Removing biofilms from stainless steel without changing surface properties relevant for bacterial attachment

Productive Biofilms are promising materials for the production of specific chemical substances.¹ The development of new bioreactors which utilize biofilms instead of suspensions of bacteria is thus an upcoming research field.² In these biofilm reactors, special solid substrates are incorporated to facilitate biofilm growth. The surface properties of these materials may influence the growth of the biofilm and its productivity in a positive or negative way.³ Because of these effects of the surface properties on the interaction with the bacteria and thus the biofilm productivity, detailed studies on this relationship are mandatory. Because technical substrates vary in their properties, in particular when they are modified by topographic structures or chemical functionalization, systematic studies should be performed on the same substrate. As the second best approach, a large number of samples may be measured for statistical reasons, but this does not allow for the study of specific effects. Thus, the cleaning of the substrates after each experiment for reuse is an important and often underestimated topic. In a previous paper, we have already shown that cleaning of unstructured and microstructured titanium substrates with oxygen plasma presents an effective cleaning procedure without changing the surface properties.⁴ 

Bioreactors are usually made of stainless steel, e.g., of the stainless austenitic steel 1.4571. Thus, plasma cleaning was also the first choice to remove biofilms of the sea water bacterium Paracoccus seriniphilus from such stainless steel surfaces.

As in the previous study, the bacteria should attach in the same way to the biofilm reactor after its cleaning because changed bacterial attachment properties could also change the productivity of the biofilm and thus its efficiency. As already known, plasma cleaning of other stainless steel (1.4501 and 1.4301) produces a high density of oxygen species on surfaces and changes the surface oxide composition and thus the surface properties.^5–7 These changes are irrepealable and not controllable. Thus, plasma cleaning was compared to the usage of basic and acidic solutions at elevated temperatures as commonly used in the food industry for stainless steel 1.4571.⁸ 

As analysis methods, x-ray photoelectron spectroscopy (XPS), contact angle measurements, zeta potential measurements, scanning electron microscopy (SEM), scanning force microscopy (SFM), and fluorescence microscopy are utilized.

P. seriniphilus bacteria were stocked with 50%-glycerol solution, stored at −80 °C. The bacteria were revivified in a preculture with 700 μl of the stock culture in 25 ml of complex medium [yeast extract, 5 g/l (Acumedia, Neogen Co.), Bacto peptone, 5 g/l (BD, Germany), sea salt, 34.3 g/l (Sigma Aldrich, Germany), and l-serine, 1 g/l (Sigma Aldrich, Germany)] at 30 °C and 150 rpm for 14–18 h. The obtained preculture was streaked on an agar plate and incubated at 30 °C for 3 days. The plate was stored at 4 °C and repitched every 4 weeks. For further cultivations, a preculture was prepared by using an inoculation loop to scrape a small quantity of P. seriniphilus in 25 ml of complex medium and kept on the shaker at 30 °C and 150 rpm for 14–18 h. For the main culture, an Erlenmeyer flask was filled with 25 ml of complex medium, and a ratio of 1:100 of preculture was taken. The flask was kept at 30 °C and 150 rpm on the shaker for the desired time.

For producing stainless steel samples with bacteria on the surface, the samples were fixed with glue (green glue, yellow-blue, Flexbar Corp. Long Island, NY) in an Erlenmeyer flask with a P. seriniphilus culture and kept at 30 °C and 150 rpm for the desired incubation time to generate a bacterial layer.

To create a real biofilm, the cultivation of microorganisms was done in an especially constructed flow cell with the dimensions 28 × 8 × 75 mm³ (width × height × length) and a reactor volume of 1.76 ml. The flow cell possesses an indentation that can house a 20 × 10 × 2 mm³ (length × width × height) substrate sample where the biofilm is grown on. A main culture in an Erlenmeyer shaking flask was inoculated to a starting OD600 of 0.2. The culture was run in a looped batch culture. A rotating displacement pump (IPC-N 8, Idex Health & Science, Lake Forest) was set to a flow rate of 5 ml/min for 15 min with a 60 min break between each interval. The cultivation time was 48–72 h.⁷ 

Polished and unpolished stainless austenitic steel samples (1.4571, X6CrNMoTi17-12-2) were obtained from the central metal workshop of the University of Kaiserslautern in the size of 20 × 10 × 2 and 5 × 5 × 1 mm³ (length × width × height). The isoelectric point was obtained using an Electro Kinetic Analyzer (EKA) from Anton Paar. The root mean square (RMS) roughness for polished and unpolished stainless steel was evaluated using SPIP (Image Metrology Denmark) based on scanning force microscopy images (MFP 3D, Asylum Research) with a scan size of 80 × 80 μm². As the RMS roughness of unpolished steel does not go into saturation even with the biggest available scan size of 80 × 80 μm², this scan size was used as the best available value. To be consistent and for better comparison, this scan size was also taken for polished steel samples although here the RMS roughness saturates at around 20 × 20 μm². The scanning force microscope was operated in the contact mode with Olympus OMCL-AC200TS-R3 cantilevers.

For O₂ plasma cleaning, the samples were precleaned with acetone (pa, 99.8%), isopropanol (pa, 99.8%), and ultrapure water in an ultrasonic bath each for 10 min [which will here be named the standard cleaning (SC) procedure] and subsequently dried with N₂. Afterward, the samples were cleaned with oxygen plasma (5 min, 50 W and 20% sccm O₂).

For base and acid cleaning, the stainless steel samples were stored in 2% NaOH-solution (JT Baker^® Organic Reagent Chemicals) for 30 min at 75 °C and in ultrapure water for 15 min at room temperature. Subsequently, the samples were immersed in 2% HNO₃-solution (65%, JT Baker^® Organic Reagent Chemicals) for 30 min at 75 °C and finally again in ultrapure water for 15 min at room temperature. Afterward, the samples were standard cleaned and dried with N₂.

The surface composition of the substrate surfaces was characterized by means of XPS. The XPS measurements for characterization of unpolished and polished stainless steel samples (see Chap. III) were performed in an UHV system with a base pressure below 2 × 10⁻¹⁰ mbar. The spectra were recorded using a commercial Phoibos 150 MCD9 hemispherical energy analyzer (Specs, Berlin, Germany). The analyzer pass energy was set to 20 eV (fixed analyzer transmission), and all spectra shown were obtained with a step size of 0.02 eV. The employed Mg Kα radiation (1253.6 eV) was provided by a standard laboratory excitation source (XR-50, Leybold, Cologne, Germany), which had an x-ray power of 100 W. By measuring the 3d5/2 peak of a Ag(111) crystal, the energy resolution was estimated to be better than 1 eV. The energy scale was aligned by initial determination of the 4p3/2, 4d5/2, 4f5/2, and 4f7/2 peaks of an Au(111) crystal.

The remaining XPS measurements of the biofilm exposed samples were carried out using an Axis Nova spectrometer (Kratos Analytical Ltd.). In this system, photoelectrons are released by a soft x-ray radiation of 1486.6 eV of a monochromatic Al Kα source. Surface charges are compensated by a charge neutralizer, consisting of a filament (I = 1.9 A and V = 3.2 V) inserted directly in the magnetic lens system, which balances the electron losses. With this charge compensation system, an energy resolution on isolators better than 0.68 eV [ester component of PET] can be achieved. The detector take off angle is 90°. Atomic concentrations at the surface were calculated from survey spectra according to standard sensitivity factors of the device manufacturer. The chemical bonding nature of the detected major elements was derived from a deconvolution of their core level spectra. The C1s photoelectron line of the omnipresent aliphatic hydrocarbon at 285 eV served as a reference of the binding energy. Pass energies of 160 eV were used in the case of survey and 20 eV in the case of core level spectra. The area of analysis was selected to be 400 × 700 μm. The XPS typical information depth is about 5–10 nm [three times the inelastic mean free path (IMFP)], which differs for different atomic species according to the kinetic energy of the released photoelectrons and depends on the material which the photoelectrons have to pass.⁹ 

Bacteria on top of the stainless steel were stained with a few drops of 4′,6-diamidino-2-phenylindole (DAPI, for nucleic acid staining; Sigma Aldrich, Germany) by an incubation for 5 min without light. After rinsing with ultrapure water (Millipore, 18.2 MΩ/cm), the samples were imaged with a fluorescence microscope (Axioskop 2 Mot, Carls Zeiss, Jena, Germany) at a wavelength of 365 nm. The Axiovision software (Carls Zeiss, Jena, Germany) was used for evaluation and imaging.

The scanning electron microscopy images were obtained using a SEM SU8000 (Hitachi, Japan) with a SE (L) secondary electron chamber detector (Everhart-Thornley detector) to get a resolution of <1.3 nm at 5.0 kV.

The contact angle of the stainless steel samples was determined by the sessile drop method by using a contact angle measuring system G2 (Krüss, Hamburg, Germany). The software controlled system places a droplet (3 μl) of ultrapure water (Millipore, 18.2 MΩ/cm) on the samples, and an image of the drop was used to determine the angle between the surface and the droplet at the three-phase point.

The zeta potential of the surfaces was measured using the EKA from Anton Paar. The samples were fixed with double sided tape in the measuring cell. The device was rinsed with ultrapure water and 1 mM KCl. The distance between the two samples was adjusted to get a flow pressure of 400 mbar at 50% of the pump power. A single measurement was performed to determine the zeta potential of the surface at the actual pH of the measuring solution (1 mM KCl). According to the single measurement result, the titration solution (0.1 M HCl or 0.1 M KOH) was chosen for obtaining the isoelectric point of the surfaces.

To investigate the influence of the cleaning procedure on stainless steel surfaces at first, the surface properties of the pristine material had to be characterized. For this purpose, their isoelectric point, surface composition, RMS roughness, and water contact angle were determined. Table I shows an overview of the results on unpolished and polished steel samples. Before each measurement, the samples were cleaned according to the standard procedure.

The main difference between the unpolished and polished steel samples is their RMS roughness. Since they have the same isoelectric point at a pH value of 4.6 and show no significant differences according to their surface chemistry, it is likely that the different water contact angles (81° and 61°) are caused by their differences in roughness. It was detected that the polished samples are more hydrophilic than unpolished steel samples. According to the Cassie-Baxter model, the water droplet can remain on top of the structures on the samples, which leads to a higher apparent contact angle for the unpolished sample.¹⁰ 

Different XPS detail spectra of unpolished and polished stainless steel can be found in the supplementary material.¹⁴ 

The surface chemistry of the unpolished and polished steel samples does not differ significantly. The 2p core level spectra of iron are dominated by the doublet structure of Fe2p3/2 at 710.9 eV and Fe2p1/2 at 724.9 eV, typical for iron bound as Fe₂O₃ (EB = 710.8 eV) with small possible contributions of Fe₃O₄ (EB = 710.4 eV). Additionally but minor prominent, the unshifted 2p doublet structure of metallic iron appears at 707 eV (Fe2p3/2) and at about 720 eV (Fe2p1/2).¹¹ 

The carbon 1s core level spectra are dominated by a photoelectron line of carbon bound in aliphatic hydrocarbon at about 285 eV. At 286.5 eV, a not completely resolved structure in the form of a shoulder appears which belongs to either CO or CN groups. Additional contributions can be described by two overlapping Gaussian peaks at 288.1 eV (O–C–O and C=O bonds) and 289.2 eV (carboxylic or ester bonds and contributions of adsorbed water).⁴ 

Different unpolished and polished stainless steel samples were standard cleaned, oxygen-plasma cleaned, and base acid cleaned (BAC). Afterward, the contact angle with water was measured after different time periods (several days) (Fig. 1). After cleaning, the samples were kept in a closed plastic box until they were investigated.

After standard cleaning, the values for polished and unpolished samples scatter in the bands of 50°–60° and 85°–95°, respectively, depending on the individual sample. The samples after base acid cleaning behave nearly the same as the standard cleaned samples. Even after a second cleaning of the same sample, the values of the contact angles remain constant (see supplementary material).¹⁴ 

Immediately after the plasma cleaning, the contact angles were near to a value of 0° and the surfaces show a complete wetting. While they are present in a closed plastic box, the contact angles increase day by day, but even after 11 days, they do not reach a constant value. Multiple plasma cleaning (not shown) leads every time to a different contact angle. Thus, oxygen plasma cleaning does not meet the requirements of constant surface properties before and after cleaning.

The base acid cleaning does not almost affect the surface hydrophobicity of polished and unpolished stainless steel samples, while the oxygen plasma cleaning has a strong influence and makes them much more hydrophilic (see Fig. 1).

XPS reveals the chemical composition of the surfaces. Figures 2 and 3 show the highly resolved core level spectra of Fe2p, O1s, and C1s and the corresponding element concentrations of the species in their different binding situations.

As already betokened, the highly resolved iron 2p core level spectra in Fig. 2 show iron mainly bound as oxide (710.9 eV) and smaller contributions of metallic iron (707 eV) in its information depth. With an IMFP length of 1.5 nm for an iron 2p core level photoelectron passing iron oxide, this information depth in the case of iron, estimated to be three times the IMFP, is about 4.5 nm.⁹ After oxygen plasma treatment, even no metallic iron is detected at the surface by XPS, while after acid base cleaning, the ratio of metallic to oxidic iron is increased. Coinciding with the Fe2p results, the O 1s core level spectra show an increased peak for the metal oxide (530.2 eV) in the case of the plasma cleaning, which is much less prominent after base acid cleaning. The contributions of hydroxyl and R-O species are nearly the same for standard cleaning and base acid cleaning, respectively. The carbon 1s core level spectra show that even after standard cleaning, the highest hydrocarbon concentrations in comparison to the other methods remain at the surface. Additionally, after base acid and oxygen plasma cleaning, lower concentrations of carbon bound in carboxylic groups are detected and compared to the standard cleaned sample (see Fig. 3). So, it seems that the base acid cleaning and oxygen plasma cleaning are able to remove impurities from the surface, which is not possible with organic solvents.

However, oxygen plasma strongly oxidizes the surface, which directly influences the contact angle. It is also known that oxygen plasma activates a surface, i.e., makes it very reactive.^6,7 Apparently, this activation leads to a change in the surface hydrophobicity, most probably due to the reaction with molecules from the ambient.

To avoid this oxidation and activation of the surface, argon plasma cleaning (5 min, 50 W, 20% sccm Ar) was also tested because it is known that it cleans by physical sputtering rather than chemical reaction.¹² However, here also, a strong influence on the contact angle was obtained. The surface is again completely wetted directly after plasma cleaning, and the contact angle does not reach the starting value after several days. It is supposed that the argon plasma cleaning is sputtering the surface in an uncontrollable way, i.e., removing small layers of the complex alloy, leaving slightly different surface compositions behind.

To check the efficiency of the base-acid cleaning for bacteria covered samples, different methods were used to verify the clean surface. An easy method is coloring the bacteria with a fluorescence active stain. DAPI is a blue fluorescent substance, which does not differ between live and dead bacteria. A one day old P. seriniphilus culture (see Sec. II A) grown in an Erlenmeyer flask on polished stainless steel was dyed before cleaning (Fig. 4, bottom left) and on another sample out of the same flask after cleaning (Fig. 4, bottom right). The comparison of the fluorescence images of the samples shows the success of the cleaning procedure (see Fig. 4, top).

A supplementary method for analyzing the same sample with bacteria before and after cleaning is scanning electron microscopy. On a polished steel sample now, a real biofilm was grown in a flow cell over 3 days and washed with ultrapure water, and various positions on the sample were imaged by SEM [Fig. 5(A)]. Afterward, the sample was cleaned by the base acid procedure and imaged again [Fig. 5(B)].

Before cleaning, the images clearly show colonies of bacteria as a biofilm attached to the stainless steel. After cleaning, only minor residues of the biofilm can be seen in the SEM images.

Additional XPS measurements were performed to quantify the amount of biofilm removed from the stainless steel by this cleaning procedure. Here, the N1s photoelectron line of nitrogen bound in an amine at 400.5 eV was used as a marker for the presence of a biofilm.⁴ For these nitrogen core level spectra, measured on unpolished steel samples using different cleaning procedures, a polished steel sample with a 7 day grown biofilm and the cleaned biofilm sample are compared in Fig. 6. Note that the biofilm sample was cleaned with base and acid, measured by XPS, again cleaned twice with base acid reagents, and again measured by XPS.

As nitrogen is a component in each biological substance, e.g., amino acids, nitrogen is good evidence for bacteria on the sample.¹¹ The peak at around 400.5 eV is typical for amide and amine and can serve as a marker.⁴ The core level spectra clearly show that the detected nitrogen is not bound in the form of nitrate (from HNO₃) (407.3 eV).¹² However, this nitrogen peak at 400.5 eV can also be found on the standard cleaned, plasma cleaned, and base acid cleaned samples, which had never been in contact with the grown up biofilm. It has to be taken into account that the stainless steel samples even after different cleaning procedures contain biological nitrogen impurities on their surfaces. This weak peak strongly increases with a biofilm on the surface and is reduced after first base acid cleaning. The second and third acid base cleaning reduces the nitrogen peak at 400.5 eV to the value of the starting position. Figure 7 shows an additional comparison of the oxygen and carbon core level spectra.

In the carbon 1s core level spectra, the increasing peaks for carboxylic species at 288.1 eV and the single bonds between carbon and oxygen or nitrogen at 286.4 eV on the stainless steel with the biofilm on top can be additionally seen. For the oxygen 1s core level spectra, the increasing peaks for O-H and O-R at 531.6 eV are notable. All the peaks responsible for biofilm element species have been significantly reduced near the starting values after base acid cleaning procedures. Figure 8 compares the element concentrations on the surfaces.

The highest amount of nitrogen can be found on the biofilm sample. But after the first cleaning with base acid, there is still a higher amount of nitrogen on the sample than on the sample before the biofilm (Fig. 8, top, main elements). Applying the cleaning procedure, the nitrogen value is almost reduced to two more times the output value. Additionally, elements like chlorine, calcium, and sulphur (Fig. 8, bottom, trace elements) are no more detected. The decreasing concentration of iron and chromium on the biofilm sample surface and the increased carbon concentration can be explained by the (carbon-containing) biofilm covering the surface.¹³ Besides nitrogen, phosphorus can also be proof for bacteria on the surfaces. As phosphorus is a part of the DNA, it could be evidence for extracellular DNA. In this case, no phosphorus was detected on any of the surfaces.

As ultimate proof of the success of the base acid cleaning, biofilms were grown on the cleaned substrates. The coverage with P. seriniphilus after 3 days of growth in the flow cell was similar to the uncleaned steel substrate.

Cleaning stainless steel bioreactor substrata without changing their surface properties is a challenging topic because only comparable surface properties and chemistry result in comparable biofilm growth, productivity, and effectiveness. These properties of the biofilm are determined by the first contact of the bacteria with the solid substrate during the development of a biofilm. Using an oxygen plasma is a good procedure for titanium samples, but not an option for stainless steel.⁴ XPS and contact angle measurements demonstrate that oxygen plasma generates a strongly oxidized surface, which changes the surface properties and thus the surface chemistry in an uncontrollable way due to the complex alloy composition. A constructive way of cleaning is a combination of base and acid at elevated temperatures that does not change the surface chemistry. It is notable that both cleaning procedures are able to remove hydrocarbon impurities in an efficient way. The efficiency of cleaning increases with the number of repetitions.

Fluorescence microscopy and SEM prove the removal of the bacteria by BAC. Using the surface sensitive analysis technique XPS, it could be shown that only very little biofilm residues remain on the surfaces. Repeating the cleaning procedure up to two times further reduces these biofilm residues. As there are nitrogen impurities on the surface, even before being in contact with biological components, it can be assumed that the biofilm is removed using this cleaning procedure and the samples are brought to their initial situation. For the purpose of reusing the substrates, this cleaning process is sufficient with special attention to the unchanged surface properties.

This work was carried out in the Collaborative Research Center CRC 926 “Microscale Morphology of Component Surfaces.” Funding by Deutsche Forschungsgemeinschaft is gratefully acknowledged. Contact angle measurements were partly carried out by Tobias Schneider and Frederik Pütz. The use of the SEM in the Nano Structuring Center (NSC) of the University of Kaiserslautern by Veronika Rink is gratefully acknowledged.