Reversible Perspiring Artificial “Fingertips”

Fingertip perspiration is a vital process within human predation, to which the species owes its survival and its biological success. In this paper, the unique human ability of extensive perspiration and controlled friction in self‐assembled cholesteric liquid crystals is recreated, mimicking the natural processes that occur in the dermis and epidermis of human skin. This is achieved by inducing porosity in responsive, liquid‐bearing material through the controlled‐polymerization phase‐separation process. The unique topography of human fingerprints is further emulated in the materials by balancing the parallel chirality‐induced force and the perpendicular substrate‐anchoring force during synthesis. As a result, artificial fingertips are capable of secreting and re‐absorbing liquid upon light illumination. By demonstrating the function of the soft material in a tribological aspect, it exhibits a controllable anti‐sliding property comparable to human fingertips and subsequently attains a higher degree of biomimicry. This biomimetic fingertip is envisioned being applied in a multitude of fields, ranging from biomedical instruments to interactive, human‐like soft robotic devices.


DOI: 10.1002/adma.202209729
manipulation and perception of the world. Our fingertips are essential for gripping objects or distinguishing different surfaces upon contact. [1] It is now understood that perspiration is equally crucial in these contact processes, as the synergy between local sudoriferous glands and fingerprint ridge apices determines the final contact properties. [2] The adaptability and control bestowed by the gripping principles of the human fingertips have inspired the use of photonic, electric, magnetic, and vacuum mechanics for robotic handling. [3] Meanwhile, the topography of human fingerprints has been intensively studied in the light of forensic science, even more so with the advent of the biometric security industry. This motivates accurate imitations of biology to be synthesized with a class of soft materials that exhibit multifunctional capabilities and complementary properties. [4] Previously, we have reported a series of fingerprint topographic coatings made of liquid crystal networks (LCNs). [5] They can be triggered by either light or electricity to alter the surface corrugation. [6] When these coatings are in touch with other objects, the surface contact area or contacting position can be fine-tuned via appropriate stimulus.
Even so, the subtleties of fingertips to be emulated are far more than the fingerprint topography. Yet, to integrate their multiple features into one object is still rather challenging. Here in this paper, we further design a new LCN coating to replicate both the unique fingerprint topography and responsive perspiration behavior observed in human fingertips. The fingerprint patterns on our fingertip coating are obtained by introducing a cholesteric liquid crystal phase. We utilize them to mimic not only the unique, intricately-arranged fingerprint topography but also the liquid secretion at ridge-located sudoriferous glands in human fingertips by involving low-molar-mass liquid crystals (LCs). [7] Then, we bolster our LCN coating with light-responsiveness by the inclusion of azobenzene moieties. The polymerization of the LCN was delicately controlled to mimic the innate properties of human fingertips. Moreover, we can surpass natural restrictions and demonstrate liquid secretion from valleys in fingertip topography, which is not innate in natural archetypes. The resultant coating displays tuneable tribological properties for practical synthetic applications via controllable perspiration. In fact, imparting these "synthetic" biological features to our LCN fingertips introduce possibilities for additional robotic functions which are essentially very human, thus offering a promising choice in relative fields such as artificial skin.
Fingertip perspiration is a vital process within human predation, to which the species owes its survival and its biological success. In this paper, the unique human ability of extensive perspiration and controlled friction in self-assembled cholesteric liquid crystals is recreated, mimicking the natural processes that occur in the dermis and epidermis of human skin. This is achieved by inducing porosity in responsive, liquid-bearing material through the controlled-polymerization phase-separation process. The unique topography of human fingerprints is further emulated in the materials by balancing the parallel chirality-induced force and the perpendicular substrate-anchoring force during synthesis. As a result, artificial fingertips are capable of secreting and re-absorbing liquid upon light illumination. By demonstrating the function of the soft material in a tribological aspect, it exhibits a controllable antisliding property comparable to human fingertips and subsequently attains a higher degree of biomimicry. This biomimetic fingertip is envisioned being applied in a multitude of fields, ranging from biomedical instruments to interactive, human-like soft robotic devices.

Introduction
The unique topography and ability for perspiration of human fingertips, both distinctive to our biological order, secured our biological success and granted us our ability to build the modern world. Characterized by their unique arrangement of ridges, human fingertips assume a pivotal function in both our

Results and Discussion
Fingerprint pattern is a known phenomenon observed in cholesteric LCs, where chiral-moiety-doped LC molecules rotate along the chiral axis (Figure 1a,b). [8] A full rotation of this axis defines the chiral pitch (P). By balancing the homeotropic anchoring force originating from the polyimide coated-glass substrate, with the torque induced by chirality, the molecular helices are wound in random directions to establish the irregular fingerprint pattern ( Figure S1 and S3, Supporting Information). [9] Moreover, when using reactive mesogens, the fingerprint alignment can be arrested by polymerizing the structure to form a liquid crystal network (LCN) ( Figure S2, Supporting Information). [10] To complete our artificial fingertip system, we introduced liquid secretion capabilities by establishing a liquid-bearing porous LCN coating of 8.4 µm thickness. We achieved this through the phase-separation process of non-reactive LCs and reactive mesogens during photo polymerization process ( Figure S4, Supporting Information). The nonreactive LC (molecule 6 in Figure 1b) acts as the porogen and establishes the formation of the initial porous LCN ( Figure S5). The photocuring kinetics determines the pore sizes of the membrane. Scanning electron microscope (SEM) images of a typical porous polymer network are shown in Figure 1c,d, which demonstrate the pore sizes of the coating ranging from 70 nm to 120 nm in diameter. To endow our LCN fingertips with light-responsiveness, we copolymerize azobenzene into the network, which allows for on-demand perspiration and variable topography by illumination with 365 nm UV light. [11] The topography of our LCN fingertip is analyzed with a crossed-polarized optical microscope (POM), where we can distinguish the black homeotropically-aligned regions from the colored planarly-aligned molecules (Figure 2b). By correlating the POM image to surface profiles, we found that the planar region forms the fingerprint ridges, and the homeotropic region constructs the valleys. Notably, in the absence of azobenzene (Figure 2c), the positive value of height difference indicates that the homeotropic regions rest at a greater profile than the planar regions. This is a result of the Marangoni effect dominating the polymerization-shrinkage effect. [8,12] Yet, upon incorporating azobenzene, polymerization rate difference according to alignment is introduced due to the dichroic nature of this additive. [13] Since azobenzene shows a larger absorption coefficient toward light when the molecule is in the x-y plane than along the z-axis, homeotropically-aligned molecules would polymerize faster than planarly-aligned molecules. The polymerizing rate difference induces the migration of 5CB as indicated by Figure 2c-e, where we found that the height difference between homeotropic regions and planar regions decrease from 776.79 nm with more azobenzene added ( Figure 2c, Table S1, Supporting Information). Whereas, when investigating the polymer network of these coatings without 5CB, the height difference between two alignment regions steadily keeps ≈780 nm as shown in Figure 2d. It means that azobenzene induces the diffusion of 5CB to planar region, and expanding these regions on the expense of the 5CB at homeotropic regions ( Figure S6, Supporting Information). We postulate this process corresponds to a "syneresis" effect due to the relatively faster gelation of homeotropic monomers, yet, in our case 5CB is still in the confinement of the more slowly crosslinked planar region. [14] Moreover, this uneven polymerization can cause rapid depletion of reactive mesogens with homeotropic alignment and thus trigger the migration of reactive mesogens toward homeotropic regions by concentration gradient. [15] This results in greater polymerization-shrinkage at homeotropic regions to further drive non-reactive LCs (5CB) toward the planar regions. [15,16] More details are summarized in Table S1 (Supporting Information) and its corresponding text.
By calculating the expansion at planar regions (or contraction at homeotropic regions) (Figure 2e), we can evaluate the amount of 5CB diffusion toward planar regions changing with azobenzene concentration (as explained in Figure S7f, Supporting Information and its corresponding text). Once the incorporated azobenzene concentration is up to or higher than 5 wt.%, the diffusion caused by "syneresis" of gelation and polymerization-shrinkage surpass the influence of the Marangoni effect, thus inducing the planar regions to protrude greater than the homeotropic regions. As a result, the protrusion of the planar regions above homeotropic regions reaches 185.49 nm (Figure 2b).
To launch perspiration, we conducted UV illumination at various intensities for a typical coating with 5 wt.% azobenzene, and we found the induction time to eject liquid (5CB) becomes shorter with increasing UV intensity, as shown in Figure 3b.
Here, the UV light intensity of illumination was optimized to a relatively low value of 6.8 mW cm 2 for better performance. The penetration of the used UV light into the coating is given in Figure S9 (Supporting Information). This illumination triggers order-parameter-reducing trans-to-cis isomerization in azobenzene. [17] As shown in Figure 3c The fingerprint valleys correspond to the black homeotropically-aligned regions, while the fingerprint ridges correspond to the colored planarly-aligned regions. c) The relative height difference between homeotropic regions and planar regions of fingerprint coatings change as a function of azobenzene concentration. d) The relative height differences between homeotropic regions and planar regions of the fingerprint coatings after extracting 5CB. e) The surface expansion at planar regions caused by 5CB diffusion.
restrictions applied by the substrate, when the order parameter of our LCN coating is reduced, the homeotropically-aligned regions contract 120.0 nm, which corresponds to 1.4% of the coating thickness. In contrast, the adjacent planarly-aligned regions display the inverse behavior, with an expansion of 121.7 nm along the z-axis. This actuation bestows an additional dimension to our LCN fingertips and ejects 1.2·10 -6 µL liquid, which is 1.2% of the total liquid in the given area (Figure 3e-g, Figure S10-S12 and Video S1, Supporting Information). We further investigated the influence of azobenzene concentration, which reveals that homeotropic regions and planar regions deform to the same degree when doping 2 wt.% azobenzene (Figure 3d). By increasing azobenzene concentration, both the contraction of homeotropic regions and the expansion of planar regions increase, while the gap between the amplitude of contraction and expansion is enlarged. This enlarged gap is attributed to the relatively denser homeotropic network formed via monomer diffusion from planar regions. Furthermore, image analysis of the typical group's liquid secretion process by an in-house algorithm was studied, showing that liquid is primarily secreted from ridges upon stimulation and is then gradually drawn to valleys ( Figure S13, Supporting Information). This is due to gravity and surface tension-induced liquid coalescence, which leads to an ≈60% drop in total droplet count on ridges and a 200% increase in liquid volume at valleys, as shown in Figure 3g,h.
Beyond mimicking the natural system of human fingertips, we can modify our LCN artificial fingertips to secrete liquid at designated valley regions (Supplementary Video S2), instead of the ridge regions which is typical of a natural fingertip. This is achieved by reversing the fingerprint topography of our coating by decreasing the concentration of  copolymerized azobenzene moiety in our material to less than 5 wt.%. Its low concentration leads to only limited polymerization rate difference and 5CB diffusion, which causes nonreactive LCs to remain more homogeneously distributed across the coating ( Figure S6b, Supporting Information). In this case, fingerprint ridges are dominated by the Marangoni effect, as deduced by their homeotropic alignment shown in Figure 4b. Upon triggering perspiration by 6.8 mW cm -2 UV illumination, the homeotropically aligned ridges contract, whereas the planarly aligned valleys expand. Both deformations count for ≈0.5% of the total thickness (Figure 4c). To investigate perspiration, we created a binary map of ridges and valleys for image analysis (Figure 4e-h). The droplets were tracked by defining them as circles, which are found to appear at the transition areas from valleys to ridges at 7.13 s. Then the liquid volume and counts increase significantly. The droplets subsequently merge and coalesce at valleys at 8.58 s. Corresponding liquid counts decrease by ≈80%, as shown in Figure 4h. As a result, the given region secretes 1.1% of stored liquid (0.94·10 -6 µL, as shown in Figure 4g). This secreting performance can be fine-tuned by various UV intensities as indicated by Figure S14 (Supporting Information). Further diverging from natural norms, all of our LCN artificial fingertips can perform reverse-perspiration processes, which are not innate in natural archetypes. Blue light irradiation of our activated coating causes the back relaxation of the azobenzene cis isomer and leads to the secreted liquid being re-absorbed by the homeotropic valley regions, as indicated by the decrease in liquid volume shown in Figure 3g and Figure 4g.
Similar to human fingertips, the biomimetic fingerprint coating can regulate its tribological properties through perspiration. The variable tribology of our LCN artificial fingertips was quantified through dynamic sliding friction tests ( Figure S15  the friction force on our LCN artificial fingertip is observed to increase by ≈70 mN within 3 s. When illuminated with blue light, the friction force falls back to almost the initial value (Figure 5a,b). We further elaborated on our experiments practically by demonstrating the variable static adhesion of our fingerprint coating to a desired position. The coating is shown to readily slide off the sloped substrate in the absence of secreted liquid, while activation of liquid secretion prevents any movement from the initial position of the sample as shown in Video S3, Supporting Information and Figure 5c-g. The friction-control mechanism involves the formation of capillary bridges at the interface ( Figure S16 and Video S4, Supporting Information), which induce an adhesion force and shear interaction between our LCN coating and the contact object. [18] Generally, the Young-Laplace equation dictates that capillary force decreases with the height of the capillary bridge between two planar surfaces, which indicates that excessive liquid secretion might be unfavorable for greater friction. [19] As for the fingerprint coating, since its surface topography entails increased roughness at the air-solid interface, capillary bridge can spread laterally from ridges. Hence, the capillary force would increase with the wetting area once the secretion volume surpasses a wetting threshold. [20] Aditionally, the moistening of ridges can decrease the modulus at contacting points, which may further increase the friction coefficient and consequently enhance the friction. [21] Figure 5. Tribological properties of our artificial fingertips. a) 3D visualization of capillary bridging between the LCN fingerprint topography and a planar surface. Insert is annotated schematics of wetting liquid bridges migrating toward local minima in gap separation due to pressure gradient. b) Dynamic friction force of the coating sliding on the horizontal glass plate under 365 nm or 455 nm light stimulation. Inserts are the surface condition of coating at each illumination stage and hypothetical application scenario. Demonstration of our LCN fingerprint coating adhering to a 60° glass slope, mimicking the friction of the human hand. c) The non-perspiring coating slides off the glass plate and the corresponding hypothetical scenario. d) The perspired coating adheres to the glass plate after UV illumination and corresponding hypothetical scenario. e) After ceasing UV illumination, the coating remains adhering to the glass plate. f) Blue-light illumination triggers reverse-perspiration and the coating slides off the glass plate.

Conclusion
In conclusion, we designed a coating to mimic the human fingertip's properties in terms of fingerprint topography and perspiration behavior. The coating is made of cholesteric liquid crystal networks containing non-reactive 5CB via polymerization-induced phase separation. Referring to authentic fingertip perspiring, localized liquid secretion at our LCN fingerprint ridges as triggered by UV light is realized. Furthermore, we can transcend natural archetypes to demonstrate liquid secretion from valleys by tuning the material composition. All these released liquids can be re-absorbed into the network by blue light illumination, given which the coating is able to regulate its tribological property by reversible perspiring, thus tuning the anti-sliding capability. Besides secreting 5CB, our coating has the potential of secreting various liquids ( Figure S17, Supporting Information), which branches out the application in the biological field. [22] We anticipate our "perspiring" LCN fingertips can be applied in robotic devices to induce greater biomimicry of the human body, thus boosting their anthropopathic function development in the future, e.g., pick and place function, self-cleaning, drugs delivery, and information transfer between robots.
Sample Preparation: Glass substrates were first cleaned with acetone and isopropanol via an ultrasonic water bath for 20 min, respectively. Then the glasses were dried up and put in UV-ozone for 20 min for surface treatment. Next, polyimide SE 5661 (Nissan Chemical industries) was spin-coated on glass substrates at 5000 rpm for 40 s to offer homeotropic alignment. The polyimide-coated glass substrates were baked at 90 °C for 10 min to evaporate the solvent and then baked at 180 °C for 90 min. (For high polar LCN fingerprint coating, the clean glass substrates were dip-coated with a 1% v/v DMOAP water solution for 15 min and baked at 100 °C for 30 min for homeotropic alignment and coupled effect.) To fabricate the fingerprint coating, the prepared monomer mixture was spin-coated on the glass substrates with polyimide layer. The spin-coating was performed at 1000 rpm (acceleration was 300 rpm/s) for 40 s. Afterward, the coating was polymerized by UV illumination at 30 °C for 90 min in N 2 environment using a mercury lamp (Omnicure EXFO S2000). A cutoff filter with transmitting light >400 nm (Newport FSQ-GG400 filter) was used during the polymerization to prevent premature isomerization of azobenzene.
Sample characterization: The fingerprint texture was observed by a cross-polarized optical microscope (Leica DM2700). Thickness and surface profile of samples was measured by interferometer (Sensofar S neox). The dynamic surface topographies were measured by Digital Holography Microscope (Lyncée Tec.). The microstructure of the LCN coating was observed with a scanning electron microscope (FEI SEM Quanta 3D FEG) in secondary electron mode. Before SEM characterization, 5CB was removed from LCNs by cyclohexene bath for 1 h. In this process, 5CB diffused to the cyclohexene, while the polymer network was firmly attached to the glass substrate without swelling or damage. After taking the sample out of cyclohexene, the residual solvent quickly evaporated, and the polymer network was left over to the substrate. The phase transition of liquid crystal monomer mixtures was measured by DSC (Q2000, TA Instruments) at a rate of 10 °C min -1 . Material distribution of fingerprint coating was characterized by Confocal Raman microscope (WITec WMT 50) at room temperature using a WITec α-300 R µ-Raman system. The temperature control for the thermal effect-induced liquid secretion was conducted by Linkam Thermal Stage. The temperature increase of coating during photo-responsive secretion was monitored by infrared thermometers (Fluke Ti32 thermal imager). Atomic force microscopy measurements were conducted on a Cypher Environmental Scanner (ES) equipped with a closed gas cell and an active heating and cooling stage. Force maps of the coating surface were recorded in contact mode using a super luminescent diode in order to reduce the signal-to-noise ratio. High-density diamondlike carbon Biosphere B20-FM probes (Nanotools, spring constant k = 2.63 N/m; r = 20 nm) were used for all measurements and thermally calibrated using the "Get Real" function in the Igor Pro software. Furthermore, a scan rate of 0.39 Hz, a Z-rate of 125 Hz, and an amplitude of 1 um were applied. After obtaining the 30 × 30 µm force maps (128 × 128 pixels), we applied the Hertz model (for spherical AFM tips) as shown in euation 1 in order to extract the Young's modulus in different areas of the coating: where F is the applied force, E s is the elastic modulus, V s is the Poisson's ratio, r the radius of the spherical AFM tip, and δ the indentation depth. All AFM measurements were acquired in air and at the machine's temperature (≈29 °C).
The friction force was measured by a home-built setup. The fingerprint coating was attached to the dock of the device with the coating surface upside. And a clean glass slide was placed on the coating with one side connected to the motor. A constant normal load of 0.5 n was placed on top of the glass slide with a window for light going through. During measurement, the motor could drive the glass slide to move at a constant speed, and the time-resolved kinetic friction force was measured by a sensor. The demonstration shown in Figure  5 was not subject to ethical approval, and the volunteer took part with informed consent.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.