Interaction of Albumin and Fibrinogen with Stainless Steel: Influence of Sequential Exposure and Protein Aggregation on Metal Release and Corrosion Resistance Open Access

The biomedical austenitic stainless steel grade Type 316L (UNS S31603⁽¹⁾) is widely used, e.g., for orthodontic or orthopedic implant materials.^1-2  Studies on metal release and biocorrosion in protein-rich solutions and environments, especially containing more than one protein, are however relatively rare. This research area has recently been reviewed.³  As the release of metals in contact with the human body can induce adverse effects,⁴  prevailing mechanisms need to be further explored. Metals are released to different extents from stainless steel in biological environments as a result of corrosion (metal oxidation), chemical or electrochemical dissolution of the surface oxide, physical processes such as wear, or their combinations.³  Several systematic studies suggest that metal release from stainless steel in simulated physiological solutions including proteins at least to some extent is governed by complexation/ligand-binding processes^5-6  and that these processes are adsorption-controlled.⁷  This has recently also been observed to be the case for stainless steel in contact with citrate that is a metal-complexing agent.⁸  The detachment of metal-bound proteins from the surface depends furthermore on, e.g., the protein solution concentration and the presence of other proteins. Theoretically, the surface/solution protein exchange increases with increasing protein concentrations in solution⁹  and upon the occurrence of the Vroman effect.¹⁰  The Vroman effect describes the displacement of surface adsorbed proteins by solution proteins of larger binding affinity (usually a larger sized protein). For example, for surfaces exposed in human blood, the adsorption of the smaller and most abundant protein albumin might already be displaced within a minute by immunoglobulin G, followed by fibrinogen, fibronectin, and kininogen.¹¹  It is therefore crucial to study the combined effect of proteins on the metal release from stainless steels. However, it is not trivial to design a metal release study investigating the Vroman effect for stainless steel, because the released amounts of metals from such alloys are very low³  and hence require longer time periods than minutes to be measured in sufficient amounts. This study therefore applied an experimental approach exposing coupons in a sequence in two solutions in which the second protein solution was introduced after 5 h. Investigations were performed using albumin (BSA), known to induce a relatively high extent of metal release for Type 316L,¹²  and fibrinogen (Fbn), a significantly larger protein (340,000 g/mol as compared to 66,500 g/mol). For comparison, a range of control experiments and the behavior of an inclusion-abundant stainless steel grade (Type 303 [UNS S30300]) were investigated in parallel. Studies of grade Type 303 are rare in the biomedical field, but exist in the case of dermal contact for which literature findings report significant amounts of nickel release in artificial sweat.¹³  This inclusion-rich stainless steel grade is expected to have a lower corrosion resistance and to be governed by different metal release mechanisms as compared to Type 316L.

The main objective of this study is to investigate whether the Vroman effect influences the extent of metal release from stainless steel in binary protein solutions relevant for human body fluids.

Massive coupons of austenitic stainless steel, grades Type 316L and Type 303, were investigated in this study. Type 316L pieces (15 mm × 15 mm × 1 mm) were cut from a cold-rolled sheet (ThyssenKrupp, Germany), and 2 mm thick discs (Ø 20 mm) of Type 303 were cut from cold drawn rods (Ugine-Savoie Imphy groupe Usinor, France). The nominal bulk composition is given in Table 1. Some microstructural information, e.g., grain size and orientations, for the Type 316L grade can be obtained from previous studies on the same grade (and batch).^14-15  To assess the presence of inclusions, coupons of Types 303 and Type 316L were polished in steps down to 0.25 μm (diamond paste) and investigated by optical microscopy (Leica DM2700 M^†) at magnifications up to 50×.

All coupons had a total geometric surface area of approximately 6 cm² to 11 cm² (each defined separately). The coupons were abraded using 1200 grit SiC paper, followed by ultrasonic cleaning in acetone and isopropyl alcohol for 7 min each, subsequently dried with cold nitrogen gas, and aged (stored) for 24±1 h in a desiccator at room temperature. During exposure in solution, the surface area to solution volume ratio was kept constant at 1 cm²/mL, which means a solution volume between 6 mL and 11 mL (depending on surface area). For the sequential exposures, the Type 316L coupons (7 cm²) were exposed in 7 mL solutions and the Type 303 coupons (10 cm²) in 10 mL. Triplicate coupons and one blank sample (test solution only) were exposed in parallel for each grade, time period, and test solution in closed acid-cleaned centrifuge tubes agitated bi-linearly (12° inclination, 22 cycles/min) at 37±0.5°C at dark conditions (Stuart platform-rocker incubator). All vessels were acid-cleaned in 10% HNO₃ for at least 24 h, rinsed four times in ultrapure water (18.2 MΩ·cm, Millipore^†) and dried in ambient laboratory air. All chemicals used were of analytical grade (p.a.) or puriss p.a. grade (in the case of nitric acid, used to acidify solution samples prior to digestion, c.f., Digestion of Solution Samples section). The pH of all test solutions was measured before and after exposure (with pH changes less than 0.06 pH).

The following solutions were prepared from analytical grade NaCl, Na₂HPO₄, KH₂PO₄, 50% NaOH, bovine serum albumin (BSA, Sigma Aldrich A7906^†), and fibrinogen from bovine plasma (Fbn, Sigma Aldrich F8630^†):

• •

  Phosphate buffered saline (PBS): 8.77 g/L NaCl, 1.28 g/L Na₂HPO₄, 1.36 g/L KH₂PO₄, adjusted with 50% NaOH to pH 7.2 to 7.4;

• •

  PBS with 10 g/L BSA;

• •

  PBS with 10 g/L Fbn;

• •

  PBS with 10 g/L BSA and 10 g/L Fbn;

• •

  PBS with 40 g/L BSA and 2.67 g/L Fbn.

Metal release studies in these solutions were conducted for 4 h and 24 h. Sequential investigations were conducted as schematically described in Figure 1 with two subsequent solutions and four time points (1, 4, 6, and 24 h). The following sequential solutions (first and second solution) were used, of which the first three combinations were investigated for comparative reasons:

• •

  PBS and PBS (denoted “PBS”)

• •

  PBS + 40 g/L BSA and PBS + 40 g/L BSA (denoted “BSA”)

• •

  PBS + 2.67 g/L Fbn and PBS + 2.67 g/L Fbn (denoted “Fbn”)

• •

  PBS + 40 g/L BSA and PBS + 5.34 (final concentration of 2.67) g/L Fbn (denoted “BSA, Fbn”).

A higher Fbn concentration was added in the final sequential test, as it was diluted twice by the BSA solution, with an initial concentration of 40 g/L BSA (0 h to 5 h) and a final concentration of 2.67 g/L Fbn and 20 g/L BSA after 5 h exposure.

After the sequential exposures, the metal coupons were removed from the solution, rinsed with ultrapure water, and stored at dry conditions in a desiccator prior to analysis by means of infrared reflection absorption spectroscopy (IRAS) and x-ray photoelectron spectroscopy (XPS).

Because of a tendency to form hydrogels, the exposed solutions with proteins were digested prior to analysis using atomic absorption spectroscopy (AAS). Before digestion, the samples were first acidified with 65% nitric acid to a pH lower than 2 and then stored frozen (−25°C). After unfreezing, between 1 mL and 5 mL of the sample solution was then diluted with ultrapure water and at least 2.5 mL 30% ultrapure hydrogen peroxide to a total volume of approximately 10 mL. The samples were then digested in an ultraviolet (UV) digester (90°C to 95°C, Metrohm 705^† UV digester). The dilution factor was calculated from the final volume (after digestion) divided by the initial volume of the added sample solution.

Released amounts of non-precipitated Fe, Cr, Ni, and Mn in solution (aqueous metals) were determined by means of graphite furnace AAS (GF-AAS; Perkin Elmer AA800^† analyst) and for some samples (for Fe) by using flame AAS. All analyses were based on three replicate readings for each sample and a quality control sample of known concentration was analyzed every fifth sample. Quality control samples in different sample matrices (different protein solutions) did not significantly deviate from the nominal concentration + blank concentration of the corresponding protein solution. The limits of detection, as determined from three times the maximum standard deviation of all blanks (solution samples without any stainless steel coupon) were 0.008 μg Fe/cm² (8 μg/L), 0.002 μg Cr/cm² (2 μg/L), 0.003 μg Ni/cm² (3 μg/L), and 0.002 μg Mn/cm² (2 μg/L). Calibration was conducted using four or five calibration standards: 0 μg/L (ultrapure water), 10 μg/L, 30 μg/L, and 100 μg/L for Ni and Mn, 0, 10, 30, 60, and 80 μg/L for Cr, and 0, 50, 100, and 200 μg/L for Fe. The blank values, mainly reflecting the metal content of the proteins, were <79 μg Fe/L, <5 μg Cr/L, <3 μg Ni/L, and <12 μg Mn/L. These observations are in agreement with previous findings for Fe, Cr, and Ni in BSA, showing Fe to be most prevalent, followed by Cr and Ni.^7,16 

The amount of released, non-precipitated metal in solution (in the unit μg/cm²) corresponds to the measured metal concentration in solution (μg/L) normalized to the exposed geometrical surface area of the stainless steel coupons (cm²) and the solution volume (L). In the case of the sequential metal release investigations, the solution volume prior to sampling was considered, e.g., 10 mL (1 h), 7.5 mL (4 h), 10 mL (6 h), and 7.5 mL (24 h), for the Type 316L coupons. All release data are presented in the unit μg/cm² as the mean value of triplicate coupons exposed in parallel, with the corresponding blank exposure concentration exposed in parallel subtracted (if > 0), and multiplied by the corresponding dilution factor. Error bars show the standard deviation between these triplicate individual coupon exposures.

Open-circuit potential (OCP) and linear polarization resistance (LPR) measurements were conducted in an alternating way of relevance for the sequential metal release investigations for the solution combinations “PBS,” “BSA,” “Fbn,” and “BSA, Fbn.” Half of the first solution volume was removed after 5 h and replaced by the second solution. The surface area of the coupons exposed in the electrochemical investigations was approximately 3 cm² and measured for each coupon individually. The solution volume was 36 mL, that is, approximately a 10-fold lower loading (surface area to solution volume ratio) as compared to the release investigations because of limitations in the electrochemical setup. Six cells were positioned in a water bath. The temperature of the cells was controlled at 37°C using a contact thermometer positioned in a reference cell filled with PBS. Because of a higher heating temperature, the first measurements in PBS and PBS + 40 g/L BSA were performed with a temperature variation of ±4°C, while the other measurements were controlled at ±1°C. All electrochemical investigations were conducted using a PARSTAT multichannel PMC Chassis^† instrument equipped with six PMC-1000^† (AC/DC) channels. Platinum wires were used as counter electrodes, and Ag/AgCl saturated KCl electrodes as reference electrodes. All exposures were conducted at aerated conditions in order to mimic given conditions for the metal release investigations. Because of the experimental setup (open cells), some solution volume evaporated during the measurements. The solution was therefore replenished with ultrapure water to the original volume prior to the LPR measurements and in the morning of the last 18 h OCP measurement. The following measurements were conducted:

• •

  1 h OCP

• •

  LPR (±15 mV_OCP)

• •

  3 h OCP

• •

  LPR

• •

  2 h OCP (half of the solution exchanged after 1 h)

• •

  LPR

• •

  18 h OCP

• •

  LPR.

The measurements were conducted in duplicates for Type 303 and Type 316L in the four solution combinations. The coupons were prepared identically as compared with the metal release measurements (ground, cleaned, and stored at dry conditions for 24 h prior to measurements). The LPR measurements were scanned between −15 mV and 15 mV of the OCP, at a scan rate of 0.1667 mV/s (step height 0.1 mV, step time 0.6 s). The corrosion resistance (R_p) was estimated with the VersaStudio 2.50.3^† software and normalized to the surface area of the coupon in solution contact (3 cm²).

IRAS was conducted using a Bruker Tensor 37^† instrument on two replicate coupons of Type 303 and Type 316L after exposure in PBS, PBS with 40 g/L BSA and 2.67 g/L Fbn, and all four sequential solutions: “PBS,” “BSA,” “Fbn,” and “BSA, Fbn.” A Pike^† IRAS accessory with an angle of incidence of 80° was used. Five hundred twelve spectra were acquired for each sample and normalized to the background spectrum (polished stainless steel) acquired prior to each measurement. Presented spectra are based on one of the measured replicates; all replicates measured for the respective exposures showed very similar results.

XPS (UltraDLD^† spectrometer, Kratos Analytical) measurements using a monochromatic Al K_α x-ray source (150 W) were performed on two separate surface areas approximately sized 700 × 300 μm² for compositional analysis of the outermost surface composition (with information depth of 5 nm to 10 nm). Measurements were performed on abraded (as described earlier) Type 303 and Type 316L coupons, and coupons (from both grades) exposed to each of the four sequential solutions (“PBS,” “BSA,” “Fbn,” and “BSA, Fbn”). Elements of the outermost surface oxide were distinguished by running a wide spectrum and high-resolution spectra (pass energy of 20 eV) for the main alloying elements: Fe2p, Cr2p, Ni2p, Mn2p, N1s, O1s, and C1s. Binding energies were corrected by the C1s peak at 285.0 eV. Figure 2 shows examples of the high-resolution spectra. Nickel was entirely found in its metallic state¹⁷  at 853.1±0.2 eV, Figure 2. Peak overlap between nickel and manganese was accounted for in the same way as described in Fredriksson, et al.¹⁸  Manganese was entirely found in its oxidized state¹⁷  at 641.3±0.3 eV, Figure 2. Iron peaks were divided in their metallic (707.8±0.7 eV) and oxidized (711.8±1.1 eV) states, as well as chromium (metallic: 574.4±0.3 eV; oxidized: 577.5±0.9 eV), as indicated in Figure 2.¹⁷  The results are presented as the relative mass content of oxidized iron, chromium, and manganese in the outermost surface oxide, e.g., [Cr_ox/(Cr_ox+Fe_ox+Mn_ox)] × 100 wt%. Furthermore, the atomic ratio of nitrogen (400.1±0.5 eV) to the sum of the carbon peaks corresponding to C–N, C–O (286.6±0.1 eV), C=C–O, and O=C–N (288.4±0.2 eV) bonds, Figure 2, denoted as N/(C2+C3), is presented. The C1 peak at 285.0 eV is not considered as it mainly derives from adventitious surface contamination.

For independent samples (normal case in this study), a student’s t-test of unpaired data with unequal variance was conducted in the software KaleidaGraph 4.0^†. If the p-value was below 0.05, the difference was considered statistically significant. The word “significant” is only used for statistically significant differences in this manuscript. In the case of the sequential tests, where the solution samples derived from the same stainless steel coupon at different time points, a student’s t-test of paired data was performed using the same software.

Type 316L is a biomedical stainless steel grade that is not expected to corrode actively at conditions tested in this study. Earlier studies have, however, shown that this grade may be releasing higher amounts of certain metals at passive conditions in biologically relevant media, as compared to other grades.^12,19-20  Furthermore, interactions with proteins may either enhance or reduce the total amount of metal release, and protein-induced reductions of metal release have mostly been shown for grades of lower corrosion resistance.^3,12  It is therefore interesting to compare Type 316L with a grade of lower corrosion resistance (Type 303) because of a relatively high number of inclusions. Optical microscopy confirmed a large amount of inclusions for Type 303 taking a considerable percentage of the surface area in claim, Figure 3(a). Only a few inclusions were observed for grade Type 316L, Figure 3(b). The nominal composition of these two grades differs mainly in the sulfur and molybdenum content, Table 1. Therefore, Type 316L is expected to have a higher pitting corrosion resistance as compared with Type 303, which is rich in MnS-inclusions.²¹ 

Total amounts of released and non-precipitated amounts of metals in solution (Fe, Cr, Mn, and Ni) after 24 h exposure to PBS, PBS + 10 g/L BSA, PBS + 10 g/L Fbn, PBS + 10 g/L BSA + 10 g/L Fbn, and the physiological concentration PBS + 40 g/L BSA + 2.67 g/L Fbn (all at pH 7.4, 37°C) are presented in Figure 4. Similar to previous findings,¹²  the presence of BSA enhances the metal release from Type 316L. The effect is significant for Fe and Ni in the case of the higher BSA concentration (PBS + 40 g/L BSA + 2.67 g/L Fbn), Figure 4(a). No effect on the extent of metal release was evident in the presence of Fbn alone when compared with findings in PBS. In the presence of both BSA and Fbn, the amount of Cr and Ni in solution was lower as compared with findings with BSA alone, Figure 4(a). Interestingly, this behavior was very different compared with findings for Type 303. Generally, the release of metals from Type 303 was dominated by Mn, most probably deriving from the large extent of inclusions.²²  The release of Fe, Cr, and Ni was furthermore lower for Type 303 as compared to Type 316L in all BSA-containing solutions, Figure 4. In contrast to findings for Type 316L, Type 303 showed the highest release of metals (dominated by Mn) in PBS containing Fbn (10 g/L), Figure 4(b). This might indicate different metal release mechanisms for Type 303 and 316L and is further discussed in the Implications, Strengths, and Weaknesses of This Study section. Measured amounts of released metals in solution after 4 h and 24 h (independent samples) are compiled in Table 2. Although there is no statistically significantly higher amount of released metals in solution after 4 h compared with 24 h, the amount determined after 4 h was often slightly higher and the variation among triplicate samples larger as compared to the amount determined after 24 h. This might be a sign of protein aggregation and precipitation, effects that would result in an underestimation of the total amount of metal release from stainless steel.

Figures 5 and 6 show released and non-precipitated amounts of Fe, Cr, Mn, and Ni in solution for the four sequential solution investigations. As evident from the measured amount of released Fe in solution from Type 316L in “BSA” and for Type 303 in “BSA, Fbn,” a strong precipitation effect was observed as implied by strongly decreasing amounts of released Fe in solution with time. This effect was mostly observed in solutions containing high concentrations of BSA (40 g/L). It was most probably also influenced by the experimental setup (Figure 1), as the solution for the measurements was pipetted from the top of the tube and protein aggregates could precipitate onto the stainless steel coupon that was removed from the solution after the exposure. Co, Cr, and Ni ions are known to induce albumin aggregation and precipitation,^23-25  effects that depend on the albumin concentration, the metal ion concentration, and the temperature. Although an underestimation of metal release in protein-containing solutions resulting from protein precipitation has previously been indicated,^23,26  the results of this sequential study show metal release levels from stainless steel Type 316L far below previously reported findings in comparable solutions (PBS with 10 and 100 g/L BSA).¹²  This is most probably related to the fact that the entire exposure solution was digested prior to analysis in previous studies, whereas solution samples in this study were pipetted from the test vessels before being analyzed, hence only determining the amount of released metals in solution without considering metals in precipitated aggregates. The substantial influence of precipitation on the total measured amount of released metals from stainless steel in this sequential study makes it difficult to draw detailed conclusions on the Vroman effect. However, it is evident from Figures 5 and 6 that: (i) the presence of proteins increases the extent of metal release in solution as compared to PBS only, (ii) the release of Fe, Cr, and Ni from Type 316L is higher in “BSA, Fbn” as compared to “BSA” after Fbn has been introduced to the solution (see time points 6 h and 24 h), and that (iii) Type 303 showed one replicate coupon out of three to corrode in the solution “Fbn,” as evident from relatively high amounts of Fe, Cr, and Ni in solution after 24 h. Compared with the reference measurements with only one solution, Figure 7, for PBS and PBS + 40 g/L BSA + 2.67 g/L Fbn, it is evident that the largest differences were observed for the protein solution and Type 316L (Figure 7[a]), most probably resulting from a precipitation effect induced by the experimental setup.

Figure 8 shows OCP and LPR measurements for Type 303 and Type 316L in the four sequential solutions at 37°C for 22 h to 24 h. Half of the solution was replenished after 5 h with the second solution. After approximately 12 h to 16 h (during the night), evaporation of the solutions became evident in the OCP measurements and had to be replenished in the morning (any resulting erroneous data were removed from the graphs). In general, most solutions and both grades revealed OCP values of approximately −0.1 V (vs. Ag/AgCl sat. KCl) at 37°C, which is comparable to a level at approximately −0.04 V at room temperature previously determined for Type 316 at similar conditions.¹²  For one out of three replicate coupons exposed in the “BSA, Fbn” solution, Type 303 revealed one noticeable corrosion event (rapidly shifting OCP to −0.4 V) before re-passivating, Figure 8(a). Otherwise, Type 303 disclosed in all solutions both higher OCP values and higher LPR values as compared to Type 316L, Figure 8. This was most evident in the “BSA” solution, in accordance with significantly lower amounts of released Fe in this solution, Figure 4. Not all LPR values could be determined for the automated alternating measuring sequence. However, measured values correspond well with the measured OCP; that is, they increase with increased and stable OCP levels, and decrease after a reduction of the OCP. Observed LPR values were, however, significantly lower than expected when compared with previous findings for Type 304 (UNS S30400) in PBS and PBS + 10g/L BSA at room temperature.¹²  It is unclear whether this may be a measurement or a temperature effect. In accordance with the metal release investigations for Type 316L, Figure 4(a), no difference in OCP was observed between exposures in the sequential solutions of PBS, Figure 8(b), and Fbn, Figure 8(d). The release of Mn and Fe from Type 303, Figure 4(b), seems furthermore not correlated to any visible changes (such as metastable pitting events) during the OCP measurements, Figure 8. This indicates that the inclusions are mainly chemically dissolved. In accordance with the sequential metal release investigations, Figures 5 and 6, one of the replicate coupons of Type 303 was actively corroding (during 6 h and 24 h) in the “Fbn” solution during the metal release investigation, and in the “BSA, Fbn” solution during the OCP measurements. This indicates test conditions that are close to the transition between passivity and active corrosion for Type 303 in this study. No general difference in OCP values was observed for Type 303 upon exposure in the Fbn and the BSA solution. A difference was only evident for Type 316L, Figures 8(c) and (d).

Figure 9(a) presents the relative metal composition (based on mass) of the outermost surface oxides of Type 303 and Type 316L after abrasion (reference) and after exposure to the four sequential solutions, as deduced by XPS. All cases revealed surface oxides composed of oxidized Cr, Fe, and Mn. No oxidized nickel was observed, only signals from its metallic state, which imply a thin surface oxide and the presence of Ni in a metal enriched layer present beneath the surface oxide.^27-31  Because of large variations among different coupons and/or lack of Mn, no statistical significance calculations could be performed for that element. In agreement with previous findings,^12,23  Cr was significantly (p < 0.05) enriched in the surface oxide on Type 316L exposed to all BSA-containing solutions when compared to the abraded reference coupons and to coupons exposed in PBS. Similarly, the amounts of released Fe in solution were substantially reduced for the same conditions. For the Fbn solution, this difference was not statistically significant as compared to the abraded reference or findings in the PBS solution. For Type 303, no metal or metal oxide signals were detected for any of the coupons exposed to the Fbn-containing solutions, which suggests a significantly thicker conformation of adsorbed proteins (>10 nm) as compared with similar conditions for Type 316L, Figure 9(a). For the coupons exposed to the BSA solution, Cr was significantly (p < 0.05) enriched and Fe significantly (p < 0.05) reduced in the surface oxide of Type 303 as compared to the reference coupons and to the coupons exposed in the PBS solution. The observed surface oxide composition for Type 316L exposed in the BSA solution was in accordance with previous findings at similar conditions, and is the result of a preferential Fe release and a general ability by BSA to complex metals from the metal oxide.^3,12  This furthermore highlights that the significantly lower extent of released metals in solution determined in this study as compared with the previous study¹²  is related to specific precipitation effects in the experimental setup and not because of an overall lower total amount of released metals. Figure 9(b) shows the atomic ratio of nitrogen to the sum of the oxidized carbon peaks predominantly related to adsorbed proteins. Some oxidized carbon may be related to surface contamination. Similar to findings of a previous study,¹²  a ratio of approximately 0.4 was determined for both grades after exposure in the BSA solution. The ratio in Fbn-containing solutions was higher (0.5 to 0.6). In the mixed “BSA, Fbn,” the ratio was 0.5 to 0.6, which means that Fbn was incorporated in the adsorbed proteins for both grades (statistically significant increased ratio as compared to the BSA solution).

Figure 10 shows IRAS spectra of the Type 303 and Type 316L coupons exposed for 24 h in different solutions. The peaks in the 1,400 cm⁻¹ to 1,750 cm⁻¹ range related to adsorbed proteins.^32-33  For the coupons exposed in PBS, these peaks were essentially missing and the small peaks in this region could be attributed to surface contamination. The broad band in the 950 cm⁻¹ to 1,150 cm⁻¹ region, present in all spectra, is a result of the presence of adsorbed phosphates (from PBS).³⁴  The amide I band (at approximately 1,670 cm⁻¹) is attributed to ν_s(C=O) and ν_s(C–N) stretches of the amide group. This band was visible for all exposure conditions with proteins. The amide II band located at approximately 1,530 cm⁻¹ was also observed on all coupons exposed to proteins and attributed to a mix of ν_s(C–N) and δ(N–H). The bands at 1,410 cm⁻¹ to 1,430 cm⁻¹ include the contribution from C–H bending and from COO⁻ groups in amino acids of the proteins. Shifting of the amide I band and its deconvolution into different peaks could provide conformational information.³⁵  However, the interaction between functional groups of the amino acids of the proteins and the alloy surface would also give rise to peak shifts.³²  These interactions could be the result of electrostatic forces or to ligand formation between the metals at the alloy surface and functional groups of the amino acids such as COO⁻.³²  It is therefore difficult to extract conformational information for proteins adsorbed to metallic surfaces based on information gained for the amide I band. It is, however, evident that the amide I band shifted (>20 cm⁻¹) with respect to the band positions of BSA and Fbn in bulk.³²  The shift of the amide I peak with respect to solution species of the proteins hence indicates a difference in the structure and binding of the adsorbed protein layer compared bulk species. A band shift would not have been detected if multilayers of proteins, not affected by the surface, were present.³² 

Overall, generated data in this study suggest that BSA at least to some extent is displaced by Fbn in mixed protein solutions, as indicated by XPS and an increased determined amount of released metals in solution. Hence, the results indicate a Vroman effect. However, its actual effect on metal release or biocorrosion was more difficult to assess as the response was strongly dependent on the stainless steel grade. Fbn interacted totally different with the Type 303 surface, forming thicker patches or layers of adsorbed proteins that enhanced the release of metals and locally induced corrosion events, while its interaction and influence with Type 316L was minor in terms of metal release and electrochemical activity. Protein adsorption and its surface interactions should in the future be further investigated by means of quartz crystal microbalance with dissipation to assess conformational properties of adsorbed proteins. Observed differences in response between Type 303 and Type 316L are most probably mainly explained by the presence of MnS-rich inclusions for the former grade. Both proteins are net negatively charged at pH 7.4 (isoelectric points of Fbn 5.5 to 5.8^36-37  and BSA 4.7 to 5.2³⁸  pH). BSA has previously been shown to adsorb as a monolayer on stainless steel Type 316L surfaces in an approximate thickness of 4 nm.¹⁶  Fibrinogen on the other hand has been reported to adsorb (after 1 h) on stainless steel Type 316L as a multilayer (side-on configuration) with an approximate thickness of 3.5 nm to 5 nm for a Fbn concentration of 1 g/L to 5 g/L in PBS.³⁹  In this study, at least for Type 303, the XPS results indicate a thickness of the Fbn layer of >10 nm. The band shifts of IRAS suggest that this layer did not achieve bulk properties, hence indicating a single layer (or patches) of Fbn. Together, the IRAS and XPS results hence point in the direction of a monolayer in which Fbn adsorbs in a configuration different from the side-on configuration. Except for the difference in MnS-inclusions, there might be other possible explanations for observed differences between the grades. One difference is the Mo content (higher for Type 316L) that could act as an inhibitor for pitting corrosion.⁴⁰  Another difference is a slightly higher Cr bulk content in Type 303, Table 1; however, this was not reflected in a higher Cr surface oxide content, Figure 9(a). It has previously also been speculated whether proteins can cover cathodic sites and therefore reduce any corrosion rate or metal release. This was discussed for iron metal at similar conditions.¹²  However, there were no such indications (such as differences in OCP, polarization resistance, or lower metal release) in this study that would support this mechanism for Type 303 at the tested conditions. Instead, the MnS-inclusions (at least for Mn release), possibly the absence of Mo as an inhibitor, and a different mechanism of protein adsorption, possibly related to the presence of inclusions, seem most probable to cause the observed differences in metal release among the two grades.

An important message from this study is that the extent of metal release in protein solutions, especially at protein concentrations of physiological relevance (high concentrations), may be strongly underestimated to an extent depending on the experimental setup and related to metal-induced protein aggregation and subsequent precipitation. This effect is easiest observed when comparing several time points during which the metal concentration in solution decreases with time. Because many metal release investigations of relevance for implant materials often are performed for only one time point, and/or the experimental setup has not been considered that important, it is recommended that protein aggregation and precipitation are considered in future metal release investigations. This is especially important because implant safety relies on accurate metal release estimations. It is furthermore evident from this study, and from findings by the authors reviewed previously,³  that the electrochemical activity does not necessarily reflect the levels of metal release for passive conditions of stainless steels in most physiological solutions. This is because of the fact that metal release investigations (at passive conditions) are governed by several non-electrochemical mechanisms, mainly surface complexation of complexing agents such as proteins and by precipitation from solution. In this study, Type 303 showed a tendency to metastable pitting corrosion in the presence of Fbn; however, this was only observed for one out of three replicate samples of the metal release and the OCP measurements. The OCP measurements did not reflect the relatively high extent of Mn release from Type 303 in a clear way. In the presence of BSA, the OCP of Type 316L tended to be lower as compared to Type 303, which was correlated with a higher metal release (of Fe, Cr, and Ni) in BSA-containing solutions for Type 316L as compared with Type 303.

A schematic illustration of the proposed metal release mechanisms of this study is given in Figure 11. Because of the lack of any visible electrochemical activity, such as metastable pitting,⁴¹  the relatively high levels of Mn release from Type 303 is proposed to be caused by chemical dissolution of the inclusions in the alloy and present at the surface. Protein adsorption was evident for both Type 316L and Type 303 in this study without any visible signs of corrosion activity (except in two cases for the Type 303 coupons), and a reduced OCP value in BSA-containing solutions for Type 316L compared with the reference solution. Based on the earlier findings with a correlation between albumin adsorption and metal release,⁷  and the known binding ability of Fe, Cr, and Ni to proteins,^25,42-44  it is proposed that there is a metal release mechanism that involves protein-adsorption-controlled metal-protein complexation followed by the detachment of these complexes into solution. This would explain observed increased amounts of released metals in solution after the displacement of BSA by Fbn. This mechanism can also influence the electrochemical activity as observed in this study (lower, but relatively constant OCP values), e.g., by reducing the thickness of the surface oxide and its simultaneous reformation. This mechanism is, however, only important for Type 316L and mostly prevailing for the release of Fe, Cr, and Ni, whereas shown in this study, the release of Mn is not affected. The determined amount of released metals in solution was strongly influenced by the extent of protein aggregation and the concomitant precipitation of these aggregates. The metal-protein complexes might have formed on the alloy surface and/or in solution. In the case of Type 303, corrosion was observed to take place for two cases (both for one coupon out of three replicates) between 6 h and 24 h of exposure in Fbn-containing solutions. This is most probably related to the formation of a patchy or relatively thick layer of Fbn formed on Type 303, which resulted in pitting corrosion and a relatively high amount of released metals in solution.

The main objective of this study was to investigate whether the Vroman effect influences the extent of metal release from stainless steel (Type 316L and Type 303) in mixed protein solutions. The following main conclusions were drawn:

• •

  Proteins were able to increase the release of metals from stainless steel, especially evident for the release of Fe, Cr, and Ni from Type 316L in BSA-containing solutions, and the release of Mn and Fe from Type 303 in Fbn- and/or BSA-containing solutions.

• •

  In mixed protein solutions, adsorbed BSA was replaced by the larger protein Fbn (the Vroman effect) for both stainless steel grades.

• •

  The Vroman effect resulted in an increased amount of released metals (Fe, Cr, Ni) in solution in the case of Type 316L and in an increased corrosion ability of Type 303 resulting from the formation of a thicker layer or possibly patches of adsorbed Fbn.

• •

  Significant metal-induced protein aggregation and precipitation was observed in solutions of relatively high physiological-relevant protein concentrations (40 g/L BSA). As a result, measured concentrations of released metals in solution underestimated the total amount of released metals at given conditions. These aspects need to be considered in future studies in protein-containing solutions.

KTH Royal Institute of Technology, the Swedish Research Council (VR, grant no. 2013-5621, and grant no. 2015-04177), Jernkontoret, Sweden, and the Erasmus+ program of the European Union are highly acknowledged for funding.