Chitosan and polycaprolactone blended PDMS coatings improve biocompatibility of magnetic elastomers

Abstract

PDMS/NdFeB composites are promising materials for soft magnetic actuators, but NdFeB particles corrode in body fluids and release toxic metal ions, limiting their biomedical use. We developed ~ 100 µm spin-coated PDMS-chitosan (PDMS-CHIT) and PDMS-polycaprolactone (PDMS-PCL) coatings that solve this problem. Over 24 weeks of immersion, these coatings reduced neodymium and iron release by more than 95%, keeping ion concentrations well below cytotoxicity thresholds. Importantly, the PDMS-PCL coating fully preserved magnetorheological actuation (ΔG’ ≈ 61 kPa under 0.5 T, comparable to uncoated composite), while PDMS-CHIT provided superior ion barrier at the cost of reduced actuation force. Biological validation confirmed cytocompatibility with fibroblasts, hemocompatibility with erythrocytes, and strong suppression of bacterial biofilm formation. These results establish a validated materials platform for biocompatible soft magnetic actuators.

Introduction

Soft magnetic composites possessing tissue-mimetic elastic moduli (10⁴–10⁹ Pa)¹, particularly polydimethylsiloxane (PDMS) reinforced with neodymium-iron-boron (NdFeB) microparticles, are a key platform for biomedical actuators. This architecture combines the flexibility and biocompatibility of PDMS (E ~ 0.05–3 MPa)²with the strong magnetic properties of NdFeB (remanence ~ 1.2–1.4 T; coercivity ~ 800–1200 kA/m)³, enabling high-force, magnetically-responsive devices. Recent demonstrations include ferromagnetic soft catheter robots for minimally invasive in vivo bioprinting⁴, magnetically actuated microcatheters with soft rotatable tips for enhanced endovascular interventions⁵, magnetically driven capsules with multimodal response enabling selective dual-drug release and sampling⁶, and implantable magnetic soft robotic systems for intelligent organ function control⁷.

However, direct exposure of NdFeB particles to physiological fluids presents a critical stability challenge⁸. Chloride-containing body fluids induce galvanic corrosion of the NdFeB alloy, preferentially dissolving Nd-rich grain boundaries and releasing cytotoxic metal ions (Nd³⁺, Fe²⁺/³⁺)^8,9. This ion leaching negates biological safety, limiting the clinical translation of devices requiring tissue contact or implantation⁹. An effective solution demands a biocompatible barrier coating that isolates the particles from the aqueous environment while preserving both magnetic responsiveness and mechanical compliance¹⁰.

Designing functional barrier coatings for these composites introduces two primary physical conflicts. The first is coating thickness: the layer must be thick enough for robust ion containment but thin enough to prevent magnetic shielding, as excessive thickness diminishes magnetic flux penetration and actuator responsiveness. The second is mechanical compatibility; the coating must deform uniformly with the soft PDMS/NdFeB core to prevent interfacial delamination or cracking during actuation. A significant modulus mismatch between a coating and the elastomeric core creates stress concentrations at the interface, leading to premature mechanical failure^11–13.

Furthermore, applications involving dermal contact, such as wearable therapeutic devices, require antimicrobial properties to prevent infection, particularly in damaged tissue^14,15.

We selected chitosan and polycaprolactone (PCL) to meet these design criteria. Both polymers have established safety profiles and can form conformal diffusion barriers in aqueous environments^16–18. Each material offers a distinct advantage. PCL provides mechanical robustness with an intermediate elastic modulus (E = 0.2–0.4 GPa), bridging the stiffness mismatch between the soft PDMS core and more rigid components¹⁹. Chitosan, a cationic polysaccharide, provides inherent antimicrobial activity through electrostatic interaction with negatively charged bacterial surfaces²⁰. However, direct application of these hydrophilic polymers onto hydrophobic PDMS results in poor adhesion and delamination risk. We addressed this by using PDMS-blended coating formulations¹⁰. This creates a graded interface, meaning a gradual transition in surface energy and mechanical properties, ensuring robust adhesion while preserving the bioactive benefits of each polymer.

This study quantifies the effectiveness of spin-coated PDMS-chitosan and PDMS-PCL coatings in simultaneously preventing metal ion release from PDMS/NdFeB composites while maintaining magnetorheological performance and biological functionality. We assess: (i) physicochemical stability at the composite-environment interface by monitoring medium pH, redox potential, and conductivity over 24 weeks; (ii) direct metal ion leaching via inductively coupled plasma mass spectrometry (ICP-MS); (iii) magnetorheological response under applied magnetic fields up to 0.5 T; (iv) microbial biofilm formation by fungi and bacteria; (v) hemolytic activity; and (vi) cytocompatibility with mammalian fibroblasts.

The selected ~ 100 μm coating architecture, optimized in our previous work¹⁰, balances effective barrier function with preserved magnetic actuation.

Results

We fabricated four composite variants to quantify coating efficacy against aqueous corrosion: Non-coated (Control), PDMS (Pure PDMS barrier layer), PDMS-CHIT (PDMS-chitosan blend), and PDMS-PCL (PDMS-polycaprolactone blend). All coatings were applied via spin-coating to both surfaces of the magnetic core, creating a layer-core-layer (sandwich) architecture with a total thickness of ~ 300–350 μm. Each individual coating layer was engineered to a uniform ~ 100 μm thickness, a dimension selected as a critical balance between achieving a robust ion barrier and preserving magnetic responsiveness.

Physicochemical stability in physiological medium

To assess the long-term stability and protective efficacy of the coatings in a physiological environment, we monitored the physicochemical evolution of the 0.9% NaCl incubation medium over 24 weeks. Figure 1 presents four complementary parameters used to track these material-fluid interactions: water absorption (panel a), pH (panel b), redox potential (panel c), and electrolytic conductivity (panel d), all measured at intervals of 1, 2, 4, 8, 12, and 24 weeks. The control condition (panels b-d) represents the 0.9% NaCl solution without any immersed sample, establishing a baseline for comparison with sample-containing conditions.

Fig. 1.

Results of physicochemical analysis of the incubation medium following samples incubation. Parameters assessed included water absorption (a), pH (b), redox potential (c), and conductivity (d). Measurements were taken at specified intervals: after 1, 2, 4, 8, 12, and 24 weeks of sample incubation in the incubation medium. Control represents 0.9% NaCl solution without sample. Data points represent mean ± standard deviation (n = 3). Asterisks (*) indicate significant difference vs. Non-coated; circles (○) indicate significant difference vs. control (p < 0.01).

Water absorption kinetics (Fig. 1a) show all samples reached swelling equilibrium within the first week. The non-coated control and the pure PDMS coating exhibited the lowest water uptake, stabilizing at ~ 0.24–0.27% and ~ 0.13–0.16%, respectively. In contrast, the functional coatings absorbed more, with PDMS-PCL showing intermediate uptake (~ 0.32–0.37%) and the more hydrophilic PDMS-CHIT showing the highest absorption (~ 0.42–0.45%).

The pH evolution (Fig. 1b) provides clear chemical evidence of coating performance. The non-coated and PDMS-only samples became alkaline (ΔpH = + 0.44 and + 0.22, respectively), a direct indicator of NdFeB corrosion forming hydroxide (OH⁻) ions. The PDMS-only sample’s partial alkalization (ΔpH = + 0.22) confirms its barrier is incomplete²¹. Conversely, the functionalized PDMS-CHIT and PDMS-PCL coatings became acidic (ΔpH = -0.41 and − 0.31, respectively), confirming the NdFeB core was isolated. This acidification is the chemical signature of the coatings themselves: protonation of chitosan’s amine groups (pK_a ~6.5) and slow hydrolytic degradation of PCL’s ester bonds releasing caproic acid (pK_a ~4.9)²². These modest pH shifts (< 0.5 units) would be readily buffered in vivo.

Redox potential (Fig. 1c) confirms the corrosion mechanism. The non-coated sample medium showed a marked increase (+ 25.7 mV), indicating a shift to an oxidative environment consistent with the generation of oxidizing metal species (Fe³⁺, Nd³⁺) during NdFeB degradation. Conversely, the PDMS-CHIT medium became more reducing (-20 mV), a behavior matching the control saline. This is attributed to chitosan’s amine (-NH₂) and hydroxyl (-OH) groups scavenging dissolved oxygen²³. Most significantly, the PDMS and PDMS-PCL coatings remained stable (210–220 mV), mirroring the baseline 214.5 mV of the control saline. This stability, derived from PCL’s hydrophobic and non-reactive character²⁴, demonstrates an effective barrier against oxidative species.

Electrolytic conductivity (Fig. 1d) directly quantifies total ionic release, providing the clearest measure of barrier effectiveness. As expected, the non-coated sample showed the largest conductivity increase (+ 2.74 mS/cm), confirming substantial leaching of ionic corrosion products (e.g., Nd³⁺, Fe²⁺/³⁺). The PDMS and PDMS-PCL coatings performed moderately, with increases (+ 2.08 and + 2.10 mS/cm, respectively) slightly above the control saline’s baseline evaporation drift (+ 1.98 mS/cm). Strikingly, the PDMS-CHIT coating exhibited the smallest increase (+ 1.58 mS/cm), a value statistically comparable to the control, indicating near-complete suppression of ion release. This superior performance is rationalized by two mechanisms: (i) a passive barrier, where chitosan’s dense network creates a tortuous diffusion path, and (ii) an active barrier, where chitosan’s amine and hydroxyl groups actively chelate metal cations, trapping them within the coating²⁵.

Quantification of metal ion release via ICP-MS

Physicochemical data provide indirect evidence of corrosion, but direct ion quantification via ICP-MS is needed to assess barrier efficacy and cytotoxicity. We analyzed the 24-week incubation media for key elements from NdFeB particles: Nd, Fe, B, and trace Nb (Table 1). The non-coated control showed extensive corrosion, releasing 10.8 ± 0.1 µg/mL Nd and 9.54 ± 0.72 µg/mL Fe. Boron (593 ng/mL) and trace Nb (0.02 ng/mL), a grain boundary modifier²⁶, were also detected. These concentrations indicate substantial material degradation and exceed cytotoxicity thresholds. The measured Nd and Fe levels are well above typical IC₅₀ values for mammalian cells (1.4–7.2 µg/mL for Nd, 0.6–2.8 µg/mL for Fe)^9,27.

Table 1.

Element release results.

Sample	Element
	Nd (µg/mL)	Fe (µg/mL)	B (ng/mL)	Nb (ng/mL)
Non-coated	10.8 ± 0.1	9.54 ± 0.72	593	0.02
PDMS	4.01 ± 0.08*	7.66 ± 0.42*	207	0.15
PDMS-CHIT	0.07 ± 0.01*#	1.19 ± 0.01*#	101	< LOD
PDMS-PCL	0.1 ± 0.40*#	0.4 ± 0.05*#	169	< LOD

Data for Nd and Fe are presented as mean ± SD (n = 3). Data for B and Nb represent single measurements (n = 1) due to high salinity matrix constraints (0.45% NaCl). Statistical significance: * < 0.01 vs. Non-coated; #p < 0.01 vs. PDMS; < LOD: below limit of detection.

The ICP-MS data confirmed that a simple PDMS barrier is insufficient. The PDMS-only coating proved only partially effective, reducing Nd (to 4.01 ± 0.42 µg/mL) and B (to 207 ng/mL). However, it failed to suppress Fe (7.66 ± 0.42 µg/mL), a concentration that remains highly cytotoxic. This confirms that unfilled PDMS networks are inherently permeable to small molecules and ions, offering incomplete protection. In stark contrast, the functionalized PDMS-CHIT and PDMS-PCL coatings demonstrated dramatically superior barrier performance. The PDMS-PCL coating was most effective for iron, achieving 95.8% Fe suppression (to 0.4 µg/mL). The PDMS-CHIT coating provided the best overall protection, reducing Nd to a minimal 0.07 ± 0.01 µg/mL (99.4% reduction) and Fe to 1.19 ± 0.01 µg/mL (87.5% reduction). Both functional coatings suppressed B (~ 100–170 ng/mL) and Nb (< LOD), indicating near-complete ion containment.

This superior performance stems from two synergistic mechanisms. First, both PCL and chitosan create a passive barrier, increasing coating density and creating a more tortuous diffusion pathway for water and ions. Second, chitosan provides a unique active barrier, where its amine (-NH₂) and hydroxyl (-OH) groups chemically chelate (coordinate) metal cations like Nd³⁺ and Fe²⁺/³⁺, immobilizing them within the coating. This active sequestration mechanism is directly evidenced by the PDMS-CHIT sample’s superior Nd suppression (0.07 ± 0.01 µg/mL). Although PCL undergoes slow degradation, its hydrophobic character maintains barrier integrity. Most importantly, the residual Nd and Fe concentrations from both functional coatings (0.07–0.1 and 0.4–1.19 µg/mL, respectively) fall well below established cytotoxicity thresholds, confirming the coatings successfully mitigate the primary toxicity mechanism^9,27.

Magnetorheology tests

The critical test was whether the ion-blocking coatings compromised magnetic functionality. All compositions exhibited a substantial magnetorheological (MR) effect—field-induced stiffening from the alignment of magnetic particles into chain-like microstructures that resist deformation²⁸.

For actuator applications, the absolute MR effect (ΔG’ = G’_0.5T − G’_0T) is the key metric, as it represents the actual stiffening force available for actuation. The uncoated composite delivered ΔG’ ≈ 62 kPa (Fig. 2i). The PDMS-PCL coating fully preserved this capability (ΔG’ ≈ 61 kPa, Fig. 2l), while PDMS-CHIT (ΔG’ ≈ 26 kPa, Fig. 2k) and pure PDMS (ΔG’ ≈ 22 kPa, Fig. 2j) reduced it by more than half.

Fig. 2.

Magnetorheological response of biocompatible-coated PDMS/NdFeB magnetic elastomers. Upper panels (a–d): Strain amplitude dependence of storage modulus (G’) under zero field (0 T, orange squares and purple circles) and applied magnetic field (0.5 T, filled and open blue squares), demonstrating the linear viscoelastic region and magnetorheological effect. Middle panels (e–h): Frequency-dependent storage modulus across 0.1–100 rad/s, illustrating the field-stiffening behavior. Lower panels (i–l): Quantitative comparison of storage modulus (G’, orange) and loss modulus (G”, blue) at three magnetic flux densities (0, 0.25, and 0.5 T), showing the magnitude of magnetorheological enhancement. Composite configurations: (a,e,i) uncoated baseline; (b,f,j) PDMS-coated; (c,g,k) chitosan-functionalized (PDMS-CHIT); (d,h,l) PCL-functionalized (PDMS-PCL). All tests conducted at 21 ± 1°C. Data points represent mean ± SD (n = 3). Asterisks denote statistically significant differences relative to zero field (p < 0.05).

All coated variants showed lower baseline stiffness than the uncoated composite (74 kPa): PDMS-PCL (39 kPa), PDMS (27 kPa), and PDMS-CHIT (26 kPa). This reduction is attributed to the compliance of the non-magnetic coating layer. The lower baseline also explains the high relative MR enhancement of PDMS-PCL (156%) and PDMS-CHIT (101%)—these values reflect baseline differences, not superior magnetic response.

All materials behaved as strong elastomers (G’/G’’ > 10), characteristic of filled polymer networks²⁹. Amplitude sweeps (Fig. 2a–d) revealed strain-softening above 5–6%, marking the end of the linear viscoelastic region. This behavior is characteristic of the Payne effect, a well-documented phenomenon in particle-filled elastomers^30–32. At low strain amplitudes, filler particles form a percolating network through particle-particle and filler-matrix interactions, contributing significantly to composite stiffness. As strain amplitude increases beyond a critical threshold, this network is progressively disrupted. Loss of interparticle contacts and disturbance of the magnetically-induced chain structures result in decreased storage modulus. The weak strain overshoot observed in G’’ at intermediate strains reflects energy dissipation associated with the breakup and reorganization of particle clusters. Importantly, the Payne effect is largely reversible upon returning to low strain amplitudes, indicating that no permanent structural damage occurs during large-amplitude deformation. Frequency sweeps (Fig. 2e–h) showed moduli increasing moderately with angular frequency, typical for filled elastomers.

Microbial biofilm formation

We assessed microbial adhesion and biofilm formation using Candida albicans (fungal, Fig. 3a) and Enterococcus faecalis (bacterial, Fig. 3b). The results showed two divergent trends. First, all composite surfaces supported substantial fungal biofilm growth, with C. albicans levels exceeding the polystyrene control by more than threefold in some cases. Second, in striking contrast, bacterial biofilm formation was minimal. All composites suppressed E. faecalis growth, but the PDMS-PCL sample was most effective, reducing biofilm to < 10% of the control. This suggests an anti-adhesive surface mechanism (e.g., hydrophobicity, surface energy) rather than a leached bactericidal agent, a hypothesis supported by subsequent extract-based assays which showed a reduced effect.

Fig. 3.

Biofilm formation levels of C. albicans 1408 (a) and E. faecalis ATCC 45,477 (b) on the surfaces of tested materials following incubation for 24, 48, and 72 h. Data are presented as mean ± SD (n = 10). The dashed horizontal line represents the biofilm formation level on polystyrene surfaces in the absence of material.

To de-convolute the effects of intrinsic surface properties versus leached chemical agents, we evaluated biofilm formation in 24-hour material extracts (Fig. 4). The antimicrobial effect of these extracts was markedly reduced compared to the direct-contact assays (Fig. 3). This strongly indicates that the bacterial suppression observed previously is not driven by a leached chemical agent but by the intrinsic surface properties of the composites (e.g., roughness, hydrophobicity). While the 50% maximum extract concentration means a minor leachable component cannot be fully excluded, the data points decisively to a surface-mediated, anti-adhesive mechanism.

Fig. 4.

Biofilm formation levels of C. albicans 1408 (a) and E. faecalis ATCC 45,477 (b) in the presence of extracts derived from the tested biomaterials. The experiment utilized extract dilutions at concentrations of 1%, 10%, 25%, and 50%. Data are presented as mean ± SD (n = 3). The dashed horizontal line represents the baseline biofilm formation level on polystyrene surfaces without the addition of the analyzed extracts.

Hemocompatibility of the tested composites

We assessed hemocompatibility using a hemolysis test with isolated erythrocytes as model host cells. The results (Fig. 5) show that tested materials are characterized by high hemocompatibility, as we detected no considerable hemoglobin release (hemolysis), either from direct contact with the composites or in material extracts. Notably, although statistical significance was observed in individual cases between the exposed samples and the negative control (p value ranging from < 0.001 to 0.9625), it should be noted that hemolysis did not exceed 6.5% in any condition.

Fig. 5.

Hemolysis levels observed following direct incubation of blood samples with the tested materials (A) and in the presence of extracts derived from the biomaterials (B). In (A), the successive colors of the bars, i.e. cream, orange and navy blue, indicate the level of hemolysis in red blood cells after 1, 6 and 24 h of incubation. In (B), the colors of the bars correspond to the final concentration of extracts prepared from biomaterials, i.e. 1% (cream bars), 10% (orange bars), 25% (light navy blue bars) and 50% (dark navy blue bars). Results are presented as mean ± SD (n = 5 and n = 3 for direct and indirect hemolysis, respectively). * indicates statistical significance (p < 0.05) when compared to PBS-exposed samples.

Cytotoxicity of tested composites

We assessed cytotoxicity using NIH/3T3 mouse fibroblasts (ATCC CRL-1658™), quantifying proliferation and morphology over 72 h (Figs. 6 and 7). Fibroblasts demonstrated sustained proliferation on all composite surfaces, confirming their fundamental biocompatibility. However, proliferation rates were significantly lower on all biomaterials compared to the rigid polystyrene control substrate (p values in ranges of < 0.0001–0.0002, 0.0126–0.0236, and < 0.0001–0.0003 for 24, 48, and 72 h time-points). This discrepancy is likely attributed to physical and mechanobiological factors, not chemical cytotoxicity. The observed lower cell density is likely a result of: (i) seeding challenges on the hydrophobic biomaterial surfaces and (ii) the profound stiffness mismatch between the soft composites (E ~ kPa-MPa) and the rigid control plastic (E ~ GPa), which is known to alter cell behavior, proliferation rates, and may induce migration to the stiffer substrate.

Fig. 6.

Proliferation of NIH/3T3 mouse fibroblast cells (ATCC CRL-1658™) on the surface of tested materials compared to plastic. Results are presented as the mean ± SD (n = 4–8). For each condition 81 microphotographs were collected from the central part of the well and nuclei were quantified. * indicates statistical significance (p < 0.05) when compared to polystyrene samples.

Fig. 7.

Proliferation capability (a) and morphology (b) of NIH/3T3 mouse fibroblast cells (ATCC CRL-1658™) on the surface of tested materials. The images display representative results from a single experiment (n = 4–5 for each condition). Due to necessity of samples fixation prior staining and imaging, each photograph presents different material from the samples’ set. Scales: 3 mm (a) or 100 μm (b).

Cell morphology analysis confirmed the composites are conducive to cell growth, with fibroblasts appearing both contracted and fully spread on all surfaces. This observed morphological heterogeneity is likely a mechanobiological artifact rather than a sign of chemical toxicity. We hypothesize this variation stems from localized stiffness gradients within the substrate. These gradients likely arise from non-uniform adhesion between the soft biomaterial sample and the rigid underlying polystyrene plate, creating micro-environments of high (well-adhered) and low (poorly-adhered) stiffness, which cells sense and respond to³³.

Discussion

Our central finding is that a passive barrier alone cannot ensure long-term biocompatibility of PDMS/NdFeB composites—an active, chemical barrier is required. PDMS-only coatings fail despite their hydrophobicity: Fe leachate reached cytotoxic levels (7.66 µg/mL). The functionalized coatings succeeded through two distinct mechanisms. PDMS-PCL increased diffusion path tortuosity, slowing ion transport. PDMS-CHIT added chemical chelation: its amine (–NH₂) and hydroxyl (‑OH) groups bind metal cations (Nd³⁺, Fe²⁺/³⁺), trapping them within the coating. The exceptionally low Nd release (0.07 µg/mL) from PDMS-CHIT confirms that active chelation is essential for ion containment in soft magnetic implants.

The second critical challenge is the trade-off between barrier function and magnetic actuation. An effective coating must be biocompatible, magnetically transparent, and mechanically compliant. The absolute MR effect (ΔG’) revealed a stark performance difference. The PDMS-PCL coating met all design requirements: it provided excellent ion protection while preserving full actuation (ΔG’ ~61 kPa vs. ~62 kPa for uncoated control). PDMS-CHIT offered superior ion containment through chelation, but reduced actuation by over 50% (ΔG’ ~26 kPa). For applications requiring maximum actuation force, PDMS-PCL is the better choice.

This work solves a key problem: traditional corrosion coatings are mechanically incompatible with soft actuators. Existing NdFeB protection methods—metallic (Ni-Cu-Ni) or epoxy coatings—were designed for rigid sintered magnets³⁴. These rigid coatings are incompatible with soft elastomers³⁵. The modulus mismatch (GPa coating vs. kPa core) creates severe interfacial stress³⁶, leading to cracking and delamination even at low strains³⁷. Our PDMS-blended graded interface addresses this failure mode. By minimizing the modulus jump, we achieved—for the first time in soft composites—a system that provides effective ion barrier, mechanical compliance, and robust interfacial adhesion under actuation.

We acknowledge several limitations of this in vitro study. First, static immersion in acellular saline does not reproduce the in vivo environment—dynamic flow, protein adsorption, and immune responses. Macrophage-generated reactive oxygen species could accelerate coating degradation. Second, we did not test cyclic loading or repeated actuation—a critical omission, since fatigue-induced microcracking is a primary failure mode that static testing cannot predict. While 24 weeks is substantial, it does not guarantee multi-year stability for chronic implants.

Furthermore, our biological validation provided only an initial safety assessment. Single cell types (erythrocytes, fibroblasts) cannot predict a full tissue response. Low fibroblast proliferation could reflect substrate stiffness (a mechanobiological effect) rather than cytotoxicity—future work must distinguish these factors. Moreover, our experiments evaluated the early-stage adhesion (up to 72 h) rather than long-term microbial colonization. As prolonged incubation periods and protein-rich environments may ultimately facilitate microbial colonization through surface conditioning, this issue should be expanded in the further research. Finally, substantial C. albicans colonization on all surfaces indicates that chronic implant designs must incorporate antifungal modifications. A further limitation concerns the distinction between the magnetorheological effect and magnetic actuation. The composites investigated here exhibit field-induced stiffening, where an applied magnetic field increases resistance to externally imposed deformation, rather than autonomous shape change. In the current isotropic configuration, the magnetic particles are randomly oriented, so the material stiffens uniformly but does not bend, stretch, or move under the field. Consequently, our biological assays evaluated cell behavior on mechanically static substrates with tunable stiffness, not on dynamically deforming surfaces. While this represents a valid first step in biocompatibility assessment, it does not capture the mechanobiological effects of cyclic substrate deformation on adherent cells, a critical consideration for implantable actuators.

To achieve true magnetic actuation with field-driven shape morphing, composites must be directionally magnetized during fabrication to create anisotropic particle alignment. Further studies should therefore address both coating durability under repeated mechanical cycling and cellular responses to dynamic mechanical stimulation in such anisotropic systems. This will be essential for applications such as peristaltic pumps, cardiac assist devices, or haptic interfaces where cells experience cyclic strain.

Conclusions

We developed biocompatible polymer coatings that resolve the cytotoxicity problem of PDMS/NdFeB soft actuators. Two barrier strategies emerged:

• i.PDMS-PCL acts as a passive barrier, suppressing ion release to sub-cytotoxic levels while fully preserving actuation force (ΔG’ ~61 kPa).
• ii.PDMS-CHIT provides an active barrier through ion chelation, offering maximum ion containment but reducing actuation force by ~ 50%.

Both coatings, based on a PDMS-blended graded interface, are hemocompatible and suppress bacterial biofilm formation. The choice between PCL (maximum actuation) and CHIT (maximum ion containment) provides a validated platform for clinically translatable soft magnetic devices. Future in vivo studies must test performance under cyclic loading and address fungal colonization risk.

Materials and methods

Preparation of composites with biocompatible layers

Elastic composites with three distinct biocompatible coatings were fabricated and characterized. PDMS/NdFeB composites were first prepared and subsequently coated with either pure PDMS, PDMS-chitosan, or PDMS-polycaprolactone (PCL) using spin-coating. The magnetic elastomer base was prepared by combining PDMS silicone (Sylgard 184, Dow Corning, USA) with neodymium-iron-boron (NdFeB) micropowder (MQFP-14-12-20000-088, Magnequench, Germany; d50 = 25 μm) at a weight ratio of 30:70 (PDMS: NdFeB). This composition was selected based on previous optimization studies³⁸. The liquid composite mixture was cast onto 60 mm diameter Petri dishes and cured. To create a sandwich structure with coatings on both surfaces of the magnetic core, a sequential two-step coating process was employed¹⁰. First, 0.5 mL of coating solution (pure PDMS, PDMS+CHIT, or PDMS + PCL) was applied to the top surface of the cured composite. Spin-coating was performed at 2000 rpm for 10 s under controlled conditions (21 ± 1 °C, 70 ± 5% relative humidity), yielding a coating layer approximately 100 μm thick. After drying at 50 °C for 1 h, the sample was inverted and the identical coating procedure was repeated on the bottom surface. Final cross-linking of the complete sandwich structure was conducted in a laboratory oven (Binder GmbH, Tuttlingen, Germany) at 50 °C for 1 h.

Three coating formulations were investigated: (i) pure PDMS, (ii) PDMS with 2 wt% chitosan, and (iii) PDMS with 10 wt% PCL (Mw = 80,000). Uncoated PDMS/NdFeB composites served as controls for comparative analysis.

Rheological measurements

Rheological measurements of the developed composites were carried out using an Anton Paar MCR 702e rheometer (Graz, Austria). Composite samples were cut for testing using a special steel punch with a diameter of 25 mm. A plate-plate arrangement (25 mm diameter) was used for the measurements. The forced oscillation method was employed, where the rotation of the upper plate of the rheometer caused a sinusoidal strain waveform. The measured stress was converted into viscoelastic moduli G’ (storage modulus - in phase with the applied strain, representing the accumulation of strain energy) and G” (loss modulus - out of phase with the applied strain, representing energy dissipation). The measurements consisted of three tests: (i) at a constant angular frequency of 6.28 rad/s and strain amplitude of 1%; (ii) at variable amplitude in the range of 0.1–100% at a constant frequency; (iii) at variable frequency of 0.1–10 Hz at a constant amplitude of 1%. The tests were performed at three values of 0, 0.25 and 0.5 T of magnetic induction generated by a custom magnetic cell manufactured by Anton Paar rheometer.

Composites incubation procedure

The incubation studies were conducted under controlled conditions. The incubation medium consisted of a 0.9 wt% sodium chloride (NaCl, Sigma Aldrich, USA) solution in deionized, ultrapure Milli-Q water (Merck Milipore, Germany). Samples were prepared as flat, circular pieces with a diameter of 10 mm. Each container held the test material and the 0.9% NaCl solution in a weight ratio of 1:10. The samples were tightly wrapped using a protective film and placed in a laboratory incubator at an internal temperature of 37 ± 0.5 °C for 5, 10, 15, 20, and 25 weeks. After the specified incubation time, the composite samples were removed and assessed for water absorption. The post-incubation solutions were then tested for changes in their physicochemical properties to assess whether the composite samples reacted with those solutions.

Water absorption (Wa)

Each sample was weighed both before and after a predetermined incubation period, and then it was dried in a moisture analyzer (Axis, Poland) in order to quantify the water absorption (Wa). A balance (Mettler Toledo, USA) with a sensitivity of 0.01 mg was used to measure the sample weight. Using the following Eq. (1), water absorption was computed as the percentage weight rise, taking into account the sample’s initial dry weight (w_d) and weight after incubation (w_i).

1

Physicochemical properties of medium after composites incubation

Following composites incubation, the physicochemical properties of the medium, including pH, conductivity, and redox potential, were assessed. A SevenMulti multifunctional ion conductometer (Mettler Toledo, Columbus, OH, USA) with dedicated electrodes was used to measure those parameters. The physicochemical tests were conducted at a temperature of 25 ± 1 °C, with each solution measured five times.

Chemical composition analysis

To quantify the rate of ions release from tested composites, the technique of Inductively Coupled Plasma Mass Spectrometry (ICP-MS) was used, and the research was conducted in accordance with the protocol presented elsewhere [6]. These studies were used to determine iron, neodymium, boron and niobium in experimental samples.

Biological analysis – sample preparation

To prepare materials for biological testing, samples were immersed in 70% ethanol for 15 min, then washed twice with sterile water and stored at room temperature in separate wells of culture plates. On the day of the experiment, samples were briefly rinsed with 70% ethanol, followed by sterile Phosphate-buffered saline (PBS, Gibco line, ThermoFisher Scientific, Waltham, Massachusetts, USA), and UV sterilized for 15 min on each side. To prepare biomaterial extracts, 50 mL of PBS or complete cell culture medium was added to each biomaterial sample and incubated for 24 h. The composition of cell culture medium was as follows: Dulbecco’s Modified Eagle Medium (DMEM) from ATCC (American Type Culture Collection, Manassas, Virginia, USA) supplemented with 10% Fetal Bovine Serum (FBS, from ATCC) and 1% PSA (Antibiotic Antimycotic 100x consisting of Penicillin-Streptomycin-Amphotericin B Suspension, from Sigma Aldrich, St. Louis, Missouri, USA). The medium was then collected and stored at -20 °C for further analysis.

Evaluation of the formation of bacterial and fungal biofilms

Laboratory strains of Enterococcus faecalis ATCC 45477 and Candida albicans 1408, purchased from Pol-Aura (Morąg, Poland) and Polish Collection of Microorganisms (Wrocław, Poland) were used to evaluate biofilm formation on the material surfaces. For this purpose, bacterial and fungal inoculums of OD ~ 1 for bacteria and OD ~ 2 for fungi were prepared from cultures in the logarithmic growth phase, diluted 1:10 in sterile TSB broth (Becton, Dickinson and Company, Franklin Lakes, New Jersey, USA). The inoculum (1000µL) was then added to culture plates with the prepared material and incubated at 37 °C on a rotor (140 rpm) for 24, 48, and 72 h. In parallel, 100 µL of bacterial/fungal inoculum was added to 96-well plates as control samples.

At designated time points, material samples were carefully rinsed with PBS and transferred to fresh culture plates. To quantify biofilm on the material surface, 500 µL of resazurin solution (Sigma Aldrich, St. Louis, Missouri, USA) at a final concentration of 200 µg/mL was added. Resazurin solution was also added to a 96-well plate with control inoculum (after removing planktonic cells). After incubation at 37 °C, 100 µL of the solution from the material samples was transferred to a 96-well plate, and fluorescence was measured at 520/590 nm. In another experimental setting, sample-derived extracts in PBS were added to the 96-well plate and adjusted to 100µL with PBS and with bacteria and fungi inoculum in TSB broth to the final concentrations of 1, 10, 25, and 50%. Upon 72-hour incubation, planktonic cells were removed and biofilm attached to the surface of the plate was washed with sterile PBS. Next, 100µL of resazurin solution (200 µg/mL) was added and left to incubate at 37 °C prior to fluorescence intensity measurement at 520/590nm wavelength. The wells without extracts added were considered as controls (biofilm formation of 100%).

To visualize biofilm formation on composite surfaces, samples incubated with bacterial or fungal inoculum for 72 h were transferred to new culture plates and stained for 15 min with the LIVE/DEAD^® BacLight™ Bacterial Viability Kit (ThermoFisher Scientific, Waltham, Massachusetts, USA) (final concentrations: SYTO9 at 5 µM and propidium iodide at 50 µM). Samples were then rinsed three times in fresh PBS to remove excess dye and any residual planktonic cells, and images were captured using a fluorescence microscope.

Assessment of the antimicrobial properties

To evaluate whether changes in microbiological biofilm formation were due to substances released into the extracellular environment, suspensions of E. faecalis and C. albicans were incubated with extracts from the biomaterials at concentrations of 1%, 10%, 25%, and 50% of the original extract for 72 h. Biofilm formation was then assessed using a resazurin-based fluorometric method.

Assessment of the hemocompatibility

To assess the direct hemocompatibility of the tested materials, purified composite samples were placed in wells of 6-well plates, and 4 mL of a PBS-diluted erythrocyte suspension (hematocrit ~ 5%) was added. For the negative control, the erythrocyte suspension was added to an empty well, while for the positive control, Triton-X100 (Sigma Aldrich, St. Louis, Missouri, USA) was added to the erythrocyte suspension to reach a final concentration of ~ 5%. Incubation was carried out for 1, 6, and 24 h at 37 °C with 5% CO₂. At each time point, 200 µL of erythrocyte suspension was transferred to a 96-well plate, centrifuged (2500 rpm, 5 min, RT), and 100 µL of the supernatant was collected for absorbance measurement at 540 nm.

For indirect hemocompatibility analysis, material extracts (at concentrations of 1%, 10%, 25%, and 50%) and 100 µL of erythrocyte suspension were added to 96-well plates, achieving a final hematocrit of ~ 5%. Hemoglobin release was measured at specified time points using a colorimetric assay.

The study was conducted with the approval of the Bioethics Committee of the Medical University of Bialystok (project title: “Evaluation of the Hemocompatibility of Modified Silicone Composites for Biomedical Applications”—approval number: APK.002.446.2023, dated 23/11/2023).

Evaluation of cytocompatibility

The toxicity assessment of the silicone composites was conducted using NIH/3T3 murine fibroblasts (ATCC CRL-1658™) purchased from ATCC (American Type Culture Collection, Manassas, Virginia, USA). For the experiment, cells (passages between 22 and 25) were maintained in a growth medium (DMEM + 10% FBS + 1% PSA) and seeded onto the sterilized material samples at a density of 20,000 cells per sample. Cells plated on culture wells without materials served as controls. Samples were incubated at 37 °C with 5% CO₂ for 24, 48, and 72 h. At each time point, the growth medium was removed, and samples were rinsed with sterile PBS. Cells were fixed in 3.7% buffered formalin, followed by staining of cell nuclei with Hoechst 33,342 (ThermoFisher Scientific, Waltham, Massachusetts, USA) (1 µg/mL, 5 min, RT) and the cytoskeleton with Alexa Fluor 488-labeled phalloidin (ThermoFisher Scientific, Waltham, Massachusetts, USA) (1 U/mL, 30 min, RT). Imaging was performed with a Leica DMi8 fluorescence microscope and quantification of stained nuclei was performed using ImageJ software. Due to necessity of sample fixation, in each time points different samples of materials were used for cellular staining and quantification.

Statistical analysis

Statistical analysis was conducted using Statistica software (TIBCO Software Inc., USA). The results are presented as the mean ± SD.

Author contributions

Conceptualization: J.M., D.Ł., A.Cz., R.B., A.M.; methodology and investigation: J.M., D.Ł., A.Cz., E.P., P.D.; writing—original draft preparation: J.M., D.Ł., A.Cz.; writing—review and editing: J.M., D.Ł., A.Cz., E.P., P.D., R.B., D.P., J.A., A.M.; visualization, D.Ł., P.D., E.P.; supervision, project administration and funding acquisition: J.M. All authors have read and agreed to the published version of the manuscript.

Funding

This research was financed by the Ministry of Science and Higher Education (Poland) under the Regional Excellence Initiative 2024–2027, grant title “Actively controlled elastic magnetic band for lymphatic drainage”, grant no MRID/WM/5/2024 (project manager: Assoc. Prof. Joanna Mystkowska), and by the Ministry of Science and Higher Education (Poland) through the project “A system for rheological measurements and observation of biological samples in a magnetic field,” contract no. 7135/IA/SP/2020 (project manager: Prof. Robert Bucki).

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.