What Welding Fume Actually Contains and What It Needs
Why this matters
Welding fume is not one substance and it does not have one control. What comes off an arc is decided by three inputs - the consumable, the base metal, and whatever is on the surface - and each constituent needs a control matched to what it does and how it gets in. The trap is that the plume's appearance is set almost entirely by iron oxide, which is the least of your problems, while the constituents that actually injure people are invisible inside it and two of them give no warning at all until hours later. So "it was not smoking much" is not an assessment, and a glove, a leather sleeve and a welding helmet are protection for a different route entirely.
The controls this card names, before the detail: local exhaust captured at the arc is the primary engineering control for the particulate fraction, and where capture is not achievable the fallback is a respirator selected and issued under a written program meeting 29 CFR 1910.134, which includes medical evaluation, fit testing and training. A welding helmet is not respiratory protection. 29 CFR 1910.252(c) in general industry, and 29 CFR 1926.353 in construction, set the ventilation and protection requirements for welding, cutting and heating, and they single out specific substances (including fluorine compounds, zinc, lead, beryllium, cadmium and mercury) plus work in confined spaces for stricter treatment. Never ventilate a space with oxygen. Oxygen enrichment turns clothing and oily surfaces into a fire that cannot be put out by stepping back, and that prohibition is explicit in 1910.252(c).
What fume actually is
An arc boils metal. The vapour leaves the pool, hits cooler air, oxidises and condenses into solid particles that are largely sub-micron, small enough to travel to the deep lung and small enough to hang in a room long after the arc is off. That is the particulate fraction, and it is metal oxides.
Riding with it is a gas fraction that no filter respirator cartridge for particulates removes. Ultraviolet from the arc makes ozone out of oxygen in the surrounding air and nitrogen dioxide out of nitrogen. Carbon dioxide in a shielding gas dissociates in the arc and produces carbon monoxide. And ultraviolet decomposes chlorinated solvent vapour into phosgene and other irritants, which is why degreasing with a chlorinated product anywhere its vapour can drift to an arc is prohibited practice rather than a judgement call.
The two fractions matter separately because they need different controls. Ventilation dilutes and removes both. A particulate filter handles one and not the other.
The three sources, and the one people blame
| Source | What it puts in the plume | Notes |
|---|---|---|
| The consumable | Manganese and iron oxide from essentially every mild steel consumable, plus fluorides from low-hydrogen coverings and fluxes, plus silicates | Manganese is present on ordinary carbon steel work, not just on exotic jobs |
| The base metal | Chromium and nickel from stainless, copper from copper alloys, aluminium and magnesium from light alloys | Hexavalent chromium is generated from stainless and from chromate-bearing material |
| The surface coating | Zinc from galvanizing, cadmium from plating, lead from old paint, hexavalent chromium from chromate primers, and the decomposition products of any organic coating | The worst constituents on a typical field job come from here |
| The process, which is not a source | Fume rate and where the plume goes | Process changes the quantity and the gas fraction, not the metals |
That last row is the one shops get backwards. Switching process changes how much fume is generated per unit of deposited metal and changes the gas fraction, but the metals in the plume come from the three real sources. A process change is not a substitute for taking the coating off.
Why appearance and smell are not a guide
Cadmium is the case that has killed people. Cadmium-plated hardware looks like bright zinc plating to the eye. Welding or cutting it produces cadmium oxide fume, which does not smell dangerous and does not sting, and the serious respiratory injury it can cause is delayed by hours, by which time the person is at home. Cadmium is regulated under 29 CFR 1910.1027 in general industry and 29 CFR 1926.1127 in construction, and 1910.252(c) gives it specific treatment. Where plated hardware of unknown origin is going to see an arc or a torch, the answer is to remove the hardware, not to ventilate harder.
Zinc is the one people notice, and it is not the worst. Metal fume fever from galvanizing gives chills, aching and fever some hours after exposure and passes. Because it is memorable and self-limiting, shops calibrate their sense of danger on it, and then apply the same instinct to a fume that does not announce itself.
Manganese gives no acute signal at all. OSHA's limit for manganese compounds and fume in 29 CFR 1910.1000 Table Z-1 is a ceiling value rather than an eight-hour average, so it is not a figure you average your way under. Independent consensus bodies publish substantially lower exposure limits than the OSHA value, and those bind only where your employer's own written program or a contract adopts them, but they are the reason a shop should not treat compliance with a decades-old ceiling as the definition of adequate.
Hexavalent chromium is a carcinogen with an inhalation route and no odour. It is regulated under 29 CFR 1910.1026 in general industry and 29 CFR 1926.1126 in construction, with its own exposure limit, and it is generated by welding and by grinding stainless and chromate primers alike. Eye protection, gloves and a face shield do nothing about it, and that pairing gets written into safety blocks constantly.
Controls, matched to route and to constituent
- Eliminate the constituent. Remove the coating back from the joint before the arc. This is the only control that works on the coating-derived metals rather than managing them, and it usually improves weld quality at the same time, because the same coating causes porosity.
- Capture at the arc. A fume extraction gun, a moveable hood positioned close, or a downdraft or backdraft table. Capture velocity falls off steeply with distance, so a hood parked three feet away is decoration.
- Position the work and the welder. The plume rises. A welder leaning over the joint puts their breathing zone directly in it, and moving the head out of the rising column is free and effective.
- General ventilation. Dilutes what capture missed, and handles the gas fraction. It is not a substitute for capture at the source.
- Respiratory protection, last. Selected for the constituent, issued under a written program meeting 29 CFR 1910.134. In a confined space, 1910.252(c) requires local exhaust or airline respirators and the space itself is governed by 29 CFR 1910.146.
Worked example: one galvanized handrail repair, assessed
A galvanized steel guard rail in a partially enclosed stairwell needs a mild steel patch welded in with a flux-cored process using a carbon dioxide bearing shielding gas. Build the constituent list from the three sources, then assign a control to each.
Constituents identified: five.
- Iron oxide, from the base metal and the consumable. This is what makes the plume visible and it is the least of the list.
- Manganese, from the consumable. Present on every mild steel job, ceiling-limited in Table Z-1, no acute warning.
- Zinc oxide, from the hot-dip galvanizing.
- Fluorides, from the flux in the cored consumable.
- Gas fraction: carbon monoxide from the carbon dioxide in the shielding gas, plus ozone and nitrogen dioxide generated by the arc's ultraviolet in the surrounding air.
Controls assigned, one per route.
- Zinc: eliminated, not managed. The galvanizing comes off both sides of the joint and back a couple of inches from it before any arc is struck, with the actual distance owned by the governing document or the coating manufacturer's guidance. That removal is itself an inhalation exposure, because grinding galvanizing puts zinc dust in the air, so it is done with capture at the tool or with respiratory protection under the 1910.134 program, not with a face shield.
- Manganese and fluorides: capture at the arc, with an extraction gun or a hood positioned close, plus head out of the plume.
- Carbon monoxide: ventilation, plus a personal monitor. A stairwell with restricted airflow is exactly where carbon monoxide accumulates, and it is odourless. If the stairwell meets the definition of a confined space, 1910.252(c) requires local exhaust or airline respirators and 1910.146 governs entry.
- Ozone and nitrogen dioxide: ventilation. No particulate filter touches them.
- Iron oxide: handled by the same capture as manganese, and it needs no separate control.
Five constituents, four distinct control types, and only one elimination. The elimination is the zinc, and it is also the one that fixes the porosity the coating would otherwise cause, so it pays twice. Nothing on this list is addressed by gloves, sleeves or the welding helmet.
Check the assessment against the rules the siblings state
- Count re-derived from the printed list, last mention first. The closing line says five constituents and four control types: the list enumerates iron oxide, manganese, zinc oxide, fluorides and the gas fraction, which is 5. The controls used are elimination, capture at the arc, ventilation, and monitoring in the enclosed space, which is 4, with iron oxide sharing capture rather than adding a fifth.
- Every control is matched to a route, not to a hazard class. Zinc, manganese, fluorides and iron oxide are inhalation and get removal, capture or both. Carbon monoxide, ozone and nitrogen dioxide are inhalation gases and get ventilation, because a particulate filter does not reach them. No entry on the list is answered with eye or hand protection.
- The hazard the instruction creates is named. The instruction to grind the galvanizing off carries its own zinc dust exposure in the same clause, with capture at the tool.
- Regulatory numbers are not restated where they are gated. The manganese figure is named as a ceiling in Table Z-1 rather than as an averaged limit, and the lower consensus limits are given their adoption path (an employer program or a contract) rather than presented as law.
- Direction check against the siblings. The porosity card has coatings as a porosity source and calls for removing the coating back from the joint before welding. This card assigns the same action for a different reason, and both name removal rather than more shielding gas or more ventilation as the primary control.
References
- 29 CFR 1910.252(c) (general industry) and 29 CFR 1926.353 (construction) for ventilation and protection in welding, cutting and heating, including confined spaces and the substances singled out for stricter treatment
- 29 CFR 1910.1000 Table Z-1 for manganese and other limits; 29 CFR 1910.1026 and 1926.1126 for hexavalent chromium; 29 CFR 1910.1027 and 1926.1127 for cadmium; 29 CFR 1910.1025 and 1926.62 for lead
- 29 CFR 1910.134 for respirator selection, medical evaluation, fit testing and the written program; 29 CFR 1910.146 for permit-required confined spaces
- ANSI/AWS Z49.1, safety in welding, cutting and allied processes, in the edition your employer's written program adopts or your contract invokes
- See related: Why Porosity Appears and What It Tells You; Setting Up a Weld So the Arc Is the Only Hazard; What a Hot Work Permit Is Actually Protecting