The synthesis route of PEG-Hep was shown in Scheme 1. In brief, 50 ml tetrahydrofuran was firstly added into a 100 ml single mouth flask and then 10 g 6-arm polyethylene glycol (PEG, Sigma-Aldrich Shanghai, China) and 0.5 g heparin (Sigma-Aldrich Shanghai, China) were added. After thoroughly stirring in ice bath for 25 min, 0.03 g DMAP (4-dimethylamino-pyridine) and 0.2 g EDC (N-(3-Dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride, Sigma-Aldrich Shanghai, China) was added to react 12 h at room temperature under stirring. After precipitating by the ice ethanol and filtration, the products were dried in the vacuum oven. The molecular structure of PEG-Hep was determined by nuclear magnetic resonance hydrogen spectrum (¹H-NMR) and ATR-FTIR.

AZ31B magnesium alloy with the diameter of 12 mm was cut into 3 mm thickness slices, followed by successively polishing using 400#, 800#, 1000#, 1500#, 2000# sandpapers. The polished samples (Mg) were ultrasonically cleaned in acetone, ethanol, deionized water for 5 min, respectively. The sample was immersed in a 12 mol/L hydrofluoric acid (HF) solution for 15 min. After washed and dried, the HF treated sample (Mg-HF) was immersed in 2 mg/ml dopamine solution (Tris buffer, pH8.5) for 12 h at room temperature. The chitosan was then immobilized on the polydopamine-modified sample (Mg-DA) by putting the sample into the chitosan solution (0.2% acetic acid, pH 5) for reacting 2 h. The chitosan-modified sample (Mg-Chi) was immersed into 2 mg/ml PEG-Hep solution to adsorb heparin through electrostatic interaction, and the resulting sample was denoted as Mg-PEG-Hep.

The surface chemical structures of the samples were characterized by ATR-FTIR (TENSOR27, Bruker of Germany) with a scanning range of 4,000–650 cm⁻¹ at room temperature, and the elemental analysis were performed by X-ray photoelectron spectroscopy (XPS, Quantum 2000; PHI Co., Chanhassen, MN) with Al Kα source (1486.6 eV). The morphology and surface wettability of the samples were observed by scanning electron microscopy (SEM, FEI, Quata 250, United States) and contact angle meter (KR ü SS GmbH, Germany) at room temperature, respectively. In order to ensure the accuracy of the water contact angle, three parallel samples were measured to calculate the average value.

The potentiodynamic polarization curves of the different samples were measured by a CHI660D electrochemical workstation (CHI Instruments, Inc., Shanghai, China) using a standard three-electrode system. The corrosive medium is simulated body fluid (SBF, composition: NaCl 8 g/L; KCl 0.4 g/L; NaHCO₃ 0.35 g/L; CaCl₂ 0.14 g/L; Na₂HPO₄ 0.06 g/L; KH₂PO₄ 0.06 g/L; MgSO₄·7H₂O 0.01 g/L; glucose 1 g/L). The sample, saturated Ag/AgCl electrode and platinum electrode are working electrode, reference electrode and auxiliary electrode, respectively. Before the test, the sample was sealed with the silicone rubber using a copper wire as the conductor. The exposed area was 1 cm². The sample was immersed into 100 ml SBF solution for 10 min to achieve a stable open circuit potential before the polarization test. The potentiodynamic polarization test was carried out at a scanning rate of 1 mV/s. The corrosion current and corrosion potential were determined according to the polarization curve.

In order to further determine the corrosion behaviors, the pristine magnesium alloy and the modified samples were immersed in SBF solution at 37°C for 1 day, 3, 7 and 14 days, respectively. Before the experiment, the sample needs to be encapsulated with silicone rubber. The sample was placed in a tube containing 20 ml SBF, and the medium was changed every 2 days. At the predetermined times, the sample was taken out, cleaned and dried, and the surface morphology was observed by a scanning electron microscopy.

At the same time, the degradable behaviors of the different samples were characterized by immersing experiment. Before testing, the samples were all encapsulated and then put them into 20 ml SBF solution with the initial pH of 7.4. The pH values of the solutions were measured by a pH meter at least three times for each time. Then the average value was obtained.

All the data of the water contact angle, pH changes, hemolysis rate and CCK-8 values in this paper were expressed by mean ± SD, and the statistical analyses were performed using SPSS 12.0. Statistically significant differences were determined by one way analysis of variance (ANOVA). A probability value (p) of less than 0.05 was considered to be statistically significant.

The chemical structure of the synthesized PEG-Hep was characterized by ATR-FTIR and hydrogen nuclear magnetic resonance spectroscopy (¹H-NMR). The results are shown in Figures 1A,B, respectively. For ATR-FTIR (Figure 1A), a broad absorption band appeared at 3,500 cm⁻¹, which is mainly due to the stretching vibration of -OH. The peaks of CH₂ and CH₃ at 2,980 cm⁻¹ and 2,890 cm⁻¹ can be clearly observed, and the peaks of C-O-C at 1275–1020 cm⁻¹ can also be detected. They are all characteristic peaks of PEG. The peaks of C-O-S at 842 cm⁻¹ and C=O at 1700 cm⁻¹ suggested that heparin was successfully introduced. In ¹H-NMR spectra, the integral intensity ratio from low field to high field is 2:3, combined with the results of ATR-FTIR, it can be known that the substance has the chemical structure of CH₂-CH₃, and the OH peak appeared at 3.42 ppm. The H1 of N-sulfated glucosamine and the methyl of N-acetylglucosamine in heparin appear at 3.27 ppm and 2.05 ppm, respectively. All these results indicated that the PEG-Hep copolymer was successfully obtained.

The surface chemical structures of the samples were firstly characterized by ATR-FTIR, and the results are shown in Figure 2A. There was no infrared absorption on the pristine magnesium alloy surface, suggesting that there were no chemical groups on the blank magnesium alloy surface. After hydrogen fluoride treatment, a chemical layer of MgF₂ was produced on the surface, leading to the occurrence of the sharp peak at 900 cm⁻¹ (Pan et al., 2017). The Mg-DA sample had new absorption peaks at 1250 cm⁻¹ and 1600 cm⁻¹, which are caused by the stretching vibration of the C-O bond and the bending vibration and shearing vibration of the N-H. After chitosan was grafted on the surface, two new absorption peaks appeared at 1670 cm⁻¹ and 1415 cm⁻¹, namely C=O and C-N, indicating that chitosan was successfully immobilized on the surface of Mg-DA sample. The Mg-PEG-Hep sample showed a sharp peak corresponding to the hydroxyl group at 3,700 cm⁻¹, and the peaks at 2,980 cm⁻¹ and 2,890 cm⁻¹ were the appearance of CH₂ and CH₃, respectively. The characteristic peak at 842 cm⁻¹ and 1280 cm⁻¹ were the appearance of C-O-S and S=O, respectively, indicating that PEG-Hep was successfully introduced on the Mg-Chi surface.

XPS was utilized to further detect the surface atomic concentrations of the different samples, and the results are shown in Figure 2B and Table 1. In addition to Mg_1s, the blank magnesium alloy surface had C_1s and O_1s peaks, indicating that the magnesium surface was oxidized and C contamination appeared. After the hydrogen fluoride treatment, the oxygen concentration was reduced, and F_1s peak appeared, suggesting that the MgF₂ chemical conversion layer on the surface was produced and replaced the previously oxide layer. The decreased O_1s peak and the occurrence of the N_1s peak on Mg-DA surface indicated that the polydopamine layer was formed on the surface, which can provide the reaction sites for the subsequent reaction. After grafting chitosan, the C_1s, O_1s and N_1s peaks were all significantly enhanced. For Mg-PEG-Hep, the C_1s and O_1s peaks on the surface were significantly enhanced, the N_1s content was reduced, and S_2p and S_2s appeared, indicating that PEG-Hep was successfully introduced on the surface. In conclusion, it can be inferred that DA, Chi and PEG-Hep were successfully introduced on the magnesium alloy.

The surface morphologies of the samples were characterized by scanning electron microscopy (SEM). As shown in Figure 3A, the surface of the magnesium alloy is relatively flat after polishing. After the hydrogen fluoride treatment, the MgF₂ layer was formed on the magnesium alloy according to ATR-FTIR and XPS, which can play protective role to prevent the corrosion of the magnesium alloys (Da Conceicao et al., 2010), and initially improve the corrosion resistance of the magnesium matrix. Dopamine (DA) is a kind of high adhesion protein inspired by mussel foot. The dopamine molecule can self-polymerize to polydopamine (PDA) in air which has a strong binding force with a variety of materials (Waite, 1983; Liu et al., 2014a). Moreover, it can also react with substances containing amino and thiol groups. From the XPS results, it can be known that the polydopamine was successfully immobilized on the Mg-HF sample. Compared with the Mg-HF, the surface of Mg-DA was rougher, which may be caused by the time-consuming immersion process and a large amount of PDA agglomeration (Ball et al., 2012; Wang et al., 2014). After grafting chitosan, white particles appeared on the surface, which is a common feature of chitosan films (Zhai et al., 2018). The positive charges of chitosan can absorb the negative-charged heparin, so that PEG-Hep can be successfully immobilized on the surface of Mg-Chi. The surface of Mg-PEG-Hep was smoother than Mg-Chi, which can protect the magnesium matrix and thus improve the corrosion resistance.

Surface hydrophilicity plays an important role in the biocompatibility of the biomaterials. Good hydrophilicity can prevent the nonspecific protein adhesion and thus improve the blood compatibility and provide good cell compatibility. The hydrophilicity of the magnesium alloy before and after surface modification was evaluated by water contact angle, and the results are shown in Figure 3B. The water contact angle of the original magnesium was 56.2°, exhibiting limited hydrophilicity. The original magnesium surface had no hydrophilic groups (as shown in Figures 2A,B) and the surface was smooth (as shown in Figure 3A), resulting in the limited hydrophilicity. After the hydrogen fluoride treatment, the water contact angle increased to 118.4°. Studies have shown that the MgF₂ chemical layer formed after the hydrofluoric acid treatment has the characteristics of low water solubility (Makkar et al., 2018), which is the reason for the increase in the water contact angle. After polydopamine deposition, the water contact angle was reduced to 23.8° because of the introduction of hydrophilic amine groups, and the enhanced surface roughness (as shown in Figure 3A) also contributed to the better hydrophilicity. Chitosan is a hydrophilic polymer (Ludwiczak and Mucha, 2010), but the Michael addition reaction between the polydopamine coating and the amino groups in chitosan formed a dense layer, resulting in a slight decrease in hydrophilicity, and thus the water contact angle was increased to 37.1°. Heparin is a natural polysaccharide, which has rich hydrophilic groups such as hydroxyl, carboxyl and sulfonic acid groups, therefore, the water contact angle of Mg-PEG-Hep was reduced to 3.9°, displaying the superhydrophilicity.

The potentiodynamic polarization curves of the magnesium alloy before and after modification in the simulated body fluid are shown in Figure 4A, and the corresponding corrosion potential and current are summarized in Table 2. Generally speaking, a high corrosion potential means that the material is thermodynamically more stable and resistant to corrosion, while a low corrosion potential means that the corrosion tendency is lager and the corrosion resistance could become worse. The surface of the unmodified magnesium alloy is active, therefore, the corrosion potential (-1.57 V) was the lowest and the corrosion current (1.419 × 10-4A·cm⁻²) was the highest among all samples, indicating that the corrosion resistance of unmodified magnesium alloy was the worst. After hydrogen fluoride treatment, a MgF₂ layer was formed. Lee et al. proved that the MgF₂ layer on the magnesium alloy has the advantages of good compactness and high bonding strength with the substrate (Makkar et al., 2018). The dense film layer is not conducive to the penetration of electrolytes, and the poor hydrophilicity can prevent water adhesion, consequently, it can reduce the corrosion rate of the magnesium alloy. After the polydopamine deposition, the polymer layer on the magnesium alloy can isolate the substrate from the corrosive medium and thus effectively protect the magnesium matrix, and the corrosion current was further reduced to ×3.37010⁻⁷ A cm⁻². At the same time, the self-polymerized dopamine-coated catechol can also undergo a Michael addition reaction with the amine group of chitosan, thus chitosan was grafted on the magnesium alloy surface. After grafting chitosan, the positive charges of chitosan can adsorb anions and hinder the direct erosion of anions to the magnesium matrix, leading to the improved corrosion resistance. At the same time, the increase of the polymer layer thickness can also improve the corrosion resistance. After PEG-Hep immobilization, the anions of heparin can combine with the positive charges of chitosan, making the surface exhibit electronegativity to prevent the adsorption of anions, so that the corrosion resistance of the magnesium alloy was further improved, and the corrosion potential increased to −1.055 V, the corrosion current was reduced to 6.650 × 10⁻⁹ A cm⁻². It can be concluded that the PEG-Hep coating can significantly improve the corrosion resistance of the magnesium alloy. It can be seen from Figure 3A that the surface film layer was relatively dense, which made the magnesium alloy difficult to corrode in the SBF and may achieve the purpose of long-term protection of the magnesium alloy.

The corrosion behaviors of the different samples were further evaluated by immersion test. The samples were immersed in SBF solution for 1 day, 3, 7 and 14 days, respectively, and then the surface morphologies of the samples before and after immersion were observed by scanning electron microscopy. The typical surface morphologies of the different samples after immersing different time were shown in Figure 4B. Elemental analysis was also performed by EDX after immersion and the results are listed in Table 3. It can be seen that the small cracks appeared on the surface of the magnesium alloy after soaking 1 day, because the pristine magnesium alloy had loose magnesium oxide, which cannot resist the erosion of anionic ions in SBF, leading to the poor corrosion resistance. Hydrogen gas was generated during pretreatment with hydrogen fluoride, resulting in a few microcracks on the Mg-HF surface. The released hydrogen during the soaking process and the presence of the corrosive Cl⁻ can weaken the adhesion between PDA and Mg, resulting in the appearance of small corrosion pits on the Mg-DA surface. However, no cracks can be observed on the Mg-Chi and Mg-PEG-Hep surfaces, indicating that the introduction of chitosan and PEG-Hep can significantly enhance the corrosion resistance of the magnesium alloy. After 3 days of immersion, the surface cracks of the magnesium alloy increased significantly. Studies have shown that the corrosion rate of the magnesium alloys is faster at 3 days as compared to 14 days (Makkar et al., 2018). On one hand, the initial larger area exposed to the solution could promote the exothermic reaction and increase the corrosion rate; on the other hand, the higher concentration of Cl⁻ and other salts may also cause the corrosion rate to increase. The surface of other modified samples was basically unchanged at 1 day, and no cracks occurred, indicating that dopamine, chitosan and PEG-Hep coatings were beneficial to improve the corrosion resistance of magnesium alloy to the different degree. On the 7th day, the corrosion on the magnesium surface became serious, and the cracks and corrosion products increased significantly. The corrosion of other samples was also more serious than before. However, there were only microcracks on the Mg-Chi and MG-PEG-Hep surface because they could cover the magnesium alloy surface and protect the magnesium alloy matrix from corrosion. After being immersed 14 days, it can be seen that white granular substances appeared on all sample surfaces (Figure 4B), and it was found that the main component may be hydroxyapatite (HA, Ca₁₀ (PO₄) ₆ (OH)₂) (Ly and Yang, 2019). As shown in Table 3, P and Ca elements appeared on the surfaces of all samples, which may be Ca-P particles formed on the surface. The surface corrosion of Mg was more serious, and the content of Mg decreased, and the content of O and C increased, indicating that these corrosion products may be carbonate. The increase of P element content on the Mg-HF surface indicates that the MgF₂ coating on the surface may be damaged. After the fixation of dopamine, the content of P and Mg further increased, and the small pits on the surface of the PDA coating caused serious corrosion of the samples. The changes of elements in other samples were not different from Mg. During the immersion of Mg-Chi, the chitosan film was slightly corroded, and a small amount of corrosion products appeared, so the elements C and Mg basically did not change, and the content of O increased. Mg-PEG-Hep surface coating had the best integrity, less corrosion products and the highest Mg content can be observed, indicating that it had the best corrosion resistance. On one hand, the coating can act as an isolating barrier between magnesium alloy and SBF, and on the other hand, heparin with negative charge properties can inhibit anion erosion.

The pH in the human body is about 7.4, which is weakly alkaline. Too high pH will damage the normal growth microenvironment of cells. The high corrosion rate of magnesium alloy can produce a large amount of OH⁻, resulting in the increase of pH. Therefore, the pH changes of the different samples soaked in SBF for 7 days were measured and the results are shown in Figure 4C. The pH value of all samples increased rapidly on the first day, among which Mg increased the fastest (from the 7.4–9.02), indicating that its corrosion resistance was the worst. After 2 days, the pH value changed slowly and became steady, implying that the degradation rate became slower. During the immersion process, the pH value of Mg was always the highest, so its corrosion resistance was the worst. After hydrogen fluoride treatment, a chemical conversion layer of MgF₂ was formed. Yuan et al. proved that the NaHCO₃, NaCl and other components in SBF can effectively prevent the dissolution of MgF₂ (Mao et al., 2013), so the pH value change tends to be flat. After grafting chitosan, the positive charge of the amine group can prevent the corrosion of Cl⁻ and other anions, so it can have a long-term protective effect on the magnesium matrix. After 4 days of immersion, a dynamic equilibrium was reached between the dissolution and corrosion products of the samples, and the corrosion rate decreased, so the pH value decreased, which was consistent with the results of the immersion experiment. The pH values of the Mg, Mg-HF, Mg-DA, and Mg-Chi samples increased slightly after being soaked for 6 days. This may be because the corrosion products had no protective effect on the magnesium matrix over time. After immersing 7 days, the pH value of Mg-PEG-Hep was the lowest. The positive charges of chitosan can react with the negative charge of heparin to form a dense protective layer, as shown in Figure 4B, which can effectively inhibit the corrosion of the magnesium alloy.

Human blood is mainly composed of red blood cells, white cells, platelets and so on. When the stent enters the human body, the red blood cells may be damaged and even ruptured, resulting in severe hemolysis. National standards stipulate that the hemolysis rate of biomaterials or devices should be less than 5%. The hemolysis rates of the different samples are shown in Figure 5A. Due to the limited corrosion resistance, the blank magnesium alloy can produce a large amount of OH⁻ during the corrosion process, which will increase the local pH. High pH will increase the binding capacity of hemoglobin and membranes, leading to the increase of hemolysis rate. Severe hemolysis can cause the hemolysis rate of magnesium alloys to reach as high as 40%. After hydrogen fluoride treatment, the hemolysis rate was 10.3%. Studies have shown that the hemolysis rate of MgF₂ formed after hydrofluoric acid treatment is 10.1% (Mao et al., 2013). This is because fluoride itself is biocompatible, and the increased corrosion resistance can also reduce the release of OH⁻. After the deposition of polydopamine, the hemolysis rate increased rapidly to 35%. The research showed that although the polydopamine layer could slow down the corrosion of the material (as shown in Figure 4A), however, due to the amine groups on its surface, more platelets and red blood cells adhere to the surface, resulting in the poor blood compatibility (Chen et al., 2018). After grafting chitosan, the hemolysis rate was about 7.9%, the negatively charged glycoprotein on the surface of erythrocytes can combine with the positively charged amino group of chitosan, resulting in the decrease of hemolysis rate, but it is still larger than 5%. After PEG-Hep was introduced, the hemolysis rate decreased to 3.8%. It can be considered that heparin can not only enhance the activity of thrombin III, thereby indirectly exerting anticoagulant effect, but also can improve the activities of protein C and stimulate endothelial cells releasing anticoagulant and fibrinolytic substances (Wu, 1996), so the hemolysis rate of Mg-PEG-Hep can meet the requirement of the national standard for hemolysis rate.

Another way of evaluating blood compatibility is platelet adhesion. When the stent enters the human body, platelet aggregation and activation may occur to promote the production of thrombin, thereby forming a thrombus. Therefore, platelets are considered to be one of the important causes of thrombus formation. The SEM images of platelet adhesion and the amount of platelets on the different samples are shown in Figure 5B. The platelets on the magnesium alloy were mostly in an aggregated state, and most of them had pseudopodia exhibiting a spreading state, indicating that the blood compatibility of the magnesium alloy was poor. After hydrogen fluoride treatment, a chemical layer of MgF₂ was formed. Mao et al. proved that MgF₂ has good anti-platelet adhesion ability (Mao et al., 2013), and the improvement of corrosion resistance (as shown in Figure 4A) can also reduce the platelet adhesion. Wei et al. (2011) and Ma et al. (2013) have also proved that the imine and quinine groups in the polydopamine layer can quickly adsorb proteins, leading to platelet adhesion and aggregation, therefore, compared with Mg-HF, the number of platelets adhered to Mg-DA was increased. In general, surfaces with good hydrophilicity have lower surface free energy which can prevent the adhesion of plasma proteins and thus inhibit platelet adhesion (Gao et al., 2019). After chitosan graft, the decreased hydrophilicity (as shown in Figure 3B) resulted in platelet aggregation, so the number of platelet adhesion did not decrease significantly. Heparin has excellent anticoagulant effect and can reduce platelet adhesion and thus only several round platelets can be found on Mg-PEG-Hep. Meanwhile, Mg-PEG-Hep had good surface wettability, which can preventthe non-specific protein adsorption, and thus further reduced the platelet adhesion and aggregation, so the number of the attached platelets decreased significantly, which also confirmed that Mg-PEG-Hep had good blood compatibility.