Underwater adhesion tends to fail for a boring reason. You press an adhesive against a submerged surface and what you actually get is a thin water film sitting between the two. Even when the bulk material is tacky, that trapped layer is enough to stop the interface from ever settling into real contact, especially on rough substrates.
A recent peer reviewed study from researchers at UNIST in South Korea approaches this with a microstructured, shape memory polymer patch that can be switched between low-adhesion and high-adhesion states underwater, making switchable underwater adhesion something they can test rather than just demonstrate. They quantify it with underwater pull-off testing, using a controlled preload and retraction protocol on a CellScale UniVert, and then repeat the same measurement across microstructure designs, temperatures, substrates, and cycling.
The mechanism is sequential. First, the micropillar surface helps push water out of the interface so dry contact regions can form. Then the polymer is heated so it can soften and adapt to surface roughness, cooled to fix that adapted shape, and reheated to recover and release. The switching ends up reading less like “temperature makes it sticky” and more like a contact-area management problem that happens to be implemented using a shape memory blend.
(a) Concept of the microstructured SMP (MSMP) patch: micropillars promote water shedding and dry contact formation, while the SMP matrix switches between a stiff and softened state with temperature. (b) Proposed adhesion sequence on a wet substrate, showing water exclusion followed by conformal contact. (c) Mold-replication fabrication workflow. (d) Photographs and SEM of the fabricated patch, including uniform micropillars (15 μm diameter, 15 μm height). Adapted from Kim et al., Small Structures, 2026. Open Access.
Underwater adhesion usually fails because water stays in the interface
When you press a patch onto a submerged surface, you are rarely pressing the adhesive directly onto the substrate. You are pressing it onto a water film. That water layer might be only a few microns thick. It can also be uneven and mobile while you load the patch. Either way, if it stays continuous, the adhesive never really gets a chance to settle into the surface.
A lot of wet adhesives try to tolerate that film, bond through it, or displace it chemically. This paper takes a more mechanical route. The patch is designed so that the first step is not bonding at all. It is getting water out of the way.
It changes what you pay attention to. The question is not whether the polymer feels tacky in air. It is whether the contact ever leaves the “wet” regime at all, and whether you can cycle in and out of that state without the result drifting every time you repeat the test.
The patch starts by pushing water out, not by bonding through it
The surface of the patch is patterned into micropillars, and the geometry is treated as an active design parameter rather than a texture choice. The authors vary micropillar spacing and compare behaviour under water to show that the microstructure controls whether the interface stays in a water-trapped state or shifts toward a dry contact state.
Micropillar geometry and why spacing ratio matters
In the experiments, spacing ratio drives changes in wetting behaviour and contact stability. If water stays trapped between pillars, the interface behaves like a lubricated contact, and you never really leave the “wet” regime. With the selected microstructure, the surface maintains a more water-repellent state and resists full wetting, which makes dry contact formation more plausible once you apply preload.
What dry interface formation looks like in practice
The most useful part here is that the authors do not rely on contact angle as a proxy for what is happening at the interface. They show images that distinguish water trapped at the interface from water expelled out of the contact region, including a comparison against a planar control surface. For a mixed audience, this is one of the easiest parts of the paper to follow because it is not abstract. You can see the difference.
(a) Water droplet images on patches with different spacing ratios (SR = s/d), showing how geometry shifts apparent hydrophobicity. (b) Measured contact angles compared with Cassie–Baxter predictions. (c) Interface schematics and microscopy showing water removal on glass: the micropillar patch expels the droplet to form a dry interface, while a planar patch leaves a trapped liquid layer. Adapted from Kim et al., Small Structures, 2026. Open Access.
Switchable underwater adhesion comes from the polymer softening, adapting, then locking in place
Once the interface can reach something closer to dry contact, the second part of the design becomes relevant. The patch is made from a shape memory polymer blend. At elevated temperature it becomes more compliant and can deform to match surface roughness. When cooled, it fixes that deformed shape, effectively holding onto the surface. Reheating drives shape recovery and reduces adhesion during release.
Stimulus-driven shape change shows up in other soft material systems as well (in different forms and timescales), and we have covered a few examples here.
This is the core of the switchable underwater adhesion idea in the paper: use temperature to change compliance and surface matching, then use recovery to get out of the high-contact state.
Shape memory polymer blending as a stiffness and shape-control knob
The authors vary blend composition and use thermal and mechanical characterization to show that the material has a temperature window where it softens and a lower temperature state where it behaves more like a fixed, load-bearing solid. It is a useful kind of switching because it does not require a complex stimulus. You heat and cool.
At the same time, thermal switching underwater is rarely as simple as it sounds. Heating rates, heat loss into surrounding water, and temperature gradients through the patch can all become limiting factors. The paper later addresses this with a heater-integrated gripper demo, but it is still worth keeping in mind. The switching works in the reported protocol. How it behaves when timing constraints tighten is a separate question.
Surface adaptation and real contact area
The switching mechanism mostly reads as a contact-area story. Warm it up and it gives. Press it in and it conforms. Cool it and it stays there. Heat it again and it starts to pull itself back out of the surface. It is a plain contact-history explanation, but it matches what they are measuring. Adhesion strength on rough wet surfaces is usually limited by how much of the interface is actually in contact.
(a) Storage modulus versus temperature for SMP blends with different PCL:PUA ratios, showing a softening window used for switching. (b) DSC thermograms used to locate thermal transitions. (c) Shape fixity and recovery ratios from cyclic thermomechanical testing. (d) Confocal topography and profiles comparing the rough substrate, the adapted ON state surface, and the recovered OFF state surface. (e) Modulus at 30°C and 65°C and recovery strength versus blend ratio. (f) Side-view image sequences showing deformation during the ON state and recovery-driven self-detachment during the OFF transition for two blend ratios. Adapted from Kim et al., Small Structures, 2026. Open Access.
Underwater pull-off testing is where the ON and OFF claims get pinned down
In this study, switchable underwater adhesion is quantified with underwater pull-off testing rather than inferred from a single gripping demo. A lot of adhesion papers lean heavily on demonstration photos. This one includes those, but it also spends time on repeatable force measurements. The authors compare ON and OFF states under a defined protocol, and then repeat the same test across microstructure designs, substrates, and cycles.
From a measurement perspective, this is where the UniVert use fits naturally. They are not doing peel tests. They are doing underwater pull-off testing: apply a preload to establish contact, then retract at a controlled rate and record the force response. That is a simple test geometry, but it is well matched to their switching claim because it gives you a consistent way to compare conditions.
If you do not spend much time interpreting force-displacement curves, this short overview on mechanical testing basics is a useful refresher.
Comparing ON vs OFF under the same preload and pull-off speed
Under water, the authors test at two temperatures corresponding to OFF and ON states. They report force-displacement curves and peak pull-off forces, then normalize by area to express pull-off strength. The important detail is not the normalization. It is that the same protocol is repeated across conditions rather than tuned each time to make the best-looking curve.
Switching ratio is really a contact story, not just a temperature story
The paper reports a large switching ratio between the OFF and ON conditions. One way to read that is as a temperature-controlled adhesive, but the experiments also make it clear that temperature is doing something more specific: it changes how well the patch can conform before it is cooled and fixed in place. If the surface does not get good contact in the ON step, there is less “stored” contact area to carry into the pull-off test. In that sense the microstructure and the surface adaptation data are doing real work in the argument, not just setting the scene.
(a) Switching concept showing ON and OFF states at 30°C and 65°C. (b) Representative underwater pull-off curves for ON and OFF states on glass. (c) Pull-off capacity and switching ratio as a function of micropillar spacing ratio. (d) Pull-off capacity on glass under dry versus wet conditions across spacing ratios. (e) Underwater pull-off capacity across substrates (glass, PET, PI, sandpaper, porcine skin). (f) Repeatability over 500 attach–detach cycles. Adapted from Kim et al., Small Structures, 2026. Open Access.
Repeatability and substrates: what holds up across cycles
Cycle testing is the obvious next question once switching is established. Heating, loading, and repeated contact can degrade interfaces quickly, even when the first few cycles look clean. Here the authors report stable behaviour over many cycles within the tested window, which helps anchor the switching claim in something repeatable rather than a single best-case result.
They also test multiple substrates, including porcine skin. That is useful, but it is best read as a stress test rather than a wearable claim. Skin is compliant, textured, and wet. If the patch can form dry contact and then adapt under those conditions, that tells you something about robustness, even if it does not resolve practical issues like safe temperature windows or long-term wear.
There are a few practical dependencies worth keeping in view. Preload matters. Roughness matters. Temperature control matters, especially underwater. The paper provides enough detail to see that the protocol was consistent, but the protocol also emphasizes that switching depends on sequence: press, heat, cool, then pull.
A heater-integrated gripper turns the same mechanism into a handling tool
The application figure shifts the work from material behaviour on the bench to how the patch might be used in an underwater handling setup. The authors build a gripper concept with integrated heating so the patch can be switched while underwater.
For robotics readers, this is the section that tends to matter most. It shows that the patch is not only strong in pull-off, but can be used to pick up objects with roughness variation and then release them.
For adhesives readers, it quietly forces a timing question. Switching in a mechanical test rig can be slow and controlled. Switching in a device often has to be fast enough to be useful, and temperature has to be delivered through a real contact. The demo does not fully answer those questions, but it does show the authors are thinking about them.
We have seen a similar pattern in other responsive materials used in robotics, where the mechanical test protocol ends up defining what “function” looks like in practice.
(a) Gripper concept integrating a heating panel and the SMP adhesive patch. (b) Switching logic for gripping and release through temperature modulation. (c) Force–displacement curves comparing grippers with and without microstructures on ground glass. (d) Temperature and normal-load timelines during ON/OFF operation. (e) Underwater lifting across surfaces spanning a roughness range (Ra ≈ 0.007–21.552 μm). (f) Pick–move–place sequence demonstrating controlled attachment and release. Adapted from Kim et al., Small Structures, 2026. Open Access.
What this work is useful for right now
In this paper, the clearest fit is switchable underwater adhesion for practical handling: picking up wet objects, holding briefly, then letting go without having to peel or pry at the edge. The design is built around switching and repeatability, not around permanent attachment.
The biomedical angle is present, but it sits a little differently. The porcine skin results show that the interface can tolerate a wet, compliant substrate, which is meaningful. At the same time, the temperature window used for switching is a practical constraint if you imagine skin contact as the primary application. That does not undermine the mechanics. It just reshapes what the work feels aimed at.
Viewed as a mechanics problem, the work is a fairly direct attempt to make switchable underwater adhesion repeatable across cycles and substrates.
Read the Full Publication
About the CellScale UniVert
The CellScale UniVert appears in this study as the tool used to quantify switching under wet conditions using a controlled pull-off protocol. The authors used a defined preload to establish contact and a defined retraction rate to measure pull-off force and pull-off strength. They then repeated the same approach across ON and OFF thermal states, across different microstructure designs, across different substrates (including porcine skin), and across repeated cycling.
That matters because switchable underwater adhesion systems can look convincing in a single photo sequence even when the performance is sensitive to contact history or test setup. Here, the pull-off data anchors the story in repeatable force-displacement behaviour, and makes it easier to interpret what the microstructure and shape memory steps are actually changing at the interface.
