Chemical-free and synergistic interaction of ultrasound combined with plasma-activated water (PAW) to enhance microbial inactivation in chicken meat and skin

In general, the poultry industry uses 0.5–1 ppm chlorine solution in the meat sanitization process. However, chlorine can react with organic material and produce halogenated organic compounds, notably chloroform, which causes bladder and rectal cancer in humans. For this reason, many industries try to avoid chlorine. This study investigated the efficacy of ultrasound and plasma-activated water (PAW) on the inactivation of Escherichia coli and Staphylococcus aureus in chicken muscle, rough skin, and smooth skin. Samples inoculated with bacteria suspension were treated by ultrasound alone and PAW–ultrasound. The Taguchi method and desirability function approach were used for the experimental design and optimization. Combined ultrasound and PAW inactivated up to 1.33 log CFU/ml of E. coli K12 and 0.83 log CFU/ml of S. aureus at a sample thickness of 4 mm, at 40 °C for 60 min, while PAW alone only reduced E. coli K12 by 0.46 log CFU/ml and S. aureus by 0.33 log CFU/ml under the same condition. The muscle topography showed a porous structure, which facilitated the penetration of PAW. The color measurements of muscle treated with ultrasound and PAW–ultrasound were dramatically different from the untreated sample, as also perceived by the sensory evaluation panel. Therefore, the synergistic interaction of combined PAW–ultrasound could be used to enhance microbial inactivation in meat.

Another low-temperature technique that has been used to inactivate bacteria is ultrasound. The advantage of ultrasound, besides the decrease in microbial load, is the increase in the water holding capacity of meat, preventing marinate and water losses in fresh meat 9 . The cavitation bubbles generated by ultrasound can have varying effects within the liquid medium, depending on whether the system is a homogenous liquid, heterogeneous solid/ liquid, or heterogenous liquid/liquid type 10 . In the homogeneous liquid phase system, the collapse of the cavitation bubbles generates extreme high-temperature and pressure conditions at low-frequency ultrasound (16-100 kHz), which produces H atoms and •OH radicals 11 . Moreover, DNA damage occurs at high temperature, due to the ultrasonic intensity above the cavitation threshold can generate RONS 12 . The accepted inactivation mechanisms of ultrasound are acoustic cavitation, micromechanical shockwaves, compression and rarefaction, and sonochemical reactions 13 . The RONS chemical reactions triggered by ultrasound-induced decomposition of water are described by Eqs. (9)- (14). The pyrolysis reactions in Eqs. (12) and (13)  The ultrasound involves heterogeneous solid surface-liquid systems, which show a different behavior from cavitation bubble collapse in bulk liquid 14 . The behavior is like a liquid jet, targeted at the surface of the solid in the liquid system. The interfacial boundary layers are disrupted by the heat and mass transfer effects at the surface of the solid 15 . Low-temperature (59 °C) and low-pressure (400 kPa) ultrasound can reduce 5 log CFU/ml of Escherichia coli K12 in apple cider 16 , highlighting the inactivation efficacy of ultrasound treatment on bacteria.
Both PAW and ultrasound are considered as environmentally-friendly and cost-effective techniques. Royintarat (2019) has been proved that arc plasma discharge combined ultrasonic can reduce bacteria on chicken meat part 17 Therefore, this research focused on using PAW from underwater jet and ultrasound to increase the efficiency of bacterial reduction in different types of chicken meat and skin. The research also investigated the pH value, electrical conductivity (EC) value, oxidation-reduction potential (ORP) value, and •OH and H 2 O 2 concentrations generated by PAW and ultrasound, which could play a significant role in the antimicrobial mechanism. This combined technology approach could be employed for bacterial inactivation in the poultry industry to replace the chlorine sanitization process. Moreover, this study will focus on using the desirability function approach (DFA) to analyze the relationship between variable factors and responses of PAW towards bacteria reduction in chicken meat and skin.

Results
Physicochemical properties of PAW and ultrasound. As shown in Fig. 1b 18 . Figure 2b indicates that the ORP value of PAW increased for DI water at 4 and 25 °C, but was stable at 40 °C. At high temperature, the chemical reactions could reduce the oxidation potential of PAW. Shen et al. 19 investigated the physicochemical properties of PAW stored at various temperature (−80, −20, 4, and 25 °C) and noticed that the ORP value was lower when the storage temperature was higher, indicating an inverse association between the temperature and ORP value, according to the Nernst equation.
The EC value demonstrated that the concentration of ions increased from 1 to around 8 μS/cm after treatment by non-thermal plasma. The ultrasound treatment increased the EC of DI water, but not for PAW. However, the temperature change was not affected by the EC value (Fig. 2c). The efficiency of the EC value for killing bacteria can be explained by a physical mechanism called electrostatic disruption 20 . The electrostatic disruption from the accumulation of charged particles disrupts the cell membrane of bacteria when the electric force overcomes the tensile strength.  In general, the collapse of large cavitation bubbles generated under ultrasound produces extremely high-temperature and pressure in the cavitation zones, generating H atoms and •OH radicals 11 . The experiment shows that the ultrasonic condition is below the threshold, so the H atoms and •OH radicals did not appear.

Desirability function Approach (DfA)
The optimum level of parameters satisfied the maximal reduction of E. coli and S. aureus. All five steps of the optimization process are presented below: Step 1: The results for chicken muscle, rough skin (Table 1), and smooth skin ( Table 2) were calculated by Eq. (27) to normalize the value in the range 0-1. Then, these normalized values were raised to the power of 0.5.
Step 2: The weighted d i values were used in Eq. 28 to obtain the d G , which was optimized in the next step. The total mean value of the d G was compared with the predicted d G in the final step. For chicken muscle, PAW = 0.4587 and PAW-ultrasound = 0.4263. The rough chicken skin of PAW = 0.5993 and PAW-ultrasound = 0.5963. The smooth, thin chicken skin of PAW = 0.4546 and PAW-ultrasound = 0.4632.
Step 3: The optimum condition was determined by considering the maximum level of each parameter. The maximum level will be the best factor, as shown in Table 1. Then, all best factors are combined to give an optimal condition.
Step 4: In this step, ANOVA was performed to investigate the significance and influence of each parameter ( Table 2). It indicated there was no significant difference in the factors between the chicken muscle and smooth skin (p > 0.05). However, rough chicken skin had significantly different skin thickness (p < 0.05). The influence of the parameters is represented as %C. For chicken muscle (PAW %C = 54%, PAW-ultrasound %C = 45%) and smooth skin (PAW %C = 98%, PAW-ultrasound %C = 99%), the temperature was the most influential parameter affecting the bacterial reduction. For rough skin, the skin thickness had the most influence on the bacterial reduction (PAW %C = 80%, PAW-ultrasound %C = 70%). The main effect plot showed a positive relationship among all the parameters with the responses, as shown in Fig. 3.
Step 5: The predicted value of the response was calculated based on the optimum level of the parameters. Table 1 provides the optimal condition for treating chicken muscle, rough skin, and smooth skin. For chicken muscle, the optimal condition was 40 °C, 60 min, and 4-mm thickness (A3B3C3), which presented an improved desirability value of 0.6484 (PAW) and 0.4763 (PAW-ultrasound). When applying the optimal condition to confirm the experiment, PAW and PAW-ultrasound reduced E. coli and S. aureus in chicken muscle by about 0.46 and 0.30, and 1.33 and 0.83 log CFU/ml, respectively. For rough skin, the optimal condition was also 40 °C, 60 min, and 4-mm thickness (A3B3C3), which presented improved desirability values of 0.5130 (PAW) and 0.5307 (PAW-ultrasound). The confirmation PAW and PAW-ultrasound experiments reduced E. coli and S. aureus on rough skin by about 0.56 and 0.43, and 1.12 and 0.86 log CFU/ml, respectively. Moreover, the optimal condition of PAW for smooth skin was 40 °C, 60 min, and 1-mm thickness (A2B2), which presented improved desirability values of 0.5000 (PAW) and 0.5000 (PAW-ultrasound). The confirmation PAW and PAW-ultrasound experiments on smooth skin reduced E. coli and S. aureus by about 0.35 and 0.08, and 0.73 and 0.10 log CFU/ml, respectively.
= . + . + . E coli reduction (log CFU/ml) 0 0105 0 0072 temp 0 0008 time (23) . = . + . + . S aureus reduction (log CFU/ml) 0 0379 0 0008 temp 0 0001 time (24) . = . + . + . E coli reduction (log CFU/ml) 0 0744 0 0139 temp 0 0017 time (25) . = . + . S aureus reduction (log CFU/ml) 0 0506 0 0011 temp (26) According to Taguchi experimental design, the regression Eqs. (15) to (26) derived from this study can be mathematically predicted the responses for reproduction of further study. The prediction of responses will be based on the changing of significant factor and the coefficient of the equation. Additional predicted value was shown as surface plot graph models in the supplementary. For instance, when the parameters of Eq. (18) Figure 4 displays the topography of the different types of chicken meat treated by DI water and PAW-ultrasound, as visualized by SEM. A comparison between the surface of chicken muscle treated with DI water only (Fig. 3a) and PAW-ultrasound treatment (Fig. 3b) showed the surface became more porous with ultrasound, which allowed PAW to penetrate the bacteria. However, the surface of rough and smooth skin was not different between the control (DI water) and PAW-ultrasound because of the denser texture and greater flexibility of the skin than muscle. Figure 5 shows the morphology of E. coli and S. aureus on muscle and skin before and after PAW treatment. In chicken muscle, both bacteria remained as a group attached to the fibrin (Fig. 5a,c,g). Additionally, some S. aureus on chicken muscle stayed separated and embedded in the tissue (Fig. 5e). After PAW treatment, the membranes of E. coli were broken, but the bacteria retained its rod shape (Fig. 5b). Severe physical damage was induced by PAW treatment on S. aureus (Fig. 5f). On the contrary, the SEM observations of E. coli and S. aureus on chicken skin indicated that both bacteria retained their shape, but the surface morphology was altered, and bubble-like protrusions appeared on the bacterial membrane (Fig. 5d,h).

Quality analysis
Color measurement. The average and standard deviation of L*, a*, b*, and ΔE are shown in Table 3. The perceivable color change can be categorized as 'very distinct' (ΔE > 3), 'distinct' (1.5 < ΔE < 3), and 'small differences' (ΔE < 1.5) 21 . In this study, the ΔE values of chicken muscle and rough chicken skin were both found to be 'very distinct' when treated by ultrasound and PAW-ultrasound, and 'distinct' for PAW treatment, whereas, the ΔE values of smooth chicken skin were 'very distinct' under all conditions. When ΔE > 3, the human eye should be able to detect the color differences distinctly 22 . However, the sensory panel did not notice differences in the color of chicken muscle due to ultrasound and PAW-ultrasound.
Hardness measurement. The hardness of chicken muscle, rough skin, and smooth skin did not change significantly due to the ultrasound, PAW, and PAW-ultrasound at 40 °C for 60 min (Table 3). This result might have been because of the low amount of energy dissipated to the system by the ultrasonic bath 23 . According to Lyng et al. 24 , the intensity of ultrasound was insufficient to change the muscle properties and may have been too low to alter the texture due to the high content of connective tissue in chicken meat.
Protein and lipid measurement. There were no significant differences in the protein and lipid contents between the non-treated chicken meat and chicken meat treated by PAW-ultrasound. In general, non-thermal plasma can decrease the protein content of a sample by heat, charged particles, the intense electric field, UV photons, and neutral reactive species. However, all five factors are presented only in gas discharge 25 .
The chicken muscle has less lipid than the chicken skin, and the rough chicken skin has more lipid than the smooth chicken skin. Samples treated with PAW-ultrasound were not significantly different from their untreated counterparts. It is likely that because of the lower lipid amount in chicken muscle than skin, the most damage induced by the ROS in PAW and ultrasound was due to the interaction of the ROS with the bacterial cell wall rather than the chicken skin.

Discussion
The key inactivation agents of non-thermal plasma are RONS. This research focuses on the ROS in PAW, notably the long-lived ROS, such as H 2 O 2 , which is responsible for the bacterial reduction. The bacteria inactivation mechanism can be explained by the following four steps. In the first step, the underwater Ar plasma jet produces Figure 5. The SEM of E. coli and S. aureus topography on chicken (a) E. coli shape on muscle wash with DI water (b) E. coli shape on muscle wash with PAW-ultrasound (c) E. coli shape on skin wash with DI water (d) E. coli shape on skin wash with PAW-ultrasound (e) S. aureus shape on muscle wash with DI water (f) S. aureus shape on muscle wash with PAW-ultrasound (g) S. aureus shape on skin wash with DI water (h) S. aureus shape on skin wash with PAW-ultrasound.

Scientific RepoRtS |
(2020) 10:1559 | https://doi.org/10.1038/s41598-020-58199-w www.nature.com/scientificreports www.nature.com/scientificreports/ high amounts of ROS. In the second step ROS, especially •OH, initiate lipid peroxidation of the lipid bilayer in the bacterial cell membrane, leading to changes in cell permeability and depolarization of the cell membrane. In the third step, ROS move through transient pores and induce oxidative stress in the cell, to increase intracellular ROS. In the forth step, intracellular ROS react with proteins, lipids, and carbohydrates, leading to alteration of molecule structures and chemical bonds. However, Zhang et al. 24 reported a reduction in ROS when adding organic matter to PAW. Therefore, excess intercellular ROS and low-pH induced redox reactions and disrupted pH homeostasis in the cell, causing cell death 26 . In turn, the ultrasound increased the penetration rate of PAW through the samples.
The bacteria morphology on chicken muscle exhibited severe physical damage as compared with the bacteria on the skin, due to more organic matter in the skin than muscle. According to Almeida et al. 27 , chicken has a high content of polyunsaturated fatty acids (PUFA) of about 21.3 ± 3.5%. A high concentration of omega-3 and omega-6 PUFAs can inhibit ROS and RNS formation 28 . The ROS target C-H bonds, especially double bonds because of the lower energy required to abstract a hydrogen atom (272 kJ/mol) than another C-H bond (422 kJ/ mol) 29 . Therefore, the efficiency of ROS to inhibit bacteria present on chicken skin was decreased, indicated by the protrusion of the bacterial membrane and not the disintegration of bacteria evidenced on the chicken muscle.
The ultrasound process caused membrane disruption, which released unsaturated lipids that reacted with free radicals in PAW. This process is called lipid oxidation. Most of lipid oxidation reaction were found in muscle foods which caused the change of color, odor, taste, and shelf life. There are many studies that found lipid oxidation such as fresh pork and beef treated with dielectric barrier discharge plasma 30 , pork loin treated with helium and oxygen plasma 31 , and also ground pork treated with plasma jet 32 .
In this study, our technique dramatically inactivated the number of E. coli 2.12 and 1.37 log CFU/ml at the initial microbial 10 5 and 10 6 CFU/ml, respectively, in the case of real situation at low concentration of microorganisms contaminated in foods. Those results agreed to the reports at the difference of initial microbial effects with the amount of bacterial inoculum with the levels of 10 2 , 10 3 , 10 4 CFU/ml of S. enterica and were reduced the survival of microorganism to 92.5%, 87.0%, 63.5% on chicken breast and 62.5%, 36.0%, 32.8% on chicken skin, respectively 6 . PAW-ultrasound 6.63 ± 0.92a 6.00 ± 1.07a 6.25 ± 0.71a 6.25 ± 0.89a Table 4. Sensory evaluation from 10 specialist using 9-point hedonic of muscle chicken meat, rough skin chicken, and smooth skin chicken after treated by different technique at 40 °C. Values are given as mean ± SD (n = 10). Means in the same column followed by the same letter are not significantly (P > 0.05) different by Turkey's test with error rate = 5.

Plasma device, PAW generation, and ultrasound treatment. A volume of 20 ml of DI water
(Milli-Q ® Direct Water Purification System, Merck KGaA, Darmstadt, Germany) was used in the experiment (Fig. 1). The single-electrode non-thermal atmospheric pressure underwater plasma jet with a direct current (DC) positive fly-back transformer was used to generate PAW in this experiment 33,34 . The device consists of a copper wire as the single electrode cover with a quartz tube and stainless-steel ground. The reactor is connected to a 1.5-kHz square high-voltage source with a 6.8-kV peak-to-peak voltage. The device is set up to discharge PAW beneath the water surface at a distance of 5 mm between the electrode and ground. This setup is the optimized condition of the device, which discharges PAW using Ar gas at a flow rate of 3 slm for 6.5 min.
The PAW volume was based on the recommendation of one of the poultry industries in Thailand. Therefore, the tissue sample at 1 (0.08 ± 0.01 g), 2 (0.15 ± 0.03 g), 3  Statistical analysis. One-way analysis of variance (ANOVA). The physicochemical properties (pH, ORP value, EC value, •OH concentration, and H 2 O 2 concentration), bacterial log reduction values, and quality analyses (color, hardness, proximate analysis, and sensory evaluation) of DI water, and DI water with ultrasound, PAW, and PAW-ultrasound were subjected to one-way ANOVA, followed by Tukey's test at a 5% significance level, using Minitab 16 statistical software, Minitab, LLC, PA, USA.
Taguchi experimental plan. The optimization of the meat conditions was investigated by the desirability function embedded in Taguchi analysis using Minitab 16 software to construct and analyze the experiments. As there are three factors for each chicken part; temperature, thickness, and treatment time, an L9 Taguchi orthogonal design was used to investigate the experimental parameters of chicken muscle and rough skin. The design summary is shown in Table 5. However, for the smooth skin, with a single thickness of 1 mm, a 2-factor, 2-level (L4) Taguchi experimental design was utilized.
Desirability function approach (DFA). To obtain the best results of all responses simultaneously, the DFA 36 described by Derringer and Suich 37 was used. The desirability function is strongly suggested for optimizing the multiple response problem 38 . It converts each response into an individual desirability function (d i ) that has values between 0 to 1 (least to most desirable, respectively). In this research, the factors are temperature, time, and thickness. The responses are E. coli and S. aureus log CFU/ml reduction. The factor setting with nominal total desirability is the optimal parameter conditions. Therefore, the DFA optimization has five steps, as follows: Step 1: Calculate the d i 35 . In this case, the maximal bacterial reduction of both E. coli and S. aureus were investigated. The equation of individual desirability of the-larger-the-better is shown in Eq. (15): (2020) 10:1559 | https://doi.org/10.1038/s41598-020-58199-w www.nature.com/scientificreports www.nature.com/scientificreports/ˆ= where y min and y max represent the lower and upper tolerance limit of ŷ, respectively; and r indicate the weights that depend on the user.
Step 2: Evaluate the composite desirability (d G ) by the following equation: where d i is the individual desirability of the property Y i ; w i is the weight of Y i , and W is the sum of the individual weights.
Step 3: Optimize the d G value. The highest d G indicates better product quality.
Step 4: Investigate the significance of each parameter by using ANOVA.
Step 5: Calculate the prediction of the optimal condition and validate the results. Quality of chicken meat analysis. Color measurement. The surface color of the muscle, rough skin, and smooth skin was measured before and after treatment by using a Minolta CR-300 colorimeter (Minolta Co., Ltd, Osaka, Japan) to record the CIE L* (lightness), a* (redness/greenness), and b* (yellowness/blueness) coordinates. ΔE is an indicator of the overall chicken sample color changes 39 , which is calculated as: where ΔL*, Δa*, and Δb* are the differences in sample color before and after treatment.  www.nature.com/scientificreports www.nature.com/scientificreports/ Hardness measurement. The hardness of chicken samples before and after treatment was analyzed using a TA.XTplus texture analyzer (Texture Technologies Corp., Stable Micro Systems, Ltd., Godalming, UK) with a strain of 70%, trigger force of 10 g, and 3.5-cm-diameter flat-faced cylindrical probe.

Microbiological analysis.
Protein and lipid measurement. The samples were analyzed for protein content using a combustion technique (Leco FP-528, USA) and adopting Jones' factor for meat (6.25) to convert the nitrogen content to protein content (Eq. (30)). Crude lipid content was determined by Soxhlet extraction with petroleum ether (Soxtec TM 8000 Extraction Unit, Foss Company, Denmark) and applying Eq. (31) to calculate the results. = − × Lipid(%) [Weight after extraction (g) Weight before extraction (g)] 100/Sample (g) (31) Sensory evaluation. The sensory attributes of appearance, color, texture, and acceptability of the product were evaluated using a 9-point hedonic scale (9 = like extremely; 1 = dislike extremely). The chicken muscle, rough skin, and smooth skin (untreated, ultrasound-treated, PAW-treated, and PAW-ultrasound-treated) were sensory evaluated by a 10-member panel consisting of staff members from the poultry industry who were previously experienced in quality evaluation.

conclusions
The desirability function with Taguchi can be used for optimizing the bacteria reduction by using PAW and PAW-ultrasound with different types of chicken meat. The SEM results showed that ultrasound could destroy the chicken muscle cells, facilitating the entry of ROS from PAW into the bacteria. Moreover, the increasing temperature could lead to a weakened cell wall of the bacteria and leave the cell membrane less protected from other treatment. The muscle meat at 4 mm thickness, soaked in PAW at 40 °C for 60 min reduced the bacterial load of E. coli by 0.46 log CFU/ml and S. aureus by 0.30 log CFU/ml. Under the same condition, PAW-ultrasound reduced the E. coli count by 1.33 log CFU/ml and S. aureus by 0.83 log CFU/ml. For the rough chicken skin, the thickness of the skin was the most important factor affecting the bacteria reduction (p < 0.05) by both PAW and PAWultrasound. The optimal condition, which was the same as that of the chicken muscle, reduced E. coli by 0.56 and 1.12 log CFU/ml and S. aureus by 0.43 and 0.86 log CFU/ml when treated with PAW and PAW-ultrasound, respectively. For smooth chicken skin, the best treatment (40 °C for 60 min) reduced E. coli by 0.35 and 0.08 log CFU/ml, and S. aureus by 0.73 and 0.10 log CFU/ml, for PAW and PAW-ultrasound, respectively. The main effect factor plot showed temperature had more influence on the physicochemical properties of PAW than the activated time (p < 0.10). The hardness, protein, and lipid measurements of the treated samples were comparable to those of their untreated counterparts. This study proved that the synergistic effect of combined PAW and ultrasound was more effective for reducing the bacterial load than either technique alone, which was significant with natural contamination.