Fridman, A. Chirokov, A. Gutsol, J. Phys. D. Appl. Phys. 2005, 38, R1–R24.

[11] C. Tendero, C. Tixier, P. Tristant, J. Desmaison, P. Leprince, Spectrochim. Acta Part B At. Spectrosc. 2006, 61, 2–30.

[12] G. Morfill, T. Shimizu, B. Steffes, H.-U. Schmidt, New J. Phys. 2009, 11, 115019.

[13] U. Kogelschatz, Plasma Chem. Plasma Process. 2003, 23, 1–46.

[14] M. G. Kong, G. Kroesen, G. Morfill, T. Nosenko, T. Shimizu, J. van Dijk, J. L. Zimmermann, New J. Phys. 2009, 11, 115012.

[15] B. Eliasson, U. Kogelschatz, IEEE Trans. Plasma Sci. 1991, 19, 309–323.

[16] U. Kogelschatz, B. Eliasson, M. Hirth, Ozone Sci. Eng. 1988, 10, 367–377.

[17] K. G. Donohoe, T. Wydeven, J. Appl. Polym. Sci. 1979, 23, 2591–2601.

[18] V. Raballand, J. Benedikt, S. Hoffmann, M. Zimmermann, a. von Keudell, J. Appl. Phys. 2009, 105, 083304.

115

[19] S. Guimond, I. Radu, G. Czeremuszkin, D. J. Carlsson, M. R. Wertheimer, Plasmas Polym. 2002, 7, 71–88.

[20] H. Muguruma, Plasma Process. Polym. 2009, NA–NA.

[21] P. Erra, R. Molina, D. Jocic, M. R. Julia, a. Cuesta, J. M. D. Tascon, Text. Res. J. 1999, 69, 811–815.

[22] E. a. Sosnin, E. Stoffels, M. V. Erofeev, I. E. Kieft, S. E. Kunts, IEEE Trans. Plasma Sci. 2004, 32, 1544–1550.

[23] E. Stoffels, I. E. Kieft, R. E. J. Sladek, L. J. M. Van Den Bedem, E. P. Van Der Laan, M. Steinbuch, Plasma Sources Sci. Technol. 2006, 15, S169–S180.

[24] M. Laroussi, D. Mendis, M. Rosenberg, New J. Phys. 2003, 5, 1–10.

[25] M. Laroussi, J. P. Richardson, F. C. Dobbs, Appl. Phys. Lett. 2002, 81, 772.

[26] D. B. Graves, J. Phys. D. Appl. Phys. 2012, 45, 263001.

[27] J. Heinlin, G. Isbary, W. Stolz, G. Morfill, M. Landthaler, T. Shimizu, B. Steffes, T. Nosenko, J. Zimmermann, S. Karrer, J. Eur. Acad. Dermatol. Venereol. 2011, 25, 1–11.

[28] G. Isbary, G. Morfill, J. Zimmermann, T. Shimizu, W. Stolz, Arch. Dermatol. 2011, 147, 388.

[29] O. Lademann, A. Kramer, H. Richter, A. Patzelt, M. C. Meinke, V. Czaika, K.-D. Weltmann, B. Hartmann, S. Koch, Skin Pharmacol. Physiol. 2011, 24, 284–8.

[30] P. Y. Kim, Y.-S. Kim, I. G. Koo, J. C. Jung, G. J. Kim, M. Y. Choi, Z. Yu, G. J. Collins, PLoS One 2011, 6, e24104.

[31] G. Isbary, G. Morfill, H. U. Schmidt, M. Georgi, K. Ramrath, J. Heinlin, S. Karrer, M. Landthaler, T. Shimizu, B. Steffes, W. Bunk, R. Monetti, J. L. Zimmermann, R. Pompl, W. Stolz, Br. J. Dermatol. 2010, 163, 78–82.

[32] M. Orazov, Y. Sakiyama, D. B. Graves, J. Phys. D. Appl. Phys. 2012, 45, 445201.

[33] D. Dobrynin, G. Fridman, G. Friedman, A. Fridman, New J. Phys. 2009, 11, 115020.

[34] R. Sensenig, S. Kalghatgi, E. Cerchar, G. Fridman, A. Shereshevsky, B. Torabi, K. P. Arjunan, E. Podolsky, A. Fridman, G. Friedman, J. Azizkhan-Clifford, A. D. Brooks, Ann. Biomed. Eng. 2010, 39, 674–687.

[35] G. Fridman, A. Shereshevsky, M. M. Jost, A. D. Brooks, A. Fridman, A. Gutsol, V. Vasilets, G. Friedman, Plasma Chem. Plasma Process. 2007, 27, 163–176.

116

[36] W. Kim, K.-C. Woo, G.-C. Kim, K.-T. Kim, J. Phys. D. Appl. Phys. 2011, 44, 013001.

[37] H. J. Lee, C. H. Shon, Y. S. Kim, S. Kim, G. C. Kim, M. G. Kong, New J. Phys. 2009, 11, 115026.

[38] S. Kalghatgi, A. Fridman, J. Azizkhan-Clifford, G. Friedman, Plasma Process. Polym. 2012, 9, 726–732.

[39] J. Köritzer, V. Boxhammer, A. Schäfer, T. Shimizu, T. G. Klämpfl, Y.-F. Li, C. Welz, S. Schwenk-Zieger, G. E. Morfill, J. L. Zimmermann, J. Schlegel, PLoS One 2013, 8, e64498.

[40] M. Vandamme, E. Robert, S. Lerondel, V. Sarron, D. Ries, S. Dozias, J. Sobilo, D. Gosset, C. Kieda, B. Legrain, J.-M. Pouvesle, A. Le Pape, Int. J. cancer 2012, 130, 2185– 94.

[41] M. G. Vander Heiden, L. C. Cantley, C. B. Thompson, Science 2009, 324, 1029–33.

[42] R. H. Engel, A. M. Evens, Front. Biosci. 2006, 11, 300–312.

[43] G. Fridman, M. Peddinghaus, M. Balasubramanian, H. Ayan, A. Fridman, A. Gutsol, A. Brooks, Plasma Chem. Plasma Process. 2006, 26, 425–442.

[44] N. S. Heo, M.-K. Lee, G. W. Kim, S. J. Lee, J. Y. Park, T. J. Park, J. Biosci. Bioeng. 2014, 117, 81–85.

[45] B. R. Locke, M. Sato, P. Sunka, M. Hoffmann, J. S. Chang, Ind. Eng. Chem. Res 2006, 45, 882–905.

[46] J. L. Brisset, B. Benstaali, D. Moussa, J. Fanmoe, E. Njoyim-Tamungang, Plasma Sources Sci. Technol. 2011, 20, 034021.

[47] H. W. Lee, G. J. Kim, J. M. Kim, J. K. Park, J. K. Lee, G. C. Kim, J. Endod. 2009, 35, 587–91.

[48] K. Vasilev, V. R. Sah, R. V Goreham, C. Ndi, R. D. Short, H. J. Griesser, Nanotechnology 2010, 21, 215102.

[49] M. Leduc, S. Coulombe, R. L. Leask, Plasma Sci. IEEE Trans. 2009, 37, 927–933.

[50] M. Mattioli-Belmonte, J. Bioact. Compat. Polym. 2005, 20, 343–360.

[51] K. Schröder, B. Finke, M. Polak, F. Lüthen, J. B. Nebe, J. Rychly, R. Bader, G. Lukowski, U. Walschus, M. Schlosser, A. Ohl, K.-D. Weltmann, Mater. Sci. Forum 2010, 638-642, 700–705.

117

[52] K. S. Siow, L. Britcher, S. Kumar, H. J. Griesser, Plasma Process. Polym. 2006, 3, 392– 418.

[53] P. K. Chu, J. Y. Chen, L. Wang, N. Huang, Mater. Sci. Eng. R Reports 2002, 36, 143–206.

[54] G. Fridman, G. Friedman, A. Gutsol, A. B. Shekhter, V. N. Vasilets, A. Fridman, Plasma Process. Polym. 2008, 5, 503–533.

[55] T. von Woedtke, S. Reuter, K. Masur, K.-D. Weltmann, Phys. Rep. 2013, 530, 291–320.

[56] M. Laroussi, IEEE Trans. Plasma Sci. 1996, 24, 1188–1191.

[57] R. Ben Gadri, J. Roth, T. Montie, K. Kelly-Wintenberg, P. Tsai, D. Helfritch, P. Feldman, D. Sherman, F. Karakaya, Z. Chen, Surf. Coatings Technol. 2000, 131, 528–541.

[58] M. Laroussi, Plasma Process. Polym. 2005, 2, 391–400.

[59] J. Ehlbeck, U. Schnabel, M. Polak, J. Winter, T. von Woedtke, R. Brandenburg, T. von dem Hagen, K.-D. Weltmann, J. Phys. D. Appl. Phys. 2011, 44, 013002.

[60] M. J. Pavlovich, Z. Chen, Y. Sakiyama, D. S. Clark, D. B. Graves, Plasma Process. Polym. 2013, 10, 69–76.

[61] K. Kelly-Wintenberg, T. Montie, C. Brickman, J. Roth, A. Carr, K. Sorge, L. Wadsworth, P. Tsai, J. Ind. Microbiol. Biotechnol. 1998, 20, 69–74.

[62] S. G. Joshi, M. Paff, G. Friedman, G. Fridman, A. Fridman, A. D. Brooks, Am. J. Infect. Control 2010, 38, 293–301.

[63] J. L. Zimmermann, K. Dumler, T. Shimizu, G. E. Morfill, a Wolf, V. Boxhammer, J. Schlegel, B. Gansbacher, M. Anton, J. Phys. D. Appl. Phys. 2011, 44, 505201.

[64] T. Maisch, T. Shimizu, G. Isbary, J. Heinlin, S. Karrer, T. G. Klämpfl, Y.-F. Li, G. Morfill, J. L. Zimmermann, Appl. Environ. Microbiol. 2012, 78, 4242–7.

[65] T. G. Klämpfl, G. Isbary, T. Shimizu, Y.-F. Li, J. L. Zimmermann, W. Stolz, J. Schlegel, G. E. Morfill, H.-U. Schmidt, Appl. Environ. Microbiol. 2012, 78, 5077–5082.

[66] R. M. Klevens, J. R. Edwards, C. L. Richards, T. C. Horan, R. P. Gaynes, D. a Pollock, D. M. Cardo, Public Health Rep. 2007, 122, 160–6.

[67] M. L. Kilgore, K. Ghosh, C. M. Beavers, D. Y. Wong, P. a Hymel, S. E. Brossette, Med. Care 2008, 46, 101–4.

[68] K. Lee, B. Joopark, D. Heelee, I. Lee, S. Ohyun, K. Chung, J. Park, Surf. Coatings Technol. 2005, 193, 35–38.

118

[69] G. Kamgang-Youbi, J.-M. Herry, T. Meylheuc, J.-L. Brisset, M.-N. Bellon-Fontaine, A. Doubla, M. Naïtali, Lett. Appl. Microbiol. 2009, 48, 13–18.

[70] R. Burlica, M. J. Kirkpatrick, B. R. Locke, J. Electrostat. 2006, 64, 35–43.

[71] G. Kamgang-Youbi, J.-M. Herry, M.-N. Bellon-Fontaine, J.-L. Brisset, A. Doubla, M. Naïtali, Appl. Environ. Microbiol. 2007, 73, 4791–6.

[72] G. Kamgang-Youbi, J.-M. Herry, J.-L. Brisset, M.-N. Bellon-Fontaine, A. Doubla, M. Naïtali, Appl. Microbiol. Biotechnol. 2008, 81, 449–57.

[73] K. Oehmigen, M. Hähnel, R. Brandenburg, C. Wilke, K.-D. Weltmann, T. von Woedtke, Plasma Process. Polym. 2010, 7, 250–257.

[74] M. J. Traylor, M. J. Pavlovich, S. Karim, P. Hait, Y. Sakiyama, D. S. Clark, D. B. Graves, J. Phys. D. Appl. Phys. 2011, 44, 472001.

[75] W. Glaze, Environ. Sci. Technol. 1987, 21, 224–230.

[76] M. Hähnel, T. von Woedtke, K.-D. Weltmann, Plasma Process. Polym. 2010, 7, 244–249.

[77] T. C. Montie, K. Kelly-Wintenberg, J. R. Roth, IEEE Trans. Plasma Sci. 2000, 28, 41–50.

[78] K. H. Becker, U. Kogelschatz, K. H. Schoenbach, R. J. Barker, Non-Equilibrium Air Plasmas at Atmospheric Pressure, IOP Publishing, Bristol, UK, 2005.

[79] Z. Machala, M. Janda, K. Hensel, I. Jedlovský, L. Leštinská, V. Foltin, V. Martišovitš, M. Morvová, J. Mol. Spectrosc. 2007, 243, 194–201.

[80] C. Bogdan, M. Röllinghoff, a Diefenbach, Curr. Opin. Immunol. 2000, 12, 64–76.

[81] C. Winterbourn, A. Kettle, Antioxid. Redox Signal. 2013, 18, 642–660.

[82] A. J. Kettle, C. C. Winterbourn, Biofactors 2005, 24, 41–5.

[83] P. Wentworth, J. Nieva, C. Takeuchi, R. Galve, A. D. Wentworth, R. B. Dilley, G. A. DeLaria, A. Saven, B. M. Babior, K. D. Janda, A. Eschenmoser, R. a Lerner, Science (80-. ). 2003, 302, 1053–1056.

[84] P. Wentworth, J. E. McDunn, A. D. Wentworth, C. Takeuchi, J. Nieva, T. Jones, C. Bautista, J. M. Ruedi, A. Gutierrez, K. D. Janda, B. M. Babior, A. Eschenmoser, R. a Lerner, Science 2002, 298, 2195–9.

[85] B. M. Babior, C. Takeuchi, J. Ruedi, A. Gutierrez, P. Wentworth, Proc. Natl. Acad. Sci. U. S. A. 2003, 100, 3031–4.

119

[86] K. Oehmigen, J. Winter, M. Hähnel, C. Wilke, R. Brandenburg, K.-D. Weltmann, T. von Woedtke, Plasma Process. Polym. 2011, 8, 904–913.

[87] T. Shimizu, Y. Sakiyama, D. B. Graves, J. L. Zimmermann, G. E. Morfill, New J. Phys. 2012, 14, 103028.

[88] Y. Sakiyama, D. B. Graves, H.-W. Chang, T. Shimizu, G. E. Morfill, J. Phys. D. Appl. Phys. 2012, 45, 425201.

[89] Y. Sakiyama, D. B. Graves, Plasma Sources Sci. Technol. 2013, 22, 012003.

[90] K. M. Robinson, J. S. Beckman, Methods Enzymol. 2005, 396, 207–14.

[91] P. Lukes, E. Dolezalova, I. Sisrova, M. Clupek, Plasma Sources Sci. Technol. 2014, 23, 015019.

[92] H. Tresp, M. U. Hammer, J. Winter, K. Weltmann, S. Reuter, J. Phys. D. Appl. Phys. 2013, 46, 435401.

[93] M. Naïtali, G. Kamgang-Youbi, J.-M. Herry, M.-N. Bellon-Fontaine, J.-L. Brisset, Appl. Environ. Microbiol. 2010, 76, 7662–4.

[94] J. O. Lundberg, E. Weitzberg, M. T. Gladwin, Nat. Rev. Drug Discov. 2008, 7, 156–67.

[95] R. Weller, M. J. Finnen, Nitric Oxide 2006, 15, 395–9.

[96] M. Wullt, I. Odenholt, M. Walder, Infect. Control Hosp. Epidemiol. 2003, 24, 765–768.

[97] Y. Kono, H. Shibata, K. Adachi, K. Tanaka, Arch. Biochem. Biophys. 1994, 311, 153– 159.

[98] J. Hoigne, H. Bader, Water Res. 1983, 17, 173–183.

[99] J. Lelieveld, P. Crutzen, J. Atmos. Chem. 1991, 12, 229–267.

[100] N. Y. Babaeva, M. J. Kushner, J. Phys. D. Appl. Phys. 2013, 46, 025401.

[101] P. J. Bruggeman, B. R. Locke, Assessment of Potential Appications of Plasma with Liquid Water, in Low Temperature Plasma Technology Methods and Applications, 2013.

[102] N. L. Oleinick, H. H. Evans, Radiat. Res. 1998, 150, S146–56.

[103] T. Dai, Y.-Y. Huang, M. R. Hamblin, Photodiagnosis Photodyn. Ther. 2009, 6, 170–88.

[104] T. Dai, G. P. Tegos, Z. Lu, L. Huang, T. Zhiyentayev, M. J. Franklin, D. G. Baer, M. R. Hamblin, Antimicrob. Agents Chemother. 2009, 53, 3929–34.

120

[105] T. Maisch, Mini Rev. Med. Chem. 2009, 9, 974–83.

[106] K. I. Møller, B. Kongshoj, P. A. Philipsen, V. O. Thomsen, H. C. Wulf, Photodermatol. Photoimmunol. Photomed. 2005, 21, 118–24.

[107] K. G. McGuigan, R. M. Conroy, H.-J. Mosler, M. du Preez, E. Ubomba-Jaswa, P. Fernandez-Ibañez, J. Hazard. Mater. 2012, 235-236, 29–46.

[108] F. Bosshard, M. Bucheli, Y. Meur, T. Egli, Microbiology 2010, 156, 2006–15.

[109] M. Berney, H.-U. Weilenmann, T. Egli, Microbiology 2006, 152, 1719–29.

[110] M. B. Fisher, C. R. Keenan, K. L. Nelson, B. M. Voelker, J. Water Health 2008, 6, 35–51.

[111] S. a Brownell, A. R. Chakrabarti, F. M. Kaser, L. G. Connelly, R. L. Peletz, F. Reygadas, M. J. Lang, D. M. Kammen, K. L. Nelson, J. Water Health 2008, 6, 53–65.

[112] C. Chatterley, K. Linden, J. Water Health 2010, 8, 479–86.

[113] A. Hamamoto, M. Mori, A. Takahashi, M. Nakano, N. Wakikawa, M. Akutagawa, T. Ikehara, Y. Nakaya, Y. Kinouchi, J. Appl. Microbiol. 2007, 103, 2291–8.

[114] K. P. Kühn, I. F. Chaberny, K. Massholder, M. Stickler, V. W. Benz, H.-G. Sonntag, L. Erdinger, Chemosphere 2003, 53, 71–7.

[115] P. Maness, S. Smolinski, D. M. Blake, Z. Huang, E. J. Wolfrum, W. A. Jacoby, Appl. Environ. Microbiol. 1999, 65, 4094–4098.

[116] H. W. Lee, H. W. Lee, S. K. Kang, H. Y. Kim, I. H. Won, S. M. Jeon, J. K. Lee, Plasma Sources Sci. Technol. 2013, 22, 055008.

[117] M. Ahro, J. Kauppinen, Appl. Spectrosc. 2001, 55, 50–54.

[118] M. A. H. Smith, V. M. Devi, D. C. Benner, C. P. Rinsland, J. Geophys. Res. 2001, 106, 9909–9921.

[119] H. Bader, J. Hoigne, Water Res. 1981, 15, 449–456.

[120] H. Bader, J. Hoigne, Ozone Sci. Eng. 1982, 4, 169–176.

[121] G. Gordon, B. Bubnis, Ozone Sci. Eng. 2002, 24, 17–28.

[122] S. A. Sattar, J. Virol. Methods 2003, 112, 3–12.

[123] L. W. Bush, L. M. Benson, J. H. White, J. Clin. Microbiol. 1986, 24, 343–8.

121

[124] M. Reichel, P. Heisig, T. Kohlmann, G. Kampf, Antimicrob. Agents Chemother. 2009, 53, 4778–82.

[125] E. J. Lowbury, H. a Lilly, Br. Med. J. 1973, 1, 510–5.

[126] N. Y. Babaeva, M. J. Kushner, J. Phys. D. Appl. Phys. 2010, 43, 185206.

[127] Z. Machala, L. Chládeková, M. Pelach, J. Phys. D. Appl. Phys. 2010, 43, 222001.

[128] B. Lange-Asschenfeldt, A. Alborova, D. Krüger-Corcoran, A. Patzelt, H. Richter, W. Sterry, A. Kramer, E. Stockfleth, J. Lademann, J. Biomed. Opt. 2009, 14, 054001.

[129] C. Edwards, R. Heggie, R. Marks, Photodermatol. Photoimmunol. Photomed. 2003, 19, 169–74.

[130] A. I. Kholodnykh, I. Y. Petrova, K. V Larin, M. Motamedi, R. O. Esenaliev, Appl. Opt. 2003, 42, 3027–37.

[131] F. Leipold, Y. Kusano, F. Hansen, T. Jacobsen, Food Control 2010, 21, 1194–1198.

[132] A. Schwabedissen, P. Łaciński, X. Chen, J. Engemann, Contrib. to Plasma Phys. 2007, 47, 551–558.

[133] C. Mayhall, Hospital Epidemiology and Infection Control, Lippincott Williams & Wilkins, 2011.

[134] M. Okano, M. Nomura, S. Hata, N. Okada, K. Sato, Y. Kitano, M. Tashiro, Y. Yoshimoto, R. Hama, T. Aoki, Arch. Dermatol. 1989, 125, 50–2.

[135] C. Gürgan, E. Zaim, I. Bakirsoy, E. Soykan, J. Periodontol. 2006, 77, 370–384.

[136] X. Thomas, P. Von Der Heyden, M.-R. De Backer-Barilly, M.-T. Bourgeois, A. Barbe, J. Quant. Spectrosc. Radiat. Transf. 2010, 111, 1080–1088.

[137] E. Wilhelm, R. Battino, R. Wilcock, Chem. Rev. 1977, 77, 219–261.

[138] M. J. Pavlovich, H.-W. Chang, Y. Sakiyama, D. S. Clark, D. B. Graves, J. Phys. D. Appl. Phys. 2013, 46, 145202.

[139] M. J. Pavlovich, Y. Sakiyama, D. S. Clark, D. B. Graves, Plasma Process. … 2013, 10, 1051–1060.

[140] G. R. Finch, D. W. Smith, M. E. Stiles, Water Res. 1988, 22, 1563–1570.

[141] N. Hunt, B. Marinas, Water Res. 1997, 31, 1355–1362.

122

[142] A. Mahfoudh, M. Moisan, J. Séguin, J. Barbeau, Y. Kabouzi, D. Kéroack, Ozone Sci. Eng. 2010, 32, 180–198.

[143] S. Moncada, R. Palmer, E. Higgs, Pharmacol. Rev. 1991, 43, 109–142.

[144] A. W. Carpenter, M. H. Schoenfisch, Chem. Soc. Rev. 2012, 41, 3742–52.

[145] C. Bogdan, Trends Cell Biol. 2001, 11, 66–75.

[146] R. Ehrlich, S. Miller, Appl. Microbiol. 1972, 23.

[147] R. L. Mancinelli, C. P. Mckay, Appl. Environ. Microbiol. 1983, 46, 198–202.

[148] K. Benedek, M. Ruffo, W. Luongo, Device and Method for Gas Sterilization, 2013.

[149] M. Cho, J. Kim, J. Y. Kim, J. Yoon, J.-H. Kim, Water Res. 2010, 44, 3410–8.

[150] J. Jeon, T. M. Rosentreter, Y. Li, G. Isbary, H. M. Thomas, J. L. Zimmermann, G. E. Morfill, T. Shimizu, Plasma Process. Polym. 2014, n/a–n/a.

[151] K. Huth, M. Quirling, S. Lenzke, E. Paschos, K. Kamereck, K. Brand, R. Hickel, N. Ilie, Eur. J. Oral Sci. 2011, 119, 204–210.

[152] A. Maiya, Int. J. Clin. Dent. Sci. 2011, 2, 23–27.

[153] S. Eick, M. Tigan, A. Sculean, Clin. Oral Investig. 2012, 16, 537–44.

[154] J. Steppan, T. Meaders, M. Muto, K. J. Murphy, J. Vasc. Interv. Radiol. 2010, 21, 534–48.

[155] H. Chen, B. Xing, X. Liu, B. Zhan, J. Zhou, H. Zhu, Z. Chen, J. Surg. Res. 2008, 149, 287–95.

[156] H. S. Kim, S. U. Noh, Y. W. Han, K. M. Kim, H. Kang, H. O. Kim, Y. M. Park, J. Korean Med. Sci. 2009, 24, 368–74.

[157] V. Travagli, I. Zanardi, G. Valacchi, V. Bocci, Mediators Inflamm. 2010, 2010, 610418.

[158] V. Travagli, I. Zanardi, A. Silvietti, V. Bocci, Int. J. Biol. Macromol. 2007, 41, 504–11.

[159] V. Bocci, I. Zanardi, V. Travagli, Med. Gas Res. 2011, 1, 6.

[160] K. Yamashita, T. Miyoshi, T. Arai, N. Endo, H. Itoh, K. Makino, K. Mizugishi, T. Uchiyama, M. Sasada, Proc. Natl. Acad. Sci. U. S. A. 2008, 105, 16912–16917.

[161] A. Garner, C. S. Croix, B. Pitt, G. Leikauf, Nat. Chem. 2009, 1, 316–321.

123

[162] K. Xu, S. Sun, J. Li, L. Li, M. Qiang, B. Tang, Chem. Commun. (Camb). 2012, 48, 684–6.

[163] P. W. M. H. Smeets, a W. C. van der Helm, Y. J. Dullemont, L. C. Rietveld, J. C. van Dijk, G. J. Medema, Water Res. 2006, 40, 3239–48.

[164] E. Katzenelson, B. Kletter, H. Shuval, J. Am. Water Work. Assoc. 1974, 66, 725–729.

[165] H. Zhou, D. W. Smith, J. Environ. Eng. 1994, 120, 841–858.

[166] S. Farooq, S. Akhlaque, Water Res. 1983, 17, 809–812.

[167] F. Liu, P. Sun, N. Bai, Y. Tian, H. Zhou, S. Wei, Y. Zhou, J. Zhang, W. Zhu, K. Becker, J. Fang, Plasma Process. Polym. 2010, 7, 231–236.

[168] R. J. Watts, D. Washington, J. Howsawkeng, F. J. Loge, A. L. Teel, Adv. Environ. Res. 2003, 7, 961–968.

[169] C. E. Bayliss, W. M. Waites, J. Appl. Bacteriol. 1980, 48, 417–22.

[170] R. Weller, R. Price, A. Ormerod, N. Benjamin, C. Leifert, J. Appl. Microbiol. 2001, 90, 648–652.

[171] O. H. and S. Administration, “Chemical Sampling Information: Ozone,” can be found under http://www.osha.gov/dts/chemicalsampling/data/CH_259300.html, 2007.

[172] T. Shimizu, 2013.

[173] J. O. Lundberg, E. Weitzberg, Nitric oxide 2010, 22, 61–3.

[174] L. Chu, C. Anastasio, Environ. Sci. Technol. 2007, 41, 3626–3632.

[175] A. Treinin, E. Hayon, J. Am. Chem. Soc. 1970, 8, 5821–5828.

[176] O. Zafiriou, M. True, Mar. Chem. 1979, 8, 9–32.

[177] M. Fischer, P. Warneck, J. Phys. Chem. 1996, 3654, 18749–18756.

[178] R. Zellner, M. Exner, H. Herrmann, J. Atmos. Chem. 1990, 10, 411–425.

[179] M. Mori, A. Hamamoto, A. Takahashi, M. Nakano, N. Wakikawa, S. Tachibana, T. Ikehara, Y. Nakaya, M. Akutagawa, Y. Kinouchi, Med. Biol. Eng. Comput. 2007, 45, 1237–1241.

[180] M. Laroussi, F. Leipold, Int. J. Mass Spectrom. 2004, 233, 81–86.

[181] R. J. Tallarida, J. Pharmacol. Exp. Ther. 2001, 298, 865–72.

124

[182] F. C. Odds, J. Antimicrob. Chemother. 2003, 52, 1.

[183] D. W. Thayer, G. Boyd, W. F. Fett, J. Food Sci. 2006, 71, R83–R87.

[184] M. Holmes, J. Chem. Soc 1926, 129, 1898.

[185] M. Morgan, P. Van Trieste, S. Garlick, M. Mahon, A. Smith, Anal. Chim. Acta 1988, 215, 325–329.

[186] W. Heaselgrave, P. W. Andrew, S. Kilvington, L. Le, J. Antimicrob. Chemother. 2010, 65, 1207–1214.

[187] G. Jiang, Z. Yuan, J. Hazard. Mater. 2013, 250-251C, 91–98.

[188] S. a Spence, V. L. Clark, V. M. Isabella, Microbiology 2012, 158, 560–70.

[189] S. J. Padayatty, A. Katz, Y. Wang, P. Eck, O. Kwon, J.-H. Lee, S. Chen, C. Corpe, A. Dutta, S. K. Dutta, M. Levine, J. Am. Coll. Nutr. 2003, 22, 18–35.

[190] C. Opländer, M. Baschin, E. Faassen, M. Born, Nitric Oxide 2010, 23, 275–283.

[191] H. Sies, Eur. J. Biochem. 1993, 215, 213–9.

[192] E. Riordan, N. Minogue, D. Healy, P. O’Driscoll, J. R. Sodeau, J. Phys. Chem. A 2005, 109, 779–86.

[193] S. Goldstein, J. Lind, G. Merényi, Chem. Rev. 2005, 105, 2457–70.

[194] W. Schroeder, P. Wilcox, K. Trueblood, A. Dekker, Anal. Chem. 1951, 23, 1740–1747.

[195] D. Moussa, A. Doubla, G. Kamgang-Youbi, J.-L. Brisset, IEEE Trans. Plasma Sci. 2007, 35, 444–453.

[196] B. Halliwell, J. Gutteridge, Free Radicals in Biology and Medicine, Oxford University Press, New York, 2007.

[197] J. H. Seinfeld, S. Pandis, Atmospheric Chemistry and Physics: From Air Pollution to Climate Change, Wiley-Interscience Publication, NY, 2006.

[198] D. a Vitturi, R. P. Patel, Free Radic. Biol. Med. 2011, 51, 805–12.

[199] A. Fridman, Plasma Chemistry, Cambridge University Press, 2008.

[200] V. N. Vasilets, A. B. Shekhter, in Plasma Bio-Decontamination, Med. Food Secur., Springer Science+Business Media B.V., 2012.

125

[201] D. Dobrynin, K. Arjunan, a Fridman, G. Friedman, a M. Clyne, J. Phys. D. Appl. Phys. 2011, 44, 075201.

[202] T. Kim, S. Song, J. Kim, R. Iwasaki, Jpn. J. Appl. Phys. 2010, 49, 126201.

[203] J. Kulla, R. Reich, S. Broedel Jr., 2009, 1–12.

[204] B. C. Rao, Technol. Soc. 2013, 35, 65–73.

[205] Y. Bhatti, M. Ventresca, Said Buisness Sch. Work. Pap. Ser. Oxford 2013.

[206] S. Ray, P. Kanta Ray, Technovation 2011, 31, 216–227.

[207] —, Green Futur. Mag. 2013.

[208] A. Akubue, J. Technol. Stud. 2000, 26.

[209] A. P. Arya, L. Klenerman, J. Bone Joint Surg. Br. 2008, 90, 1414–6.

[210] N. Ravi, J. Prabhu, S. Ajuja, Jugaad Innovation: Think Frugal, Be Flexible, Generate Breakthrough Growth, John Wiley And Sons, 2012.

[211] —, Econ. 2011.

[212] S. Jarosławski, G. Saberwal, BMC Health Serv. Res. 2013, 13, 199.

[213] T. Lancet, Lancet 2012, 380, 447.

[214] S. Ballati, J. Bird, A. G. Hegedus, 2013.

[215] D. G. Nocera, Daedalus 2012, 141, 45–52.

[216] A. Gadgil, Annu. Rev. energy Environ. 1998, 23, 253–286.

[217] G. Whitesides, Econ. 2011.

[218] N. De Geyter, R. Morent, Annu. Rev. Biomed. Eng. 2012, 14, 255–74.

[219] F. Rossi, M. Moisan, M. Kong, M. Laroussi, Plasma Process. Polym. 2012, 9, 559–560.

[220] K.-D. Weltmann, K. Fricke, M. Stieber, R. Brandenburg, T. von Woedtke, U. Schnabel, IEEE Trans. Plasma Sci. 2012, 40, 2963–2969.

[221] B. Langlais, D. A. Reckhow, D. R. Brink, Ozone in Water Treatment: Application and Engineering, American Water Works Association Research Foundation, Denver, CO, 1991.

126

[222] M. H. Kim, J. H. Cho, S. B. Ban, R. Y. Choi, E. J. Kwon, S.-J. Park, J. G. Eden, J. Phys. D. Appl. Phys. 2013, 46, 305201.

[223] Y. Raizer, Gas Discharge Physics, Springer-Verlag, Berlin, 1997.

[224] M. Janda, Z. Machala, IEEE Trans. Plasma Sci. 2011, 39, 2246–2247.

[225] Z. Machala, I. Jedlovský, V. Martisovits, IEEE Trans. Plasma Sci. 2008, 36, 918–919.

[226] L. Yu, C. O. Laux, D. M. Packan, C. H. Kruger, J. Appl. Phys. 2002, 91, 2678.

[227] Z. Machala, E. Marode, C. Laux, C. H. Kruger, J. Adv. Oxid. Technol. 2004, 7, 133–137.

[228] —, “An Introduction to Indoor Air Quality (IAQ): Nitrogen Dioxide (NO2),” can be found under www.epa.gov/iaq/no2.html, 2012.

127