Self-cleaning glass does not repel water. It does the opposite — and that is the whole trick

Ask most people what a self-cleaning surface does and they will describe water hitting glass and bouncing straight off it, the way rain sits on a lotus leaf in perfect silver beads. It is a good description of a real phenomenon. It is also very nearly the opposite of how the self-cleaning glass actually installed on office towers and skylights works — that glass is engineered so water will not bead. It spreads out into an even sheet with no droplets in it at all.

Both surfaces are sold as self-cleaning, both work, and they achieve it by contradictory means. Understanding why there are two routes rather than one is the fastest way to tell which of the two you are being offered, and to know what it will and will not do once it is on your building.

What a self-cleaning surface actually has to do

Dirt on an outdoor surface is two things stuck together. There is loose particulate — dust, pollen, soot, the cement fines that settle on everything near an Indian construction site — and there is an organic film of oils, exhaust residue and biological matter that glues the particulate down. A surface that only sheds loose dust will still go grey, because the film holds the next layer.

So self-cleaning has two jobs: break the bond holding the dirt on, and then transport the dirt off the surface. Rain is the only transport mechanism available for free, which means every self-cleaning technology is really a way of manipulating what water does when it lands. The two routes differ in which manipulation they choose.

Route one: make water refuse to touch the surface

This is the famous one, and it is copied from a plant. A lotus leaf is covered in micrometre-scale bumps, and each bump is itself coated in nanometre-scale wax crystals. A water droplet arriving on that structure cannot sink into it; it rests on the tips of the bumps with air trapped in the spaces underneath, touching perhaps a few per cent of the actual surface area. This is called the Cassie–Baxter state, and it is why the droplet holds an almost spherical shape — a contact angle above 150°, against roughly 20–30° for ordinary clean glass.

A near-sphere sitting on a few contact points has almost nothing holding it. Tilt the leaf a couple of degrees and it rolls, and as it rolls it picks up dust particles, which adhere more strongly to the water than to the waxy surface. That is the lotus effect, described properly by Barthlott and Neinhuis in the 1990s: the leaf is not repelling dirt, it is handing the dirt to a passing droplet.

Reproducing it in a coating means reproducing both ingredients — a low-surface-energy chemistry, usually fluorinated or silicone-based, laid over a two-scale roughness fine enough to hold the air pockets. That is entirely achievable in a lab, and the resulting surfaces are genuinely spectacular. What happens next is the problem.

The number in the brochure is the wrong number

Almost every superhydrophobic product is sold on its contact angle: 155°, 160°, sometimes higher. On its own that figure tells you close to nothing about whether the surface will clean itself.

The property that decides whether a droplet rolls is contact angle hysteresis — the difference between the angle at the front of a droplet as it advances and the angle at the back as it recedes — or, equivalently, the sliding angle, the tilt at which a drop of stated volume begins to move. A true lotus-like surface has a sliding angle under about 10°, often under 5°.

That the two can come apart is not a theoretical worry; it has a name. The petal effect, described on rose petals, is a surface with a contact angle above 150° whose droplets stick so firmly you can turn the petal upside down without them falling. The roughness there is coarser, so water partly penetrates it and pins in place. High contact angle, zero self-cleaning. A specification sheet quoting only the static angle cannot distinguish that case from a working one.

The second and more consequential failure is the Cassie-to-Wenzel transition. The trapped air that holds the droplet up is a metastable arrangement. Pressure, impact from heavy rain, condensation forming inside the texture overnight, or a surfactant lowering the water's surface tension can all push water down into the grooves. Once it is in, the droplet is in the Wenzel state — fully wetted, gripping a surface area larger than a flat one would offer, and therefore stuck harder than it would be on plain glass. The surface has not merely stopped working; it has become worse than doing nothing.

And the roughness that makes all of this possible is, structurally, a field of tiny fragile protrusions. A fingertip, a cloth, blown grit, a cleaner's brush — anything that flattens the texture destroys the effect permanently. The lotus leaf lives with the same fragility and solves it in a way no coating can: it grows a new wax layer.

Durability, not performance, is the unsolved problem in superhydrophobic coatings. Making a surface that repels water beautifully is a solved exercise. Making one that still does it after two years on a real façade is where the field's actual research effort sits.

Route two: make water spread out completely

The opposite approach abandons droplets altogether, and it is what commercial self-cleaning window glass uses.

A very thin layer of titanium dioxide — anatase phase, on the order of fifteen nanometres, fired into the glass on the float line rather than painted on afterwards — does two separate things when ultraviolet light falls on it, and the coincidence of those two effects in one material is why this route exists at all.

The first is photocatalysis. TiO₂ has a band gap of about 3.2 eV, so a photon below roughly 388 nm — present in ordinary daylight — lifts an electron into the conduction band and leaves a hole behind. Those charge carriers reach the surface and generate reactive species from adsorbed water and oxygen, which slowly break down the organic film that glues dirt to the glass. Slowly is the honest word: this is a process measured in days of daylight, not minutes.

The second is photoinduced superhydrophilicity, reported in Nature in 1997 and still slightly odd. Under the same UV illumination, the TiO₂ surface's affinity for water rises dramatically — the contact angle falls towards zero. Water arriving on it cannot form droplets; it spreads into a continuous sheet.

That sheeting behaviour is the actual cleaning mechanism, and its advantage over the beading route is specific. A droplet that dries on glass leaves its dissolved load behind as a ring — the spots and streaks on every window after rain. A sheet of water runs off carrying the loosened film with it and leaves nothing to dry into a mark. The glass is not repelling the dirt; it has digested the glue and then washed the residue off in one film.

Its limits are equally specific, and any supplier who does not state them is not being straight with you. It needs UV, so it is an outdoor technology and does poorly on a shaded north face or behind UV-blocking glazing. It needs rain, or periodic wetting, because it has no other transport mechanism. It is slow. And photocatalysis acts on organic matter, so it does nothing whatsoever about cement splash, hard-water scale or mineral dust — which, in much of India, is most of what is actually on the window.

Against that, it has the property the other route lacks: it is a hard inorganic layer chemically integrated with the glass, not a fragile texture sitting on top of it, so it survives being cleaned and lasts the life of the pane.

Where this is worth money

Solar panels. This is the application with real economics behind it. A photovoltaic module loses output in direct proportion to the light that dust stops from reaching it, and in dry pre-monsoon weeks across northern and western India the accumulation is fast enough that measurable output is lost within days of a clean. Published soiling studies in Indian conditions vary widely by site and season, but they agree on the shape of the finding: soiling is one of the largest controllable loss terms in an Indian solar plant, and manual cleaning costs both labour and water, in places where water is the scarcer of the two. Anything that extends the interval between cleans is worth its cost quickly.

Building glazing and façades. The saving here is access equipment as much as labour — the cost of cleaning the outside of a tall building is dominated by getting a person to it.

Bathroom glass and sanitaryware, where the enemy is soap scum and limescale, and where a hydrophobic coating that stops scale adhering in the first place is the more useful of the two routes.

Anti-fogging optics, which is the superhydrophilic effect used for a different end. Fog is not a film of water, it is thousands of tiny droplets scattering light; a surface water cannot form droplets on cannot fog. Mirrors, goggles and camera housings use exactly this.

Textiles and anti-graffiti coatings, where the requirement is that a liquid — spilled coffee, spray paint — never gets to bond to the fibre or the stone at all.

What to ask before buying either kind

1. Which route is this — hydrophobic or hydrophilic? They fail differently, need different conditions, and suit different dirt. If the sales material does not make it obvious, ask outright.
2. What is the sliding angle or the contact angle hysteresis? For a water-repellent coating this is the number that predicts self-cleaning. A supplier who quotes only a static contact angle has either not measured it or would rather not say.
3. Does it need sunlight to work? Photocatalytic products do. This decides whether it can go indoors, on a north elevation, or behind laminated glass.
4. What abrasion testing has it had? Taber cycles, wash cycles, wipe tests — on your substrate. For textured hydrophobic coatings this is the whole question, since the effect dies with the texture.
5. What does it do to appearance? On glass, ask for haze and light-transmission figures before and after. A coating that costs a per cent of transmission on a solar module has to earn that back before it earns anything.
6. What does it explicitly not do? A self-cleaning coating is not an antimicrobial one — keeping a surface visually clean and reducing microbial load are separate claims with separate test standards, and conflating them is the most common piece of loose language in this market.

To be concrete about what a range looks like: Smart Warrior Coatings, part of Reinste Nano Ventures — which, in the interest of disclosure, belongs to the same group as this publication — sells its water-behaviour product, Warrior Aquaperl Surface Shield, as a separate line from its silver-based antimicrobial shields rather than as one product doing both jobs. That separation is the correct shape, and it is a reasonable thing to expect of any supplier: the TDS should tell you which route the product takes, which substrates it is qualified on and how it was tested, and the questions above are answered there or not at all.

Why it matters for students and researchers

This is an unusually good field for a student, because the central problem is openly unsolved and everyone in it says so.

The durability of superhydrophobic surfaces is that problem. Two decades of papers report ever more impressive contact angles, and the number of those surfaces that survive commercial use remains small. The research responses are worth knowing: self-healing coatings that hold a reservoir of low-surface-energy material which migrates to repair a damaged surface; bulk-structured materials where the texture runs through the whole thickness, so abrading the top merely exposes more of the same; and a slow, useful pressure on the field to report abrasion data alongside wetting data.

On the photocatalytic side the live question is light. Anatase TiO₂ needs UV, which is a few per cent of sunlight and almost none of indoor lighting, so visible-light-active photocatalysts — nitrogen-doped titania, and a growing family of alternatives — are a large and active area with a real prize attached. There is also a measurement problem worth a thesis on its own: laboratory wetting figures and field cleaning performance correlate far less well than the literature's confidence implies, and standard, comparable field protocols barely exist.

The lesson underneath all of it is one that generalises well beyond coatings. Nature's solution — the lotus leaf — is not more durable than ours. It is more repairable. Copying the structure without copying the repair mechanism is why the imitation lasts two years and the original lasts a summer and then grows another one.

Frequently asked questions

Is self-cleaning glass actually self-cleaning?

Partly, and the qualifier matters. It substantially extends the interval between cleans on rain-exposed outdoor glass by decomposing the organic film and sheeting water off without leaving spots. It does not remove mineral deposits, cement splash or hard-water scale, it needs daylight and rain to do anything, and sheltered panes under a deep overhang see much less benefit. Treat it as a reduction in cleaning frequency, not an elimination of cleaning.

Why does a rough surface repel water better than a smooth one?

Because the water is not really touching it. In the Cassie–Baxter state the droplet rests on the peaks of the texture with trapped air beneath, contacting only a small fraction of the surface, so there is very little adhesion for gravity to overcome. The catch is that this arrangement is metastable — if water gets pushed into the grooves, the same roughness that was repelling it now increases the contact area and makes the drop stick harder than it would on flat glass.

Can I apply a self-cleaning coating myself?

Spray-on hydrophobic treatments for cars, glass and fabric are widely sold and do work, with a life usually measured in months rather than years because the fragile texture wears. Photocatalytic self-cleaning glass is generally a factory product — the titania layer is deposited on the float line at high temperature, which is precisely what makes it durable. Retrofit photocatalytic sprays exist; ask specifically about adhesion, expected life and whether they change light transmission.

Do these coatings kill germs?

Different question, different test. Photocatalytic surfaces do have antimicrobial activity under UV, since the reactive species they generate damage cells as readily as they damage an oil film — but that is a light-dependent effect, measured against specified organisms under ISO 21702 or ISO 22196, and a self-cleaning claim is not evidence for it. Hydrophobic coatings do not kill anything at all; they make a surface harder for anything to stick to, which is a genuine but different benefit.

Which is better, hydrophobic or hydrophilic?

Neither, on its own. Choose by the dirt and the conditions. Outdoors, rain-washed, organic grime, long service life wanted: the photocatalytic hydrophilic route. Indoors, or against liquid staining, scale and spills, or where there is no UV: hydrophobic. Where optical clarity through condensation is the point, hydrophilic wins outright, because it is the only one of the two that prevents fogging.