What the number describes
An adult at rest exhales roughly 15 to 20 litres of carbon dioxide per hour. In a 25 m3 bedroom with the door and the window shut, that volume has nowhere to go. Outdoor air now sits at around 420 ppm, parts per million. A single-occupancy bedroom with no air change routinely passes 1,200 ppm by morning. With two people, 2,000 ppm is unremarkable. We have recorded above 3,000 ppm in recent, tightly built flats where the ventilation had been switched off because of its noise, or blocked by a clogged extract grille nobody had cleaned in a decade.
At these levels CO2 is not a poison. Its usefulness lies elsewhere: it is the least expensive marker of air change available. Where CO2 accumulates, everything else accumulates too, moisture, volatile organic compounds released by furniture and paint, particles, allergens, and depending on the ground beneath the building, radon, the second leading cause of lung cancer. The sensor measures none of those. It measures the condition that lets them settle, and it does so continuously, which no spot sample can.
Three thresholds of three different kinds
| Value | What it is | What to do with it |
|---|---|---|
| 800 ppm | A comfort target used in several indoor air quality frameworks | A reasonable aim for a bedroom, reachable almost anywhere |
| 1,000 ppm | An old reference, inherited from Pettenkofer’s 19th century work and carried into many guidelines | Above it, ventilation is not keeping up with occupancy |
| 1,400 ppm | In France, the confinement index value above which an investigation is required in buildings open to the public | An action trigger, not a health limit |
None of these numbers describes a hazard. They describe insufficient air flow, expressed in the easiest unit to read.
On the direct effects of CO2 itself, restraint is in order. Chamber studies have measured reduced performance at high concentrations. Others, designed to separate CO2 from the organic compounds people themselves emit, found nothing. The question is open. What holds up slightly better concerns sleep: in protocols where bedroom air change is altered without the sleeper’s knowledge, better ventilation goes with sleep rated as more restorative and marginally better next-morning test scores. Sample sizes are small, exposures short, and part of the outcome is self-reported. We treat this as a consistent signal, not as a demonstration.
What a 100 euro sensor actually shows
Two families of device sit side by side on the shelf and have nothing in common. Non-dispersive infrared sensors, labelled NDIR, measure CO2. Devices labelled eCO2, or CO2 equivalent, measure volatile organic compounds and infer a figure by correlation. In a bedroom that figure drifts the moment someone opens a bottle of perfume or wipes down a surface. The datasheet has to say NDIR. Below 80 euros, it rarely does.
Placement matters as much as the instrument: head height, 50 cm to a metre from the pillow, never in the direct path of exhaled breath, never on the window sill. Record five to seven consecutive nights, noting each evening whether the door was open or shut, so there is something to compare against.
The curve then reads in three parts. The slope after lights out gives the real tightness of the room. The plateau, where one appears, shows the balance between production and air change. The morning peak gives the worst moment of the night, which is the one that counts. Finally, the drop that follows the door opening measures the actual flow through that opening, which no theoretical calculation delivers as quickly.
What can be fixed without building work
Most of the bedrooms we measure come back below 1,000 ppm with three actions, none of which needs a contractor.
- Leave the door ajar by 10 to 15 cm. In our records the morning peak typically falls by 300 to 600 ppm. It is the highest yield correction available, and it costs nothing.
- Check the undercut of the door. A gap of 1 to 2 cm below the leaf is what allows air to sweep from the room towards the extract points. Thick carpet or a draught excluder cancels it.
- Clean the trickle vents in the window frames and the extract grilles in the bathroom and the toilet. A clogged grille can move almost no air with nothing to signal it. A sheet of paper held against the grille by suction is a crude but immediate check.
Then comes the window. Opening it 2 to 3 cm, or tilting it, almost always solves the problem and creates another. On a street facade, opening exposes the room to road traffic noise, whose cardiovascular risk is established from 53 dB. Depending on the season it also admits cold air and pollen. That trade-off is settled at the address, not in a guide.
When heat recovery ventilation is justified
A balanced mechanical ventilation system with heat recovery supplies filtered fresh air and reclaims heat from the extracted air. It solves CO2 without opening anything. It becomes justified when several conditions stack up:
- a genuinely airtight dwelling, confirmed by a blower door test, in practice a recent build or a deep retrofit
- a facade exposed to noise beyond the reference above, or to loaded outdoor air. Fine particles are the leading environmental risk factor worldwide, so fine filtration has a value of its own here, independent of CO2
- an occupancy pattern that rules out opening windows: seasonal allergies, a child’s bedroom, daytime sleep after night work.
In a retrofit, expect something like 5,000 to 12,000 euros for a flat, depending on duct runs and the ceiling space available. The counterpart is real maintenance: filters twice a year, ducts to clean, and air flows commissioned by a professional after installation rather than assumed before it. A poorly balanced system performs worse than a window left ajar and costs a hundred times more. Conversely, in an older building with leaky joinery, measurement often shows there is nothing to solve.
What we do
Seven nights of recording at the sleeping position, alongside temperature and relative humidity, which we aim to hold between 40 and 60 per cent. We report the curve, the peak, the time spent above 1,000 ppm, then rank the corrections by measured effect and by cost. Lif8 measures places. We are not a healthcare provider and we make no diagnosis: sleep that stays unrefreshing over time is a matter for a doctor, not for a sensor.
