Enhancing anti-adhesion properties by designing microstructure - the microscopy and spectroscopy study of the intercellular bacterial response

Abstract

This study is the first one that investigates in detail the bacterial intercellular response to the high density of crystallographic defects including vacancies created in Cu by high pressure torsion. To this aim, samples were deformed by high pressure torsion and afterward, their antibacterial properties against Staphylococcus aureus were analyzed in adhesion tests. As a reference an annealed sample was applied. To avoid the influence of surface roughness, specially elaborated conditions for surface preparation were employed, which do not introduce defects and assure comparable surface roughness. The analysis of the chemical composition and thickness of passive layers by X-ray photoelectron spectroscopy showed that they were comparable for nanostructured and micrograined samples, consisting of Cu₂O and CuO, and a thickness of 6 nm. The interface bacterium-substrate was prepared by a focused ion beam and further analyzed by scanning transmission electron microscopy and energy dispersive spectroscopy. High pressure torsion processed Cu shows enhanced anti-adhesion properties while in contact with S. aureus than micrograined Cu. There is a linear correlation between luminous intensity and grain size^−0.5. The bacterial intercellular defence mechanism includes the creation of Cu₂O nanoparticles and the increased concentration of sulphur-rich compounds near these nanoparticles.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-024-75045-5.

Introduction

Antibacterial properties are mainly assigned to the chemical composition of metals and their alloys rather than to their microstructural features. Cu, Ag, and Zn have remarkable antibacterial properties among metallic materials¹. Cu has several advantages, e.g. it is relatively cheap and easy to process using traditional metal working techniques. Its application in hospitals has the potential to reduce bacterial infections². It was proved that the increase of Cu content in Cu alloy was correlated with the reduction of S. aureus viability in the contact with alloy. For C19700 (99% Cu) alloy, all bacteria on the material surface died after only 1.5 h, whereas for C24000 (80% Cu) alloy, all bacteria died after 4.5 h. Additionally, for C77000 (containing only 55% Cu) alloy, at least 50% of all bacteria were still viable after 6 h³. For this reason, much research to improve the antibacterial properties of materials is focused on adding Cu particles to the matrix⁴.

It is, however, well known that the properties of materials are controlled not only by their chemical composition but their microstructure plays an important role. However, research on the impact of microstructural features e.g. grain boundary density, dislocations density, and vacancy concentration on the antibacterial properties of materials are scarce^5–8. Recently, it has been proven that the high dislocation density introduced in Cu by cryo-rolling considerably enhanced the antibacterial properties of Cu compared with pristine and cold-worked Cu⁵.Unfortunately, grain size was not taken into consideration in this study and the focus was on differences in dislocation density. Authors, however, have calculated the crystallite using X-ray diffraction Willamson-Hall technique and showed that crystallite size decreased with increased dislocation density. However, the crystallite as the region of a coherently scattered X-ray refers to a subgrain or cell size bounded by dislocation boundaries⁹. In another study, the impact of grain size reduction on the antimicrobial activity of pure Cu was considered⁷. The grain size was reduced from 85 μm to 20 nm by rotation-accelerated shot peening (RASP). However, at the same time, the roughness of surfaces increased and reached almost Ra = 30 μm. This treatment improved the antimicrobial properties of the RASP sample for Escherichia coli in comparison with the initial material. Considering the bacteria size investigated e.g. E.coli (2 μm in length), the roughness of the surface plays an important role in bacterial cell attachment and screens the real impact of the grain size reduction. It has been shown that without changing the chemical composition of austenitic stainless steel, known for lacking in antibacterial properties, simply by the proper surface modification, the steel can become an antibacterial material¹⁰. Argon plasma etching with direct current discharge led to the creation of nanopillars which in turn led to the destruction of the cell membrane. It is well known that the surface roughness even at the nano scale, can influence the process of bacterial attachment. It has been proved that the nanoroughness in the range of 45 nm to 85 nm created on stainless steel can prevent bacterial adhesion by reducing the contact area between the bacteria and the surface¹¹. It is, therefore, important in studies concerning the impact of grain size on the antibacterial properties of investigated materials to ensure the same surface roughness. Recently a study on cold sprayed Cu, which attempted to reveal the impact of the grain size on antibacterial properties led to the conclusion that the surface roughness of nanograined material was greater than for the conventional one and the real antibacterial effect came from both factors, meaning, the percentage of grain boundaries and surface roughness⁸.

One should expect that because the grain size refinement leads to a higher density of grain boundary and triple junctions, which serve as electrochemically active sites, the ion release rate should be greater for nanograined materials¹². The release of Cu ions is one of the main factors responsible for bacteria death in the case of bulk Cu material. The Cu ions trigger an avalanche of reactions which may slightly vary for bacteria strains due to differences in their structure but finally, it leads to bacteria membrane depolarisation^13,14. The electrical potential differential between the inside and outside of metabolically active bacteria with intact membranes is approximately 100–200 mV, with the inside of the cells displaying a lower negative potential. Cu ions decrease the potential difference that causes membrane depolarization by binding to negatively charged regions on the cell membrane. Thus, when the potential difference falls to zero, leakiness or rupture of the membrane is observed.

It is thus of great interest to investigate the effect of microstructural features, i.e. grain size as well as dislocation density, on antibacterial properties in controllable conditions (having other parameters such as surface roughness stable). One of the methods that allow the modification of the material microstructure in a broad range is high-pressure torsion (HPT)¹⁵. During HPT the high strains imposed lead to the creation of a high density of defects which rearrange to form nanograins with non-equilibrium grain boundaries^16,17. Excess grain boundary energy, long-range elastic strains, and increased free volumes are characteristics of such boundaries. These unique microstructural features give the materials produced by HPT a wide range of exceptional properties in many fields. Sometimes the term ‘super-functional properties’ is applied since they are characterized by improved mechanical properties, especially high strength¹⁸, improved radiation resistance with the possibility of application for diagnostic mirrors in fusion reactors¹⁹, hydrogen storage capabilities²⁰ and improved magnetic properties²¹. However, no investigations up to now have been performed on the analysis of the antibacterial properties of HPT-processed materials in adhesion tests. The unique feature of the microstructures produced by HPT is the high density of vacancies and dislocations up to the saturation levels which assures exceptional diffusivity^22,23. This may, in turn, lead to an unprecedented response of bacteria at the interface to the produced by this technique material.

This work aims to investigate:

• (i)the correlation between crystallographic defects and anti-adhesion properties of HPT-processed Cu with a special emphasis on the impact of vacancies since their role has never been considered;
• (ii)the analysis of the intercellular bacterial defence mechanism of the impact of HPT-processed Cu on bacteria in adhesion tests by observing cross sections of a selected bacterium-surface interface prepared by the focused ion beam (FIB) using high-resolution scanning transmission electron microscopy (STEM) equipped with energy-dispersive X-ray spectroscopy (EDX) detector.

To avoid the influence of surface roughness, specially elaborated conditions for surface preparation were employed, which do not introduce defects and assure comparable surface roughness.

Results

The Cu samples were deformed by HPT in 1 or 5 rotations. Further in the text, for simplification, samples are marked HPT_1_0, HPT_1_1.25, HPT_1_3.75, and HPT_5_3.75, where the first number indicates the number of rotations and the second the distance from the sample center (in mm). Additionally, to be able to compare the impact of microstructure refinement, Cu was annealed at 600 °C for 2 h. This sample was used as a reference sample and is marked as AN.

Roughness analysis

The average roughness (Ra) of HPT-processed samples was approximately 4 nm. A slightly higher Ra was measured for AN approximately 7 nm. This is because variously oriented grains might have slightly different heights. However, it is important to point out that Ra was the same within one grain as for HPT-processed samples. The representative images showing the surface roughness of annealed and HPT-processed samples are presented in the Supplementary information (Figure S1).

Microhardness analysis

After one rotation microhardness increased considerably from 45 HV0.2 for AN to 138 HV0.2. However, after one rotation the microhardness differed in the diameter of the sample from 120 HV0.2 in the middle of the sample, 138 HV0.2 for 1.25 mm and 146 HV0.2 for 3.75 mm from the sample center. The microhardness distribution was much more uniform in the case of HPT_5, as the average was 151 HV0.2 but 3.75 mm from the sample center reached 154 HV0.2.

Microstructure analysis

A microstructure of the Cu AN sample is presented in Fig. 1a). To determine the grain size of analyzed microstructures the equivalent diameter (d₂) was applied. It is specified as the diameter of a circle having an area equal to the surface area of a given grain. The grains of AN were fully recrystallized and the average equivalent diameter was 27.7 μm while taking into calculation grain boundaries and 9.6 μm concerning calculation of grain boundaries and twin boundaries. In the case of HPT_1_0, the microstructure consisted of elongated bands, parallel to the sample surface of an average thickness of 300 nm (Fig. 1b). While the deformation increased on diameter, the bands started to divide into grains as presented in Fig. 1c)and d). In the case of HPT_5_3.75, the average diameter reached the lowest value of 140 nm (Fig. 1e).

Fig. 1.

a) An exemplary microstructure of a Cu AN – SEM image in BSE mode, microstructures of HPT-processed Cu samples b) HPT_1_0, c) HPT_1_1.25, d) HPT_1_3.75 and e) HPT_5_3.75 – TEM images in BF mode.

X-ray measurements revealed that the dislocation density increased from 9.18 × 10¹¹m⁻²for AN to 4.68 × 10¹⁴and 5.15 × 10¹⁴ forHPT_1_0 and HPT_1_3.75, respectively.However, after further deformation, the dislocation density slightly decreased to 3.37 × 10¹⁴ m⁻². The slight decrease of the value between HPT_1_3.75 and HPT_5_3.75 samples is associated with the dislocation rearrangement to create additional grain boundaries after imposing higher strains.

Bacterial adhesion analysis

Tests of bacterial adhesion revealed that HPT-processing affected the Staphylococcus aureus behavior on Cu surfaces (Fig. 2). Although in all cases the bacteria were predominantly alive (as shown by green fluorescence), fewer bacteria were attached to HTP-processed samples in comparison with AN ones. Besides, bacteria present on AN surfaces were mostly gathered in clusters while single bacteria were prevalent on e.g. HPT_5_3.75 surfaces (Fig. 2a). Interestingly, the fibers (probably polysaccharides) were visibly attached to bacterial clusters on the AN sample, suggesting the initial stage of biofilm formation (as shown on the SEM image). In the contrary, such fibers were practically absent on HPT_5_3.75 surfaces (Fig. 2a). It was also observed that the density of adhered bacteria and the number of clusters decreased with the distance from the center of the HPT-processed sample (as shown for HPT_1 surface; in the following order: HPT_1_0 > HPT_1_1.25 > HPT_1_3.75). For better clarity, the luminous intensity of AN and HPT-processed samples was quantified (Fig. 2b).The luminous intensity in the function of (d₂ ^−0.5) is presented in Fig. 2c). The results suggest that the number of applied revolutions and finally the distance from the HPT processed Cu sample center were important for bacterial adhesion. It can be concluded that greater deformation degree of the sample reduces the bacterial adhesion, division and biofilm formation.

Fig. 2.

a) Fluorescence images showing the viability and adhesion of S. aureus to Cu surfaces of AN and HPT-processed samples. SEM image shows the presence of biofilm formed on AN sample; the area was selected randomly, b) luminous intensity of AN and HPT-processed samples, c) luminous intensity (d₂ ^−0.5).

To better understand the intercellular bacterial response to nanostructures, for further analysis AN and HPT_5_3.75 samples were selected. To this aim, additional research has been conducted for these samples such as surface vacancy concentration analysis, chemical composition and thickness of passive layers, and SEM/STEM surface observations of adhered bacteria and bacterium-surface interface.

Sub-surface vacancy concentration analysis

DB-VEPAS and PALS measurements were performed on AN, and HPT_5 samples to reveal the impact of the surface preparation technique on vacancy concentration as presented in Fig. 3. The depth distribution of defects is very homogenous across the sample thickness as indicated by a monotonic decay of the S(E_p) curves.

Fig. 3.

a) Low e⁻momentum fraction, b), c) positron lifetime components and d) their relative intensities as a function of positron implantation energy and corresponding depth < z > for Cu AN and HPT_5 samples.

In HPT_5 the overall density of point defects c_V,_{HTP_5} ~ 1.2 · 10²⁰ cm⁻³ is much larger than in AN sample, c_V,_AN ~ 2 · 10¹⁶ cm⁻³. Due to the relatively long positron diffusion length, resulting in the positron back diffusion to the surface, the lifetimes in the subsurface region (E_p ≤ 7 keV) can be affected by the surface states. Nevertheless, the subsurface depth differs between HPT_5 and AN samples concerning the positron lifetimes (defect size) and their relative intensities (relative defect density). The major defect type (τ₁, I₁ = 43–75%) in the AN sample is found in the range of single-vacancy and a mixture of single- and double-vacancy for the bulk and subsurface regions, respectively^24,25. Since τ₁ increases for the HPT_5 sample agglomeration of point defects in the sub-surface is expected, however, it is unlikely that their size exceeds a bi-vacancy. The second lifetime component τ₂ represents larger vacancy agglomerations of about 11 to 13 vacancies for the AN sample, increasing to about 21 vacancies within a cluster after HPT. The vacancy agglomerations are typical for grain boundaries and their intersections, which are clearly within the diffusion reach for positrons as well as statistically sufficiently abundant (the average grain size of about 140 nm). The relative intensities, although, slowly decaying in the subsurface region as a consequence of partial positron annihilation with the surface states (due to the positron implantation profile and relatively long diffusion length), saturate in the bulk region providing a good clue about a level of defect density.

Chemical composition and thickness of passive layers

The high-resolution Cu 2p and Auger Cu LMM spectra were used to get an insight into the chemical state of the passive layers formed. The Cu 2p_3/2 peak observed at 932.5 eV (Fig. 4a and b) for the AN and HPT_5 samples can be attributed to Cu¹⁺ and/or Cu⁰ states. Additionally, a shoulder Cu 2p_3/2 peak at 934.0 and weak shake-up satellites indicate the presence of Cu²⁺ species, originating from the CuO component of the passive layer on Cu. The intensities of the signals associated with Cu²⁺ decrease after 550 s of ion sputtering as seen in Fig. 4a) and b) suggesting that as etching time increased, the Cu²⁺ proportion in the surface layer decreased. This confirms the partial removal of copper oxide layer from the samples’ surface. It was estimated that after the Ar⁺ etching, the Cu²⁺ proportion in the oxide film decreased from 17.9% of the total integrated intensity to 4.7% for AN and from 17.4 to 6.8% for the HPT_5 sample, respectively.

Fig. 4.

XPS survey spectra of Cu AN and HPT_5 samples surface; XPS Cu2p (a, b) and corresponding spectra Auger Cu L₃M_4,5M_4,5 (c, d) spectra for the AN and HPT_5 samples; XPS depth profile of passive oxide layer on e) AN and f) HPT_5 samples. Black lines and symbols were taken from the fresh surface, while red ones were taken after 550 s of Ar⁺ sputtering.

Figure 4c) and d) show the changes observed in the Auger Cu L₃M_4,5M_4,5 peak before and after 550 s of Ar⁺ sputtering for AN and HPT_5 samples. The peak shift to higher kinetic energies after ion etching suggests the presence of Cu in the metallic state after the 550 s of Ar⁺ etching. The calculated Auger parameter of the main Cu 2p_3/2 peak before ion sputtering is equal to 1850.0 eV, which is close to Cu¹⁺, and after 550 s of ion sputtering– 1851.1 eV, which is close to Cu⁰. Hence, one can conclude that after argon ion etching Cu is mainly in a metallic state.

Figure 4e) and f) present the XPS depth profile of the passive oxide layer on investigated surfaces. The results show that for both AN and HPT_5 samples, the passive layers of about 6 nm in thickness were created consisting of a mixture of Cu₂O and CuO oxides. This is consistent with data from the literature stating that the native oxide layer on Cu can reach a thickness of 8–10 nm and is best described by a model structure made up of a duplex-type oxide layer with an inner Cu₂O layer at the interface with the underlying Cu metal and an outer layer of CuO in direct contact with the gas atmosphere²⁶.

SEM/STEM surface observations and bacterium-surface interface analysis

For these samples, SEM as well as STEM/EDX observations were performed as can be seen in Figs. 5, 6, 7 and 8. While applying the higher magnification that is used in fluorescence microscopy, one can confirm after the adhesion test the presence of the adhered bacterial cells on both tested surfaces. On the surface of HPT_5_3.75 (Fig. 6a)) in comparison to AN (Fig. 5a)), fewer bacteria were visible. Moreover, fewer bacteria during the division stage were observed. The etched areas visible in Fig. 5a) and 6a) are the result of the fixing procedure, probably the reaction of Cu, bacteria products, and glutaraldehyde since they were not observed before experiments.It cannot be excluded that the beginning of Cu etching could have started during adhesion experiments. More etched areas in HPT_5 confirm the more excessive ion release from this sample.

Fig. 5.

a) SEM surface observations of Cu AN after adhesion tests; yellow circles indicate dividing bacteria, white arrows show single bacteria, b) BF-STEM image of bacteria just after division and a magnified image of the contact of the bacterium-Cu AN surface, c) BF-STEM and HAADF images of a cross-view of a selected bacterium-Cu AN interface.

Fig. 6.

a) SEM surface observations of Cu HPT_5_3.75 after adhesion tests; yellow circles indicate dividing bacteria, white arrows show unpaired bacteria b) BF-STEM and HAADF images of a cross-section view of a selected bacterium-Cu HPT_5_3.75 interface; the original surface level label is “a guide to the eye”.

Fig. 7.

Chemical element distribution (S, O, Cu, P, N and C) at a cross-section of a) a selected bacterium-Cu AN interface, b) a selected bacterium-Cu HPT_5_3.75 interface.

Fig. 8.

Cu-rich nanoparticles at a cross-section of a selected bacterium-Cu HPT_5_3.75 interface with a Fast Fourier Transform (FFT) of the HR-image of Cu-rich nanoparticles, confirming the presence of Cu₂O. Exemplary Cu₂O nanoparticles are marked in yellow circles.

To understand better what has occurred at the interface between bacteria and the surface, a cross-section of the bacterium-Cu interface was prepared by FIB for AN (Fig. 5b and c) and HPT_5_3.75 (Fig. 6b) samples. The diameter of S. aureus is approximately 0.5 μm. In the case of bacteria adhering to the AN surface, the bacterium maintains its round shape, and the surface beneath the bacteria is plain. There is no interaction between the bacterium and the substrate, which is confirmed by the STEM-EDX map showing the distribution of Cu, C, O, S, P, and N at the interface bacterium- AN (Fig. 7a). These conditions enable the bacteria to move to the division phase (Fig. 6b). In the case of bacteria adhering to the HPT_5_3.75 surface, the bacterium is flattened (Fig. 7b). This image suggests that the bacterial cell wall and membrane have been at least partially damaged during contact with the surface. The surface beneath this bacterium is characterized by a high roughness which is caused by the release of Cu ions into the surrounding space (in particular, to the adhered bacteria). It is well-presented in the STEM-EDX map of the distribution of Cu, C, O, S, P, and N in Fig. 7b. The EDX results show that Cu is distributed within the bacterial cell next to the Cu surface. It is accompanied by an increased concentration of O. The thickness of the “layer” containing Cu and O in the bacteria cell varies slightly along its length and the rate of grain dissolution also varies from grain to grain. This is because the rate of Cu atom release depends on the orientation of nanograins, the planes with high atom density are characterized by increased reactivity²⁷. While looking at the images at higher magnifications (Fig. 8) one can notice that the “layer” consists of nanoparticles rich in Cu and O of less than 10 nm in diameter which were revealed to be Cu₂O phase by high-resolution STEM. Additionally, S can be found inside bacteria above the Cu-rich nanoparticles.

Discussion

This study investigated the antibacterial properties of Cu substrates of various microstructures in adhesion tests against S. aureus, which is a Gram-positive bacterium, negatively charged, of 0.5 μm in diameter²⁸. It is covered by a peptidoglycan lattice overlayer. Peptidoglycan consists of glycan strands and peptides, which play the role of cross-links between glycan strands²⁹. Beneath that lattice layer, there is a lipid membrane²⁹.To verify the response of S. aureus to various nanostructural features, surfaces of the same roughness but differing in nanostructural features were prepared to guarantee the same contact area between the bacterium and the surface at the initial period of contact. It is important to point out that the roughness of the surfaces was at a low level of 4 nm, impossible to achieve using conventional polishing techniques. Moreover, conventional grinding and polishing generate defects in the subsurface layer increasing the similarity of subsurface layers of deformed and undeformed materials in terms of the vacancy and dislocations density¹⁹. Additionally, passive layer chemical composition and thickness were investigated to evaluate the impact of the chemical state of the surface. Even though the samples differed in nanostructure, the chemical composition as well as the thickness of the passive layers were comparable. It has been noticed that AN and HPT_5 passive layers consisted of CuO and Cu₂O and an overall thickness of 6 nm. Previous authors noticed that the passive layer of micrograined and nanostructured cold-sprayed surfaces differed considerably, the former was covered by CuO, and the latter by Cu₂O passive layer³⁰. Therefore, this study makes it possible to compare the impact of nanostructural features on anti-adhesion properties rather than the impact of the passive layer thickness and chemical composition. The difference between AN and HPT_5 samples is neither the surface roughness nor the chemical composition and thickness of the passive layer, but mainly the microstructure: (i) vacancy concentration, (ii) dislocation density, and (iii) grain boundary density.

At the beginning of the contact bacterium-substrate, the bacterium touches the passive layer which consists of CuO and Cu₂O. It has previously been proved that Cu₂O in humidity partially changes to hydroxide Cu-O- which reacts with peptides²⁹. This leads to the degradation of the peptidoglycan lattice. Consequently, Cu ions, which can easily leave the substrate since the thickness of the passive layer is only 6 nm, bind to negatively charged domains, which are located on the cell membrane leading to the reduction in the potential difference causing depolarization of the bacterium. Analyzing Fig. 6b one can see the changes in the HPT_5 sample beneath the bacterium up to approximately 40 nm in depth. This suggests that what matters is the substrate microstructure, not the passive layer since the thickness of the passive layer is only 6 nm. The microstructure features like vacancies, dislocations, and grain boundaries can impact the rate of Cu ions release, being often treated as a determinant of antibacterial properties^31,32. In previous works, it has been proposed that the effective diffusivity of Cu ions depends on the diffusivity of dislocation pipe diffusion^33,34. Others combined it with the average grain size and sub-grain size^30,34. However, these calculations were made for cold-sprayed Cu layers. These layers vary in the microstructure considerably in comparison with HPT-processed Cu samples³⁵ since dynamic recrystallization is observed in such layers. It means that the vacancy concentration, density of dislocations and grain boundaries in such samples are much lower than in HPT-processed samples. Moreover, their roughness is also much greater than the roughness applied in this study, which makes it difficult to distinguish the impact of the microstructural elements. Interesting is also the fact, that after applying RASP to Cu samples, annealed Cu, and RASP Cu of high surface roughness, no obvious difference was noticed in antibacterial properties, tested with S. aureus⁷. The Cu ions concentration resulting from the density of defects and surface roughness was not sufficient to bring the differences to evidence.

In this study, it seems that the rate of Cu ions release depends on the synergetic effect of the concentration of vacancies, the density of dislocation, and grain boundaries. For grain boundaries, one can notice a linear correlation between luminous intensity and d₂ ^−0.5 in the nanoscale regime. The correlation between luminous intensity and d₂ ^−0.5 is much weaker than between microhardness HV0.2 and d₂ ^−0.5 in the Hall-Petch equation. There is no evident correlation between luminous intensity and the density of dislocations/ the square root of dislocation density in the nanograin regime. For HPT_5_3.75 the dislocation density slightly decreased in comparison with HPT_1_3.75. It can be justified by the fact that while the degree of deformation increases, for a certain refined grain size the reached dislocation density starts to decrease, because the dislocations rearrange to form new grain boundaries. However, the microhardness is slightly higher for HPT_5_3.75 than for HPT_1_3.75. This suggests that the increase in the concentration of vacancies and density of grain boundaries rather than the density of dislocations in the nanosize grain regime is meaningful. The correlation between luminous intensity and d₂ ^−0.5 comes from the fact that grain size refinement/ high density of grain boundaries plays a prevailing role in the diffusivity of Cu ions. Apart from the impact of the Cu ions from the surrounding medium the direct contact between the bacterium and substrate matters. When grains are much bigger than the contact area between the bacterium and the substrate, mainly the dislocation density plays an important role in the improvement of antibacterial properties and it is an essential factor. However, when the grains become smaller than the contact area between the bacterium and the substrate, the density of grain boundaries starts to have an important impact on antibacterial properties. It is, not only the density of grain boundaries or dislocations but also the concentration of vacancies that is essential. In HPT_5 the overall density of point defects is 6000 larger than in AN sample while the grain size decreased nearly 70 times and the dislocation density increased nearly 380 times. Generated vacancies accelerate the release of Cu ions. They cannot be neglected in the case of the analysis of the impact of nanostructure on the anti-adhesion properties of HPT-processed Cu. In the case of cold-sprayed Cu, when there is dynamic recrystallization during the process, their impact is probably negligible. The nanograined Cu during direct contact with the bacterium slowly disrupts the structural integrity of the bacterium and decelerates bacterium division and biofilm creation.

To better understand the impact of the microstructural features of the substrate on the anti-adhesion properties of the substrate the intercellular bacterial response mechanism was analyzed. To the best knowledge of the authors, the direct observation of the bacterial intercellular defense mechanism in the case of HPT-processed Cu in contact with S. aureus has never been observed before. The kinetics of bacteria depolarization depend on the microstructural features of the substrate. It has been revealed by EDX analysis that Cu₂O nanoparticles (confirmed by high-resolution STEM analysis) were present at the interface-bacterium-HPT_5.In the previous study³⁷a similar phenomenon was observed for Cu DHP and Cu5Zn5Al1Sn samples with predeposited artificial sweat and after wet/dry cyclic exposures in contact with Bacillus subtilis. However, the origin of the nanoparticles was uncertain as the samples were peroxidized. In this study, the samples were not artificially peroxidized. On their surface, only a passive layer of the thickness of 6 nm was created. It is important to notice that the thickness of the impacted by bacterium area below the surface is approximately 40 nm. It, therefore, suggests that Cu-rich nanoparticles have been created by the Cu ions dissolution from the surface into the cell, not due to the entry of Cu oxides formed in advance at the Cu surface. Probably just afterward, the Cu ions are transformed by bacteria into less harmful nanoparticles (Cu₂O)³⁸. Additionally, sulphur can be found inside bacteria above the Cu-rich nanoparticles, present within 30 nm close to the bacterium-Cu plate interface. One of the possible explanations of sulphur presence inside the cell is the bacterial response to the presence of reactive oxygen species (ROS). ROS appear in bacteria as a consequence of their contact with Cu and uptake of Cu ions. ROS can be also created during the contact of bacteria with Cu nanoparticles³⁶. Cysteine and methionine residues-containing compounds, thioredoxin, glutathione and low molecular weight (LMW) thiols containing the reducing sulfhydryl groups are elements of bacterial antioxidant defence system (ROS scavengers)^39–42. All these compounds contain sulphur in their molecules. They are probably transported within bacterial cell to the area of the highest accumulation of ROS and, thus the highest risk of oxidative damage. Therefore, their migration to ROS-rich areas of bacteria may be manifested by sulphur detection. The interaction bacterium-HPT-processed copper might be more complex than described and other processes e.g. direct or indirect Fe-S clusters degradation can also take place⁴³.

Materials and methods

Material fabrication

High-purity electrolytic tough pitch (ETP) Cu, delivered as a 10-mm-diameter rod, was the material applied in this investigation. The microstructure of Cu was refined by HPT which was performed at the University of Vienna, at the Faculty of Physics. Before experiments, the material was cut into disks of 10 mm in diameter and 0.8 mm in thickness. The disks were subjected to torsional straining for 1 and 5 revolutions with a speed of 0.2 rpm at a constant pressure of 6 GPa at room temperature. The strain defined as simple shear, γ, was calculated according to the equation γ = 2π × r × n/t, where r, n and t are the distance from the torsion axes, the number of applied revolutions and the mean thickness of the sample, respectively. The equivalent strains εeq = γ/√3 calculated in 1.25 mm and 3.75 mm from the central point of the sample after 1 rotation were equal to 6 and 17, respectively. The equivalent strains εeq = γ/√3 calculated 3.75 mm from the central point of the sample after 5 rotations were equal to 85.

Roughness measurements

Roughness was measured for each sample in five randomly located areas using an Optical profilometer WYKO NT 9300 and expressed by arithmetical mean roughness Ra parameter (Ra_avg). Additionally, the standard deviation (SD) has been calculated. The investigations were performed at the Warsaw University of Technology, at the Faculty of Materials Science and Engineering.

Microhardness measurements

Zwick microhardness tester was employed to measure the microhardness of samples under a load of 200 g and a loading time of 10s, operated at the Warsaw University of Technology, at the Faculty of Materials Science and Engineering. The values of Vickers microhardness, HV, were collected along 3 diameters at precise locations: in the sample center, 1.25 mm from the center and 3.75 mm from the center. Afterward, the mean value and standard deviation were calculated for AN, HPT_1_0, HPT_1_1.25, HPT_1_3.75 and HPT_5_3.75.

Microscopy observations

A cross-section of AN sample was prepared using Ion Milling Hitachi IM 4000 and it was further observed using a scanning electron microscope (SEM) SU8000 Hitachi in backscattered electron (BSE) mode at 5 kV electron accelerating voltage. A retractable Photodiode BSE (PDBSE) detector was used which collects low, medium and high-angle BSEs.

Cross-sectional lamellae from HPT_1_0, HPT_1_1.25, HPT_1_3.75 and HPT_5_3.75 were prepared by FIB Hitachi NB5000, parallel to the radius direction of the sample. Before cutting lamellae, a tungsten layer of 2 μm thick was deposited on the selected area on the substrate surface to protect it from the Ga ions impact. Microstructures of samples were revealed using transmission electron microscopy (TEM) JEOL1200 at 120 kV. The observations were performed in a bright field (BF) mode. Stereological and image analysis methods were applied for the quantitative^44,45. To determine their size equivalent diameter (d₂) was used. Moreover, the grain boundary area in the unit volume (S_v) was measured by counting the intersection points of the test lines with the grain boundary network. The investigations were performed at the Warsaw University of Technology, at the Faculty of Materials Science and Engineering.

Surface observations of samples after adhesion tests were performed using a SEM SU8000 Hitachi in a secondary electron (SE) mode at 5 kV electron accelerating voltage. Cross-sectional lamellae from the bacterium-Cu surface interface were prepared by FIB, parallel to the radius direction of the AN and HPT_5_3.75 samples. Before the FIB cutting gold of thickness 5 nm was deposited on top of the samples to protect the sample from Ga-ion sputtering during initial tungsten deposition; to avoid Ga-ion sputtering during FIB sample preparation, a 2 μm thick tungsten layer was deposited on the selected area on top of the substrate surface earlier covered with gold. The lamellae were observed using STEM Spectra 200 by Thermo Fisher Scientific in BF-STEM mode and in high-angle angular dark field (HAADF) mode. The EDX analysis was performed at 200 kV.

Doppler broadening variable energy positron annihilation spectroscopy and variable energy positron annihilation lifetime spectroscopy measurements

Doppler broadening variable energy positron annihilation spectroscopy (DB-VEPAS) measurements have been conducted at the apparatus for in-situ defect analysis (AIDA)⁴⁶ of slow positron beamline (SPONSOR) in HZDR, Germany⁴⁷. Positrons were accelerated and implanted with defined kinetic energies into samples in the positron implantation energy range of E_p = 0.05-35 keV, which translates to a depth profile. A mean positron implantation depth was approximated using a simple material density (ρ) dependent formula⁴⁸:

1

The formalism does not account for the positron diffusion hence only approximates the depth. Implanted into a solid, positrons annihilate in delocalized lattice sites or localize in vacancy-like defects and their agglomerations emitting at least two anti-collinear 511 keV gamma photons once they meet electrons. Since positrons are thermalized and exhibit nearly no momentum compared to the electrons, a broadening of the 511 keV distribution is observed which is mostly due to the momentum of the electrons. The broadening is characterized by a characteristic parameter S, a fraction of the annihilation line in the middle of the spectrum (511 ± 0.93 keV). The S-parameter represents a population of positrons annihilating valence electrons with low momentum and accounts for vacancy type defects and their concentration⁴⁹.By fitting the S(E_p) dependence it is possible to calculate the so-called effective diffusion length of positrons, L₊, and resulting defect concentration, c_V⁵⁰:

2

where N = 6.51 × 10²² at/cm³ is atomic density of Cu, ν_V≈ (1.2 ± 0.2) 10¹⁴ at s⁻¹ is a specific positron trapping rate (trapping coefficient)²³, τ_B ≈ 114 ps is the bulk positron lifetime⁵¹ and L_+,B =192 ± 2 nm the diffusion length calculated from the pristine sample. The effective positron diffusion length, L₊, which is inversely proportional to defect concentration was calculated using the VEPFit⁵². The code permits to fit S(Ep) curves for multilayered systems and delivers thickness, L₊, and specific S-parameters for each layer within a stack and at the surface.

Variable energy positron annihilation lifetime spectroscopy (VEPALS) measurements were conducted at the Mono-energetic Positron Source (MePS) beamline at HZDR, Germany⁵³ using a CrBr₃ scintillator detector coupled to a Hamamatsu R13089-100 photomultiplier (PMT) tube and a processing unit the SPDevices ADQ14DC-2X digitizer⁵⁴. The time resolution function was in the range of 0.250 ns. All spectra contained at least 1·10⁷ counts. A typical lifetime spectrum N(t), the absolute value of the time derivative of the positron decay spectrum, is described by:

3

where k different defect types contributing to the positron trapping are related to k + 1 components in the spectra with the individual lifetimes τ_i and intensities I_i (ΣI_i=1)⁵⁵. All the spectra were deconvoluted using a non-linear least-squares fitting method employed within the fitting software package PALSfit⁵⁶ into 3 discrete lifetime components, which represent localized annihilation at 2 different defect types (sizes; τ₁ and τ₂) and negligible third background contribution. The corresponding relative intensities reflect relative variations of defect concentration of each defect type (size) as long as the size of compared defects is in a similar range. In general, positron lifetime is directly proportional to defects size, i.e., the larger is the open volume, the lower is the probability the longer it takes for positrons to be annihilated by electrons^55,57. The positron lifetime and its intensity have been probed in function of positron implantation energy E_p. The average positron lifetime τ_av is defined as τ_av = ∑_iτ_i·I_i.

X-ray photoelectron spectroscopy measurements

The chemical composition of the surface of the AN and HPT_5 samples was investigated using X-ray photoelectron spectroscopy (XPS) (Microlab 350 spectrometer, Thermo Electron) located at the Institute of Physical Chemistry Polish Academy of Sciences. A non-monochromatic X-ray source (Al Kα) with an energy of 1486.6 eV for the most intense line and a power of 300 W was used for XPS measurements. Spectra were obtained with a 100 eV pass energy for the survey scan and 40 eV pass energy for narrow regions (high resolution spectra). The spectra were fitted using Avantage Surface Chemical Analysis software (ver. 5.41) using a smart-type background and a Gaussian/Lorentzian function. Element peak intensities were corrected by Wagner sensitivity factors. The binding energies measured were corrected in reference to the energy of C 1s peak at 284.8 eV. The composition depth profiles were measured using an Ar + ion gun (EX05—Microlab). The following parameters of ion etching were applied: 10–40 s etching steps, ion energy 2 keV, beam current 1.3 µA, crater size 36 mm².After each sputtering session, the XPS spectra were acquired at an energy of 1486.6 eV.

X-ray measurements

The X-ray diffraction (XRD) method was used to determine the Cu samples’ dislocation density (ρ). XRD has been carried out on a Bruker D8 Discover diffractometer, located at the Warsaw University of Technology, at the Faculty of Materials Science and Engineering. equipped with filtered Co Kα radiation (λ = 0.17902 nm) and working with Bragg-Brentano geometry on a point beam with a diameter of approximately 1 mm. X-ray patterns were gathered from an area of approximately 1 mm in diameter, which center was located 3.75 mm from the sample center and in the case of the HPT_1 sample also in the sample center. The conditions were as follows: accelerating voltage = 35 kV, current = 40 mA, angular range of 2θ from 35° to 120°, scanning step size Δ2θ = 0.025°, and the counting time = 6 s. The ρ was calculated from XRD peak broadening using a modified Williamson-Hall (W-H) plot:

4

where K for fcc materials equals 14.4 with the Burgers vector of dislocations,

b, along 〈111>, ε is the lattice strain evaluated from W-H plot, b is the Burgers vector for Cu - b = 0.2556 nm.

Bacterial adhesion test

Before adhesion tests, samples were mechanically grounded and polished following the same procedure for all samples using firstly diamond suspension with decreasing diamond powder diameter (3 and 1 μm) followed by 2-hour vibratory polishing in the Mastermet by Buehler suspension consisting of SiO₂ of 0.06 μm in size. The suspension did not show any antibacterial properties as presented in the Supplementary information in Figure S2. The sample preparation was performed at the Warsaw University of Technology, at the Faculty of Materials Science and Engineering.

Prior to the experiments, performed at the Medical University in Lublin, the samples (in triplicate) were sterilized in 98% ethanol solution in the 12-well plates (Costar, Corning Inc., USA), for 15 min., with subsequent drying under UV lamp (OSRAM 15 W, at a distance of 15 cm from the samples; 30 min.). Next, samples were incubated in 1.5 mL of 3.0 × 10⁸ CFU/mL Staphylococcus aureus ATCC 25 923 suspensions (in Mueller–Hinton broth (Biomaxima, Lublin, Poland) diluted 100-fold in sterile 0.9% NaCl). The plate was protected with stripes of parafilm (Bemis, Neenah, WI, USA) to prevent evaporation of the liquid phase and incubated for 2 h at 25 °C, at 100 rpm (in Innova 42 incubator; New Brunswick Scientific, Enfield, CT, USA). Afterward, the samples were transferred into the fresh 6-well plate and gently washed using 0.9% NaCl (5 mL 0.9% NaCl, 3 times), to remove all unbound bacterial cells.

The adhered bacterial cells were detected on the sample surfaces after staining with Viability/Cytotoxicity Assay Kit for Bacteria Live & Dead Cells (Biotium, Fremont, CA, USA), which enabled to differentiate live (green) and dead (red) bacterial cells. The visualization and cell counting were performed using fluorescence microscopy (in Olympus BX41 microscope equipped with UC 12 Soft Imaging System camera) using CellSens Dimension 1.12 software (Olympus, Japan). For further quantification of the results, the areas of the image occupied by fluorescent light (indicating the presence of bacteria) were calculated using the ImageJ program, and luminous intensity, defined as the surface fraction of pixels above a certain intensity level, was calculated.

After the experiment, samples were fixed in 2.5% glutaraldehyde (15 min), followed by washing in PBS pH 7.4 (three times) and dehydration in subsequent ethanol solutions (30%, 50%, 70%, 80% and 99.8%, 15 min. per each step). Surface observations of samples after adhesion tests and cross-section observations of the bacterium-substrate interface were performed using SEM, FIB and STEM as explained in the Microscopy observations subsection.

Conclusions

This work investigated the impact of crystallographic defects introduced by HPT in Cu on its anti-adhesion properties and the intercellular bacterial response in comparison with free-from-defect samples.

1. The average grain size of 9.6 μm of annealed Cu was reduced to 140 nm after HPT in 5 rotations. The density of dislocations of 9.2 × 10¹¹ m^-2and 3.4 × 10¹⁴ m^-2, respectively for annealed and HPT-processed Cu and point defect concentration of ~ 2 · 10¹⁶ cm^-3and 1.2 · 10²⁰ cm^-3, respectively for annealed and HPT-processed Cu were measured in the subsurface of vibratory polished samples.

2. The passive layer formed on the surface of HPT-processed and annealed samples, investigated by X-ray photoelectron spectroscopy, is of the same chemical composition, consisting of Cu₂O and CuO, and a thickness of 6 nm. Moreover, the surface roughness of these samples was comparable.

3. HPT-processed Cu shows enhanced anti-adhesion properties while in contact with S. aureus compared to micrograined Cu. There is a linear correlation between luminous intensity and d₂^-0.5.

4. The study of the intercellular bacterial response to HPT-processed Cu suggests that Cu-rich nanoparticles (Cu₂O), present in the bacterium, have been created by the Cu ions dissolution from the Cu surface into the bacterium cell. Consequently, the concentration of sulphur, suggesting the presence of cysteine and methionine residues-containing compounds, thioredoxin, glutathione, and low molecular weight thiols, increases in the vicinity of Cu-rich nanoparticles.

Author contributions

A.T.K. Conceptualization, Methodology, Research, Validation, Writing - Original Draft, Supervision, Funding acquisition; A.M. Research, Writing - Original Draft; P.S. Research; K.B. Research; A.R. Research, Writing - Original Draft; M.K. Research; D.S. Research; M.S. Research; B. (A) Research; M.O.L. Research, Writing – Original Draft; M. (B) Research; E.R. Research; E.H. Research; A.W. Research; M.L. Validation, Writing – Review and Editing; A.B. Research, Methodology, Validation, Writing – Review and Editing. All authors reviewed the manuscript.

Data availability

The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.