Hygromachines: Humidity-Powered Wheels,Seesaws,and Vehicles

Abstract

Hygroscopic soft actuators offer an attractive means to convert environmental energy to mechanical motions as they use water vapor, a ubiquitous substance in the atmosphere. To overcome the limits of existing hygroactuators, such as simplistic actuation mode, slow response, and low efficiency, here we present three kinds of humidity-powered soft machines adopting directionally electrospun hygroresponsive nanofibrous sheets. The wheels, seesaws, and vehicles developed in this work utilize spatial humidity gradient naturally established near moist surfaces such as human skin, so that they operate spontaneously, realizing energy scavenging or harvesting. We also constructed a theoretical framework to mechanically analyze their dynamics, which allowed us to optimize their design to obtain the highest motion speed physically possible.

Introduction

Transport of water in soft media can induce volume change and consequently power movements in artificial (e.g., hydrogels) and natural (e.g., plants) actuators. In particular, water vapor, which is ubiquitous in the atmosphere, is utilized by many plants to drive self-propulsion,^1–3 self-sealing,⁴ and seed dispersal.^5,6 Those natural actuators commonly adopt a bilayer configuration, where one layer is hygroscopically active, while the other layer is inactive. For example, seed awns of wild wheat, Erodium and Pelargonium species, exploit hygroresponsivity of polymers such as lignin and hemicellulose in the walls of cells that constitute the active layer. On the contrary, the walls of cells serving as an inactive layer are tightly bound by inextensible cellulose microfibrils.^7–10

Although the hygroscopic actuators in nature have served their functions for survival and reproduction over millions of years,^11,12 several issues remain to be solved before the biological principle can be translated to practical applications including soft robotics and energy harvesters. First, most of the botanical actuators rely on changes of humidity over a long timescale as days or even seasons. Thus, the spatial, as well as temporal, gradient of humidity needs to be exploited as a practically reliable energy source, which should drive actuators at a feasible level of frequency (of the order of 0.1–1 Hz). Second, various modes of actuation beyond what are exhibited by plants should be devised using the spatial gradient of humidity. Third, the design of actuation system should be optimized to maximize the efficiency.

Spatial gradient of humidity is naturally established near moist surfaces, including human skin, wet soil, shores, and kitchen and laboratory benches, where the high humidity level drops down to the atmospheric dry level over a distance on the order of 1 cm.¹³ To exploit the spatial gradient, we need to either make hygroscopic actuators continually move between the regions of high and low humidity or devise a way to continually change the humidity level around the actuators. The former scheme was used by a so-called hygrobot, whose bilayer actuator crosses the humidity boundary layer by bending–unbending cycles.¹⁴

The latter scheme was realized by deploying a great number of actuators in parallel to scale up the actuation force to drive the mechanical system.¹⁵ Scaling up through increasing the number of actuators, however, grows the size and mass of the machines, which may prevent them from utilizing humidity gradient established only within ∼1 cm from natural moist surfaces. Despite those early attempts, diverse design options to exploit the naturally established spatial humidity gradient are wanted to cope with a range of practical situations.

Just as most of the mechanical systems that harvest or scavenge environmental energy, the power conversion efficiency and energy density play a critical role for the success of hygroresponsive actuators. Although many previous researches on soft actuation with environmental humidity focused on the development of novel materials^16–21 and elucidation of mechanical principles,⁷ design optimization through rational analysis has been rare.^22–25 Seeking the optimized machine design is important because it enables not only a dramatic improvement in energy conversion performance, but also exploitation of a naturally occurring humidity gradient on wet surfaces without unnecessary enlargement of the system.¹⁴

To address these issues, here we introduce three types of machines—hygrowheels, hygroseesaws, and hygrovehicles—that are spontaneously driven by spatial humidity gradient without controlled external stimuli. We mechanically analyze their dynamics, and optimize their design to ensure sufficient force to drive centimetric machines. We start with explaining our bilayer actuator, an engine that drives our machines. We term all of these machines hygromachines, for they commonly convert the chemical potential energy of environmental humidity to mechanical work.

Materials and Methods

Fabrication of hygrowheels

The rim and struts of hygrowheel were made of 100-μm-thick polyvinyl chloride (PVC) film, which was cut by a film cutting machine (Silhouette portrait). To prevent bilayers from directly contacting the substrate, we attached polyethylene terephthalate films to the free end of bilayers. The wet substrate was realized by damped filter paper (GB005 Blotting Paper; Whatman). The width of bilayer was 8 mm and the inactive layer comprises a 35-μm-thick polyimide (PI) film and a 25-μm-thick silicone adhesive.

Fabrication of hygroseesaws

The transparent rod of hygroseeaw was made of 200-μm-thick PVC film. The central axis and fixed rods were made of acrylic shaft of 2-mm diameter. The sliders were made of white V4 resin using a stereolithography (SLA)-type 3D printer (Formlabs; Form 2). To reduce friction of sliders, we lubricated the contact surface between machine parts with a lubricant (WD-40 Multi-Use Product). The electromagnetic generator consisted of a circular magnet plate coiled with copper wire. The magnet used was neodymium magnet 10 mm in diameter and 3 mm in height, which has a shaft at the center. Copper coils were formed by winding a total of 300 m of 40-gauge magnet wire.

To connect the axes of seesaw and generator, we used plastic gear and pinion, which have 40 and 8 teeth, respectively. The width of bilayer was 24 mm and the inactive layer comprises a 50-μm-thick PI film and a 30-μm-thick silicone adhesive.

Fabrication of hygrovehicles

The transparent chamber, shutter, and wheel of hygrovehicle were made of 200-μm-thick PVC films. The gears, axles, and pins were made of white V4 resin using an SLA-type 3D printer (Formlabs; Form 2). To reduce friction of shutter and cylindrical gears, we lubricated the contact surface with a lubricant (WD-40 Multi-Use Product). Two pairs of magnets were made of magnet paper (Canon magnetic photographic paper MG-101). The wet floor was realized by damped filter paper (GB005 Blotting Paper; Whatman). The radius of wheel was 15 mm and the distance between the bottom window and wet floor was maintained at 2 mm. The width of bilayer was 32 mm and the inactive layer comprises a 50-μm-thick PI film and a 30-μm-thick silicone adhesive.

Results

Hygroengines: Hygroresponsive bilayer actuators

Machines powered by environmental humidity variation need an engine to convert the surrounding chemical potential energy into the mechanical work, and a mechanical structure to translate the cyclic motion of engine into a desired motion. We adopt a phytomimetic bilayer actuator as an engine of our machines. The active layer that expands via absorbing environmental humidity is fabricated by electrospinning nanofibers of polyethylene oxide (PEO). PEO was selected as the humidity-responsive material due to its superior hygroexpansivity, low toxicity, and ease of electrospinning.²⁶ As shown in Figure 1A, a 10 wt% aqueous PEO solution is pumped through multiple metallic nozzles, at whose tips nanojets are ejected onto a rotating drum collector under a high electrical field (1.5 kV/mm). By tuning the spooling rate to the linear speed of the jets, we get a sheet of directionally aligned nanofibers, whose scanning electron microscopy image is shown in Figure 1B.

FIG. 1.

Hygroengine made of a hygroscopically responsive bilayer. (A) A schematic of directional electrospinning process to make a wide nanofibrous sheet. (B) A schematic of the bilayer and the coordinate system. The SEM image shows the active layer consisting of PEO nanofibers aligned in one direction. (C) Experimental image of a bilayer actuator going through a bending cycle with variation of environmental humidity. The thicknesses of active and inactive layer are identically 50 μm, and the length from the clamped to free end is 20 mm. PEO, polyethylene oxide; SEM, scanning electron microscopy.

The aligned microstructure in the active layer, as inspired by seed awns of Erodium¹ and Pelargonium² species, is known to enhance both the actuation speed and magnitude.¹⁴ By moving the nozzles parallel to the drum axis using a linear stage, as shown in Figure 1A, we can obtain a wide fibrous sheet (5 cm in maximum width). The active layer thickness can be varied by adjusting the jetting duration. The empirically measured properties of active PEO layer are listed in Supplementary Table S1. A humidity-insensitive PI film is attached to the hygroscopically responsive PEO layer to result in a bilayer structure (Fig. 1B). The bonding of the active and inactive layers is achieved via a 25-μm-thick silicone adhesive film having an adhesive shear strength of 0.5 MPa.

When a bilayer is exposed to humid air, the water vapor diffuses into the active layer and the water concentration ϕ(z,t) follows the one-dimensional unsteady diffusion equation: ∂ϕ/∂t = D∂²ϕ/∂z² with D being the vapor diffusivity in the porous fibrous layer. In our previous research, we empirically determined the water vapor diffusivity in the directionally electrospun PEO sheet.¹⁴ With the initial uniform water concentration ϕ₀, the initial condition is given by ϕ(z,0) = ϕ₀. Under the atmospheric humidity of ϕ_∞, the boundary conditions are given by ϕ(0,t) = ϕ_∞ and ∂ϕ(h,t)/∂z = 0, where we considered no moisture flux at the interface of active and inactive layers (z = h). As the bilayers situated in our hygromachines do not cross the humidity boundary layer, we assume that the humidity surrounding the active layer is invariant (Supplementary Text S6 and Supplementary Fig. S5).

Solving the equation gives the spatiotemporal evolution of the humidity in the active layer, which allows us to find the strain tensor: ɛ (z,t) = ɛ ₀ − ζ κ − ɛ _h. Here, ɛ ₀ is the reference strain in the reference plane, ζ is the distance from the reference plane, κ is the bending curvature, and the hygroscopic strain is ɛ _h = αϕ with α being the hygroscopic expansion coefficient vector, or strain increase per 1% of relative humidity (RH) increase at room temperature. Integrating the local stress, σ = $\bar{D}$ ɛ , over the bilayer thickness, we get the resultant force N = ∫ σ dζ and the bending moment M = ∫ σ ζdζ, where $\bar{D}$ is the stiffness matrix. In our modeling, we balance the force due to internal stress of composite laminate and external force the actuator needs to overcome to move the machine.

The equilibrium equations for force and moment lead us to predict the bilayer curvature as a function of time t and external loads, the detailed procedure explained in Supplementary Text S1. Figure 1C shows the experimental images of a bilayer actuator clamped at one end, which is initially uncurved as fabricated at ϕ_∞ = 0.4, bends downward when exposed to humid air (ϕ_∞ = 0.8), and upward when dry (ϕ_∞ = 0.2).

Hygrowheels

A hygrowheel is a circular rim to which multiple bilayer actuators and their guiding struts are attached, as shown in Figure 2A. The wheel is placed on a filter paper under which is a bath of 40°C warm water. As shown in the upper right panel of Figure 2A, the high humidity near the substrate drops down to the atmospheric value over ∼2 cm, as measured by a humidity sensor (Sensirion SHT85). The bilayers are curved close to the rim when dry, but bend away from the rim when wet. The rolling sequence is depicted in Figure 2B, which starts when a strut A entering the humidity boundary layer touches the substrate. The bilayer actuator at B bends to lift and rotate the entire wheel with point A as a pivot.

FIG. 2.

Hygrowheels. (A) Schematics of a hygrowheel, and the experimentally measured humidity distribution over the moist substrate. (B) The rolling sequence of hygrowheel. (C) Experimental images of a hygrowheel climbing 12° uphill. (D) Contour map of the hygrowheel's translational velocity as a function of the active layer thickness and the tilt angle. The rim diameter is 25 mm, and the bilayer is 8 mm long with its inactive layer 35 μm thick. (E) Speed of hygrowheel with the active layer 40 μm thick as a function of the tilt angle. (F) Speed of hygrowheel on a zero tilt substrate as a function of the active layer thickness. In (E, F), red lines are theoretical predictions, and blue circles correspond to experimental results with the error bar being the standard deviation of 5 measurements.

When the center of mass of the wheel crosses the vertical pivot line (blue dashed line), a unit step of rolling is completed. Because dry bilayers keep entering the humid region with wheel rolling, our hygrowheels can be continually driven by the spatial humidity gradient (Supplementary Video S1). The wheel can even run uphill as long as the bilayer bending can produce sufficient force to rotate the wheel beyond the pivot line, as shown in Figure 2C.

We can calculate the bending rate of bilayer within the humidity boundary layer and the time taken by the wheel's center of mass to cross the pivot line, τ_w, as discussed in Supplementary Text S2 and Supplementary Figure S1. Since the wheel rotates an angle of α_w = 2π/n, when the pivot line is crossed, the linear translation speed of the wheel is given by U_w = 2l_wsin(α_w/2)/τ_w, where n is the number of struts and l_w the distance of pivot from the center of wheel. We can find the maximum tilt angle of substrate the wheel can climb (Supplementary Text S2 and Supplementary Fig. S1). The wheel speed is a function of the substrate tilt angle and the geometry of bilayer and wheel. We plot the dependence of wheel translational speed on the thickness of active layer and the tilt angle in Figure 2D.

We compare our theoretical predictions with the experimental data at a fixed active layer thickness and at a zero tilt angle in Figure 2E and F, respectively, to verify our model. Figure 2E shows that the speed decreases with the increase of tilt angle because the horizontal distance from the wheel center to the pivot line increases. Figure 2F shows that there exists an optimal thickness of active layer that maximizes the wheel speed. If the active layer is too thin (thinner than 17.4 μm as calculated in this plot), it cannot generate enough force to push the wheel to rotate. Too thick an active layer needs too long a time for vapor to diffuse into the soft medium, slowing its response.

The theoretical maximum speed (26 mm/s) at 28 μm thickness is 4.4 times higher than at 55 μm thickness, clearly demonstrating the importance of mechanical analysis for efficient design of hygromachines.

Hygroseesaws

A hygroseesaw is a rod with the axis of rotation at the center and two hygropistons loaded at two ends, as depicted in Figure 3A. The hygropiston consists of a bilayer actuator, one end of which is fixed to the rod tip, while the other end is attached to a cylinder of mass m_s capable of sliding along a slot in the rod. The hygropiston pushes the cylindrical slider as the bilayer is straightened out under high humidity, but pulls the slider as the bilayer bends with drying. Figure 3B schematically shows the working principle of a hygroseesaw along with the corresponding experimental images.

FIG. 3.

Hygroseesaws. (A) Schematics of hygroseesaw and hygropiston. (B) Working principle of hygroseesaw with experimental images. (C) Contour map of the hygroseesaw frequency as a function of the active layer thickness and the slider mass. The bilayer length is 36 mm when straight and its inactive layer is 50 μm thick. The seesaw rod is 134 mm long. (D) Seesawing frequency with the active layer 55 μm thick as a function of the slider mass. (E) Seesawing frequency with the slider mass 0.39 g as a function of the active layer thickness. In (D, E), red lines are theoretical predictions, and blue circles correspond to experimental results with the error bar being the standard deviation of 7 measurements. (F) Image of a hygroseesaw connected to an electromagnetic generator. The rotation axis is 50 mm above a warm water surface. Insets, two LEDs that blink alternately by the AC current from the generator. (G) Voltage measured across a load resistor of 10 kΩ. (H) Voltage measured across an RC circuit with bridge rectifier, which has a peak duration of 2 s. In (G, H), the active layer thickness is 55 μm and the slider mass is 1.9 g. Scale bars in (B, F), 2 cm. AC, alternating current; LED, light emitting diode; RC circuit, resistor-capacitor circuit.

In the experiment, a transparent chamber houses vapor supplied from the underlying water bath. At stage 1, the hygropiston at the left end, which has just entered the humid chamber, pushes the slider closer to the center of rod. At the right end in the low humidity region, the hygropiston pulls the slider away from the center of rod. When the mass of a bilayer is m_b and the chord length difference between two bilayers is Δl, the resulting torque imbalance is calculated as ΔT = (m_s + m_b/2)gcos θΔl with θ being the tilt angle of the seesaw, and g the gravitational acceleration. Because of intrinsic friction at the rotation axis, ΔT should exceed a critical value ΔT_c determined by the friction at the rotation axis and the seesaw weight, to rotate the seesaw.

Through an intermediate rotation stage 2, which takes much shorter than stage 1, the seesaw enters stage 3 where the hygropiston at the right end is under high humidity while it is dry at the left end. Now the right piston pushes and the left piston pulls their own slider, causing the seesaw to rotate in the opposite sense in stage 4 and to return to stage 1. In this manner, the seesaw can continually rotate reciprocally as driven by the spatial gradient of humidity (Supplementary Video S2).

By computing the sliding distances of both the hygropistons as a function of time, we can find the time taken for ΔT to exceed ΔT_c, which corresponds to a half period of seesawing cycle (Supplementary Text S3 and Supplementary Fig. S2). The period is a function of the active layer thickness h_a and the mass of slider m_s, once the bilayer length, its inactive layer thickness, and the rod length are given. We plot the seesawing frequency as a function of h_a and m_s in Figure 3C. We compare our theoretical predictions with the experimental data at fixed h_a and m_s in Figure 3D and E, respectively, to verify our model. Figure 3D shows that the frequency increases with the slider mass because larger m_s facilitates the tipping of seesaw by breaking the torque balance more easily.

However, m_s cannot be increased indefinitely because too heavy a slider overstretches or overcurls the hygropiston due to gravitational effects, which was observed when m_s exceeded 1 g in our experiments. Figure 3E shows that there exists an optimal thickness of active layer that maximizes the seesawing frequency. Just as in the hygrowheel, too thin an active layer cannot generate enough force to move the hygropiston overcoming the slider weight, and too thick a layer needs excessively long time for vapor to diffuse into the medium. The theoretical maximum frequency (0.076 Hz) at 56 μm thickness is 1.8 times higher than at 100 μm thickness.

Our hygroseesaw can continually convert the chemical potential energy associated with spatially varying environmental humidity into electrical energy by connecting a simple electromagnetic generator, as shown in Figure 3F and Supplementary Video S3. The axes of seesaw and generator are linked with gears of 5:1 ratio to increase the rotating angle and angular velocity of a magnet in the generator. We used different electric circuits to measure the instantaneous voltage induced on a load resistor, to rectify the alternating current to direct current, and to power two light emitting diodes (LEDs), with details given in Supplementary Figure S3.

The peak voltage across a load resistance over 10 kΩ was measured to be 3.0 V as shown in Figure 3G, but the maximum power was measured to be 2.5 mW with a load resistance of 500 Ω. The rectified output had a duration of 2 s, Figure 3H. The LEDs blinking according to the seesaw rotation are indicated in Figure 3F.

Hygrovehicles

A hygrovehicle consists of two pairs of wheels that carry an engine bay between them, as schematically shown in Figure 4A. The linear reciprocal motion of hygropiston within the bay is converted to rotation of an axle via a one-way rotating gear. Two pins moving with the hygropiston slide back and forth on spiral grooves of cylindrical gears, causing the gears to rotate in one way (Supplementary Video S4). To harness the spatial gradient of humidity while keeping the hygroresponsive bilayer's relative vertical distance from the underlying moist surface, the humidity around the hygropiston is periodically varied by using a hygrochamber, which is operated as depicted in Figure 4B and Supplementary Video S5. The chamber has a shutter that can simultaneously open the top window and close the bottom window, or vice versa, when a critical force from the bilayer is applied.

FIG. 4.

Hygrovehicles. (A) A schematic of hygrovehicle (left) and its working principle (right). (B) Working principle of hygrochamber with experimental images and RH-stroke (Δx) diagrams. (C) Movement of hygrovehicle on a wet filter paper for one cycle of hygrochamber. (D) Contour map of the hygrochamber frequency as a function of the active layer thickness and the magnetic force. The bilayer length is 43 mm when straight and its inactive layer is 50 μm thick. (E) Hygrochamber frequency with the active layer 36 μm thick as a function of the magnetic force. (F) Hygrochamber frequency with the magnetic force 5.3 mN as a function of the active layer thickness. In (E, F), red lines are theoretical predictions, and blue circles correspond to experimental results with the error bar being the standard deviation of 15 measurements. Scale bars in (B, C), 1 cm. RH, relative humidity.

Stage 1 starts as soon as the shutter has opened the bottom window and closed the top window. Then the humid air enters the chamber through the bottom window, increasing the interior humidity. Although the active layer in the bent bilayer absorbs water vapor in this stage, the bilayer cannot unbend until its force on the shutter gets strong enough to overcome the resisting force provided by a pair of magnet 1 fixed to the chamber bottom and magnet 2 on the shutter. Upon the force from the bilayer exceeding the magnetic force F_m, the released shutter quickly opens the top window and closes the bottom window, which corresponds to stage 2.

At stage 3, the dry air above enters the chamber through the open top window, drying the active layer while the shutter is locked by a pair of magnet 3 on the chamber bottom and magnet 4 on the shutter. When the force produced by the dried bilayer exceeds the magnetic force, identical to F_m above, the shutter is released to close the top window and open the bottom window, corresponding to stage 4. We plot the diagram of RH in chamber versus the hygropiston stroke, which completes a cycle, in Figure 4B.

In stages 1 and 3, we can calculate the time taken for the bilayer to overcome the magnetic force F_m and the curvature change at the end of each stage by combining the vapor diffusion and elasticity models, as given in Supplementary Text S1. In the model, we neglect the time taken for the vapor to fill the initially dry chamber or escape from the humid chamber because it is much shorter than the time for diffusion in the active layer (Supplementary Text S5 and Supplementary Fig. S4). Furthermore, stages 2 and 4 are much shorter than the other stages, and thus, a cycle approximately takes a sum of durations of stages 1 and 3.

The distance the hygrovehicle travels through a hygrochamber cycle, as shown in Figure 4C, is given by L = Rβ, where R is the radius of the wheel and β is the rotation angle of the axle per cycle. The average speed of the vehicle is $\bar{u}$ = Lf_c, where f_c is the shutter opening frequency of hygrochamber.

Because the vehicle's linear speed can be varied by altering the gear size and wheel radius, we rather plot the hygrochamber frequency as a function of the active layer thickness and F_m in Figure 4D. We compare our theoretical predictions with the experimental data at fixed h_a and F_m in Figure 4E and F, respectively, to verify our model. Figure 4E shows that the frequency decreases with the magnetic force because a stronger magnetic resisting force requires a bilayer to absorb vapor longer to build up enough force to overcome F_m. Figure 4F implies that a thicker active layer helps to move the shutter by easily overcoming the magnetic forces. However, too thick an active layer tended to limit the hygropiston stroke by reducing the amount of curvature change.

In our model, active layers thicker than 50 μm failed to fully open and close the shutters although the magnetic forces were overcome.

Discussion

We have designed, fabricated, and demonstrated three machines that can harness the environmental humidity energy to generate mechanical motions. Furthermore, we have analyzed the machine dynamics, to optimize their design and obtain the best performance within the given parameter space. The designs have commonly revolved around the idea that the spatial gradient of humidity should be used to continually drive the machines without human intervention. Using bilayers consisting of hygroscopically active and inactive layers as an engine, our machines exhibit diverse motions, that is, rolling of the entire machine (hygrowheels), reciprocal rotation (hygroseesaws), and linear locomotion by wheels (hygrovehicles). The hygrowheels and hygroseesaws periodically change the distance of the bilayers from the moist substrate to realize repeated cycles of bending and unbending.

The hygrovehicles periodically change the humidity level surrounding the bilayer whose vertical position is kept constant, using a hygrochamber. The hygroseesaws and hygrochambers commonly use the bilayers as a piston that pushes a slider when wet, but pulls when dry. The mechanical analyses of the machine dynamics allow us to optimize the performance of those machines to obtain speeds or rates of reciprocation several times higher than those poorly designed.

Unlike other soft actuators that respond to external stimuli controlled by artificial means, for example, electric^27–29 and magnetic fields,^30,31 light intensity,^32,33 temperature,³⁴ and pH,³⁵ our hygromachines work spontaneously under the spatial humidity gradient. From the thermodynamic point of view, our machines produce work as the chemical potential energy flows from the humid down to dry regions, just as heat engines produce work as the thermal energy is transported from a hot to cold reservoir. If the machines work in the spatial gradient of humidity naturally established around human skin, wet soil, shores, and kitchen or laboratory benches, they can be regarded as energy harvesters or scavengers, as exemplified by our hygroseesaws lightening up LEDs. Environmentally sustainable and scalable energy generation can be achieved through the utilization of multiple hygroseesaws; batteries can be used to store the generated power.

A variety of machines are possible by connecting shafts, gears, or cargos to our hygromachines for a range of purposes including those attempted in this work. In particular, these machines can be utilized to explore small inaccessible spaces without the risk of energy depletion.

In Figure 5, we compare the speed of the hygromachines with those of the autonomous and spontaneous robots,^14,15,36 as well as robots that are powered by controlled external stimuli.^{14,23,24,37–43} The hygrowheel demonstrates the highest speed (in body length/min) among the autonomous robots, and the hygrovehicle registers a speed that is comparable with the other robots. When the hygroseesaw is used for electric power generation, it produces a voltage peak of 3 V and power peak of 2.5 mW. Compared with the power generation device of Chen et al.,¹⁵ the seesaw's power peak is 50 times higher and the voltage peaks are comparable.

FIG. 5.

Velocity of various soft robots measured in BL per minute versus BL. The red data points represent autonomous robots; robots powered by external stimuli are represented in other colors. BL, body length.

Despite the advantages in exploiting the ubiquitous energy source of humidity in the atmosphere, there are limitations to be considered in designing hygromachines. First, the actuation speed and the mechanical stiffness of hygroresponsive materials can hardly go together because higher porosity, desirable to expedite vapor diffusion, tends to lower the stiffness. Increasing the hygroscopic response speed of recently reported strong hygroresponsive materials^16–19 through porosity control^44,45 may resolve this problem. Second, the spatial gradient of humidity is established over a distance on the order of centimeter, which may restrict the size of the actuators. We have shown how to make larger machines using centimetric actuators with hygrowheels and hygroseesaws, promoting more ideas for other purposes. Third, the durability of polymeric actuators can be an issue. It was reported that the bilayer actuators made of PEO can endure 10⁴ cycles of RH variation between 0.2 and 0.8 at 25°C.¹⁴

One way to address this duration issue is by using cheap, disposable, and environmentally benign polymeric materials and fabrication processes. Overall, the future studies to widen the application area of hygroscopic actuation would commonly benefit from our theoretical framework to analyze and optimize the machines driven by hygroresponsive engines.

Footnotes

Authors' Contributions

H.-Y.K. conceived and supervised the research. M.C. and B.S. designed and fabricated the hygromachines, analyzed the experimental data, constructed theoretical models, and performed numerical computations. All the authors discussed the results and prepared the article and figures.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This work was supported by the National Research Foundation of Korea (NRF) grants funded by the Korean Government (MSIT) (Grant Nos. NRF2018R1A3B1052541 and NRF2021M3F7A1017476). H.-Y.K. acknowledges administrative support from SOFT Foundry Institute.

Supplementary Material

References

Evangelista

, Hotton

, Dumais

. The mechanics of explosive dispersal and self-burial in the seeds of the filaree, Erodium cicutarium (Geraniaceae). J Exp Biol, 2011; 214:521–529.

Jung

, Kim

H-Y

. Self-burial mechanics of hygroscopically responsive awns. Integr Comp Biol, 2014; 54:1034–1042.

Elbaum

, Zaltzman

, Burgert

, et al. The role of wheat awns in the seed dispersal unit. Science, 2007; 316:884–886.

Katifori

, Alben

, Cerda

, et al. Foldable structures and the natural design of pollen grains. Proc Natl Acad Sci USA, 2010; 107:7635–7639.

. How pine cones open. Nature, 1997; 390:668.

Armon

, Efrati

, Kupferman

, et al. Geometry and mechanics in the opening of chiral seed pods. Science, 2011; 333:1726–1730.

, Choi

, Shin

, et al. Hygroresponsive coiling of seed awns and soft actuators. Extreme Mech Lett, 2020; 38:100746.

Rafsanjani

, Stiefel

, Jefimovs

, et al. Hygroscopic swelling and shrinkage of latewood cell wall micropillars reveal ultrastructural anisotropy. J R Soc Interface, 2014; 11:20140126.

Abraham

, Tamburu

, Klein

, et al. Tilted cellulose arrangement as a novel mechanism for hygroscopic coiling in the stork's bill awn. J R Soc Interface, 2012; 9:640–647.

10.

Burgert

, Fratzl

. Actuation systems in plants as prototypes for bioinspired devices. Philos Trans R Soc A, 2009; 367:1541–1557.

11.

Dumasis

, Forterre

. “Vegetable Dynamicks”: The role of water in plant movements. Annu Rev Fluid Mech, 2012; 44:453–478.

12.

Forterre

. Slow, fast and furious: Understanding the physics of plant movements. J Exp Bot, 2013; 64:4745–4760.

. Evaporative microclimate driven hygrometers and hygromotors. Europhys Lett, 2014; 107:64002.

, et al. Hygrobot: A self-locomotive ratcheted actuator powered by environmental humidity. Sci Robot, 2018; 3:eaar2629.

, et al. Scaling up nanoscale water-driven energy conversion into evaporation-driven engines and generators. Nat Commun, 2015; 6:7346.

16.

Rüggeberg

, Burgert

. Bio-inspired wooden actuators for large scale applications. PLoS One, 2015; 10:e0120718.

. Hygromorphic materials for sustainable responsive architecture. Constr Build Mater, 2015; 98:570–582.

, et al. Analysis of hygroscopic self-shaping wood at large scale for curved mass timber structures. Sci Adv, 2019; 5:eaax1311.

19.

, Guo

, Anderson

, et al. Bio-inspired polymer composite actuator and generator driven by water gradients. Science, 2013; 339:186–189.

, et al. An autonomous actuator driven by fluctuations in ambient humidity. Nat Mater, 2016; 15:1084–1089.

, et al. Actuation and locomotion driven by moisture in paper made with natural pollen. Proc Natl Acad Sci USA, 2020; 117:8711–8718.

, et al. Multiresponsive actuators based on modified electrospun films. RSC Adv, 2018; 8:10302.

, et al. Facile fabrication of moisture responsive graphene actuators by moderate flash reduction of graphene oxides films. Opt Mater Express, 2017; 7:2617–2625.

, et al. 3D printing graphene oxide soft robotics. ACS Nano, 2022; 16:3664–3673.

25.

, Wang

, Xue

, et al. Programmable humidity-responsive actuation of polymer films enabled by combining shape memory property and surface-tunable hygroscopicity. ACS Appl Mater Interfaces, 2021; 13:38773–38782.

, et al. The effect of processing variables on the morphology of electrospun nanofibers and textiles. Polymer, 2001; 42:261–272.

. Agile reversible shape-morphing of particle rafts. Soft Matter, 2021; 17:7554–7564.

, et al. Photothermal nanocomposite hydrogel actuator with electric-field-induced gradient and oriented structure. ACS Appl Mater Interfaces, 2018; 10:7688–7692.

, et al. Soft robotic manipulation and locomotion with a 3d printed electroactive hydrogel. ACS Appl Mater Interfaces, 2018; 10:17512–17518.

30.

, Lum

, Mastrangeli

, et al. Small-scale soft-bodied robot with multimodal locomotion. Nature, 2018; 554:81–85.

, et al. Printing ferromagnetic domains for untethered fast-transforming soft materials. Nature, 2018; 558:274–279.

, et al. Photoinduced multi-directional deformation of azobenzene molecular crystals. J Mater Chem C, 2019; 7:503–508.

, et al. Photo-mechanical effects in azobenzene-containing soft materials. Soft Matter, 2007; 3:1249–1261.

, et al. Atomic scale processes of phase transformations in nanocrystalline NiTi shape-memory alloys. Acta Mater, 2017; 123:90–101.

, et al. An aptamer-functionalized chemomechanically modulated biomolecule catch-and-release system. Nat Chem, 2015; 7:447–454.

, et al. Soft phototactic swimmer based on self-sustained hydrogel oscillator. Sci Robot, 2019; 4:eaax7112.

, et al. Flexible and super-sensitive moisture-responsive actuators by dispersing graphene oxide into three-dimensional structures of nanofibers and silver nanowires. ACS Appl Mater Interfaces, 2020; 12:3245–3253.

, et al. Selective decrosslinking in liquid crystal polymer actuators for optical reconfiguration of origami and light-fueled locomotion. Angew Chem, 2019; 16:5386–5391.

, et al. Light-driven bimorph soft actuators: Design, fabrication, and properties. Mater Horiz, 2021; 8:728.

, et al. 4D Printing of seed capsule-inspired hygro-responsive structures via liquid crystal templating-assisted vat photopolymerization. Adv Funct Mater, 2023; 33:2211602.

, et al. Multi-stimuli-responsive actuator based on bilayered thermoplastic film. Soft Matter, 2022; 18:5052–5059.

42.

, Zhang

, Yang

, et al. A bioinspired multilegged soft millirobot that functions in both dry and wet conditions. Nat Commun, 2018; 9:3944.

43.

, Sang

, Liu

, et al. Dual-mode biomimetic soft actuator with electrothermal and magneto-responsive performance. Comp Part B Eng, 2022; 238:109880.

, et al. Novel approach to fabricate porous sponges of poly(D,L-lactic-co-glycolic acid) without the use of organic solvents. Biomaterials, 1996; 17:1417–1422.

, et al. Porous gelatin hydrogels: 1. Cryogenic formation and structure analysis. Biomacromolecules, 2007; 8:331–337.

46.

Christensen

. Mechanics of Composite Materials. Wiley: New York;. 1979.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

4.21 MB

0.93 MB

0.90 MB

0.89 MB

0.90 MB

0.01 MB

7.84 MB

0.01 MB

3.83 MB

8.05 MB

6.47 MB

7.16 MB

6.73 MB

0.00 MB