Creating robotic intelligence using multistimuli-responsive cobalt-doped manganese oxide

The study addresses the need for intelligent, compact, material-driven microrobots capable of complex, feedback-controlled motions without cumbersome external mechatronic integration. Conventional smart actuators often require temperature/humidity stimuli or high-intensity IR light and integrate separate sensors (resistive, capacitive, or opto-electronic), complicating system design. The research proposes cobalt-doped manganese oxide (Co-MnO₂) as a multistimuli-responsive artificial muscle that can be actuated by visible or NIR light and humidity while simultaneously providing an electrical resistance change as an intrinsic feedback signal. The goal is to realize fast, large, low-power actuation with tunable behavior and built-in sensing to enable intelligent functions such as self-adapting load lifting, object sorting, and on-demand structural stiffening in microrobotic devices.

Prior work has explored various smart actuating materials for microrobotics, including systems actuated by temperature, humidity, and light. Light-driven systems offer advantages in wireless triggering and spatially resolved control. Previous sensing-actuating systems often required external resistive, capacitive, or opto-mechanical/electronic sensors, increasing complexity and sometimes lacking simple electrical readouts for feedback control. Recent material-centric approaches have demonstrated multifunctional structures (e.g., origami metallic backbones with resistive heating, strain sensing, antennas) and carbon-based actuators with self-sensing for humidity, thermal, or piezo-induced resistivity changes. However, these typically rely on temperature/humidity changes or strong IR illumination (>1 sun), and often exhibit limited deflection (<180°) or slow response (>10 s). This work fills the gap by providing a visible light-driven, dual-responsive actuator with fast, large deformation and built-in electrical feedback.

Fabrication of actuators: A bilayer architecture was used with an active Co-MnO₂ layer electrodeposited atop a passive Au/Ni supporting layer. A Ni layer and then a Au layer were first electrodeposited (details in Supporting Text S1, Fig. S1). Co-MnO₂ was then electrodeposited on the Au side from mixed baths: solution A (0.15 M CoSO₄, 0.15 M CH₃COONa, 0.15 M Na₂SO₄) and solution B (0.15 M C₄H₆MnO₄·4H₂O, 0.15 M Na₂SO₄) at A:B ratios of 1:5, 1:2, and 2:1. An anodic current density of 0.545 mA/cm² was applied for 15 min under mild stirring. All reagents were reagent grade (Sigma-Aldrich). Activation treatment: Fresh actuators were subjected to cyclic voltammetry (CV) in 1 M NaOH to activate light/humidity actuation. In the Experimental section: −0.2 to 0.5 V at 25 mV/s for five cycles. In Results describing activation for light/humidity response: −0.5 to 0.2 V at 25 mV/s. Activation caused irreversible uncurling toward the Ni support, and post-activation, the actuator expanded (backward curl) at high RH (~95%) or in water and contracted (forward curl) at lower RH (~65%). Actuation tests: Activated actuators were tested at room temperature and RH ~95% under NIR or visible (Vis) light from a xenon arc lamp (Zolix HPS-500XA). Humidity-driven actuation was observed by changing RH. Electrochemical actuation was also tested in 0.5 M Na₂SO₄ via CV. Sensing characterization and feedback control: A self-integrated Au–Ni/Co-MnO₂/Au–Ni sensing-actuating unit was made by attaching a Au–Ni electrode onto a Co-MnO₂ actuator (electrolyte Co:Mn = 5:1). A constant current of 1 µA was supplied (Corrtest CS350), and voltage was recorded to compute resistance and its rate of change. For feedback-controlled devices (e.g., smart load-lifting system; robotic finger with multiple pathways), the resistance of a sensor comprising two facing Co-MnO₂/Au–Ni bilayers was continuously measured. An Arduino calculated the resistance decrease rate |R| and adjusted light intensity via simple proportional control by tuning LED current using a digital pin and a 2SD526 transistor to match a setpoint C_IR. Material characterization: SEM (Hitachi S-4800) and TEM (FEI Tecnai G20 STEM) characterized morphology. FTIR (Bruker Tensor 27) probed bonding/adsorbed species. Contact angles of DI water on Co-MnO₂ were measured (Sindatek 100SB). AFM (Bruker Multi-Mode 8) examined surface features. Surface temperature under illumination was measured by a FLIR SC7700M IR camera. In-plane grazing incidence XRD (Rigaku SmartLab, 0.5° incidence, Cu Kα) and Raman spectroscopy (Spectra Pro HRS-300) assessed phases/crystallinity. XPS (Kratos Axis Ultra, Al Kα) analyzed oxidation states. Temperature-dependent conductivity was evaluated; photothermal heating under illumination was assessed. Data analysis: Actuation magnitude was quantified via curvature κ and angular deflection; actuation/recovery speeds were computed in deg/s. CV data identified ion insertion/extraction processes. FTIR peak intensities and shifts, contact angle changes, XRD/Raman patterns, and XPS valence shifts elucidated activation mechanisms. Thermal and resistance measurements linked illumination-induced heating to resistance changes.

• Multistimuli/multiresponsive behavior: Activated Co-MnO₂ responds to visible and NIR light, humidity, and electrochemical stimuli. Post-activation, the actuator expands at high RH (~95%) or in water and contracts at lower RH (~65%). Humidity-driven actuation achieved ~1260° angular deflection for a 20-mm actuator.
• Built-in self-sensing: Under Vis illumination at ultralow intensity (e.g., 8 mW/cm²), the electrical resistance of Co-MnO₂ decreases, providing a convenient feedback signal. A constant-current circuit (≈1 µA) allowed resistance tracking; illumination on/off produced reversible ΔR/R. The resistance decrease is consistent with a positive temperature coefficient of conductivity and photothermal heating during illumination; direct heating experiments reproduced resistance reductions.
• High actuation performance: For a 20-mm actuator (Co:Mn 1:2), NIR light at 550 mW/cm² produced a curvature up to ~1.75 mm⁻¹ and cumulative angular deflection of ~2000° (≈4.5 loops). Actuation speed reached ~1000°/s with recovery speed ~900°/s, completing one loop in ~100 ms for initial loops under 24 °C and RH 95%.
• Low power visible-light actuation: For Co:Mn = 2:1, actuation was triggered at ~5 mW/cm² Vis light, yielding ~720° tip deflection. At 50 mW/cm² Vis light, curling into one loop occurred within ~0.136 s.
• Tunability via Co doping: Actuation magnitude and responsiveness are tunable by the Co:Mn ratio in the electrodeposition bath (A:B mixtures of Co and Mn precursors), enabling devices with multiple selectable configurational pathways.
• Activation mechanism and hydrophilicity: Activation via CV in 1 M NaOH increased hydrophilicity (contact angle decreased from ~40° unactivated to <5° activated) and water adsorption/intercalation (FTIR peak at ~1630 cm⁻¹ increased; 3200–3400 cm⁻¹ blueshift indicating more free water). XRD/Raman showed no new phases but reduced crystallinity. XPS indicated oxidation from Co(II) to Co(III) and Mn(II)/Mn(III) to Mn(III)/Mn(IV), with Na⁺ incorporation into turbostratic MnO₂ basal planes (Na-birnessite-like), facilitating water uptake. Thermogravimetric evidence suggested more water retained in Co-doped samples than pure MnO₂.
• Electrochemical actuation: CV in 0.5 M Na₂SO₄ showed Na⁺ insertion/extraction-driven volume changes consistent with actuation, without irreversible expansion seen during NaOH activation.
• Intelligent device demonstrations: Using resistance as a feedback signal and tunable actuation, microrobotic devices were built that self-sense Vis light (~4 mW/cm²), perform complex motions along selectable pathways, and demonstrate self-adapting load lifting, object sorting, and on-demand structural stiffening.

The findings demonstrate that Co-MnO₂ can function as an integrated actuator-sensor material, addressing the challenge of compact, intelligent microrobotics without complex external sensor integration. Light (visible/NIR) and humidity stimuli induce rapid, large deformations, while a concurrent decrease in electrical resistance under illumination provides a straightforward electrical feedback variable for closed-loop control. Activation enhances hydrophilicity and water intercalation, enabling humidity/light-driven volume changes, and CV evidence supports Na⁺ insertion/extraction as a key actuation mechanism. Co doping enables tuning of actuation characteristics, which, together with embedded sensing, allows devices to autonomously modulate motion, select among multiple configurational pathways, and adapt mechanical stiffness for tasks such as load lifting and object sorting. The material’s responsiveness at low visible light intensities and high actuation speeds significantly improves upon prior systems that required high IR intensities, offered limited deformation, or had slow response times.

This work introduces cobalt-doped manganese dioxide as a multistimuli-responsive, self-sensing artificial muscle material that delivers fast, large, and energy-efficient actuation while providing an intrinsic resistive feedback signal. Activation via NaOH CV imparts strong humidity and light responsiveness through enhanced hydrophilicity and Na⁺-assisted water intercalation. Co doping tunes actuation behavior, enabling microrobotic devices capable of intelligent functions including feedback-controlled motion along multiple pathways, self-adapting load lifting, object sorting, and on-demand structural stiffening. The results open a pathway toward creating robotic intelligence by leveraging multistimuli-responsive materials. Future work could optimize doping/composition and device architectures for lower power operation in broader environmental conditions, integrate more sophisticated control strategies, and explore scalability and durability for practical applications.

• Environmental dependence: Peak performance and demonstrations are conducted in humidity-rich conditions (e.g., RH ~95%); actuation in drier environments is weaker before activation, and humidity strongly affects behavior.
• Activation requirement: Actuators require an electrochemical activation step in 1 M NaOH with CV, which irreversibly alters the device and may add processing complexity; the paper notes two potential windows (−0.2 to 0.5 V; −0.5 to 0.2 V) associated with activation.
• Light intensity for fastest response: While visible-light actuation occurs at ultralow intensities (~5–8 mW/cm²), the highest speeds/deflections reported use higher NIR intensities (e.g., 550 mW/cm²).
• Material changes: Activation reduces crystallinity and alters oxidation states; long-term stability, cycling durability, and potential degradation under repeated light/humidity cycling are not fully detailed in the provided text.
• Integration details: Although sensing is built-in, external circuits (constant current source, Arduino control, LED driver transistor) are needed; full on-board integration and wireless operation are not described.