Why Are There Grey Shadows on the Ceiling Tiles Above the Desks?

“It looks like someone’s been up there with dirty hands.”

That was a practice manager in Hatton Garden, standing under a run of tiles in EC1N with grey bands running along every grid line and a soft dark halo around each supply diffuser. She had it in her head that the cleaners had been leaning on the ceiling. Nobody had touched it. The marks had arrived by air, over about four years, and they had arrived in a pattern so regular that once you know what you are looking at you can read the building’s ventilation from the floor.

What she was looking at is usually called ghosting, or plate-out, and it is one of the few problems in this trade where the honest answer is that cleaning is the wrong tool.

What is putting the grey there?

Fine airborne particulate landing on a surface that is very slightly colder than the air around it.

The mechanism is thermophoresis. Sub-micron particles get pushed down a temperature gradient, from warm towards cool, because the gas molecules on the hot side of a particle strike it harder than those on the cold side. Small force, slow work – which is exactly why the result is a faint even grey rather than a splash. The particles are too light to settle out under gravity in any useful time, so they drift until they find somewhere colder to stick.

Once a particle lands on a fibrous surface it is not coming off again in the way dust comes off a desk. It has keyed into the texture. The tile is the filter.

Why does it follow the grid lines and the diffusers?

Because those are the cold spots and the turbulent spots.

The exposed grid is steel. Steel conducts heat far better than a mineral fibre tile does, so the grid and the strip of tile immediately adjacent to it run measurably cooler than the middle of the tile, and the deposition concentrates there. That gives you the grid-following bands, and it is the same physics that produces thermal tracking lines on plasterboard ceilings above unfilled joists in flats.

Around the diffuser you get a second effect on top of the first. Air leaving a supply diffuser is cooler than room air and it is moving fast, so you have both a temperature gradient and turbulent impaction driving particles onto the tile face in the throw pattern of the jet. The halo takes the shape of the diffuser’s air pattern rather than a neat circle, which is why four-way blow gives you a rough cross and a linear slot gives you a stripe.

Stand under it and look at which way the smudge points. You will know within ten seconds whether the grille is throwing evenly.

Where is all that particulate coming from?

Three sources cover almost every London office I inspect, and they are not equally weighted.

Combustion products from outside dominate. Diesel particulate is overwhelmingly sub-micron, it comes in through the fresh air intake and through every door onto the street, and in a building on Clerkenwell Road with a taxi rank outside there is a constant supply of it. This is the one nobody can do much about beyond filtration, and it is the one that produces the broad even greying across a whole floorplate.

Then there is whatever the occupants are burning. Candles are the great unacknowledged culprit in small offices and in reception areas, and a cheap paraffin candle with an oversized wick produces staggering quantities of soot. Reed diffusers and oil burners are worse than people think. In Hatton Garden specifically the picture is different again, because the workshops upstairs are running polishing motors that throw rouge and tripoli compound into the air alongside the burnout kilns, and that workshop load finds its way into the ceiling void of the office below through service penetrations that were never sealed.

Laser toner is the third, and it is the one people always name first.

Is the printer to blame?

Partly, and less than its reputation suggests.

A laser printer emits ultrafine particles during fusing, mostly volatile compounds from the toner resin that condense into particles rather than toner powder itself escaping. The emission is real and it is measurable. What it produces on a ceiling is a localised darkening in the bay above and slightly downstream of the machine, rather than the floor-wide greying most clients blame on it.

If the staining is uniform across 3,000 square feet, the printer is not your problem and moving it will not help. If there is one noticeably darker bay above the print room door, move the machine and put it somewhere with a local extract.

Why does cleaning a mineral fibre tile usually wreck it?

Because you are cleaning a soft, absorbent, air-permeable board that was never designed to be wet.

A standard mineral fibre tile is compressed mineral wool with a binder, faced with a thin painted or fissured decorative layer. In a plenum return system, air passes through the tile itself. The tile is the filter, and it has been filtering for years, so the soiling sits within the surface layer rather than on top of it.

Wet cleaning does four things, all bad. Water swells the binder and the tile sags in the grid, sometimes visibly within a day. The surface nap lifts and the fissured texture roughens, so the tile holds soil faster afterwards. You get tide marks at the wet edge that are more conspicuous than the original stain. And if you lift enough soil to make a difference, you lift some of the surface coating with it, leaving a patch of different sheen.

Then there is the performance question that nobody asks. That tile has a declared fire classification and a declared sound absorption value, both of which depend on its surface and its perforation. Coating it, painting it or soaking it changes both. A ceiling that has been rolled with emulsion to hide staining has usually lost a chunk of its acoustic absorption and may no longer meet the classification it was specified under.

What about dry sponge and bleach pens?

Dry chemical sponge, the vulcanised rubber block, is worth a try and is the only thing I would put near a stained tile as a first move. It lifts loose surface soot without moisture and it will visibly improve a tile that has light deposition. On plate-out that has been building for three years it does very little, because the particles are keyed into the fibre rather than resting on it.

Bleach and oxidising pens will lighten a mark. They will also lighten it past the surrounding tile, and a bright white oval on a four-year-old ceiling reads as damage rather than cleanliness. I have watched a client spend a morning doing this and then ask us to make it look normal again, which is not a thing that can be done.

So what should you order instead?

New tiles, in most cases, and this costs me the survey and the labour I would otherwise be quoting for.

Take one tile down and turn it over. The manufacturer, product name and edge detail are printed on the reverse face, along with a batch reference. That is the whole specification and you need all of it, because ordering the right product with the wrong edge detail gives you a box of tiles that will not sit in your grid.

What you are reading off the back is a family name and an edge type. In UK offices you will most often find Zentia, which is the business that was formerly Armstrong’s UK ceilings operation, with Dune eVo or Ultima+ being the common tiles. Rockfon Sonar and Blanka turn up constantly. AMF Thermatex less often. The edge detail will be Board, which is a square edge sitting on a 24mm exposed grid, or Tegular, which is rebated and steps down below the grid, or MicroLook, which is the Tegular equivalent for a narrow 15mm grid. Get that wrong and nothing else matters.

Check the module too. Most floorplates are 600 by 600, some are 1200 by 600, and a surprising number of older buildings are still on imperial modules that need cutting from oversize boards.

How do you avoid a chequerboard?

By replacing whole bays rather than individual tiles.

An installed tile yellows under UV and ages towards cream. A new one out of the box is bright white. Drop one new tile into a run of eight-year-old ones and it announces itself from across the room, and you have swapped a grey mark for a white one.

Replace a defined area to a grid line, so the eye reads a boundary rather than a patch. Where a client has spare tiles in a store cupboard from the original fit-out, those are gold, because they have aged in the dark and will still be a shade off but far closer than new stock. Rotating tiles from a plant room or a store area into the visible floor and putting the new bright ones where nobody looks is the oldest trick on the list and it works.

How do you stop the new tiles going the same way?

Deal with the air, because the tile is the filter and a new one starts filtering the moment you fit it.

Get the fresh air filters checked against what the system was specified for. Filter classification moved to ISO 16890 in 2018, so an old maintenance schedule may still be describing grades that the replacement cartridges no longer use, and buildings quietly drop a class when somebody orders on price. Then put the diffuser faces on the routine cleaning schedule rather than the annual list, because a grille loaded with felted particulate throws unevenly and concentrates the deposition further.

And the holes nobody sealed?

Those matter more than the filters in a mixed-use building.

Service penetrations between an upper-floor workshop and the ceiling void below – riser openings, cable drops, the gap around a soil pipe – carry more contaminated air than any intake does. The Hatton Garden ceiling took forty-one tiles and half a day. Sealing the penetrations took a morning with expanding collar and fire-rated sealant, and that was the part that mattered.