Introduction

Corrosion damage is a ubiquitous issue for metallic structures submerged in the sea, as the high-salt environment causes surface material erosion and crevice corrosion by dissolution of the metal ions during complex electrochemical and anaerobic corrosion processes1,2,3. The calculated rates of steel corrosion below sea level are between 0.08–0.14 mm yr−1 and 0.07–0.21 mm yr−1 in splash and tidal zones3, while the global cost of corrosion is estimated to be ca. US$2.5 trillion per year in 20134. The useful lifespan of metals becomes significantly reduced when exposed to saltwater, requiring frequent maintenance, which is especially costly for petroleum, offshore construction and sea transport industries. By implementing existing anti-corrosion protection methods, the cost could be reduced by 15–35%, or as much as US$875 billion annually4. Among these strategies, the addition of chemical compounds to the medium in contact with the surface, called corrosion inhibitors, is one of the most convenient methods to reduce the corrosion rate of steels. Most efficient commercially-used inhibitors are organic compounds, such as azole derivatives5,6, Schiff bases7,8,9, phenolic compounds3, amine derivatives10,11, thio-compounds11,12 and pyrimidine derivatives11,13. These inhibitors usually contain electronegative atoms, heterocyclic compounds containing polar functional groups and conjugated double bonds in their molecular structures14, through which they absorb on metal surfaces by forming an adhesive film that restricts the exposure of the metal to corrosive media. Although these synthetic compounds exhibit good corrosion inhibition performance, they bring concerns with regard to how they may affect living organisms15. Thus, there is a strong push to develop bio-based corrosion inhibitors that are chemically and mechanically robust but do not release non-natural organic chemicals into the environment.

Protein and its amino acids are compounds that contain amino and carboxyl groups gathering as a feature of their molecular structure. This unique structure proposes that they would be potential applicants as corrosion inhibitors16. In fact, amino acids have been systematically researched as corrosion inhibitors in corrosive media to control the inhibition of corrosion reactions of metals16,17,18. Of note, mussel adhesive protein (Mefp-1) derived from the marine mussel Mytilus edulis has been shown to exhibit good corrosion inhibition on carbon steel19,20,21,22.

Here, we report our findings on the anti-corrosion properties of a recombinant protein, rMrCP20, derived from the natural adhesive cement protein (MrCP20) of a common marine biofouler, the barnacle Megabalanus rosa (M. rosa). Barnacles are sessile crustaceans that strongly adhere to solid immersed substrates. Blackwood et al.23 and Eashwar et al.24 have reported that barnacle attachment to stainless steel substrates in seawater leads to more substantial corrosion events under dead barnacles, while crevice corrosion of steel substrates underneath live barnacles was less than 4%. Barnacles have been demonstrated to adhere strongly to metal substrates25 and to minimize corrosion events in such conditions of strong attachment. Hence, drawing inspiration from the structural adaptation of M. rosa and these reported observations, we posited that the adhesive protein MrCP20 may help prevent steel corrosion provided it binds strongly to the substrate. Our team has previously unveiled the secondary and tertiary structures of rMrCP2026 and identified that its underwater adhesive properties stem from its high content of cysteines (Cys) and charged amino acid residues.

Herein, we show that rMrCP20 protein exhibits concentration-dependent anti-corrosion properties, which as we demonstrate is linked to its adhesive characteristics in conjunction with its ability to capture Fe ions released from the steel surface. Corrosion inhibition effect of rMrCP20 was demonstrated through time-resolved observations of corrosion on the alloy – mild steel (AH36) coupon, whereas the iron weight loss from the corrosion events was measured via inductively coupled plasma optical emission spectroscopy (ICP-OES). A series of electrochemical studies provided insights into the increased impedance, corrosion inhibition process and redox reactions of the protein, while small angle X-ray scattering (SAXS) confirmed the protein’s oligomeric conformation at the optimal corrosion inhibition concentration. Quartz crystal microbalance with dissipation monitoring (QCM-D) and nanoindentation studies indicate that rMrCP20 strongly adhered to steel substrates, with the presence of Fe ions resulting in compaction of the nanoscale protein film. Fourier Transformed Infrared Spectroscopy (FTIR) in conjunction with molecular dynamics (MD) simulations identified the main molecular interactions between rMrCP20 and Fe ions, in particular indicating that Fe3+ ions form ionic bridges and coordination bonds with the negatively charged side chain and histidine (His) residues of rMrCP20, respectively. Finally, fluorescence microscopy of tagged rMrCP20 in conjunction with Auger electron spectroscopy (AES) and X-ray photoelectron spectroscopy (XPS) surface mapping indicated preferential adsorption of the protein at the grain boundary (GB) of the steel surface, which we suggest helps mitigating GBs-initiated corrosion.

Results and discussion

Anticorrosion studies

rMrCP20 was first expressed and purified using the same protocol as previously described26. To examine the corrosion-inhibition properties of rMrCP20 protein, epoxy-embedded polished AH36 steel coupons were immersed in a pH 8.3 buffer containing 150 mM NaCl and 20 mM Tris(hydroxymethyl)aminomethane (Tris), in the absence or presence of rMrCP20 at different concentrations (see Materials and Methods). Figure 1a, b illustrate representative pictures of AH36 coupons at various time points and their respective immersion solutions after 24 h, at a range of rMrCP20 concentrations from 0 to 10 mg mL−1.

Fig. 1: Time-resolved study of corrosion on AH36 coupon.
figure 1

a Representative pictures of AH36 coupons showing the concentration-dependent anti-corrosion effect of rMrCP20 at immersion time intervals of 0, 1, 3, 7 and 24 h, and after cleaning the coupons with 1% HCl. b Respective buffer solutions after 24 h of coupon immersion. c Field emission scanning electron microscope (FESEM) images and overlay of energy dispersive X-ray spectroscopy (EDX) map of pitting corrosions before and after removal of corrosion products on surface of coupons, with a pitting of ca. 300 μm diameter outlined. d ImageJ analysis to determine percentage of corroded area of AH36 coupons. Error bars represent the standard error (SE) from triplicate tests. e Weight loss of AH36 over a 24 h incubation period and corresponding inhibition efficiency at increasing concentrations of rMrCP20. Error bars represent the SE from triplicate tests.

Full size image

In the absence of rMrCP20, the coupons underwent corrosion, with surfaces covered with rust and other corrosion products, and the respective buffer solution turning into intense brown color. Pitting corrosion was observed at different sites of the coupon substrate within the first hour and propagated as the immersion time increased, with the confined environment beneath the initial corrosion products appearing to accelerate pitting. Figure 1c illustrates a deep pitting site imaged by scanning electron microscopy (SEM). At a low concentration of ca. 0.1 mg mL−1, rMrCP20 was previously found to accelerate the corrosion of steel coupon surfaces27. However, at increasing concentration of rMrCP20, corrosion was reduced considerably and little to no corrosion was observed on coupons incubated in 5 mg mL−1 and 10 mg mL−1 of proteins (Fig. 1a). These pits did not propagate further as the immersion time increased and their respective post-incubation buffer solutions had a light brown color (Fig. 1b). Further analysis was performed using ImageJ software to estimate the corroded area and the results are shown in Fig. 1d. In the absence of rMrCP20, on average 91% of AH36 coupon surfaces was corroded after 24 h. However, at increasing protein concentration surface corrosion was reduced drastically, down to 2% at 10 mg mL−1, thereby demonstrating a concentration-dependent corrosion inhibition effect of rMrCP20 on AH36 steel.

Weight loss measurements are routinely employed to evaluate the effectiveness of metal corrosion inhibitors28. Fe content from the incubation solutions and respective coupon surfaces were measured by ICP-OES to reflect the total weight loss for each sample at concentrations of rMrCP20 from 0.1 mg mL−1 to 10 mg mL−1, and to calculate the inhibition efficiency using Eqs. (1) and (2)29,30.

$${CR}=\frac{8.76\times {10}^{4}\times \Delta m}{\rho {AT}}$$
(1)
$$\eta \, \left( \% \right)=\theta \times 100=\left(1-\frac{{{CR}}_{1}}{{{CR}}_{0}}\right)\times 100$$
(2)

Where CR is the corrosion rate (mm yr−1), Δm is the total weight loss (g), ρ is density of the metal samples (7.86 g cm−3), A is the surface area (cm2), and T is the exposure time (h). η is the corrosion inhibition efficiency (%), θ is the degree of surface coverage, while CR0 and CR1 are the weight losses of the coupons in the absence and presence of rMrCP20, respectively. The corrosion rate calculated via Eq. (1) decreased significantly with increasing the concentration of rMrCP20 (Supplementary Table 1), with the lowest corrosion rate of 0.12 mm yr−1 and a high corrosion-inhibition efficiency η of 88.48% (Fig. 1e). The observed enhanced corrosion-inhibition efficiency is attributed to the strong adhesive properties of rMrCP20, whereby protein adsorption to the coupon surfaces –and hence the degree of coupon surface coverage increased at higher protein concentrations.

Electrochemical studies

To further investigate the impedance brought about by different concentrations of rMrCP20, electrochemical impedance spectroscopy (EIS) measurements were performed for AH36 samples after 24 h incubation in buffer solutions containing various concentrations of rMrCP20 protein. A protein concentration of 5 mg mL−1 was selected as the optimal concentration for further characterizations as it exhibited good corrosion inhibition performance, whereas increasing the amount of protein to 10 mg mL−1 did not significantly affect the results.

Figure 2a–c show the Nyquist plots (Fig. 2a), Bode modulus (Fig. 2b) and phase angle representations (Fig. 2c) from EIS measurements. The Nyquist plots show single semi-circle loops from high to mid-frequency range, indicating that the corrosion of AH36 was controlled by a charge transfer process (Fig. 2a)31,32,33. Tailing was observed at the lower frequency range of each Nyquist plot, and these imperfections can be attributed to the roughness and heterogeneity of the AH36 working electrode32,34, distribution of active center34,35, adsorption of inhibitor molecules36 and accumulation of corrosion products on the working electrode35,36. The diameter of the Nyquist plot is directly correlated to the impedance of the system, hence providing a suitable comparison of the resistance of the Fe substrate exposed to different concentrations of rMrCP20. The diameters increased at higher concentration of rMrCP20, suggesting that rMrCP20 adsorbed onto AH36 surfaces increases the charge transfer resistance and hence imparts corrosion resistance properties.

Fig. 2: Electrochemical impedance and potentiodynamic polarization (PDP) spectra after 24 h immersion.
figure 2

a Nyquist plots with inset displaying the plots at higher frequencies between Z’ 0 to 1500 Ω and equivalent circuit diagram as inset. b Bode modulus plots. c Phase angle plots. d Summary table of electrochemical impedance parameters for AH36 in buffer solution, at various concentrations of rMrCP20 after 24 h immersion. Data are presented as average ± SE. e PDP curves in the potential range from −0.25 V versus OCP to −0.4 V versus Ag/AgCl with the sweep rate of 10 mV min−1 in aerated condition at 25 °C. The inset displays the plots in the range from −0.72 V to −0.60 V. f PDP parameters after 24 h immersion. Data are presented as average ± SE.

Full size image

Bode modulus plots (Fig. 2b) provide a more comprehensive representation of the impedance across the range of frequencies measured, while phase angle plots (Fig. 2c) indicate the phase shift across the range of frequencies and the plausible components in the respective equivalent electrical circuits (EEC). At higher protein concentrations, the Bode modulus plots show an increase in impedance values at low frequencies, where corrosion events are usually observed, while the phase angle plots show an increase in the maximum phase angle value. The higher impedance is attributed to the formation of more homogenous rMrCP20 protein films at higher protein concentrations. Furthermore, the diameter of the Nyquist semicircles also increased with exposure time at higher rMrCP20 concentrations (Supplementary Fig. 1), which is caused by the formation of a mixed surface layer of corrosion products (iron oxide) and adsorbed proteins36. The evolution of the surface film most likely involved complexation between rMrCP20 and Fe ions (Fe2+ and Fe3+), which blocked the charge transfer path. This observation also highlights the stability of the adsorbed rMrCP20 molecules on the steel surface.

To further interpret the EIS results, an equivalent electrical circuit (EEC) model was chosen to extract the relevant electrochemical parameters (Fig. 2a inset). In the circuit, Rs represents the solution resistance, Rct the charge transfer resistance, CPEdl and CPEf the constant phase elements of the double layer and protein film respectively, and Rf the resistance of the adsorbed rMrCP20 film. The double layer capacitance (Cdl) values were calculated from the expression31:

$${C}_{{dl}}=\frac{{Y}_{0}{\omega }^{n-1}}{\sin \left[n\left(\frac{\pi }{2}\right)\right]}$$
(3)

where Y0 is the CPE constant, n is the CPE exponent, ω is the angular frequency \((\omega =2\pi {f}_{\!\!\max })\) in rad s−1, and \({f}_{\!\!\max }\) is the frequency at which the imaginary component of the impedance is highest. The corrosion inhibition efficiency (% η) was calculated using Eq. (4)37:

$${\eta }_{{EIS}}\left( \% \right)=\frac{{R}_{{ct}}^{0}-{R}_{{ct}}^{{\prime} }}{{R}_{{ct}}^{0}}\times 100$$
(4)

where \({R}_{{ct}}^{0}\) and \({R}_{{ct}}^{{\prime} }\) are the charge transfer resistances in the presence and absence of rMrCP20, respectively. The values of the electrochemical parameters derived from the Nyquist plots are listed in Fig. 2d and demonstrate that adding rMrCP20 to the corrodent led to a reduction in Cdl and an increase in Rct and Rf, which became more apparent as the rMrCP20 concentration increased. As the concentration of rMrCP20 increased from 1 to 5 mg mL−1, the Rct value increased from 926 Ω cm2 to 4721 Ω cm2, while the corresponding η value increased from 19.8% to 84.3%. The value of phase shift (n) for CPEdl also increased from 0.79 without rMrCP20 to 0.85 in the presence of 5 mg mL−1 rMrCP20, suggesting a decrease in the heterogeneity of the coupon surface arising from adsorption of rMrCP20. Cdl for the control sample was 757 μF cm−2, which was significantly higher than the corresponding value for the system incubated in rMrCP20, indicating growth in the electrical double layer thickness. These results suggest that the adsorption of rMrCP20 at the metal/electrolyte interface impedes charge transfer due to greater resistance. Hence, at higher concentration of rMrCP20, the AH36 steel surface is better protected against corrosion.

To obtain further qualitative understanding of the corrosion reactions, potentiodynamic polarization measurements (PDP) measurements were carried out on AH36 samples after 24 h incubation in the respective solutions, and the polarization curves are shown in Fig. 2e. The polarization curves present two opposing reactions, the anodic branch corresponding to Fe dissolution and the cathodic branch due to hydrogen evolution. The obtained polarization parameters, namely corrosion potential (Ecorr), corrosion current density (icorr), anodic and cathodic Tafel slopes (βa and βc) measured by the Tafel extrapolation method at the segment 20–50 mV from Ecorr are presented in Fig. 2f. The corrosion rate (CR) and corrosion inhibition efficiency \(({\eta }_{{PDP}})\) of rMrCP20 was calculated using Eqs. (5) and (6)37:

$${\eta }_{{PDP}}\, \left( \% \right)=\left(1-\frac{{i}_{{corr}}^{0}}{{i}_{{corr}}^{{\prime} }}\right)\times 100$$
(5)
$${CR}=3.17{E}^{-9}\times \frac{M}{{nF}\rho A}{i}_{{corr}}$$
(6)

where \({i}_{{corr}}^{0}\) and \({i}_{{corr}}^{{\prime} }\) are the corrosion current densities in the presence and absence of rMrCP20, respectively, 3.17 × 10−9 is conversion factor, the ratio M/n is the equivalent weight, F (96485 C mol−1) is the Faraday constant, \(\rho \, (7.86 g{{cm}}^{-3})\) is the density of the metal samples, and A is the area of the sample \(({{cm}}^{2})\). As shown in Fig. 2e, both the anodic and cathodic icorr decreased in the presence of rMrCP20 and Ecorr slightly shifted towards the cathodic direction, indicating that rMrCP20 has a stronger effect on the cathodic reaction than on the anodic reaction. This result was complemented with cyclic voltammetry (CV) measurements (Supplementary Fig. 2). Since the peaks of the redox reactions measured by CV were not shifted at different concentrations of rMrCP20, the reaction mechanism appeared to remain consistent. In addition, the peak currents observed in CV scans for both the oxidation and reduction reactions progressively decreased with the addition of rMrCP20, further confirming the lowered occurrence of redox reactions, and hence the corrosion inhibition properties of rMrCP20. Furthermore, the peak currents from CV measurements remained constant in the presence of rMrCP20 with repetitive scans, whereas in the absence of the protein the peak currents were much greater for the second scan.

As shown in Fig. 2f, icorr decreased from 29.5 μA cm−2 without rMrCP20 down to 4.0 μA cm−2 with 5 mg mL−1 of rMrCP20. The lower values of icorr in the presence of rMrCP20 can be attributed to protein adsorption onto the sample’s surface and hence to the formation of a protective layer, which hinders the electron transfer process at the interface and reduces the rate of corrosion reactions. Figure 2f also illustrates the effect of rMrCP20 on the Tafel slopes values βc, which were independent of the rMrCP20 concentration, indicating that the presence of rMrCP20 can suppress the corrosion process by blocking reaction sites without affecting the kinetics of the cathodic reactions38. Meanwhile, a gradual decline in βa was observed with increasing rMrCP20 concentration, from 244 mV dec−1 in buffer to only 140 mV dec−1 in the presence of 5 mg mL−1 rMrCP20, indicating a change in the Fe dissolution mechanism. The change in Fe dissolution is attributed to the formation of metal-ion protein complexes as the protein interacts with the steel surface, which was further investigated by attenuated total reflection Fourier transform infrared spectroscopy (ATR-FTIR), QCM-D and MD simulations (discussed below). Accordingly, the corrosion-inhibition efficiency improved at increased rMrCP20 concentration, from 15.9% (1 mg mL−1) to 86.5% (5 mg mL−1), respectively. In summary, the corrosion inhibition values η obtained from all three methods (weight loss, EIS and PDP measurements) are consistent (Supplementary Fig. 3) and all indicate enhanced corrosion-inhibition activity as the concentration of rMrCP20 increased.

Adsorption of rMrCP20 and metal ion – protein interaction studies

QCM-D measurements were carried out on Fe sensors to examine the adsorption properties of rMrCP20, followed by subsequent addition of FeCl3 analyte as a source of free Fe3+ to investigate the Fe3+/adsorbed protein layer interactions. The changes in resonance frequency (Δf) and dissipation (ΔD) of the sensor surface were measured simultaneously. Figure 3a shows the results obtained for the 5th overtone as a function of time for the following steps: flowing of rMrCP20, rinsing, flowing of FeCl3 solution and second rinsing (see Materials and Methods). To allow direct comparison of the Δf and ΔD shifts measured at different harmonics, the net shifts of the 5th and 11th overtones are displayed in Fig. 3b, c. Upon injection of rMrCP20, an exponential increase in Δf intensity was measured (Fig. 3a), followed by a plateau, indicative of a rapid initial adsorption of rMrCP20 onto the Fe sensor surface followed by saturation at the surface. Δf recorded was ca. −40.0 Hz and −39.2 Hz upon adsorption saturation for the 5th and 11th overtones, respectively. The significant decrease in Δf was accompanied by a proportional increase in ΔD to 1.62 × 10−6 and 1.55 × 10−6 for 5th and 11th overtones, respectively. Fast initial adsorption rates suggests that rMrCP20 has a high affinity for the Fe sensor surface, which can be driven by complementary electrostatic and hydrophobic adsorption39,40,41. The changes in Δf and ΔD due to rMrCP20 binding to the Fe sensor surface were generally similar amongst different overtones, indicating that the protein formed a compact and rigid layer39. While the inversely proportionate changes in ΔD and Δf denote the strong binding kinetics of rMrCP20 on Fe substrate, ΔD decreased gradually while Δf plateaued, suggesting that the adsorbed protein layer packed more tightly over time. The change of Δf was similar at different rMrCP20 concentrations, indicating that rMrCP20 formed a mono-adlayer (Supplementary Fig. 4c). Upon rinsing, a slight decrease in Δf intensity was detected due to the removal of excess unbound proteins, but most proteins remained strongly bound to the Fe surface.

Fig. 3: QCM-D measurements tracking real-time adsorption of rMrCP20.
figure 3

a QCM-D frequency (Δf) shifts (left axis) and dissipation (ΔD) shifts (right axis) for 5th overtone as a function of time. The inset displays an illustration of protein structural changes upon adsorption and interaction with Fe3+. Changes in frequency (b) and dissipation (c) obtained from the 5th and 11th overtones at different steps in QCM-D measurements. Error bars in the figure represent the SE from triplicate tests.

Full size image

The introduction of FeCl3 induced large shifts in the QCM-D signals, splitting the response across the different harmonics (Supplementary Fig. 4a, b). The increase in Δf intensity indicates significant binding interaction between the analyte and adsorbed protein layer, which can occur via formation of Fe3+ metal-protein complexes and electrostatic interactions. The proportional increase in ΔD implies a more loose packing of the adlayer, which allows more ingression of water molecules that remained trapped within the adlayer40. Upon the final rinsing step to remove any loosely bound material, only a slight decrease in Δf intensity was observed, suggesting that Fe3+ ions were strongly bound to the pre-adsorbed rMrCP20 layer, whereas the slight decrease in ΔD indicates stiffening of the surface layer.

The mode of interaction between rMrCP20 and the metal surface can be deduced from the adsorption isotherms, which were obtained based on the results from weight loss measurements (Supplementary Table 1) to determine the adsorption characteristics of rMrCP20 on AH36 steel in buffer solution. The degree of surface coverage (θ) as a function of concentration (C) of the protein was fitted into various adsorption isotherm models, including Langmuir, Temkin, Freundlich, Flory–Huggins, Frumkin and El-Awady to identify the best fit based on the obtained R2 values. The best fit was obtained for the Freundlich model (R2 = 0.96, Supplementary Table 2), which is generally applicable for multilayer adsorption events on heterogeneous surfaces, with the assumption of a large number of different types of binding sites acting simultaneously42,43,44. The linear form of the Freundlich model can be written as42:

$$\log \theta =\frac{1}{n}\log {C}_{{inh}}+\log {K}_{{ads}}$$
(7)

where Kads represents the Freundlich adsorption capacity (L mg−1) and n describes the heterogeneity of the system related to the adsorption intensity. A larger n value connotes a more heterogeneous system, and n > 1 suggests a favorable adsorption process45. Freundlich constants Kads and n values were \(1.33\times {10}^{-2}\) L mg−1 and 2.18, respectively. The value of Kads was then used to calculate the standard free energy of adsorption \(({\Delta G}_{{ads}}^{0})\)31:

$${\Delta G}_{{ads}}^{0}=-{RT}{{{{\mathrm{ln}}}}}\left(1\times {10}^{6}{K}_{{ads}}\right)$$
(8)

where R is the universal gas constant, T is the absolute temperature, and 106 is the concentration of water molecules expressed in ppm. The calculated \({\Delta G}_{{ads}}^{0}\) value of −23.52 kJ mol−1 indicates spontaneous interaction between the protein and the surface of AH36 coupons, and is within the range for mixed adsorption involving both physisorption driven by electrostatic interactions as well as chemisorption caused by charge sharing or charge transfer from the protein molecules to the metal surface33,46,47.

Since the QCM-D measurements have large dissipation values (ΔD > 0), a viscoelastic model should be used to obtain more accurate adsorbed masses in QCM-D. However, analysis of soft and hydrated protein adlayers using the Voigt viscoelastic model has been shown to significantly deviate from values measured by surface plasmon resonance, a direct analysis to gain thickness from adlayer mass48. Therefore, to determine the thickness of rMrCP20 and its metal complex layer we carried out nanoindentation measurements (Fig. 4a) on rMrCP20 coated over a QCM-D sensor. Samples were equilibrated in buffer, and a cube-corner tip was used to indent the samples via a displacement-controlled indentation of 200 nm in depth. Figure 4b shows the surface roughness profile of the samples and Fig. 4c the histogram of the average layer thickness, indicating significant difference between samples. The layer thickness was inferred from the slope change during the loading step of indentation cycles, as illustrated in the load vs. displacement curves shown in Fig. 4d. The presence of a layer ca. 74 nm thick was consistently observed on the substrate when hydrated and was attributed to a thin oxide layer. Upon adsorption of rMrCP20, an indentation depth of ca. 82 nm was measured before the tip contacted the substrate, indicating that the protein formed a layer of ca. 8 nm (subtracting the thickness of the oxide layer). Subsequent indentation of the protein layer after interaction with FeCl3 showed a layer of thickness ca. 79 nm, indicating that the protein layer became slightly thinner (ca. 5 nm).

Fig. 4: Nanoindentation of rMrCP20 layer formed on substrate.
figure 4

a Illustration of indentation carried out with a cube-corner tip. b Scanning probe microscopy images of sample surfaces. c Bar plot of average values of layer thickness obtained by nanoindentation on substrate, substrate with adsorbed rMrCP20 layer, and substrate with rMrCP20 layer after interaction with FeCl3. The average for each sample group is shown and SE indicated by error bars. All statistical significances difference between two groups are represented by P-value asterisk indicators with P < 0.0001 (****) and P < 0.001 (***). d Load-displacement (loading and unloading) curves with insets showing magnified views of the loading portion of indentation curves, where the change in slope was used to detect the substrate surface and infer the thickness of the protein film.

Full size image

To assess the effect of Fe ions on rMrCP20 secondary structure and identify the type of binding interactions between Fe ions and the protein’s side chains, ATR-FTIR spectroscopy measurements were conducted (Fig. 5). With rMrCP20 kept at 5 mg mL−1 concentration, different concentrations of FeCl3 were added as indicated in Fig. 5a. The most significant changes in the spectra were observed at the protein:FeCl3 molar concentration ratio of 1:60. At low FeCl3 concentration, minimal changes were observed compared to rMrCP20-only spectra. The amide I band was further deconvoluted to obtain a semi-quantitative estimate of the secondary structural content, as shown in Fig. 5c. The initial β-sheet content of rMrCP20 protein was ~53% at 5 mg mL−1. At the 1:60 protein:FeCl3 concentration ratio, the amide I maximum of rMrCP20 spectra shifted from 1641 cm−1 to 1631 cm−1 (Fig. 5b), indicating a transition towards anti-parallel β-sheet structures. Similarly, the appearance of 1553 cm−1 in amide II confirm that FeCl3 at high concentrations induced changes in the protein’s conformation. The amide III peaks at 1262 cm−1 and 1296 cm−1 were likewise more noticeable at high FeCl3 content.

Fig. 5: FTIR-ATR investigation of the effect of FeCl3 on rMrCP20 protein structure.
figure 5

a ATR-FTIR spectra of rMrCP20 and FeCl3 at different molar concentration ratios, illustrating the spectral changes and bond assignments. b Shift in amide I band, indicating a transition of the protein secondary structure transiting towards anti-parallel β-sheets at higher FeCl3 concentrations. c Semi-quantitative secondary structure based on amide I band deconvolution.

Full size image

New peaks were noticeable at higher concentrations of FeCl3, indicating that new bonds formed between the protein and Fe3+. The appearance of 800 cm−1 and 910 cm−1 bands were assigned to the stretching of aromatic –CH groups and C=C, respectively, attributed to their interaction with Fe3+. The distinct splitting of the band centered at 1043 cm−1 to 1038 cm−1 and 1053 cm−1 can be attributed to chelation of Fe3+ ions by the imidazole side chain of His49. The band at 1403 cm−1, assigned as S = O bond appeared in the presence of Fe3+, and was attributed to the oxidation of thiol functional groups of cysteine residues. The appearance of 1138 cm−1, 1262 cm−1 and 1296 cm−1 bands were assigned to new interactions involving the protein’s side chain C OH, C–O/–CH3 and amide bond C = O functional groups, respectively. The intensity of the COO− stretching band of acidic residues at 1398 cm−1 observed for rMrCP20 decreased, concurrently with the appearance of a peak at 1437 cm−1 upon introduction of high FeCl3 concentrations, is assigned to the formation of ionic bridges or coordination bonds between the carboxyl side chains and Fe3+ 49,50.

Small angle X-ray scattering and MD simulation studies of rMrCP20 and Fe ions

Small angle X-ray scattering (SAXS) was used to determine the radius of gyration (Rg) of rMrCP20 protein in our first steps towards developing a mechanistic model for the protein’s adsorption to substrate surfaces (Supplementary Fig. 5a–c). The oligomeric state of rMrCP20 was obtained from SAXS experiments. SAXS profile of rMrCP20 were collected at 1.3, 4.1, and 6 mg mL−1, respectively (Supplementary Fig. 5a). The scattering pattern of rMrCP20 at protein concentration of 1.3 mg mL−1 had a low signal to noise ratio and thus, no data pre-processing was performed with this dataset. Rg values using Guinier approximation of rMrCP20 at protein concentration of 4.1 and 6 mg mL−1 (Supplementary Table 3) showed a slight increase in molecular size, suggesting that the shift is most likely due to interparticle interaction between rMrCP20 molecules. To exclude the concentration-dependent effect, the results at protein concentration of 4.1 mg mL−1 were used for further data analysis. The overall structural parameters are recorded in Supplementary Table 3. Based on the Porod-volume (Vp) and DAMMIF-excluded volume (Vex), determined from the scattering pattern at protein concentration of 4.1 mg mL−1, the molecular weight (MW) of rMrCP20 was calculated to be 42 ± 4.2 and 45 ± 4.5 kDa, respectively (Supplementary Table 3). Knowing the calculated monomeric MW based on the protein sequences is 21 kDa, the data indicate that rMrCP20 adopts a dimeric state in solution at 4.1 mg mL−1.

To investigate the distribution of Fe ions interacting with the protein and to calculate the number of Fe ions bound to each residue, MD simulations were performed for rMrCP20 protein dimer in the presence of a high Fe ion concentration. Two types of initial structures were considered. One is the NMR structure of rMrCP20 (PDB ID: 6LEK), another one is the structural model built by AlphaFold 251. SAXS data showed rMrCP20 adopts a dimeric state in solution. Thus, the dimeric forms of both models were compared with the SAXS profile, and one of these dimers from AlphaFold 2 models showed a better fit, giving a χ2 value of 1.76 (Supplementary Fig. 5d, e). This dimeric model was selected for subsequent MD simulation.

From the MD, the largest cluster structure with 54 close-contacting Fe ions around the protein dimer and the inset highlighting the interactions of Fe ions at rMrCP20 interface are shown in Fig. 6a, b. The average number of Fe ions bound to the residues is shown in Fig. 6c. The high root-mean-square-deviation (RMSD) obtained in MD simulation may reflect the intrinsic flexibility of the dimers in the presence of Fe ions (Supplementary Fig. 6), which predominantly interact with the negatively charged residues aspartic acid (Asp) and glutamic acid (Glu), binding up to more than 4 Fe ions, as shown in Fig. 6c. These Fe-protein interactions, especially those on the chain interfaces, may help to enhance protein-protein interactions, as experimentally observed from the spontaneous aggregation of rMrCP20 upon addition of FeCl3 (see optical micrograph in Supplementary Fig. 7). The results corroborate the FTIR data (Fig. 5a), notably the significant shift of the carboxyl band at high FeCl3 concentration attributed to the interaction of Asp and Glu side chains with Fe ions, occurring in conjunction with changes in the protein’s secondary structural conformation.

Fig. 6: MD simulation of rMrCP20 dimers in the presence of Fe ions.
figure 6

a Representative largest cluster structure of the protein dimer with each chain represented by a different color, illustrating the distribution of close-contacting Fe ions (orange spheres). b. Zoom-in view of Fe ions at the rMrCP20 interface. c Plot of average number of Fe ions bound to the residues, with color-coded protein sequence indicated below the plot. Error bars in the figure represent the SE.

Full size image

Proposed multimodal anti-corrosion mechanism

To further understand the adsorption behavior of rMrCP20, the AH36 substrate was incubated with fluorescein isothiocyanate (FITC)-labeled rMrCP20 and observed by fluorescence microscopy (Fig. 7a). Stronger fluorescence signals were observed at the GBs, indicating that more dye-protein conjugate accumulated at the GBs. Further investigations by AES and XPS were performed to confirm these observations. AH36 samples were incubated with rMrCP20 in 150 mM NaCl, pH 8.3 for 15 min, which were then rinsed with 150 mM NaCl solution and dried in air before analysis. Distinct AH36 grains and GBs were observed under SEM and the corresponding AES elemental mapping profiles (C, N, O) with topographical corrections are shown in Fig. 7b. Surface topography corrections were carried out using the algorithm (P–B)/B52, where B is the background signal and P is the Auger peak (Supplementary Fig. 8). We note that the signal intensities of the elements of interest were more pronounced at the GBs, indicating a higher concentration of rMrCP20 at these sites. GBs, which are inherently defective regions associated with grain misorientation, are usually less dense on atomic scale and tend to have higher free energies, thus potentially acting as preferential adsorption sites. We also conducted XPS mapping and verified the adsorption of the protein on a steel coupon (Fig. 7c), which can be inferred from the emergence of new peaks commonly associated with proteins. Further deconvolution of the high-resolution C 1s, N 1s, O 1s, and S 2p core-level spectra (Supplementary Fig. 9) revealed expected functional groups present for a protein, such as C = O amide bonds or C–S bonds associated with Cys residues, which are in close agreement with those obtained from the ATR-FTIR spectra (Fig. 5a). Overall, these data suggest preferential adsorption of rMrCP20 at the GBs, hence providing enhanced protection against aggressive corrosive ions.

Fig. 7: Optical and electron microscope measurements of rMrCP20 layer formed on AH36 substrate.
figure 7

a Fluorescence image of FITC-labeled rMrCP20 layer on substrate. b SEM image and AES elemental mapping profiles (C, N, O) of adsorbed rMrCP20 layer. c The XPS survey spectra of rMrCP20 adsorbed layer and pristine metal. All scale bars are 50 μm.

Full size image

Based on these findings, we propose a multimodal anti-corrosion activity by rMrCP20. rMrCP20 protein is highly negatively charged, comprising clusters of charged amino acids with a large number of 32 Cys residues that contribute to binding interactions with substrates, thereby allowing barnacles to adhere to charged surfaces such as metals26,53. The thiol side chain on Cys residues is easily deprotonated at seawater pH, leading to a negatively charged sulfhydryl group54, which then binds to charged surface ions via electrostatic interactions55 to form more stable thiolate-metal ion complexes56. The side chains of Asp and Glu residues are also deprotonated and negatively charged at seawater pH, hence favoring electrostatic interactions with cations. A multimodal corrosion inhibition mechanism is thereby illustrated in Fig. 8 to explain our findings. Corrosion of metal surfaces often occurs on or adjacent to GBs due to the different electrode potentials of the grain and boundary, making them susceptible to redox reactions that further lead to intergranular corrosion in the presence of electrolytes57, while other forms of corrosions such as uniform, crevice and pitting corrosions also occur concurrently at the substrate-electrolyte interface58,59. rMrCP20 protein readily adsorbs onto metal substrate, preferentially filling GBs and coating the substrate surface thereby preventing direct contact between the substrate and seawater. The protein adheres strongly to the metal substrate through iron-protein interactions, hence blocking charge transfer path and functioning as a protective coating to inhibit different forms of corrosion.

Fig. 8: Proposed multimodal corrosion inhibition action of rMrCP20 protein on steel substrates.
figure 8

The corrosion inhibition action of rMrCP20 is attributed to the adhesive property of the protein and its ability to spontaneously bind with free Fe ions.

Full size image

Conclusion

rMrCP20 protein exhibits a concentration-dependent anti-corrosion property to effectively inhibit corrosion of AH36 steel substrates in solution at rMrCP20 concentrations above 5 mg mL−1. The natural adhesive protein was shown to adsorb rapidly onto substrate surfaces, forming a relatively homogeneous layer that increases the impedance of the coating and reduces the propensity for spontaneous corrosion. The protein layer readily incorporates Fe ions, predominantly through electrostatic interactions with negatively charged residues at seawater pH conditions, concurrent with conformational changes in the protein’s secondary structures, resulting in a thin proteinaceous film that enhances impermeability against seawater, thereby shielding the underlying substrate against aggressive corrosive ions. Our findings indicate that the barnacle adhesive protein rMrCP20 has promising potential for further scaling-up development into an effective and green anti-corrosion additive for steels in the marine environment. Alternatively, the study provides molecular-level guidelines to develop artificial protein-based corrosion inhibitors inspired by the primary and secondary structures of rMrCP20.

Materials and methods

rMrCP20 expression and purification

The detailed expression and purification protocol of rMrCP20 was described in Mohanram et al.26. Briefly, an ampicillin - resistant pET-22b(+) vector with cloned cDNA of CP20 (Uniprot Q9GRC4) and C terminal his-tag was purchased from Bio Basic (Singapore). The plasmid was then transformed into chemically competent Escherichia coli BL21 (DE3) cells. A single colony transformants of rMrCP20 were grown in Luria-Bertani medium with 100 μg mL−1 ampicillin at 37 °C and expressed using 1 mM IPTG (Isopropyl thio-β-D-galactoside) at 20 °C. The harvested cells were then lysed via probe sonication (40% amplitude, 1 s pulse + 1 s pause, 20 min) and the crude lysate after high-speed centrifugation was purified by affinity chromatography (Ni-NTA affinity column, GE Healthcare Life Sciences, HisTrap, 5 mL). The eluted protein was further purified using size-exclusion chromatography (SEC) (Superdex 75, 16/600) using isocratic elution in 20 mM Tris, 150 mM NaCl, at pH 8.3. The fractions containing rMrCP20 were concentrated to the required concentration using Vivaspin column (M.W. cutoff 5 kDa). The protein concentration measurements were done using absorbance spectrometry at 280 nm (Nanodrop 2000c).

Preparation of steel coupons

AH36 steel coupons cut to 10 × 10 × 2 mm were purchased from Ebenezer Excel Engineering, Singapore. Chemical composition (in wt %) of the AH36 material was Nb 0.03, Cr 0.03, Mn 0.2 and the balance Fe. The coupons were embedded in cold mount epoxy (EpoFix, Struers), while for electrochemical studies, coupons were welded with an electrical wire before embedding in epoxy. Prior to all experiments, grinding was performed manually with a series of grit paper (# 800 to # 2000) to expose a surface area of 1 cm2 and then polished to a mirror-like finish using a diamond suspension (Ø = 3 µm, Struers) to alleviate the effects of surface roughness on protein absorption60. The samples were then washed with Milli-Q water (18.2 MΩ cm, MilliporeTM), cleaned with ethanol in an ultrasonic bath, washed again with Milli-Q water and finally dried with nitrogen gas36.

Time-resolved corrosion study of on AH36 coupon

AH36 steel samples were immersed in a buffer solution with different rMrCP20 protein concentrations (0.1, 0.5, 1, 3, 5 and 10 mg mL−1) and incubated at ambient temperature, on an orbital shaker set at 150 rpm for 24 h. AH36 steel samples were removed at stipulated time points for visual inspection, and subsequently removed after 24 h for collecting corrosion-products on the coupons and from buffer solution. Experiments were triplicated, using freshly prepared AH36 steel coupons.

Weight loss measurements

To quantify the total Fe mass loss after incubation, both Fe content from corrosion products bound to sample surfaces and Fe content dissolved in incubation solutions were collected and measured using inductively coupled plasma optical emission spectrophotometer (ICP-OES) (PerkinElmer® Optima™ 8300). Corrosion products bound to the sample surfaces were removed with 10 min of ultrasonication in 5 mL of 1% HCl solution containing 100 mM hexamethylenetetramine (Sigma-Aldrich) for preventing additional Fe dissolution from the coupons27. 1 mL of the respective Fe solutions were dissolved separately in 9 mL of 10% HNO3, then further diluted by 10x in Milli-Q water and filtered through 0.45 µm membrane prior to ICP-OES measurements. Calibration for Fe concentration standard curve was first performed using 5, 10, 25, 50 and 100 ppm standard solutions, and the wavelength for metal detection was Fe 238.86 nm. All experiments were performed in triplicates.

Electrochemical measurements

Electrochemical measurements including electrochemical impedance spectroscopy (EIS) and potentiodynamic polarization (PDP) were conducted using PGSTAT302N Autolab (Metrohm) with NOVA 2.1 software. The conventional three-electrode system was used, in which the reference and working electrodes were saturated Ag/AgCl and AH36 steel with a single side of 10 × 10 mm exposed square area, respectively. A platinum sheet of 20 × 10 mm was chosen as the counter electrode as the area needs to be larger than the exposed working electrode to guarantee uniformity of the electrochemical reaction61. The working electrode was exposed to buffer solution containing various rMrCP20 concentrations. All electrochemical measurements were performed at room temperature and repeated in triplicate.

Prior to measurements, the working electrode was incubated in a test solution for 30 min under stagnant condition until a stable open circuit potential (OCP) was attained. EIS measurements at different time intervals were conducted with a perturbation amplitude of 10 mV at OCP over a frequency range of 0.01–10000 Hz. The NOVA 2.1 software was employed for analysis of the EIS spectra. Nyquist and Bode plots were drawn by fitting the impedance data using equivalent circuit.

Potentiodynamic polarization (PDP) experiments were carried out after 24 h of exposure, in the potential range from −0.25 V versus OCP to −0.4 V versus Ag/AgCl with the sweep rate of 10 mV min−1. The corrosion potential (Ecorr) and corrosion current density (icorr) were calculated by extrapolating the linear Tafel segments of the anodic and cathodic curves. The data processing was performed using the NOVA 2.1 software.

Quartz Crystal Microbalance with dissipation monitoring, QCM-D

The adsorption of rMrCP20 on Fe substrate and its interaction with Fe3+ were analyzed using Quartz Crystal Microbalance (QCM) with Dissipation monitoring (Q-Sense E4 QCM, Sweden), connected with IPC peristatic pump (Ismatec SA, Switzerland). The iron-coated quartz crystals with a fundamental frequency of 4.95 MHz were purchased from Q-sense (QSX 319, Biolin Scientific). All QCM sensor crystals were cleaned according to a standard protocol whereby the surface is immersed in 1% Hellmanex III (Sigma-Aldrich) solution for 30 min, then ultrasonicated in 99% ethanol (Merck) for 10 min, followed by rinsing with Milli-Q water and finally dried with nitrogen gas. These measurements were carried out at a constant flow rate of 100 µL min−1 and the temperature of the measuring cell was controlled at 22 °C. The resonant frequency of the oscillator (f) and the energy dissipation value (D) were recorded simultaneously as a function of time. The system baseline was established for 1 h using the buffer solution without protein. After stabilization, 23 µM of rMrCP20 solution was injected and the adsorption took place for 45 min. The cell was then rinsed with buffer until stable frequency and dissipation values were attained. 100 µM of FeCl3 solution was injected and the adsorption of FeCl3 on pre-adsorbed protein layer took place for 35 min. Buffer was then re-introduced to wash out the weak bound complexes.

Attenuated total reflectance Fourier-transform infrared spectroscopy (ATR-FTIR)

Samples were prepared at 1 and 5 mg mL−1 rMrCP20 protein concentration with a series of FeCl3 concentrations in separate vials of 100 µL volume and incubated at ambient temperature for 1 h, by dipping the vials in liquid N2 for 5 min and lyophilized immediately. Attenuated total reflection fourier-transform infrared (ATR-FTIR) spectroscopy measurements were performed on a Bruker Vertex 70 (Massachusetts, USA) instrument equipped with a PIKE Technologies MIRacle attenuated total reflection (ATR) Diamond-ZnSe 3-reflection accessory and a LN2 cooled MCT detector. Scans were obtained at ambient temperature over the range of 4000 to 750 cm−1 with a resolution of 2 cm−1, averaged over scans obtained at 20 kHz for 1 min. All spectra processing were performed on OPUS 6.5, in the sequence of water vapor subtraction, background subtraction of buffer, baseline correction, and min-max normalization using amide I band. Amide I band was deconvoluted by secondary derivation, with peak fitting performed using 100% Gaussian curves with individual FWHM kept relatively consistent. The deconvoluted peaks were then assigned to the respective secondary structures40,51,52,53.

Nanoindentation

QCM sensors (same as above) were first cleaned with ethanol and Milli-Q water, then dried with nitrogen gas. Each sensor was then incubated separately in 5 mL of 5 mg mL−1 rMrCP20 protein solution for 50 min in 12-well plates, followed by 5 mL of buffer for 50 min. Half of the sensors were then incubated in 4 mL of 50 μM FeCl3 for 40 min, followed by 4 mL of buffer for 40 min. All incubations were performed on a shaker. Nanoindentation was performed with a Hysitron TI 950 Triboindenter (Minnesota, USA), using a Berkovich (cube-corner) indentation tip. The tip was calibrated with a fused quartz standard sample of E̅ = 69.9 GPa ± 10%. Sensors were attached to magnetic stainless steel specimen discs and placed on a standard stage, then tip to optic calibrations were performed prior to lifting the tip 100 μm above sample and adding a drop of water on the sample, allowing the tip to submerge and equilibrate. After 30 min, indentations were performed with displacement-controlled setting of 200 nm depth. Load–hold–unload indentation function setup was applied for all indents, with segment intervals of 5 s (load), 2 s (hold), and 5 s (unload). At least 30 indents were obtained for each sample. Slope tangents were obtained separately from contact with sample layer and substrate, and intersections of the tangents indicated the sample layer thickness. Statistical significance between groups were acquired via two-sample t-test assuming unequal variances. P-value of P < 0.0001 (****) and P < 0.001 (***) represented significance difference between two groups.

Field emission scanning electron microscopy – energy dispersive X-ray (FESEM-EDX)

AH36 coupons were adhered to sample stubs with adhesive carbon tape, and copper tape. Without coating, imaging was performed using JEOL JSM-FESEM 7800F PRIME (Massachusetts, USA), at SEI-mode, 20 kV, and 92 µA emission current. EDX maps were acquired at a 20 keV, 1024 resolution, with fixed duration of 3 frame counts, and process time 4, and pixel dwell time of 20  µs.

SAXS data collection and analysis

SAXS data for rMrCP20 were collected with a Xenocs Nano-inXider SAXS instrument equipped with a microfocus sealed-tube X-ray source (Cu, 30 W, 40 μm focus) and a Dectris Pilatus 3hybrid pixel detector. The X-rays are filtered through the two-dimensional single-reflection multilayer optics and collimated by a three-pinhole system. The sample-to-detector distance was set at 0.94 m, and the sample chamber and X-ray paths were evacuated prior to usage. This setup covers a range of momentum transfer of 0.08 < q < 4 nm−1 [q = 4πsin(θ)/λ, where 2θ is the scattering angle]. SAXS experiments of rMrCP20 were carried out at room temperature in the buffer composed of 20 mM Tris-HCl, pH 8.3 and 150 mM NaCl, using the low-noise flow cell. The protein concentrations used were 1.3, 4.1 and 6 mg mL−1. The data were collected for 45 min, and for each measurement a total of nine frames at 5 min intervals were recorded. The scattered X-rays detected by a two-dimensional area detector were converted to one-dimensional scattering using the built-in SAXS software (Xenocs).

All the data processing steps were performed using the program package PRIMUS62,63. The scattering of the buffer was subtracted from the data. The experimental data obtained were analyzed for aggregation using a Guinier plot64. The forward scattering, I(0) and the radius of gyration, Rg, were computed using the Guinier approximation, which assumes that at very small angles (q < 1.3/Rg), the intensity is represented as \(I(q)=I(0)\times \exp [-(q{R}_{g\left)\right.})^{\frac{2}{3}}]\)64. These parameters were also computed from the extended scattering patterns using the indirect transform package GNOM65,which provides the distance distribution function, P(r), and hence the maximal particle dimension, Dmax, and the radius of gyration, Rg. The hydrated volume, Vp, which was used to estimate the molecular mass of the protein, was computed using the Porod invariant. The theoretical scattering profiles were generated from the atomic resolution structure of rMrCP20 (PDB ID: 6LEK state 1) by considering both monomeric and dimeric structures, and the simulated models generated in this study. All generated theoretical scattering profiles were evaluated against the experimental scattering profile using CRYSOL66.

Molecular dynamics simulations. Initial Structure

AlphaFold2 server67 was used to predict the structure of rMrCP20 in monomeric and dimeric forms. One of the dimeric form was found to be better consistent with the SAXS data in that the theoretical scattering profile of the dimeric structure calculated using CRYSOL63 fitted well with the experimental profile (χ2 = 1.76). Thus, this dimeric structure was taken as the initial structure for the following molecular dynamics simulation.

The rMrCP20 dimer was represented by AMBER99SB-ILDN force field68. The system were solvated with 161,769 TIP3P69 water molecules, 802 Fe3+ ions and 2354 chloride ions were added to neutralize the system. The system was put in a cubic box of 173 Angstroms with periodic boundary condition. The molecular dynamics (MD) simulation used GROMACS70 5.1.2 software. The LINCS71 algorithm was used to constrain bonds between heavy atoms and hydrogen to enable a timestep of 2 fs. A 1.2 nm cutoff was used for Van der Waals interaction and short-range electrostatic interactions calculations, and Particle Mesh Ewald method was implemented for long range electrostatic calculations. Simulation temperature was maintained at 300 K using a V-rescale thermostat72 and 1 bar pressure using Parrinello-Rahman73 barostat. A production run of 100 ns was obtained.

Fluorescence microscopy

rMrCP20 was labeled with FITC (fluorescein isothiocyanate, Sigma-Aldrich) using the same protocol as previously described74. Briefly, 3 mg mL−1 of FTIC was dissolved in acetone and 200 μL was slowly added to 4 mL of rMrCP20 at 7 mg mL−1 in 150 mM NaCl, 20 mM Tris, pH 8.3 buffer. The reaction mixture was incubated overnight at 4 °C. The FITC-rMrCP20 mixture was then subjected to a PD-10 gel filtration column (GE healthcare life sciences) and eluted with buffer solution (150 mM NaCl, 20 mM Tris, pH 8.3). To view the GB under microscopy, AH36 substrate was polished to mirror surface quality and slightly etched with 2% Nital solution. The FITC-rMrCP20 solution was then cast on an AH36 etched substrate and rinsed with buffer to remove the loosely bound proteins. The images were obtained with an inverted microscope (Zeiss Axio Observer, Z2).

Auger electron spectroscopy (AES) data were obtained using a field-emission Auger microprobe (JEOL JAMP-7830F) with a primary electron beam having an accelerating voltage of 10 keV and a probe current of 10 nA. The sample was tilted at 30° throughout the analysis. AES elemental mapping was collected with a pixel density of 256 × 256 and a dwell time of 10 ms. The mapping was acquired by setting the analyzer to a specific Auger peak (P) energy and background (B) energy, operated at a constant retarding ratio of 0.6 for high sensitivity analysis. To correct for the topographical effects, the data were processed using the (P − B)/B algorithm.

X-ray photoelectron spectroscopy (XPS) analysis was performed using an AXIS Supra spectrometer (Kratos Analytical, UK) equipped with a hemispherical analyzer and a monochromatic Al K-alpha source (1487 eV) operated at 15 mA and 15 kV. The XPS spectra were acquired from an area of 700 × 300 μm2 with a take-off angle of 90°. Pass energy of 160 eV and 20 eV were used for survey and high-resolution scans, respectively. A 3.1-volt bias was applied to the sample to neutralize charge build up on the sample surface. The binding energies (BEs) were charge-corrected based on the C 1s of adventitious carbon at 284.8 eV. The data processing was performed using the ESCApe software.