Bacterial decontamination of process liquids and paints in E-coating lines by pulsed electric ﬁeld treatment

Cultivation-based and DNA-based methods for determining the bacterial load and the composition of the bacterial spectrum have been successfully established for media in electrodip painting, and used for the detailed analysis of the contamination situation in an E-coating system of an automobile plant in Germany. Dominating representatives of the genus Microbacterium spp., the orders Burkholderiales and Pseudomonadales , the family Cytophagaceae and the genera Corynebacterium spp . , Sphingomonas spp., and Stenotrophomonas spp . were used for inactivation experiments. Different pulsed electric ﬁeld (PEF) parameters were studied for an effective and target-directed inactivation of deﬁned bacterial suspensions containing mixtures of Gram-positive as well as Gram-negative bacteria, but also single species suspensions in adequate liquids. PEF treatment with pulse durations longer than 1.0 l s effectively killed bacteria even in low conductivity media, regardless of whether the pulses were unipolar or bipolar, indicating that the choice of pulse shape does not limit the design of the PEF system. Model calculations showed that for efﬁcient treatment in bypass mode, a high treatment ﬂow rate is required rather than a high inactivation efﬁciency of the PEF treatment. By using speciﬁc treatment parameters, such as bipolar pulses of 50 k Vcm (cid:2) 1 and a treatment energy of 40 J mL (cid:2) 1 , a signiﬁcant reduction in both Gram-negative and Gram-positive bacteria (> 2 log 10 reduction) can be achieved while minimizing electrode corrosion and coating degradation. PEF treatment proves to be an effective alternative to the use of biocides in an E-coating system and can help maintain a bacteriostatic environment in the system by operating at different points, in transfer ﬂow or bypass mode, ensuring biocide-free


Introduction
As electrophoretic coating (E-coating) technology continues to improve, new waterborne paints have been developed that help to make processes more environmentally friendly.5][6] Disinfection measures, paint properties and the problem of contamination must be coordinated as part of a hygiene concept. 7n both anodic dip painting (ADP) and cathodic dip painting (CDP), the parts to be coated first pass through various preparation processes that require large quantities of demineralized water. 2,8As the water is pumped in and cascades through the circuit, the tanks in the E-coating line and the rinse water become contaminated, leading to a significant bacterial contamination over time.In the past, the paint mixtures killed these bacteria during the dipping process, but due to changing environmental requirements and the near-complete substitution of organic solvents and heavy metals, paints are becoming more and more microbiologically sensitive. 4An increased bacterial load can lead to a reduction in the quality of the coating and an increase in water consumption coupled with a reduction in the service life of the E-coating system due to more frequent cleaning. 9In the worst case, the entire contents of the E-coating dip tank can be lost.Therefore, targeted disinfection measures are necessary to limit the uncontrolled bacteria proliferation.
The regular way to reduce bacterial proliferation is to use certain chemical biocides in the pretreatment and E-coating dip tanks.This is done at regular intervals and on demand.However, not every disinfectant is up to the task; moreover, some products are completely incompatible with individual paint mixtures or cleaning baths. 10For this reason, careful selection and dosing of biocides are necessary to prevent coating failure.In addition, there is a risk that biocide-resistant bacteria will emerge as a result of the extensive use of biocides. 1,7This would render the biocides useless, with unpredictable consequences for maintaining troublefree operation.1][12] In the future, legislative restrictions on biocide concentrations are expected to increasingly limit the use of biocides.It is therefore essential to search for alternative methods to ensure sustainable operation of the paint shops while acknowledging environmental considerations.
An alternative method for the reduction of bacterial contamination in liquids is the pulsed electric field (PEF) treatment.PEF treatment is a physical method wherein the cell membrane undergoes permeabilization due to an applied external field. 13,14The external field induces high electric field strengths across the cell membrane, resulting in forming of aqueous pores into the lipid bilayer. 15This phenomenon, termed electroporation, finds applications in various fields, such as medicine, genetics, food production, and biotechnology. 16,17In principle, the increased permeability of cell membranes facilitates the molecular exchange between cytoplasm and external medium across the lipid membrane.6][37][38] Once these pathogens enter surface waters, the spread of bacteria is irreversible and the risk of contamination of water resources increases with the diversity of harmful bacteria. 23acterial inactivation of more than 5 log 10 was achieved by a combination of heating to 55 °C and PEF treatment. 37As with industrial fluids, wastewater contains different mixtures of bacteria and the fluid properties, such as conductivity and temperature, can vary, requiring a high degree of flexibility in the technology.It was also shown that a large variety of undesirable microorganisms, among them Gram-negative, Gram-positive bacteria, and fungi, were effectively reduced by PEF treatment. 31he aim of this study within the framework of an R&D project was to develop resource-efficient water management and plant concepts by integrating PEF technology, with the objective of maintaining a bacteriostatic milieu in the system that guarantees biocidefree and trouble-free operation. 39As part of the study, PEF treatment was used for the first time to decontaminate E-coating fluids.This purely physical and automated process can significantly reduce the use of chemicals and is effective even for turbid liquids with chemically sensitive components, such as cathodic dip painting.

State of the art: analysis of bacterial contamination in an automotive paint shop
For this study we used bacteria strains identified along the E-coating process sequence in a paint shop from the automotive industry, as reported in DiWaL Report 2021, BMBF Grant. 39For a better understanding of the sites identified as sources of contamination, a simplified schematic of the process sequence is shown in Fig. 1.The process is divided in three main zones: pretreatment (VBH-07 to VBH-11), E-coating (KTL-01), and post rinse zone (KTL-02 to KTL-05).From the entire pretreatment process this scheme shows only the phosphating (VBH-07) and cleaning stages (VBH-08 to VBH-11).In the study (DiWaL Report 2021), 11 pretreatment dip tanks (VBH-01 to VBH-11), the E-coating dip tank (KTL-01), and the four post rinse stages (KTL-02 to KTL-05) were investigated.Bacteria of the family Burkholderiaceae, Sphingomonadaceae, and Micrococcaceae were identified in the pretreatment zone, whereas only Burkholderiaceae spp. was found in the post rinse zone. 39The last tank (deionized water rinsing tank VBH-11) in the pretreatment zone is prone to bacterial contamination, particularly from dripping water from car bodies and from deionized water inlet.This tank was identified as the main source of bacterial contamination of the E-coating tank, with bacterial concentrations of $ 10 3 cfu mL À1 .According to this study, no bacteria could be cultivated in samples from the E-coating tank, KTL-01, or in the two subsequent post rinse tanks (KTL-02 and KTL-03), i.e., no CFU could be determined.While no bacterial DNA could be isolated in KTL-01, DNA was extracted from the samples of the two subsequent stages.This resulted in a very narrow spectrum of bacterial colonization, predominantly belonging to the order Burkholderiales.In contrast, a very high bacterial load with over 10 5 cfu mL À1 was found in the post rinse tanks KTL-04 and KTL-05, which was supposed to be the second main contamination source of the E-coating tank, due to the permeate backflow.

Identification and cultivation of bacteria
Bacterial population in the 200 mL water samples of the different tank systems was investigated by conducting PCR-DGGE analyses targeting the V1-V3 region of the Eubacteria 16S rRNA genes with an amplicon size of 509 bp (base pair) after initial membrane filtration (0.2 lm Nucleopore) and DNA extraction using a commercial kit system (FastDNA TM SPIN Kit for Soil, MP Biomedicals, Santa Ana, USA).Each PCR reaction included a 2.5 lL buffer (Extra Buffer, 15 mM MgCl2, VWR Life Science, Germany), 0.5 lL of deoxyribonucleotide triphosphate (DNT, 10 lM, VWR Life Science, Germany), 0.25 lL of each primer (40 nM, Eurofins Genomics, Ebersberg, Germany).The primers sequences according to Muyzer et al. with GC 27F (GC-Clamp) 5-AGAGTTT-GATCCTGGCTCAG-3 containing a GC-clamp (i.e., 5-CGC CCG CCG CGC CCC GCG CCC GTC CCG CCG CCG CCC CCG CCC C-3) and 518R 5-ATTACCGCGGCTGCTGG-3, 0.125 lL of Taq polymerase (TEMPase Hot Start DNA Polymerase, VWR Life Sciences, Germany), and 1 ng lL À1 template (10 ng DNA per sample) were used, and the volume was adjusted to 25 lL by adding sterile PCRwater (DNase-and RNase-free water, MP Biomedicals, Santa Ana, USA). 40The temperature profile of PCR thermocycler (C1000 Touch, Bio-Rad, Feldkirch, Germany) consists of 3 min at 95 °C followed by 30 times at 95 °C for 30 s, 56 °C for 1 min, and 72 °C for 2 min.The prokaryotic 16S rDNA PCR amplicons were controlled by a 1% agarose gel electrophoresis (Serva, Heidelberg, Germany), together with reference bacterial 16S rRNA gene amplicons, using SYBR Gold (Invitrogen, Karlsruhe, Germany) fluorochrome for DNA band visualization with F1 Lumi-Imager (Roche Diagnostics, Mannheim, Germany).
The denaturing gradient gel electrophoreses (DGGE) gel was made with a linear urea gradient from 40-70%.Reference markers [Escherichia coli (DSM 1103), Pseudomonas aeruginosa (DSM 1117), Enterococcus faecalis (DSM 20478), Enterococcus faecium (DSM 20477), Staphylococcus aureus (DSM 2569), and Stenotrophomonas maltophilia (SMK279a)] were generated and used as a ladder to enable meaningful comparisons across different DGGE gels.A maximum of 15 lL of the PCR samples, containing equal amounts of PCR product, were loaded onto the gel.The running time of the gel was 17 h at 70 V and 56 °C.The DGGE gel was analyzed via an F1 Lumi-Imager workstation (Roche Diagnostics, Mannheim, Germany) using the Lumi-Imager software (LumiAnalyst 3.1).The excised DNA band samples from the DGGE were given to Eurofins Genomics (Ebersberg, Germany) for sequencing using MiSeq, with 2 9 300 bp targeting the V1-V3 16S rRNA gene region.The sequences were aligned with public rDNA data bank for taxonomic identification. 41,42n the cultivation approach, bacterial quantification was carried out by determining the colony-forming units (CFU) on two agar nutrient media in order to be able to detect the widest possible spectrum of bacteria.For this purpose, a nutrient-rich medium (LB broth, Luria/Miller, Roth, Karlsruhe, Germany) and a nutrient-reduced mineral medium (R2A-Agar, Merck, Germany) were used.As is known, only a fraction of the bacteria present can be cultivated on synthetic culture media.

Sampling and preparation of bacterial model suspensions
Bacteria of the families Burkholderiaceae, Sphingomonadaceae, and Micrococcaceae were selected.The selection was made based on the concentration of bacteria and the bacteria identified as sources of bacterial contamination along the E-coating process.From these families the representative bacterial species, Sphingomonas gimenis, Gram-negative, Microbacterium maritypicum, Gram-positive, and Burkholderia cepacia, Gram-negative, were isolated and used to inoculate model media.Prior to inoculation into the sample, bacteria were cultivated for 24 h at 30 °C on LB medium, on shaker, to reach a bacterial concentration around 10 8 cfu mL À1 .
Model suspensions were used to test the bacterial inactivation efficiency of PEF treatment of representative bacteria in different process samples and growth media.One reason for using model suspensions is that some of the process samples from the paint shops, identified as a source of bacterial contamination, have a relatively low bacterial concentration below 10 3 cfu mL À1 , which is not sufficient to accurately study inactivation rates above 3 log 10 units.Secondly, systematic studies on PEF parameter dependency with scattering microbial composition, for example due to different sampling points and times, are not expedient.Samples were collected from the pretreatment zone (VBH-11 tank) and the post rinse zone (KTL-02 to KTL-05), along with several deionized water tanks used to compensate for any water loss.Native samples from VBH-11 and KTL-04 tanks were stored in a refrigerator at 4 °C and tempered at room temperature before PEF treatment.To prepare the model suspensions, liquid samples from paint shops were each enriched with one of the representative bacterial species up to a concentration of 10 6 cfu mL À1 (see Table 1).Prior to enrichment with externally cultured bacteria, the sample matrices were sterile filtered through a 0.2 lm pore syringe filter (VWR Life Science, Germany).In addition, bacterial suspensions in LB medium were prepared using the same procedure.For the preparation of model suspensions, representative bacteria, cultivated as described before, were centrifuged at 10,000 9 g for 5 min (Biofuge Pico, Heraeus, Germany) to remove the growth medium.The resulting bacteria pellet was resuspended in model media or LB medium.Table 1 shows the model suspensions from selected tanks that were identified as sources of bacterial contamination and the isolated bacteria used to enrich the model media.

Determination of bacterial viability
The viability of bacteria after PEF treatment was determined by colony counting on agar plates (LB-Agar, Roth, Karlsruhe) after incubation for 24-48 h at 30 °C.After serially diluting of untreated and treated samples, aliquots of 200 lL were plated on agar media (CASO, casein-soja bean agar, Merck, Germany).After incubation, the numbers of CFU on agar plates were counted.The CFU was calculated as an average of CFUs of at least three plates.Bacterial inactivation (I) after PEF treatment is given as log 10 reduction of viability: I = log 10 (N total /N viable bacteria ).

PEF treatment
Experimental determination of PEF treatment parameters required for effective inactivation of bacteria in aqueous process media and growth media were carried out using a homemade laboratory transmission line pulse generator.The transmission line pulse generator based on 50 X coaxial cables (RG 213, Belden, Villingen-Schwenningen, Germany) delivered squarewave pulses with pulse durations ranges from 100 ns to 10 ls and pulse amplitudes up to 25 kV, as described by Frey et al. 38 Commercially available electroporation cuvettes (BTX Instrument Division, Holliston, MA, to achieve parallel resistances of 52.0 X.The electrolyte was circulated by a peristaltic pump (40 mL min À1 , Ismatec ecoline, Germany) during operation through an inlet and an outlet attached to the outer cylinder to avoid resistance changes due to Joule heating.The capacity volume of the circulated electrolyte was 5 L. The pulse repetition rate was either 1 Hz or 5 Hz.The specific PEF treatment energy (W) was set to 40 ± 2 J mL À1 and 80 ± 3 J mL À1 by varying the number of pulses (N).For this, the number of pulses, for given pulse duration (t p ), conductivity of the suspension (j e ), applied electric field strength (E), and required specific treatment energy (W), was calculated using equation (1).
Bipolar square-wave pulses were generated by a bidirectional pulse forming line.For this special form of transmission line, after charging the line, one end of the line is short-circuited by means of a spark gap at the temporal start of the pulse.Thus, a pulse shape is formed between the two inner conductors of the coaxial cable which, viewed in time, exhibits a polarity change in the middle of the pulse.As a result, in the further course of the pulse, a first positive half-wave is followed instantaneously by another half-wave with reversed polarity but the same amplitude.For the PEF treatment, bipolar pulses with the same specific pulse energy (40 J mL À1 and 80 J mL À1 ), pulse amplitude (50 kV cm À1 ), and pulse duration (1 ls and 2 ls, were considered as the sum of the two half-waves, ) as the unipolar pulses were generated.Pulse parameters for treating the model suspension with unipolar and bipolar pulses are given in Table 2.

Operation modes
In addition to defining the parameter range for PEF treatment, a key task was to find the way of operation and the sites of integration of the PEF technology into the ongoing E-coating process, considering inactivation requirements and ensuring trouble-free operation.The required inactivation performance affects the technical requirements of the PEF system, in particular the specific treatment energy and flow rate, which in turn influence the required performance of the pulse generator.The choice for the site of operation is designed to protect the E-coating dip tank from bacterial contamination and growth, thus maintaining its bacteriostatic state.Contaminating bacteria are introduced in various ways: directly into the E-coating dip tank by dripping water from the car bodies during transfer from pretreatment (VBH-11 tank), by the supply of deionized water used to compensate for water evaporation, and by the backflow from the permeate rinse tanks used to recycle water and paint (tank KTL-02 to KTL-05).In this context, two operating modes were defined to calculate the required inactivation performance of the PEF treatment: (I) in ''transfer-flow'' (Fig. 2a), to prevent the direct contamination by bacterial transfer from a source of contamination (e.g., deionized water, or backflow from permeate rinse tanks) to the E-coating dip tank, and (II) in ''bypass''(Fig.2b), to reduce the bacterial load in a certain tank, e.g., treatment of the E-coating dip tank or the VBH-11 tank.
Mathematical modeling to determine the required inactivation performance I. PEF treatment in transfer-flow For this operating mode, the required inactivation performance of the PEF treatment (I PEF ) depends on the level of bacterial contamination in the donor tank (N donor ) and the maximum level of contamination that is acceptable for trouble-free operation (N max ) in the receiver tank, see equation (2).The maximum bacterial load in the Ecoating dip tank was determined by the paint company (PPG Deutschland Sales & Services GmbH) and paint shop operator (BMW Group) based on operating experience and should not exceed 10 3 cfu mL À1 .On the other hand, no bacteria could be grown in the paint samples and no CFU could be detected in the Ecoating dip tank (KTL-01) and the two downstream permeate rinse tanks, KTL-02 and KTL-03, while a very high bacterial load, over 10 5 cfu mL À1 , was found in the next post-treatment zones (KTL-04 and KTL-05).By assuming that the backflow into the E-coating tank has the same bacterial concentration (N donor ) as in KTL-04, then the required minimum inactivation performance is I PEF = log 10 (10 5 AE10 À3 ) ‡ 2 log 10 , as calculated from equation ( 2).This inactivation performance is also sufficient for the PEF treatment of freshwater and deionized water feeds, as the usual contamination level of cultivable bacteria in these samples is around 10 3 cfu mL À1 .
II. PEF Treatment in bypass Contrary to the operation mode in the transfer-flow, the required inactivation efficiency of the PEF treatment in operation in the bypass (Fig. 2b) is determined by several process parameters.One of the most important parameters is the volumetric flow rate ; ¼ _ V of the PEF treatment system, as it influences both the temporal inactivation efficiency in the treatment tank and the design of the PEF system.A second important parameter is the growth rate (l) of the bacteria in the treated tank.This parameter results from the bacterial reproduction rate in medium and the accumulation rate caused by dripping liquids from car bodies or backflows as well as the water inflow.The time dependency of the concentration of viable bacteria C(t), as a function of volumetric flow rate (Ø), tank volume (V 0 ), and inactivation efficiency of the PEF treatment (I PEF ) can be calculated by solving the following differential equation: where dC l (t) is the bacterial change due to growth and accumulation of bacteria in the tank and dC PEF (t,Ø,I PEF ) is the bacterial change due to inactivation by PEF treatment.Details of the calculation can be found in the Appendix I.The solution of equation (3) gives: where C 0 is the concentration of viable bacteria in the tank at beginning of the PEF treatment.Figure 3 shows the time course of the bacteria concentration C(t) on a logarithmic scale during the operation of the PEF treatment in the bypass mode.The calculation was carried out at three different ratios of volumetric  flow to tank volume (Ø/V 0 : 0.05 h À1 , 0.10 h À1 and 0.15 h À1 ) and three inactivation rates of the PEF treatment (I PEF : 1 log 10 , 2 log 10 and 3 log 10 ).A value of 0.03 h À1 was assumed for the bacterial growth and accumulation rate (l), based on the long-term monitoring of the bacterial growth (accumulation) rate in automotive series painting (BMW Group, DiWaL Report 2021). 39According to the model calculation the Ø/V 0 ratio is the most important factor for effective bacteria reduction in the treated tank when the PEF treatment is operated in bypass mode and a bactericidal effect is to be achieved.Even more, this applies to all decontamination methods that work in bypass, even filtration, where all bacteria would be filtered out.In this case the inactivation, I, for each pass through the filter would be I > > 1, and therefore, it applies: And from equation (4) follows: Equation (5) shows that with a higher inactivation efficiency (e.g., I PEF > 2 log 10 ) of the system operating in bypass mode, the resulting microbial reduction in the treated tank is only dependent on the flow rate of the decontamination system, in other words how fast the tank contents can be pumped through the decontamination system.This has a direct impact on the design and power requirements of the PEF system, as a higher flow rate requires a higher power supply to maintain the same treatment intensity.
In the next calculation example, it is assumed that bacterial load in the target tank (e.g., E-coating dip tank), noted as initial bacterial concentration C 0 , has reached a critical concentration ( ‡10 5 cfu ml À1 ) and must be reduced to less than 10 3 cfu ml À1 as quickly as possible, without additional chemical treatment (e.g., biocides).According to equation ( 4) and the processing parameters shown in Fig. 3, the bacterial concentration in the target tank can be reduced from 10 5 bacteria per mL to a noncritical concentration of 10 3 bacteria per mL (see › arrow in Fig. 3) within $ 39 h of PEF treatment with an inactivation efficiency of I PEF ‡ 2 log 10 at a volume flow rate ratio Ø/V 0 of 0.15 h À1 .
Increasing the volumetric flow rates shortens the operating time required to reduce the number of active bacteria in the target tank to a noncritical concentration, while the inactivation performance of the PEF treatment (I PEF ) plays a subordinate role, only affecting bacterial reduction in the targeted tank when it increases from 1 log 10 to 2 log 10 .Higher inactivation rates, e.g., I PEF > 3 log 10 , almost do not affect the reduction of bacteria in the treated tank (Fig. 3, mmarks) as compared to the 2 log 10 PEF inactivation efficiency (Fig. 3, s, I PEF = 2 log 10 ).This shows once again the importance of the flow rate, as only by increasing the flow rate ratio, in this case above 0.15 h À1 (Ø ‡ 15% of the target tank volume per h), is it possible to reach the noncritical bacteria concen- tration in less than 39 h.It also shows the technical requirements for the PEF system.For example, to treat the E-coating dip tank, with a volume of 230 m 3 , the necessary bypass flow rate for this example can be calculated as follows: 0.15 h À1 AE230 m 3 = 34.5 m 3 h À1 , for which already few PEF systems are available on the market. 43n summary, a bacterial inactivation efficiency of the PEF treatment of 2 log 10 is sufficient to cause at least a stagnation of bacterial growth (bacteriostatic effect) in the E-coating dip tank (KTL-01), both in bypass and transfer-flow treatment mode.

Bacterial inactivation by PEF treatment
The results reflect bacterial inactivation entirely due to the impact of the PEF and not promoted by electrically induced thermal heating.The PEF treatment energies of 40 J mL À1 and 80 J mL À1 would contribute to a temperature rise of approximately 10 K and 20 K, respectively, under adiabatic conditions, i.e., no heat transfer in the environment and electrodes.In the case of a continuous flow treatment and an initial medium temperature of 25 °C, the medium temperature would rise to a maximum of 45 °C under adiabatic conditions, which is still below the inactivation temperature of bacteria (< 70 °C).The measured temperature increases in the cuvettes immediately after PEF treat-ment were less than 2.5 K under our experimental conditions.
In a first approach PEF treatment was performed on native samples (without bacterial enrichment) with a specific treatment energy of 150 J mL À1 and in addition 80 J mL À1 for VBH-11 samples, to test the inactivation performance at a high energy input.Above a treatment energy of 150 J mL À1 , impairment of the coating was observed.Native samples from VBH-11 and KTL-04 tanks were tempered at room temperature prior PEF treatment.No bacterial enrichment was performed in this case.These samples were contaminated with a mixed bacterial population.The predominant bacterial species are as shown in Table 1.After the PEF treatment the bacterial population in the native VBH-11 samples was completely eradicated, when treated with 150 J mL À1 indicating an inactivation efficiency of more than 3 log 10 (n.d.: not detected bacteria in the treated samples), while the bacterial population in the KTL-04 samples was also reduced by more than 3 log 10 from 10 6 cfu mL À1 to 10 2 cfu mL À1 (Fig. 4).At the lower treatment energy of 80 J mL À1 , an inactivation rate of 2 log 10 was achieved with both pulse durations of 1 ls and 2 ls.No bacteria could be cultivated from the samples of the two downstream tanks of the post rinse zone (KTL-02 and KTL-03); i.e., no CFU could be detected.Although bacterial DNA could be isolated, it was not possible to grow bacteria in this medium.It is assumed that certain bacterial species are inhibited in their growth, by diluted biocides or paint components (e.g., organic solvents).This is based on the observation that the bacteria concentration in the downstream process is higher in the diluted permeate of the KTL-05 and LKTL-04 tanks than in the tanks closer to the E-coating dip tank, KTL-02 and KTL-03.

Bacterial inactivation by PEF treatment with unipolar and bipolar pulses
Our second approach aimed to decrease the specific treatment energy and the pulse duration while ensuring a bacterial inactivation efficiency above 2 log 10 , as required by the model calculation.Shorter pulses are expected to be less prone to electrode charge accumulation, thereby reducing paint degradation by minimizing electrochemical reactions.For comparison, we applied square-wave pulses with pulse durations of 1 ls and 2 ls to model suspensions enriched with selected bacteria, as described in Table 1.The PEF treatment procedure is detailed in Table 2.Although short pulses ( £ 1 ls) were likely to prevent paint degradation, we decided to test the higher energy input of 80 JAEmL -1 with longer unipolar pulses of 2 ls to compare inactivation efficiency as this protocol was expected to provide the highest inactivation rate.Figure 5a shows the results of PEF inactivation of sterile filtered samples from the VBH-11 tank enriched with B. cepacia (gray bars), M. maritypicum (blue  2. Each experiment was repeated at least twice in triplicate.The used bacterial suspensions contained a mixture of previously identified and enriched Gram-positive and Gram-negative bacteria bars), and S. gimenis (green bars) as a function of specific treatment energy and pulse duration.Figure 5b shows the results of PEF inactivation of model suspensions prepared with samples from the KTL-04 tank inoculated with B. cepacia.For comparison, the bacterial inactivation was also tested in the nonselective LB medium enriched with each of these bacterial strains (gray bars with white stripes).The lowest inactivation, around 2 log 10 , was achieved with Gramnegative bacteria (B.cepacia and S. gimenis) when treated in VBH-11 model medium with lower specific treatment energies (40 JAEmL -1 ).For all other PEF treatments, the inactivation rate was greater than 2 log 10 .Contrary to expectations the highest inactivation was achieved for the Gram-positive bacteria M. maritypicum, when treated in the model suspension based on VBH-11 samples (blue bars).In general, bacterial inactivation was higher in model media than in LB media.We believe that this is again due to residual biocides or other chemical compounds that inhibit bacterial growth or have bactericidal properties, especially in model media from KTL-04 tanks.
PEF treatment with unipolar pulses leads to buildup of charges on the electrodes and partial electrolysis of the solution.Ions such as hydroxide (OH À ) or hydrogen cations (H + ) are forming by water dissociation during the pulse application. 44,45This may result in coagulation of the coating, in corrosion of the electrode and in introduction of small particles of electrode material in the liquid.Thus, the E-coating process is affected.The use of bipolar pulses has the advantage over unipolar pulses, that the ions formed during the first half-wave are neutralized by the polarity reversal during the second half-wave. 39This prevents charge buildup on the electrodes and limits side effects such as coating deposition on the treatment electrodes and coagulation. 39,45,46However, a requirement for successful bacterial inactivation is a complete charging of the cell membrane, which according to our previous results, Frey et al. can only be achieved with longer pulses (> 0.5 ls) at small cell diameters (Ø bacteria: $ 1 lm) and low conductivity (< 0.5 mS cm À1 ). 38his study has shown that bipolar pulses with a halfwave of less than 0.6 ls result in poor inactivation performance despite the same treatment.With the short bipolar pulse of 1 ls (2 9 0.5 ls) used in our study, the exposure time of the electric field of the halfwaves (0.5 ls) is below this time limit, so that a poor inactivation efficiency was to be expected.Even with this disadvantage, short bipolar pulse of 1 ls (2 9 0.5 ls) duration were also tested, as this treatment cause less electrochemical reactions at the electrodes, thus preventing damage to the CDP or ADP. 45As the model calculations have shown, the required inactivation rates should be higher than 2 log 10 , which can be achieved by increasing the specific energy input.Although the strategy followed in this approach was to reduce the specific energy input to 40 J ml À1 in order to improve the efficiency of the method, we tested for 1 ls (2 9 0.5 ls) bipolar pulses also higher energy inputs (80 J ml À1 ).It was therefore necessary to test the bacterial inactivation efficiency of PEF treatment with bipolar pulses and compare it with the inactivation achieved by treatments with unipolar pulses of the same pulse length.8][49] From the results in Fig. 5, the inactivation rates in the model media are higher than in LB medium.It is assumed that this is due to residual biocides or other bactericidal components (e.g., organic solvents) in these samples, which could affect the survival rate of bacteria after PEF treatment.But the fact that the bacterial inactivation rate in LB medium is greater than 2 log 10 indicates that PEF treatment can provide the required inactivation efficiency even in biocide-free E-coating systems.However, in a biocidefree operation, the absence of biocides would affect the growth and accumulation rate of bacteria (l) in the Ecoating tank.The low value of l = 0.03 h À1 used in our calculations is related to the intensive use of biocides and organic solvent residues in paints, which have an antibacterial effect.The actual growth factor could not be estimated for this type of operation as the E-coating system was never operated in a biocide-free mode.Therefore, at higher bacterial growth rates, an antibacterial effect of PEF treatment in bypass mode can only be achieved by increasing the treatment flow rate.Consequently, a coating bath maintenance method based on PEF treatments of critical tanks and backflow circuits should consider each coating system individually.Extrapolation of the results is only possible to a limited extent, and it is recommended to devise tailored solutions.If the E-coating system shows significant contamination, it is necessary to inspect the hygiene standards in place.The presence of high levels of bacteria indicates inadequate cleaning procedures and poor hygiene standards.PEF treatments with specific treatment energy of 40 J mL À1 can inactivate both Gram-negative and Gram-positive bacteria selected from this specific E-coating line to a degree high enough (> 2 log 10 ) to ensure a bacteriostatic state in the system.Therefore, PEF treatment can be effective as a biocide-free bath maintenance method for the treatment of contamination hotspots in E-coating lines for high-throughput car body paint shops.

Conclusions
The results of this study prove that PEF treatment with pulse durations longer than 1.0 ls resulted in efficient bacterial inactivation even in media with low conductivity, regardless of the pulse shape, unipolar or bipolar.Thus, the design of the PEF system is not limited by the choice of pulse shape.The model calculations show that the bypass treatment mode requires a high treatment flow to be efficient: Ø ‡ 15% of the treated tank volume per h.For such applications, the development of a suitable pulse generator based on solid states switches capable of treating high volume flow rates (> 10,000 L h À1 ) is required.By using appropriate treatment parameters, such as bipolar pulses of 40 kV cm À1 to 50 kV cm À1 and a treatment energy of 40 J mL À1 , a 2 log 10 reduction in both Gramnegative and Gram-positive bacteria can be achieved while limiting electrode corrosion and coating degradation.In summary, PEF treatment is an effective alternative to biocide application that can be used in an E-coating system to maintain the bacteriostatic status of the system by operating at different spots, in transfer-flow or bypass mode, ensuring biocide-free operation.
within the paper and the Appendix.Should any raw data files be needed in another format they are available from the corresponding author upon reasonable request.Source data are provided with this paper: https://doi.org/To calculate the evolution of the concentration of viable bacteria over time, C(t), the following assumptions and notations were used: N(t) and N(t + dt) are the number of viable bacteria in the tank at timepoint t and t + dt, respectively.Assuming that the number of viable bacteria in the tank will change by the accumulation of bacteria (given by the growth and accumulation rate, l) and by bacterial inactivation (I PEF ), due to PEF treatment, then, at timepoint t + dt the number of bacteria in the tank will change accordingly to: where N dead is the number of bacteria inactivated during an infinitesimal time dt in the PEF system.In addition to the meaning of equation ( 2) as the efficiency of the PEF treatment, I PEF also stands for the bacterial inactivation rate, which is defined as follows: In equation ( 7) N total is the total number of bacteria treated during dt and can be expressed as: where dV is the infinitesimal volume, treated during an infinitesimal time dt in the PEF system.V 0 represents the tank volume.The ratio dV AE V 0 À1 determines the number of bacteria treated during dt.From equations ( 7) and ( 8) and the meaning of N dead = N total -N viable it follows for N dead at time t + dt: The flow rate through the PEF system is considered to be constant and is given by: Thus, it follows for N dead at time t + dt: By rearranging equation ( 6) and using equation ( 11) we obtain: The infinitesimal change in concentration, dC(t), is obtained by dividing the equation ( 12) by the volume of the tank, V 0 , and considering that dN = N(t + dt)-N(t) = V 0 AEdC(t) and N(t) = V 0 AEC(t): By integrating the equation ( 14), the time dependence of the concentration of viable bacteria can be calculated: Equation ( 13) can also be written as follows: Thus, the change in bacterial concentration in the tank is the difference between the changes in bacterial accumulation and growth, dC l (t), and the bacteria inactivated by PEF treatment, dC PEF (t, Ø, I PEF ), as given by equation (3).It can be written: where and dC PEF t; ;; Finally, equation ( 19) can be written as follows: II. Pulse shape of the unipolar and bipolar square pulses delivered by the home-built transmission line pulse generator: See Fig. 7.

Fig. 1 :
Fig. 1: Schematic outline of a car body E-coating line (based on a technical drawing from the BMW Group).The simplified process line shows the pretreatment zone, starting with the phosphating in VBH-07 dip tank followed by dip rinse tanks (VBH-08, VBH-09), passivation (VBH-10) and deionized water rinsing (VBH-11), E-coating dip tank (KTL-01), and the spray rinse zone with four stages (KTL-02 to KTL-05).The car bodies flow is from VBH-07 to KTL-05.In the post rinse zone, the backflow of the permeate is from KTL-05 to KTL-01 tank (light blue arrows).The blue arrows indicate the inlets for the deionized water and the red arrows indicate the inlets for the chemical additives

Fig. 2 :
Fig. 2: Schematic view of the operating modes of PEF treatment: (a) in transfer-flow, from a donor tank, the source of contamination, to the recipient tank, typically the E-coating dip tank and (b) in bypass, to reduce the bacterial load in a process tank, e.g., E-coating or VBH-11 tank.Goal of both operating modes is to keep the bacterial concentration over time (N(t)), below a certain contamination level (N max ), to maintain the bacteriostatic state

Fig. 3 :
Fig. 3: Time kinetic of bacterial concentration in the target tank during PEF treatment in bypass mode, at three distinct volume flow ratios (Ø/V 0 = 0.05 h 21 , 0.10 h 21 and 0.15 h 21 ) and three inactivation rates of PEF treatment (I PEF : 1 log 10 , 2 log 10 and 3 log 10 ).The initial bacterial concentration in the treatment tank is C 0 = 10 5 cfu mL 21 and the bacterial growth and accumulation rate is l = 0.03 h 21 .The bacterial concentration in the tank is reduced from 10 5 to 10 3 bacteria per mL (means 2 log 10 , › arrow) after approx.39 h of PEF treatment with an inactivation efficiency of 2 log 10 at a volume flow rate ratio of 0.15 h 21

Fig. 4 :
Fig. 4: Bacterial inactivation in native samples from VBH-11 and KTL-04 tanks after PEF treatment with unipolar square-wave pulses (1 ls and 2 ls) using a specific treatment energy of 150 J mL 21 and 80 JAEmL 21 (at 50 kV cm 21 ) adjusted by the number of pulses.The conductivities of the native samples are 0.05 mS cm 21 (VBH-11) and 0.08 mS cm 21 (KTL-04) as listed in Table 2.Each experiment was repeated at least twice in triplicate.The used bacterial suspensions contained a mixture of previously identified and enriched Gram-positive and Gram-negative bacteria

Fig. 5 :Figure 6
Fig. 5: Inactivation efficiency as a function of treatment energy and pulse duration for PEF treatment with unipolar squarewave pulses.(a) Inactivation of B. cepacia (gray bars), M. maritypicum (blue bars), and S. gimenis (green bars) inoculated in sterile filtered samples from VBH-11 tank and for comparison bacteria in LB medium (gray diagonal strips); (b) Inactivation of B. cepacian in sterile filtered samples from KTL-04 tank and for comparison bacteria in LB medium (diagonal strips).All experiments were repeated at least twice, each time in triplicate (Color figure online) Figure 6 illustrates the inactivation efficiency of PEF treatment with bipolar pulses as a function of the specific treatment energy and the pulse duration, which is given as twice the half-wave width (e.g., 2 9 0.5 ls).Figure 6a illustrates the inactivation rates of B. cepacia (gray bars), M. maritypicum (blue bars), and S. gimenis (green bars) in VBH-11 model media.Meanwhile, Fig. 6b shows the inactivation rates of B. cepacia in KTL-04 model media.It can be observed that inactivation using bipolar pulses is less effective compared to unipolar pulses of the same duration and amplitude.Moreover, short bipolar pulses (half wave £ 0.5 ls) are ineffective, for instance, to reduce the M. maritypicum Gram-positive bacteria by 1 log 10 (90%) and therefore not recommended for use in industrial applications.To ensure an inactivation efficiency of the PEF treatment, I PEF ‡ 2 log 10 , as required by the model calculation for all operation mode, bipolar pulses of a pulse duration longer than 2 ls (i.e., 1.0 ls for each half-wave) are required.Although it was expected that Gram-positive bacteria would be more robust to PEF treatment than Gram-negative bacteria, we observed that by selecting appropriate treatment parameters, inactivation in the range of 1 log 10 -2 log 10 rate reductions can be achieved even with low specific treatment energies (e.g., 40 J mL À1 ).[47][48][49]From the results in Fig.5, the inactivation rates in the model media are higher than in LB medium.It is assumed that this is due to residual biocides or other bactericidal components (e.g., organic solvents) in these samples, which could affect the survival rate of bacteria after PEF treatment.But the fact that the bacterial inactivation rate in LB medium

Fig. 6 :
Fig. 6: Inactivation efficiency as a function of treatment energy and pulse duration for PEF treatment with bipolar squarewave pulses.(a) Inactivation of B. cepacia (gray bars), M. maritypicum (blue bars), and S. gimenis (green bars) inoculated in sterile filtered samples from VBH-11 tank; (b) Inactivation of B. cepacian in sterile filtered samples from KTL-04 tank.All experiments were repeated at least twice, each time in triplicate (Color figure online)

Fig. 7 :
Fig. 7: Examples of measured temporal shapes of pulses obtained with a transmission line generator (a) unipolar square-wave pulses of 1 ls duration and 10 kV pulse amplitude.For an electrode gap of 2 mm, the electrical field strength was 50.00 ± 0.35 kV cm 21 .(b) Bipolar square wave of 2.0 ls pulse duration, considered as the sum of the two half-wave pulses, generated by a bidirectional pulse forming line.The electric field strength in the treatment chamber was the same as for unipolar pulses

Table 1 :
Model media and their sampling location, used to prepare model suspensions Gram-negativeThe model media were inoculated with each of the predominant bacteria identified in the process tanks of the E-coating line.In addition, LB medium was used as a control USA) with a gap size of 2.0 ± 0.1 mm and a volume of 400 lL were used in all experiments as treatment chambers.The pulse rise time of the transmission line generator was 16 ns.The pulse durations were set by the length of the coaxial cable to 0.5, 1.0, and 2.0 ls, while the pulse amplitude achieved in the electroporation cuvettes was 50.00 ± 0.35 kV cm À1 .

Table 2 :
Pulse parameters for treating the model suspension with unipolar and bipolar square-wave pulses Two specific treatment energies, 40 J mL À1 and 80 J mL À1 , were used for comparison of PEF treatment efficiencies of unipolar and bipolar pulses.The specific PEF treatment energy was set by the number of pulses.Model suspensions were inoculated with bacteria according to Table1(b) (a)N(t) < N max 10.2314/KXP:1778016480 Conflict of interest All authors certify that they have no affiliations with or involvement in any organization or entity with any financial interest or nonfinancial interest in the subject matter or materials discussed in this manuscript.Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material.If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.