A blade across a self-cleaning pane leaves the glass looking perfect. Clear, bright, no scoring, nothing a client would ever query on handover. The photocatalytic layer that cost the developer a premium per square metre is gone in that stroke, and the building will carry on looking fine for about eight months until somebody notices the north elevation is streaking in a way it didn’t used to.
You won’t see it happen.
What Is Sitting on the Outside of Glass in a London New-Build?
Coatings on architectural glass fall into two families, and the distinction governs everything a cleaner is allowed to do.
Pyrolytic or hard coats are applied online, during float production, while the glass is still hot enough that the coating fuses into the surface. Pilkington’s Activ self-cleaning product and K Glass are the familiar examples. These are durable, can be toughened and handled like uncoated glass, and can sit on the weather face of a unit – surface 1, in the numbering that runs from outside in.
Sputtered or soft coats go on cold in a vacuum chamber, magnetron-deposited in layers including metallic silver. Solar control and high-performance low-E products – Planitherm, SunGuard, Suncool and their equivalents – are usually soft coats. They are physically delicate and are normally sealed inside the cavity on surface 2 or 3, where nobody can touch them.
Usually. That word is doing a lot of work, and section three is about what happens when it doesn’t hold.
The candle test, which takes eleven seconds
Hold a flame up to the glass at a slight angle and count the reflections. A double-glazed unit gives you four images. If one of them is a different colour from the others – typically a slightly odd violet or coppery cast – there’s a low-E coating, and the position of the odd reflection in the sequence tells you which surface it’s on. A handheld coating detector does the same job more reliably for about £150, reads through the unit in a second, and every commercial operator working new-build stock should own one and carry it in the van rather than in the office.
That tells you a coating exists. It does not tell you the product, the warranty terms or what the manufacturer’s care guidance permits, which is why the coating schedule from the glazing subcontractor matters more than any test you can do on site.
Why Does a Scraper Destroy Self-Cleaning Glass Without Marking It?
Self-cleaning glass works on two mechanisms sitting in a titanium dioxide layer only nanometres thick. Ultraviolet light drives a photocatalytic reaction that breaks down organic soiling on the surface. The same layer is hydrophilic, so rainwater sheets across the pane rather than beading, carrying the loosened material off in a film rather than leaving it in dried droplet rings.
That layer is measured in nanometres. A carbon steel blade held at any angle you like will take it off in a single pass without leaving a mark visible to anyone, because the coating is thinner than the wavelength of the light you’re using to inspect it. Same for abrasive pads, fine wire wool, cerium oxide polish, and any of the scratch-removal products a restoration contractor might reach for.
The chemical list is longer than the mechanical one. Silicone residue from sealant work spreads across a photocatalytic surface and stops it functioning. So do water-repellent treatments of the Rain-X type, which are hydrophobic and therefore doing the precise opposite of what the coating exists to do. Waxes and polishes are out, along with anything marketed as leaving a protective film. Strong alkaline traffic-film removers will attack it. Mild detergent in clean water on a soft cloth or a soft brush is the whole permitted method, and pure water through a water-fed pole is well suited to it.
The bit where a new façade isn’t working yet
A newly installed self-cleaning elevation needs UV exposure before the coating becomes active – manufacturers generally talk about a period of days to a few weeks depending on light levels. Cover it in protective film, hoarding or scaffold sheeting through a long fit-out and the clock hasn’t started.
Which means the builders’ clean on a new London block frequently happens on glass that is technically self-cleaning and functionally inert, and the man being paid to get plaster splash and cement runoff off it has no reason to suspect he’s working on anything special. That single day of work is where most of the coating damage in this city happens.
Which Coatings Sit on the Room Side, and Who Cleans Those?
Here is the part that surprises facilities managers. Not every high-performance coating is buried in the cavity.
Surface 4 low-E coatings, applied to the internal room-facing pane, have become common in residential specification because they improve U-values in retrofit and in slim units. They are on the surface your internal cleaning contractor wipes twice a week. Some are hard coats and reasonably tolerant. Some are not, and the care guidance for them is more restrictive than anything applied to the outside of the building.
Then there’s applied film – retrofit solar control and privacy films, everywhere in London commercial fit-out because they solve an overheating problem after the fact for a fraction of the cost of reglazing. Film is polyester with an adhesive layer, it scratches under anything harder than a microfibre cloth, and most film warranties specify a cure period of about thirty days before the first clean, a neutral detergent, and no ammonia at all. Half the proprietary blue glass cleaners in a cleaning cupboard are ammonia-based.
There is a related trap in commercial towers where anti-reflective glass has been used at ground floor for display or reception glazing. Anti-reflective coatings sit on the outer face by definition, they are optically delicate, and they turn up in exactly the locations that collect fingerprints and the adhesive residue of whatever the last event contractor stuck to the glass. Nobody puts a sign on it.
The contractor who was never told
Internal glass cleaning is usually bundled into the daily office cleaning contract at a rate that assumes nothing more complicated than a spray bottle. The façade cleaner gets a method statement, a coating schedule and a technical conversation. The night cleaner gets a trolley.
I’ve watched a surface 4 coating come off a partition-height pane in a Greenwich Peninsula apartment because somebody used a cream cleaner on a stubborn mark near a door handle. The resident thought the glass had gone cloudy. It had gone bare, and no polish or treatment puts a sputtered coating back.
You won’t see it happen.
Does Self-Cleaning Glass Earn Its Money in London?
Here’s the opinion, and I get argued with about it regularly. On a large share of London elevations, self-cleaning glass is a specification-sheet feature that never delivers what the brochure suggests, and specifying it commits the building to a restricted cleaning regime for the sake of a benefit the site conditions won’t support.
The mechanism needs two inputs: ultraviolet light and rain reaching the glass. A north elevation in a dense block off Wembley Park’s high street gets meaningfully less UV than the south face of the same building. Recessed glazing behind a balcony soffit gets almost no direct rainfall, which is why you see clean sheeted glass above and a filthy band below the overhang on the same façade. Add a canyon street in Nine Elms with towers either side and the geometry works against you twice over.
Then there’s what London air deposits. Photocatalysis breaks down organic soiling. It does nothing for the inorganic component – brake dust, mineral spotting from hard water runoff, cement leachate off concrete elements above – and inner London delivers plenty of that.
Where it does work
South and west elevations with unobstructed sky, and anything sitting above the immediate street canopy on an exposed site. Barking Riverside and the exposed river frontages generally do far better than anything in a tight inner-London street pattern. The technology is sound. The London application of it is uneven, and nobody involved in the specification is around to see which elevations underperform.
What Does a Warranty Claim Turn On?
Documentation, and almost nothing else.
Coating performance warranties typically run ten to twelve years and are conditional on the glass being maintained in accordance with the manufacturer’s published care guidance. When a claim goes in for a self-cleaning elevation that has stopped self-cleaning, the processor’s first question is what has been put on it and by whom. If the answer is a cleaning contract with no named products and no method statement, the claim is over before it starts.
What defends a claim is dull: a copy of the coating schedule identifying each product by elevation, a method statement naming those products and referencing the relevant care guidance, records of what solution was used at what dilution, and a note of every other trade that has worked on the glass since practical completion.
The clause worth adding to the cleaning specification
Require the contractor to confirm in writing that no blades, abrasives, silicone-bearing products or hydrophobic treatments will be used on any coated surface, and require them to flag any pane they find with bonded contamination they can’t remove by permitted means rather than dealing with it quietly.
That second half is the one that matters. A cleaner who finds cement splash welded to a self-cleaning pane has two options: report it and look ineffective, or reach for a blade and look excellent. The contract has to make the first option the safe one.
I’d tell any developer or managing agent to get the coating schedule out of the glazing subcontractor at handover and hand it to whoever cleans the building, and to strike out any cleaning tender that won’t name products in its method statement. That excludes a certain number of firms who would otherwise have quoted, and it should. The schedule for the block I clean in Colindale runs to two pages. It has been used twice in four years, both times to tell somebody no.




