Cold atmospheric pressure plasma and low energy electron beam as alternative nonthermal decontamination technologies for dry food surfaces : A review

Background: Dry food products are often highly contaminated, and dry stress-resistant microorganisms, such as certain types of Salmonella and bacterial spores, can be still viable and multiply if the product is incorporated into high moisture food products or rehydrated. Traditional technologies for the decontamination of these products have certain limitations and drawbacks, such as alterations of product quality, environmental impacts, carcinogenic potential and/or lower consumer acceptance. Cold atmospheric pressure plasma (CAPP) and low energy electron beam (LEEB) are two promising innovative technologies for microbial inactivation on dry food surfaces, which have shown potential to solve these certain limitations. Scope and approach: This review critically summarizes recent studies on the decontamination of dry food surfaces by CAPP and LEEB. Furthermore, proposed inactivation mechanisms, product-process interactions, current limitations and upscaling potential, as well as future trends and research needs for both emerging technologies, are discussed. Key findings and conclusions: CAPP and LEEB are nonthermal technologies with a high potential for the gentle decontamination of dry food surfaces. Both technologies have similarities in their inactivation mechanisms. Due to the limited penetration depth of both technologies, product-process interactions can be minimized by maintaining product quality. A first demonstrator with Technology Readiness Level (TRL) 7 for LEEB has already been introduced into the food industry for the decontamination of herbs and spices. Compared with LEEB, CAPP is at the advanced development stage with TRL 5, for which further work is essential to design systems that are scalable to industrial requirements.


Introduction
Low moisture or dry food ingredients and products have been a significant component of the human diet for thousands of years. These food ingredients have a low moisture content, such as spices, nuts and cereals, or they undergo a drying process, e.g., nuts, herbs and fruits. Dry products can be contaminated with food-spoiling bacteria, pathogens and bacterial spores, especially herbs and spices, which can often be highly contaminated with viable counts > 10 7 CFU g −1 (Schweiggert, Carle, & Schieber, 2007). The contamination source is often related to the pre-harvest environment and can be transferred during post-harvest processing, which can also be conducted in an open-air environment. Contamination may also occur as a consequence of recontamination issues during processing and poor manufacturing practices. Dry products are often inappropriately assumed to be microbiologically safe since the low moisture content represents a significant barrier to the growth of microorganisms (Gurtler, Doyle, & Kornacki, 2014). However, dry stress-resistant microorganisms, such as bacterial spores or some types of Salmonella, are still viable and can multiply if the product is rehydrated or incorporated into high moisture products with a sufficient amount of available nutrients. For instance, the contamination of nuts and nut products with pathogens is a reoccurring concern in the food industry. Between 2001 and 2016, approximately 82 recalls of nuts and nut products were reported in the United States, with quantities of up to 148.000 tons of products, due to pathogen contamination (Palumbo, Beuchat, Danyluk, & Harris, 2015). Salmonella strains are responsible for most of the reported outbreaks associated with the consumption of nuts and nut products (Gurtler https et al., 2014). Furthermore, in recent decades, cases of food-borne diseases and intoxicants have increased due to contaminated herbs and spices, which were incorporated in different food products (Buckenhüskes & Rendlen, 2004). Moreover, the use of spoiled herbs and spices can also drastically reduce the shelf-life of food products. The inactivation of microorganisms on dried products is challenging, mostly due to the increased resistance of microorganisms in low-water activity foods in comparison with water-rich foods (Fine & Gervais, 2005). Common technologies for the decontamination of dry products are fumigation with ethylene or propylene oxide, saturated steam or dry heat, microwave treatment, high energy electron beams and Gamma-rays (Pan, Bingol, Brandl, & McHugh, 2012;Schweiggert et al., 2007). However, these technologies have certain limitations and drawbacks, such as alterations of the aroma, odor and color (Schweiggert et al., 2007). Irradiation with Gamma-rays can only be applied in authorized facilities and in controlled doses. Moreover, this process has a poor consumer acceptance in the EU. The main drawbacks of Gamma-irradiation are difficulties in transporting sources and licensing new facilities. Fumigation with ethylene oxide and propylene oxide is widely used in the USA; however, it is prohibited in many countries due to its carcinogenic potential to humans (Farkas & Mohácsi-Farkas, 2014;Schweiggert et al., 2007).
Such drawbacks and limitations, as well as consumers' demand for high quality and safe foods, require the development and application of emerging nonthermal technologies for the gentle decontamination of dry food products. Cold atmospheric pressure plasma (CAPP) and low energy electron beam (LEEB) are two promising innovative technologies for microbial inactivation on dry food surfaces. This review will discuss and compare the application of both nonthermal technologies for microbial inactivation; dealing with bacteria, molds, and bacterial spores; proposed inactivation mechanisms, product-process interactions, current limitations and upscaling potential; and future trends and research needs.

Cold atmospheric pressure plasma technology
Plasma can be characterized as an at least partially ionized gas and is a complex mixture of different components, such as charged particles (electrons & ions) and neutral species (atoms & molecules), in addition to radicals, UV photons and irradiated heat. In general, plasma can be classified according its temperature into thermal and nonthermal plasmas. A thermal plasma is an almost completely ionized gas, whereby the temperatures of the charges and neutral species are approximately equal, with temperatures typically reaching at least 15,000 K (Eliasson & Kogelschatz, 1991). In comparison to thermal plasma, nonthermal ones are only partially ionized, indicating that the number of neutral species is much higher than the number of charged species, whereby the temperature of the different particles is not equal. The temperature of electrons is still in the range of several thousand Kelvin, but the temperature of the neutral species and ions can be close to ambient temperature. Thus, nonthermal plasmas are also termed cold plasmas.
For the generation of cold plasma energy need to be supplied to a gas, electric energy sources have been shown to be the most convenient. The lifetime of the particles inside the plasma is quite small due to energy loss by collision processes, and therefore energy must be supplied continuously for plasma applications. The generation of cold plasma can be achieved under atmospheric pressure and/or lower pressure conditions. Hence, plasma generation at atmospheric pressure allows continuous processing, which is a clear advantage for food process applications. A manifold number of different plasma systems are available because of various possible electrode configurations (geometry, number, location). Fig. 1 schematically illustrates two of the most commonly used plasma systems: plasma jet and dielectric barrier discharge (DBD) systems. However, CAPP technology is not yet standardized. Research groups are often using plasma systems, which are available in their country and/or custom built or in-house manufactured systems. Furthermore, these plasma systems are designed mostly for laboratory applications.

Microbial inactivation mechanisms of cold atmospheric pressure plasma
There is continued and growing interest in CAPPs for microbial inactivation in the food sector, whereby the possible applications are manifold, such as the treatment of fresh and dry products, or the in package treatment of food (Bourke, Zuizina, Han, Cullen, & Gilmore, 2017;Niemira, 2012a;Schlüter et al., 2013). As stated in chapter 2.1, CAPP is a complex mixture of different generated components, such as UV photons, charged particles, radicals and other reactive nitrogen, oxygen and hydrogen species (RNS, ROS & RHS), such as nitrogen oxides (NO • and NO x ), peroxynitrite (ONOO − ), atomic oxygen (O), ozone (O 3 ), singlet oxygen ( 1 O 2 ), superoxide anion (O 2 − ), hydrogen radicals (H • ), hydroxyl radicals (OH • ) and/or hydrogen peroxide (H 2 O 2 ). These reactive species, which can act individually and/or synergistically, are responsible for the antimicrobial effect of CAPP. The plasma system and the applied operating parameters (e.g., process gas, moisture, and energy input) affect the composition of the generated plasma and consequently also the antimicrobial efficiency of the plasma treatment (Bourke et al., 2017;Niemira, 2012a). Whereby, the plasma chemistry is complex and can involves hundreds of different species and thousands of possible reactions (Lu & Wu, 2013;Sakiyama, Graves, Chang, Shimizu, & Morfill, 2012). Another aspect that influences CAPPbased microbial inactivation is the manner in which the plasma will be applied to a surface, such as direct, semi-direct or indirect application (Table 1). Direct plasma application to a surface results in an interaction of possible generated UV photons, charged particles, and radicals, among others, with microorganisms on the surface. Compared with direct CAPP treatment, the antimicrobial effect is based on long-lived reactive species if CAPP is applied semi-directly or indirectly. The mechanisms responsible for microbial inactivation by CAPP have not been elucidated in detail. Fig. 2 illustrates the mechanisms that could be involved in CAPP-mediated microbial inactivation. However, the inactivation mechanisms can vary between microorganisms, e.g., bacterial spores are more resistant than bacteria (Hertwig, Reineke, Ehlbeck, Knorr, & Schlüter, 2015a). During the treatment, microorganisms are exposed to continuous bombardment with different reactive components of the generated plasma. Thereby, atomic and molecular radicals and excited molecules can cause an erosion of the microbial cell, atom by atom, through etching (Moisan et al., 2001). The reactive components will be adsorbed onto the microorganism surface, causing the formation of volatile compounds via chemical reactions and openings and lesions in the cell membrane. Consequently, Gram-negative bacteria are believed to be more easily inactivated by etching due to their thinner membrane structure compared with Grampositive bacteria (Stoffels, Sakiyama, & Graves, 2008). Fröhling, Baier, Ehlbeck, Knorr, and Schlüter (2012) showed that Gram-negative Escherichia coli are more sensitive to CAPP than Gram-positive Listeria innocua. Bacterial spores can also be inactivated by etching caused by oxygen atoms and radicals (Park et al., 2004). Microbial support structures, such as biofilms, can also be degraded by the erosion of organic material due to the breakage of chemical bonds (Bourke et al., 2017). Different reactive species inside the CAPP, such as RNS and ROS, can also interact with various cellular macromolecules, such as membrane lipids, proteins and DNA. Oxidative DNA damage caused by ROS can lead to microbial inactivation (Li, Sakai, Watanabe, Hotta, & Wachi, 2013). Oxidative damage of membrane lipids affects their ability to regulate mass transport in and out of the cell (Laroussi & Leipold, 2004). Furthermore, the generated reactive species may also diffuse into the microbial cell, which can result in a decrease in intracellular pH. If a bacterial cell cannot maintain pH homoeostasis, it will be inactivated (Booth, 1985;Padan & Schuldiner, 1987).
The cell membrane can also be damaged via electrostatic disruption due to the accumulation of charged particles on the microorganism surface (Laroussi, 2002). Such electrostatic forces caused by charge accumulation on an outer membrane surface can overcome the tensile strength of the membrane, resulting in its rupture, as shown by Mendis, Rosenberg, and Azam (2000).
UV photons emitted by CAPP are able to induce the dimerization of thymine bases in bacterial DNA strands, which impedes the ability of bacteria to replicate (Laroussi, 2002). Furthermore, UV photons can also damage the DNA of bacterial spores (Setlow, 2007). Faster spore inactivation with increasing UV emission of the used CAPP has been reported by Boudam et al. (2006) and Reineke, Langer, Hertwig, Ehlbeck, and Schlüter (2015). Hertwig et al. (2015b) showed an increase in DNA damage in spores with increasing UV emission. UV photons can also induce intrinsic photodesorption, an erosion of the cell, atom by atom, due to the breakage of chemical bonds (Moisan et al., 2001).
Bacterial spores can also be inactivated by CAPP due to damage to the inner spore membrane and key germination proteins (Wang, Doona, Setlow, & Li, 2016). Hertwig, Reineke, Rauh, and Schlüter (2017a) showed that different spore properties, the outer spore coat, dipicolinic acid (DPA) level inside the spore core and the DNA saturation with small acid soluble proteins (SASPs), contribute to the resistance of Bacillus subtilis spores to different generated plasma components. Nevertheless, identifying the main mechanism responsible for microbial inactivation is challenging, since the different interactions can occur simultaneously and also post mortem.
The humidity of the process gas also has an effect on CAPP-based inactivation, whereby the specific impact is still part of ongoing research. Muranyi, Wunderlich, and Heise (2008) treated B. subtilis and Aspergillus niger spores with air-plasma up to 8 s and increased the relative process gas humidity up to 80%. For A. niger spores an improved inactivation effect was observed, with maximum inactivation at 70% relative humidity. This effect was enhanced after prolonged CAPP treatment. However, B. subtilis spores showed slightly lower inactivation with increased relative humidity, especially for shorter treatment times. In contrast, Jeon, Klaempfl, Zimmermann, Morfill, and Shimizu (2014) and Patil et al. (2014) reported an increased inactivation for Geobacillus stearothermophilus and Bacillus atrophaeus spores with increasing process gas humidity. Both research groups suggested that the effect was probably due to a higher generation of ROS, such as hydrogen peroxide and hydroxyl radicals. In a humid environment reactive CAPP components, such as ROS and RNS, are able to initiate various chain reactions. This can results in the generation of a larger diversity of reactive species (Surowsky, Schlüter, & Knorr, 2014).

Cold atmospheric pressure plasma treatment of dry food surfaces
The application of CAPP is a promising technology for the decontamination of dry food products. A comprehensive overview of studies investigating microbial inactivation on dry food products by CAPP is presented in Table 2, which briefly summarizes the plasma sources and process gases used to inactivate various microorganisms and the obtained results. The summarized literature overview shows the high diversity in plasma processing with respect to the plasma source, energy input, process gas and kind of application. The large number of CAPP process variables involved, as well as food matrix-related parameters and methodological differences, complicate the comparability of these studies. The results showed that CAPP is capable of decontaminating plant and animal-based products; however, the largest proportion of the work was performed using plant products. Choi, Puligundla, and Mok (2016) investigated the application of CAPP to dried, shredded Alaska pollock, a typical snack food in China, Korea and other Asian nations. For the treatment, they used a corona discharge plasma jet driven by air and reported an inactivation of the native microbial flora of 2.5 log 10 after a 20-min treatment. Bußler et al. (2016) focused on an alternative and sustainable food source: edible insects. They treated mealworm larvae flour with air-plasma generated by a surface DBD up to 15 min Fig. 1. Schematic of a plasma jet system (left) and dielectric barrier discharge system (right).

Table 1
Overview of different types of cold plasma applications. (Adapted from Schlüter et al., 2013).

Type Description Examples
Direct Plasma is in direct contact with the substrate Plasma jet Interaction based on irradiation (ultraviolet (UV), vacuum UV (VUV)), charged molecules, radicals, and reactive particles Dielectric barrier discharges (DBD)

Semi-direct
Distance between plasma and substrate much larger than the mean free particle path Surface DBD with gap No interaction with charged particles Sterrad process with Antimicrobial effect based on irradiation, long-lived radicals, metastable and inhibitory substances plasma-activated hydrogen peroxide Indirect Irradiation with VUV, UV UV lamps No reaction with plasma particles Ozone generator Plasma is used to treat gas or liquids Plasma-processed air (PPA) Plasma-processed water (PPW) C. Hertwig et al. Trends in Food Science & Technology 77 (2018) 131-142 and inactivated the native microbial load by 3.0 log 10 . Sun, Anderson, and Keller (2014) inoculated black peppercorns with a cocktail of different Salmonella strains and reported an inactivation of 5.5 log 10 following 80 s of air-plasma treatment. The antimicrobial efficiency of a direct argon plasma jet and indirect air-plasma treatment was compared by Hertwig et al. (2015a). Direct plasma treatment resulted in much lower inactivation, probably due to the involvement of different inactivation mechanisms. The indirect plasma treatment inactivated Salmonella enterica, B. subtilis and atrophaeus spores by 4.1, 2.4 and 2.8 log 10 , respectively, after a 30-min treatment. Hertwig et al. (2015c) studied the inactivation of the native microbial flora of black peppercorns, crushed oregano and paprika powder following an indirect treatment with air-plasma. The authors reported an inactivation of > 4.0 and > 3.0 log 10 for black peppercorns and paprika powder, respectively, after 60 min. A lower inactivation of 1.6 log 10 was determined for crushed oregano after 90 min, probably due to the much lower initial native microbial load. Deng et al. (2007) treated almonds inoculated with E. coli and showed an inactivation of 5.0 log 10 after 30 s of DBD plasma treatment. Niemira (2012b) inoculated almonds with different types of Salmonella and E. coli O157:H7 and used a plasma jet with the process gases air and N 2 . The treatment with air-plasma was more efficient, the highest inactivation was achieved for E. coli O157:H7 after 20 s with 1.2 log 10 . Air-plasma was also more efficient in inactivating Salmonella Enteritidis PT30 on almonds compared with plasma generated with N 2 , O 2 , CO 2 and CO 2 in an admixture with argon (Hertwig et al., 2017b). After a 15-min treatment between two diffuse surface barrier discharge plasma plates, > 5.0 log 10 S. Enteritidis PT30 was inactivated.
In addition to bacteria, CAPP can also be applied to inactivate molds on nuts. Dasan, Boyaci, and Mutlu (2017) used an atmospheric pressure fluidized bed plasma generated with air and N 2 for the treatment of dehulled hazelnuts. The authors reported an inactivation for both process gases for A. flavus and parasiticus spores of > 4.0 log 10 after 5 min. Similar results were obtained by Dasan, Boyaci, and Mutlu (2016) by treating maize using the same plasma setup and mold spores. The application of CAPP is also able to reduce the native microbial flora of sprout seeds. Contaminated seeds are one of the major concerns related to sprout-associated outbreaks. The native microbial flora of chickpea seeds could be inactivated by 2.0 log 10 following 5 min of airplasma treatment (Mitra et al., 2014). Butscher, Van Loon, Waskow, von Rohr, and Schuppler (2016a) treated E. coli-inoculated alfalfa, onion, radish and cress seeds using argon-plasma and reported an inactivation between 1.4 and 3.4 log 10 , depending on the seeds, after 10 min. Air-plasma generated with a corona discharge plasma jet inactivated 2.3 log 10 of the broccoli seed native microbial flora after 3 min (Kim, Puligundla, & Mok, 2017). The same research group also investigated the antimicrobial efficiency of different CAPP treatments on a dried laver. A DBD air-plasma treatment of 10 min could inactivate the native microbial flora up to 2.5 log 10 .
In addition to microbial inactivation, CAPP can also be applied to degrade mycotoxins on dry food products. Siciliano et al. (2016) inoculated aflatoxins onto dehulled hazelnuts and treated them with DBD N 2 -plasma. After a 12-min CAPP treatment, the aflatoxin concentration was reduced by up to 70%. Similarly, a > 80% reduction of aflatoxins on corn following a 10-min air-plasma treatment with RH of 40% was reported by Shi, Ileleji, Stroshine, Keener, and Jensen (2017). Furthermore, direct and indirect CAPP treatment were equally effective for the degradation of aflatoxin. Ten Bosch et al. (2017) studied the degradation of different mycotoxins produced by Fusarium, Aspergillus and Alternaria species using DBD air-plasma and showed a varied degradation rate depending on the mycotoxin structure.
CAPP affects not only microorganisms, but it also interacts with the treated surface, which could lead to plasma-product interaction. However, due to the limited penetration depth of the plasma components, such plasma-product interactions are commonly restricted to the product surface. Bußler et al. (2016) reported an increased oil-binding capacity after air-plasma treatment of mealworm flour, whereas the water binding capacity decreased. In contrast, Thirumdas, Deshmukh, and Annapure (2016) observed an increase in water holding and binding capacity in basmati rice flour after an air-plasma treatment. CAPP could also lead to undesirable plasma-product reactions. Hertwig et al. (2017b) investigated the impact of different process gases on the almond surface color. The use of process gases containing nitrogen (air and N 2 ) resulted in a significant browning of the almonds, whereas the other applied process gases (O 2 , CO 2 , 90% CO 2 + 10% Ar) did not considerably alter the color. Indirect air-plasma treatment of black  Table 2 Overview of studies investigating the decontamination of dry food products by cold atmospheric pressure plasma (CAPP). Some values/numbers have been approximated from graphs provided in the cited literature.
(CFU: Colony forming unit; DBD: Dielectric barrier discharge; DCSBD: Diffuse coplanar surface barrier discharge; RH: Relative humidity).  peppercorns, oregano and paprika powder resulted in a significant loss of paprika powder redness, but the color of the other two products was only slightly affected (Hertwig et al., 2015c). The impact of direct and indirect CAPP treatment on quality parameters of black peppercorns, volatile oil content and the main aroma compound piperine, has been shown by Hertwig et al. (2015a). Neither plasma treatment substantially affected the surface color or the quality parameters. The impact of different CAPP applications on dried seafood products after microbial inactivation has also been investigated (Choi et al., 2016;Kim et al., 2015aKim et al., , 2015b. The researchers reported no significant changes in physicochemical properties, such as color, total phenolic content or radical scavenging activity, after the CAPP treatment. CAPP can interact with the surface of seeds to enhance germination and seedling growth without affecting physico-chemical and sensory characteristics of the grown sprouts (Butscher et al., 2016a;Dobrin, Magureanu, Mandache, & Ionita, 2015;Kim et al., 2017;Mitra et al., 2014;Zahoranová et al., 2016). CAPP treatment can change the surface characteristics of the seeds, which increases their wettability, as shown by a decreased contact angle and improved water uptake of the seeds (Randeniya & De Groot, 2015).

Limiting factors
The structure of a surface is one main parameter limiting the antimicrobial effect of a CAPP treatment due to a non-uniformity of the treatment. Hertwig et al. (2015a) and Butscher, Zimmermann, Schuppler, and Rudolf von Rohr (2016b) reported that a well-structured surface, i.e., a corrugated surface, can negatively affect microbial inactivation. Hertwig et al. (2015b) showed how the microbial inactivation of a CAPP is significantly reduced by the treatment of surfaces with a complex structure, such as on black peppercorns. The researchers investigated the inactivation of B. subtilis spores inoculated with a comparable spore density of 4 × 10 6 spores cm −2 on a simple flat glass surface, a spherical model (glass beads) and a real food matrix (black peppercorns). Black peppercorns have a well-structured surface characterized by cracks, grooves and pits, which might cause shadow effects for the different generated components of the plasma. Components such as UV photons and radicals cannot interact with the microorganisms, thus reducing the inactivation efficiency of the CAPP treatment. Another parameter affecting the plasma-based inactivation is the microbial load on the treated surface. Yu et al. (2006) investigated the inactivation of E. coli with surface densities between 10 7 to 10 11 CFU cm −2 and reported a decreasing inactivation with an increasing cell surface density. Deng, Shi, Shama, and Kong (2005) and Fernández, Shearer, Wilson, and Thompson (2012) reported a slower inactivation with an increased initial microbial load for the treatment of B. subtilis spores and Salmonella Typhimurium. The decreased inactivation at a higher cell surface density probably results from stacking and aggregating of the microorganisms. The top layer of multilayered cell structures, even if the microorganisms are inactivated, could form a physical barrier to shield the microorganisms beneath the top layer from generated reactive plasma components and inactivation (Deng et al., 2005). This process can be attributed to the limited penetration depth of plasma components, which depends on their half-lives and the treated material, e.g., the penetration depth of UV photons can exceed 1 μm depending on the wavelength and material (Lerouge, Wertheimer, Marchand, Tabrizian, & Yahia, 2000). In the case of cell agglomeration, the inactivated top cell layer must be decomposed due to processes such as etching and photodesorption, so that the reactive plasma components can inactivate cell layers beneath the top one.
The surface-to-volume ratio of a product could also influence the inactivation efficiency. Hertwig et al. (2015c) reported that for a CAPP treatment of products with a high surface-to-volume ratio, such as powdered products, the plasma is more likely to interact with the food surface itself rather than with the microorganisms on that surface.

Low energy electron beam technology
Electron beam (EB) technology belongs to the group of ionizing radiation, such as X-rays and γ-rays. All three technologies depend on the generation of high energy ionizing radiation to inactivate microorganisms. However, compared with X-and γ-rays, for which inactivation is due to energetic photons, EB utilizes high energy electrons for microbial inactivation. Unlike radiation with γ-rays, EB does not use radioisotopes. The electrons are generated via electricity and can be switched on or off as needed. Depending on the kinetic energy of the electrons, EB can be distinguished between high energy (> 300 keV) and low energy electron beams (< 300 keV). The penetration depth of the electrons is controlled and determined by their energy and mostly by the density of the treated material. After the electron penetrates the product, it will lose its kinetic energy due to collisions with the product particles. The higher the kinetic energy, the deeper the electron penetrates into the product (Urgiles et al., 2007). High energy electrons can effectively penetrate food products up to several centimeters, whereas the penetration depth of LEEB is limited to the micrometer scale (Urgiles et al., 2007). High energy EB already proved to be an efficient technology for the decontamination of dry products and shelf life extension of fresh produce, such as melons and berries (Pillai & Shayanfar, 2018). Whereby, a uniform treatment can be ensured by an optimized process configuration and by an optimized packaging of the product to be treated (Pillai & Shayanfar, 2015). Furthermore, the high penetration depth of the high energy electrons mitigate a non-uniformly dose distribution inside the product. The dose uniformity within a batch is controlled with dosimeters placed in several different locations of the batch (Pillai & Shayanfar, 2015).
LEEB technology deposits electron energy close to the surface where microorganisms are present, resulting in an extremely high efficiency for surface decontamination (Urgiles et al., 2007). LEEB was introduced in 2002 in the agricultural sector as a replacement for the chemical dressing of seeds. Currently, LEEB systems for seed treatment with a capacity of up to 30 tons per hour are available (Röder et al., 2009). In 2012, Tetra Pak introduced LEEB for the sterilization of food packaging material as replacement for hydrogen peroxide (Comet Group, 2012a). Recently, the first demonstrator of LEEB was introduced in the spice and herb industries (IIA, 2017). A sealed EB lamp is used as a LEEB source (Fig. 3). The low energy electrons are generated by a cathode made of tungsten filaments, sitting at negative high voltage electrodes inside an ultrahigh vacuum. If a high voltage is applied, the electrons start to "boil off" the tungsten wires, and electrostatic optics focus and guide them to the window (Comet Group, 2012b). Such LEEB systems offer different advantages, such as a low or negligible thermal impact on the treated product, compact size and simple integration into existing production processes (Chalise, Hotta, Matak, & Jaczynski, 2007;Comet Group, 2012a). Compared with other technologies using ionizing radiation, the shielding infrastructure of LEEB machinery can be maintained at a minimum level due to the lower penetration depth of the electrons and reduced X-ray scattering.

Microbial inactivation mechanisms with low energy electron beams
The mechanisms of microbial inactivation with LEEB are not completely understood, but it is believed that they are similar to those associated with other types of ionizing radiation (X-and γ-rays) (Chalise et al., 2007). The main target of ionizing radiation is DNA (Moeller et al., 2008;Nicholson, Munakata, Horneck, Melosh, & Setlow, 2000). LEEB can inactivate microorganisms due to direct or indirect interactions, whereas the final inactivation is an outcome of both processes. The mechanisms involved in microbial inactivation by LEEB are illustrated in Fig. 2. The direct interactions result from energy transfer of the electrons to target molecules, e.g., DNA, RNA, enzymes and C. Hertwig et al. Trends in Food Science & Technology 77 (2018) 131-142 membrane proteins. Thus, the cellular DNA is damaged due to disintegration, such as single or double-strand breaks, as well as base modifications. The induced DNA damage is the main mechanism responsible for microbial inactivation by direct effects (Tahergorabi, Matak, & Jaczynski, 2012). Indirect effects of LEEB are caused by the ionization of water and/or oxygen molecules, leading to the generation of different reactive species. Since bacterial cells consist of up to 70% water, the generated electrons will be absorbed, generating highly reactive radicals, such as hydroxyl and superoxide radicals, affecting inter alia DNA and cell membranes (Chalise et al., 2007;Tahergorabi et al., 2012). Ghomi et al. (2005) treated E. coli with a pulsed LEEB of 5 μs generated by a secondary emission electron gun and suggested that cell death occurred by irreversible electroporation of the bacterial cells. However, spectroscopic and scanning electron microscopy analyses indicated no damage to the outer cell structure. The antimicrobial efficiency can vary between microorganisms, e.g., bacterial spores are more resistant than bacteria. Gram-negative E. coli is less resistant to LEEB than the Gram-positive B. subtilis (Chalise et al., 2004;Rahman et al., 2006). Low energy electrons generate secondary electrons with lower energy levels on their way to the treated surface due to ionization processes by inelastic collisions. This process can result in the creation of a so-called "electron cloud", which can also penetrate and cover the pores and cracks of structured surfaces (Bugaev et al., 1994). Nikjoo and Lindborg (2010) proposed that low energy electrons can generate more secondary electrons than high energy ones. Furthermore, it is well known that secondary electrons with lower energy levels damage DNA (Brun, Cloutier, Sicard-Roselli, Fromm, & Sanche, 2009;Folkard et al., 1993;Nikjoo, O'Neill, Goodhead, & Terrissol, 1997).
An important parameter to compare the effects on biological material of different ionizing radiations, such as different energy levels of EB, is the relative biological effectiveness (RBE), which is a dimensionless value that describes the effectiveness of one type of ionizing radiation in comparison to a reference radiation. The higher the RBE of an ionizing radiation type, the more effective it is compared to the other one. Bellamy, Puskin, Hertel, and Eckerman (2015) reported an RBE > 1 for low energy electrons compared with the reference of 1 MeV electron radiation, implying a higher antimicrobial efficiency of low energy compared with high energy EB. Table 3 provides a comprehensive overview about the application of LEEB for microbial inactivation on dry food products. The first studies were carried out by researchers in Japan at the end of the 20th century (Hayashi et al., 1997(Hayashi et al., , 1998a(Hayashi et al., , 1998bHayashi & Todoriki, 1999;Todoriki & Hayashi, 2000). The authors showed that LEEB using electron energies between 60 and 210 keV enables inactivation of the native microbial flora of different kinds of seeds (radish, alfalfa, etc.), varieties of rice and wheat, as well as herbs and spices (black and white peppercorn, coriander, basil). Hayashi et al. (1997) treated different rice varieties, wheat and buckwheat, and they reported a minimum electron energy for microbial inactivation between 75 and 160 keV and a necessary dose to reduce the native microbial load < 10 CFU g −1 between 6 and 20 kGy depending on the product. For the treatment of black peppercorns, a dose of 10-15 kGy and electron energy of 210 keV was  Table 3 Overview of studies investigating the decontamination of dry food products by low energy electron beam (LEEB). Some values/numbers have been approximated from graphs provided in the cited literature. • 10 min treatment, 7.5 kGy reduced the microbial load < 10 CFU g −1 Baba et al. (2004) Wheat, brown rice Soft Electron Processor  with 500 kg h −1 • BL: reduced by ∼2.6 log 10 after 12 min C. Hertwig et al. Trends in Food Science & Technology 77 (2018) 131-142 needed to reduce the native microbial load < 10 CFU g −1 . The decontamination of white peppercorns, coriander and basil required a minimum electron energy of 100 keV, but increasing the electron energy to 210 keV reduced the treatment time by a factor of 6 (Hayashi et al., 1998b). Complete microbial inactivation of alfalfa and radish seeds required an electron energy of up to 75 keV (Hayashi & Todoriki, 1999), whereas an energy of up to 190 keV was necessary for inactivation of the native microbial flora on Adzuki bean, pot herb mustard and black gram seeds (Todoriki & Hayashi, 2000). Baba, Kaneko, and Taniguchi (2004) used a Soft Electron Processor with a throughput of 500 kg h −1 for the treatment of wheat and brown rice and reported a required dose of 14 and 12.5 kGy to reduce the microbial load by 90%, respectively. Trinetta, Vaidya, Linton, and Morgan (2011) inoculated S. enterica Poona on cantaloupe and tomato seeds and E. coli O157:H7 on lettuce seeds and treated them with LEEB (150 keV, 7 kGy). The authors reported an inactivation depending on the treated seeds, with the highest inactivation determined for tomato seeds. LEEB can also be applied to inactivate seed-borne fungi that cause smut and bunt diseases (Röder et al., 2009). Furthermore, this technology also offers the possibility of inactivating the eggs of insect pests. A treatment dose of 0.48 kGy with an electron energy of 60 keV was sufficient to inactivate the eggs of Callosobruchus chinensis on the surface of adzuki beans, whereas the larvae inside the beans were not completely inactivated (Imamura et al., 2004).

Low energy electron beam treatment of dry food surfaces
The limited and controlled penetration depth of the low energy electrons reduced their effect on the product surface. Thus, it can be assumed that product-process interactions, which could affect the product quality, can be minimized. Hayashi et al. (1997) investigated the viscosity of a grain suspension after LEEB treatment. The researchers reported a decrease in viscosity with an increasing electron energy. However, the viscosity of grain suspensions treated with LEEB doses, necessary to reduce native microbial load < 100 CFU g −1 , was higher than those of grains exposed to γ-rays. The results suggested that the low energy electrons degraded only starch molecules near the surface. Todoriki, Kikuchi, Nakaoka, Miike, and Hayashi (2002) compared the gelatinized properties of soymilk of beans decontaminated with LEEB and temperature-treated soymilk, whereas the gelatinized properties were higher for LEEB treated samples. Hayashi et al. (1998a) also reported increasing TBA values for brown rice with an increasing electron energy, indicating lipid oxidation. The germination behavior of LEEB-treated seeds was also investigated, but no effect of the treatment on germination was shown (Fan et al., 2017;Todoriki & Hayashi, 2000;Trinetta et al., 2011).

Limiting factors
The main limiting factor of LEEB technology is probably the limited penetration depth of the low energy electrons, as well as limited knowledge in the science and industry community. Ghomi et al. (2005) investigated the antimicrobial efficiency of pulsed LEEB (80 keV) by treating different concentrations of E. coli. The authors reported a reduction of bacterial inactivation with increasing initial concentrations, which was attributed to the formation of cell agglomerates for higher concentration. The top microorganism layer was inactivated, but it also worked as an attenuator of the low energy electrons, thus reducing the antimicrobial effect, similarly to other surface decontamination technologies such as CAPP. Urgiles et al. (2007) compared LEEB (100 keV) and EB (10 MeV) by treating Bacillus pumilus, megaterium and subtilis spores. For LEEB, D-values of 1.34, 3.46 and 1.01 kGy were reported, respectively. Applying EB with 10 MeV resulted in much higher D-values of 2.12, 4.11 and 2.05 kGy. These results show the high potential of LEEB for surface decontamination since the electron energy is deposited on the surface. However, 100 keV has been reported to cause complete inactivation of the spores following a dose > 30 kGy for LEEB (Urgiles et al., 2007). The FDA (2017) has recommended a minimum dose of 44 kGy for the sterilization of food products by ionizing radiation used solely for space flight programs. Hayashi et al. (1997) and Hayashi et al. (1998b) reported different required electron energies and doses for the decontamination of different rice varieties, wheat, buckwheat and black and white peppercorns. Similar results were reported by Trinetta et al. (2011) andFan et al. (2017) for the treatment of different seeds. These findings indicated an impact of the treated surface morphology on the inactivation efficiency of LEEB, demonstrating that a complex surface structure can lower the antimicrobial efficiency of the treatment. The distance between the point of electron emission to the atmosphere and the product can also impact the inactivation process. The passage of electrons through the atmosphere to the target reduces their kinetic energy due to collisions with atmospheric molecules. Electrons with energies of 80 keV have a maximum "lifetime" in air of 7.1 cm, but during their traversal of this pathway, electrons can lose energy due to collisions (Ghomi et al., 2005).

Upscaling, consumer acceptance, regulatory and future research needs
To date, no commercial CAPP process for the decontamination of dry products has been applied in industry because no adequate CAPP systems are currently available that are validated, scalable to industrial requirements and cost-effective. The research community is focusing on process parameters, such as plasma sources and process gases, for maximization of microbial inactivation and minimization of the impact on the product quality, since a minimal impact on product quality is an essential factor for the implementation and acceptance of a novel food processing technology (Pankaj, Wan, & Keener, 2018). However, focusing on the application of air-plasma is probably recommended since the use of air is much more cost-effective than noble gases. Niemira (2012a) evaluated the process gas costs regarding the scale-up to industrial systems and showed the significant difference in process costs of CAPP systems using a noble gas such as helium, or air as well as mixtures of oxygen and nitrogen. The process gas costs for 1000 h of operation, calculating with a gas consumption of 500-4000 l min −1 , or 300-2400 m 3 h −1 , would be: noble gas helium $636,000-$9,096,000; nitrogen $9000-$72,000; oxygen $18,000-$144,000 and air ∼$0. Even though the addition of noble gases as a minor component of the process gas can enhance the inactivation efficiency, the specific application must be verified to justify the higher process gas costs (Niemira, 2012a). In addition, plasma equipment that is tailored for the food industry must be developed and designed, which can be easily integrated into existing process lines. Such equipment must be practical to operate, provide competitive economics compared with established technologies and guarantee safety by adequate insulation, grounding and shielding. An ideal industrial scale plasma system should be able to continuously process products with different characteristics, such as powders, granular and/or flat materials. Furthermore, a uniform treatment even for complex shaped products must be ensured. A key factor in this regard is a homogenous spatial distribution of the reactive plasma components. CAPP applications in the material science and medical sector demonstrated that non-uniformity treatment can be mitigate by the arrangement and geometry of the CAPP sources (Homola et al., 2017;Nie, Cao, Ren, Wang, & Kong, 2009). A primary focus for CAPP process validation is the complex plasma chemistry. Airplasma can generate more than 75 reactive species with approximately hundreds of simultaneous reactions (Keener & Misra, 2016;Sakiyama et al., 2012), which could lead to potentially undesirable product-process interactions. The regulatory approval for a CAPP process will require a large amount of data and time efforts because the complex plasma chemistry with its possible effects on the product must still be analyzed and assessed. Acceptance of data or conclusions reached from a regulatory review may significantly differ between countries due to the different regulatory requirements for such novel food processes (Keener & Misra, 2016). The most effective way to validate a plasma-C. Hertwig et al. Trends in Food Science & Technology 77 (2018) 131-142 based decontamination process would by a case-by-case assessment of a certain product. However, the manner in which consumers will react to products treated with cold atmospheric pressure plasma technology is still not clear. According to the Technology Readiness Level (TRL), an established system to estimate the technology maturity, CAPP can be rated with TRL 5, for which key elements have been demonstrated in a relevant environment. Moreover, the high variety of CAPP sources and set-ups opens the possibility of potential combination processes, such as simultaneous decontamination and drying processes. Future research should also focus on the promising results for the degradation of mycotoxins on different dry products to understand and optimize the degradation mechanisms. LEEB belongs to ionizing radiation regulation, and thus this technology is subject to the same legal regulations as other irradiation technologies considering the treatment of food products and labeling. In the United States, a maximum irradiation dose of 30 kGy is allowed for dry and dehydrated aromatic substances such as herbs and spices. In the European Union, the application of irradiation to dried aromatic herbs, spices and vegetable seasoning is harmonized, and a maximum average absorbed dose of 10 kGy is allowed (Pillai & Shayanfar, 2017). Although ionizing radiation has proven to be efficient, environmentally clean and energy-effective, it is rarely used in the EU because of its poor consumer acceptance in Europe (Schweiggert et al., 2007). Erroneously, the main consumer concerns in the EU have focused on the perceived "carcinogenicity" of irradiated foods and the association with 'radioactivity' (Frewer et al., 2011). However, a joint FAO/IAEA/WHO Study Group on High-Dose Irradiation (JSGHD, 1999) concluded that, from the perspective of human safety, any food may be irradiated at any dose, as reflected in a revision of the Codex General Standard for Irradiated Foods" (CAC, 2003) The first prototype "Soft Electron Processor" for the decontamination of wheat and rice with a performance of 500 kg h −1 was constructed by Nissin-High Voltage Co. in Japan (Baba et al., 2004) in 2004. However, to reduce 90% of the native microbial load, a dose of > 10 kGy was necessary, making this LEEB application not suitable for commercial use in the European Union. Nevertheless, it has recently been shown that LEEB is able to fulfill industrial requirements for the spice and herb industries, and a first industrial scale prototype, with TRL 7, was introduced in the second quarter of 2017 on the European market, demonstrating a capacity of one ton per hour (IIA, 2017). Although LEEB is already used on an industrial scale, studies investigating the impact on food products, and the potential for mycotoxin degradation are scarce. However, food products produced under established Good Manufacturing Practice and treated with doses greater than 10 kGy can be considered nutritionally adequate and safe (FAO/IAEA/WHO, 1999).

Conclusions
The demand by consumers for safe and high quality food requires the research community and industry to develop new technologies with clear pathways into the market. CAPP and LEEB are emerging nonthermal technologies with high potential for the gentle decontamination of dry food surfaces. Due to the limited and controlled penetration depth of both technologies, product-process interactions can be minimized by maintaining the product quality. LEEB has already been successfully introduced into the seed dressing as well as packaging industries and the first demonstrator for the decontamination of herbs and spices has been installed in the market. Compared with LEEB, CAPP is at an advanced development stage with TRL 5, and scalable systems for industrial requirements are needed. LEEB is at an early demonstration phase where research on several food commodities might be still necessary.
There is an urgent need for the introduction of novel technologies to consumers, since consumers are not only concerned regarding the safety and quality of their products but also regarding the applied technologies. The inclusion of all stakeholders in discussions based on scientific evidence and data could lead to successful market launches of these promising emerging surface decontamination techniques in the very near future.