Chemistry and Science Behind Carbon Monoxide Poisoning

In 1895, the Scottish physiologist John Scott Haldane published a 17-page paper in the Journal of Physiology (volume 18, pages 201-217, read at the Physiological Society on 18 July 1895) that established carbon monoxide as the gas killing miners in "afterdamp" disasters and demonstrated that pressurized oxygen reverses the binding. The chemistry behind his result is the reason a concentration the lungs barely notice can become dangerous quickly: CO holds onto red blood cells much more strongly than oxygen does, so even small amounts in the air translate to large drops in the oxygen the body can use. This page explains the chemistry in plain language for homeowners and renters: why your senses cannot warn you, why a pulse oximeter is not a CO meter, what the molecule does inside furnaces and generators, and what alarms exist for.

Quick Science Summary

• CO holds onto red blood cells far more tightly than oxygen does, so a small amount in the air leads to a large drop in usable oxygen.
• The remaining oxygen on bound cells also releases less readily to tissues, which compounds the effect.
• A standard pulse oximeter cannot tell CO-bound hemoglobin from oxygen-bound and can read normal during a real exposure.
• Children, elderly, and pregnant occupants are general vulnerable groups; any suspected exposure during pregnancy is a reason to seek emergency medical evaluation right away.
• UL 2034 alarms are tuned to the time-and-concentration combinations where symptoms become likely, which is why they sound earlier at higher concentrations and tolerate brief low-concentration excursions.

What Carbon Monoxide Does Inside Cells

Hemoglobin is the iron-containing protein inside red blood cells that picks up oxygen in the lungs and releases it to tissues. Carbon monoxide attaches to that same iron site. Every site filled with CO is a site that cannot carry oxygen, so the blood reaching the body delivers less. The bond is also stable enough that fresh air alone takes hours to undo it. The tissues that consume oxygen fastest, the brain and the heart, feel the loss first, which is why the early signs of CO exposure are headaches, dizziness, and confusion rather than something more specific.

Why CO Binds So Much Tighter Than Oxygen

The reason CO is dangerous at concentrations that other gases would barely register is structural: the CO molecule slips into the same iron site oxygen uses, but with a much stronger grip. Sources that have tried to quantify the difference give a range rather than a single number, because the exact figure depends on experimental conditions and the species of hemoglobin studied. The practical point is the one that matters at home: once CO occupies the binding site, it does not come off quickly without help. Room air clears it over hours; an emergency department can speed clearance with high-flow oxygen or, in severe cases, with specialized treatment. The slow release is the chemistry behind every safety rule on this site: leave the exposure first, get medical care if symptoms warrant, and have the source identified and fixed before reoccupying the space.

What This Chemistry Means at the Detector

UL 2034, the US performance standard for residential CO alarms, sets time-to-alarm windows tuned to the concentration-and-duration combinations where symptoms become likely. The alarm must trigger at 70 ppm between 60 and 240 minutes, at 150 ppm between 10 and 50 minutes, and at 400 ppm between 4 and 15 minutes. The thresholds are equipment specifications, not personal-decision triggers. They exist so the alarm sounds before sustained exposure has become serious, while tolerating brief low-concentration excursions that healthy occupants would not feel.

A fingertip pulse oximeter does not give the same protection. The device is optimized to measure oxygen-bound hemoglobin and cannot reliably distinguish it from CO-bound hemoglobin, so it can read normal during a real exposure. Treat the pattern of symptoms (worsening indoors, improving outside, multiple people affected) as the actionable signal, not the pulse-ox reading.

The Legislatures Reached the Same Conclusion by Refusing to Write One

The mixing behaviour above has a consequence you can test outside the laboratory: if carbon monoxide diffused into a stable layer, a legislature drafting an alarm requirement would have to say where in the room the layer sits. We hold the carbon monoxide instrument for all 51 US jurisdictions and went looking for the number. Not one of them writes a mounting height into law. Nine records raise the subject at all, and every one hands it to the manufacturer or to the adopted code rather than fixing it: New Mexico puts it plainly, that mounting height is set by the manufacturer rather than by the code, so the manual that came with the unit governs the last few inches. Oklahoma is the clearest negative, because its health rule requires a detector in each qualifying guest room and then stops, prescribing no placement height, no listing standard and no power source.

The search was run against other phrasings before that absence was believed, since an empty result proves nothing until it has shown it can find something. Fourteen records contain the word height, and once the nine are set aside every remaining use is about the height of the building, not the detector: Hawaii and Pennsylvania defining a dwelling as not more than three stories, Mississippi's test that turns on how tall the property is, Ohio confirming that its retrofit duty does not depend on building height at all. The 93 municipal codes were scanned the same way and return two measurements in inches, both about the room rather than the device: Salt Lake City lets a corridor ceiling drop to 6 feet 4 inches on condition that a smoke and a carbon monoxide detector are installed, which trades headroom for detection, and Lawrence, Kansas fixes an 84-inch ceiling inside a habitable-room specification that happens to sit beside its alarm rule.

Read that as convergence rather than as a gap. The physics says the height of the device is not what decides whether it sees the gas; the drafting, across fifty-one independent instruments and ninety-three municipal codes, declines to legislate a number that would not do any work. What the same instruments do legislate is horizontal distance from where people sleep, which is the measurement that actually changes how long the gas takes to reach the sensor.

Where the Chemistry Happens in Real Appliances

Carbon monoxide is the partial-combustion partner of carbon dioxide. When combustion is complete, the carbon in the fuel ends up as CO2 plus heat plus water vapor. When combustion is incomplete (not enough oxygen, fuel-air mixture wrong, flame temperature too low, exhaust pulled back into the burner), the carbon stops at CO. Each common residential source fails in a specific way:

• Furnace with a cracked heat exchanger. The heat exchanger is supposed to keep combustion gases inside the metal pathway and dump heat into the airflow on the other side. A crack lets CO migrate into the conditioned air, which then distributes the gas throughout the home through the duct system. This is why annual furnace inspections check for hairline cracks.
• Water heater with a blocked or backdrafting flue. The combustion chamber relies on draft to pull exhaust up the flue. A blocked vent (bird nest, ice cap, structural damage) or a depressurized utility room (range hood, dryer, attic fan pulling air faster than makeup ventilation supplies) lets exhaust spill into the room. Many fatalities involve water heaters in unfinished basements where CO accumulates and migrates upward through floor penetrations.
• Gas range used for heating. Stovetop and oven burners are tuned for short-duration cooking. Run for hours, even a well-maintained burner produces enough CO to push a poorly ventilated kitchen past detector thresholds.
• Portable generator. Small engines run rich (extra fuel) at low load, and CPSC puts their CO emission rates "on the order of hundreds of times the CO emission rates of gasoline powered automobiles". CPSC counts "at least 1,332 CO-related consumer deaths involving portable generators" from 2004 through 2021, "an average of about 74 lives lost annually, with thousands of non-fatal poisonings of consumers per year", and notes that fatalities have risen in recent years. ANSI/PGMA G300-2018 requires a shutoff that stops the engine before CO measured an inch or two above the machine passes 400 ppm as a 10-minute rolling average, or 800 ppm in a single reading; older units lack that interlock.
• Charcoal grill. Smoldering charcoal (not flaming) produces the highest CO yield of common cooking fuels. This is why charcoal grills must never be used indoors, in a garage, or on a covered porch, and why "the fire is almost out" is the most dangerous phase.

Common Misconceptions, Rebuilt

• "A normal pulse oximeter reading means I am fine." Standard pulse oximeters cannot distinguish CO-bound hemoglobin from oxygen-bound. A normal reading does not rule out CO exposure when the pattern of symptoms fits.
• "CO is heavier than air and sinks to the floor." Molecular weights are CO 28.01 g/mol versus air about 28.97 g/mol, a difference under 4 percent. CO mixes through indoor air rather than stratifying. Place alarms per manufacturer instructions on each level.
• "If I cannot smell anything, the air is safe." CO has no odor at any concentration; the smell of incomplete combustion (smoke, exhaust) often accompanies CO but does not have to. The smell-like page covers this in more detail.
• "Low levels of CO are not a big deal." Sustained low-level exposure can still produce real symptoms, and vulnerable groups (children, elderly, pregnant occupants) feel it sooner. Detector alarm thresholds are set so brief excursions do not trip the alarm, not so chronic exposure goes undetected.
• "Once I feel better, the CO is gone." Symptoms can ease while CO is still bound in the blood, which is why removing the exposure plus medical evaluation (rather than just feeling better in fresh air) is the right response after a real event.

Glossary

• Carbon monoxide (CO): A gas without odor, color, or taste produced by incomplete combustion.
• Hemoglobin: Iron-containing protein in red blood cells that transports oxygen.
• Hypoxia: Reduced oxygen delivery to tissues.
• Pulse oximeter: A device that estimates oxygen saturation by light absorption; cannot reliably distinguish CO-bound hemoglobin from oxygen-bound and is not a CO detector.
• UL 2034: The US performance standard for residential CO alarms; defines time-to-alarm thresholds the device must meet.
• NFPA 72 (formerly NFPA 720): The installation code that defines residential CO alarm placement, in Chapter 29. NFPA 720 was the CO-specific standard, published for the last time in 2015 and withdrawn by a Standards Council decision of 14 August 2018 that moved its contents across. The requirements have appeared in NFPA 72 since the 2019 edition.
• Incomplete combustion: Fuel burning that produces CO instead of fully converting to CO2.
• Backdrafting: Combustion gases flowing back into indoor space instead of venting outdoors.

Sources & References

• CPSC, Safety Standard for Portable Generators, supplemental notice of proposed rulemaking, 20 April 2023, for the generator CO emission rate compared with automobiles and the ANSI/PGMA G300-2018 shutoff threshold
• ATSDR/CDC, Toxicological Profile for Carbon Monoxide
• Sturrock and Brown, John Scott Haldane: The Father of Oxygen Therapy (historical biography)
• CDC, Carbon Monoxide Poisoning Prevention (overview)
• UL 2034, Single and Multiple Station Carbon Monoxide Alarms, the standard itself, which UL sells
• CPSC, Carbon Monoxide Alarm Conformance Testing to UL 2034, Table 3, the free source for the time-to-alarm windows quoted above
• NFPA 72, National Fire Alarm and Signaling Code (Chapter 29, residential CO alarm placement)
• CPSC, Safety Standard for Portable Generators, 88 FR 24336 (20 April 2023), for the generator death counts and the emission-rate comparison quoted above

Last updated: May 17, 2026