Porosity in Welding: Causes, the Three Gases, Location Diagnosis, and Fixes

Porosity in welding is gas that gets trapped inside the molten weld pool and freezes there as a void — a bubble that lost its race against solidifying metal. That is the entire defect in one sentence. The rest of this article covers where that gas comes from, how to read the pattern of holes left behind, what the codes permit, why you cannot simply weld the problem away, and how to stop it from starting.

Last February a two-man fabricator I know got a call about a stainless rinse tank that had failed its fill test. The shop had built a 240-gallon 304 tank for a poultry processor, and every fillet on the outside looked textbook: consistent ripple, clean color, no undercut. The welder had twenty-two years in food-grade work and nobody questioned the beads. Then they filled the tank, marked the seams with chalk, and within ten minutes a dark wet line appeared along one sidewall-to-shell joint. They cut the seam out with a plasma torch, and the cross-section looked like bread — a chain of small round cavities running through the root pass, invisible from either side before cutting.

The first argument in the shop was about the machine. Somebody wanted the voltage turned up. Somebody else blamed the wire feed speed. The welder quietly suspected the draft from the roll-up door the shop had started leaving open that week. He was right, but only after they had wasted an afternoon swapping settings and running test beads on scrap that all came out fine — because the tests were run with the door closed. When they welded a coupon with the door open, the porosity came right back, scattered through the bead like pepper.

That story contains the two habits that will save you the most time with this defect. First: the answer is almost never in the machine dials. Second: the pattern of the holes tells you where to look before you touch anything. A porous weld is not a mystery — it is a record of gas that arrived faster than it could escape, and that record is readable by anyone who learns the vocabulary.

This is the most common gas-related defect in arc welding, and the one most argued about on inspection reports and shop floors. It shows up in every process, on every metal, at every skill level — and once you can name the gas source, the fix is usually cheap.

What Porosity Actually Is, Physically

Solubility is the whole story. Molten steel will dissolve far more gas than solid steel can hold. While the puddle is liquid, hydrogen, nitrogen, and oxygen sit comfortably in solution, dissolved between the iron atoms like sugar in warm water. The moment freezing begins, the metal’s capacity to hold those gases drops sharply and continuously as the temperature falls through the freezing range. Gas that was happy a second ago is now being squeezed out of a solution that can no longer keep it. It has to go somewhere, and where it goes is the solidification front: it forms bubbles at the advancing boundary between liquid and solid.

Everything comes down to one race. A bubble forming in the puddle can escape. Bubbles are buoyant; they rise, they reach the surface, they pop, and nothing is wrong. The defect only exists when the metal around that bubble locks solid before the bubble gets out. Faster gas arrival, slower bubble escape, or quicker freezing — any of the three tips the race toward a trapped void, and most welding problems that cause porosity do it by pushing one of those three levers without ever announcing it.

Because of that, porosity is often invisible where you are standing. The external bead can be perfect — correct profile, correct color, no visible pit — while the interior is riddled with cavities. An inspector looking at the finished weld with a flashlight is seeing the cover, not the contents. That is why volumetric methods like radiography exist, and it is why a welder who is proud of a bead can still be genuinely shocked when the film comes back showing a constellation of dots through the center of every pass.

Porosity vs. other defects. Keep the categories straight. Porosity is a rounded or elongated gas cavity; slag inclusion is trapped flux, not gas; a crater pipe is a shrinkage void at the bead end; lack of fusion is a planar defect where the weld never wetted the base metal. Each responds to different fixes — lump them together as “bad welds” and you will chase your tail.

Why Aluminum Punishes You Harder Than Steel

Aluminum’s problem with porosity is structural, not temperamental. The ratio of hydrogen solubility in liquid aluminum to its solubility in solid aluminum is roughly 0.05, while the same ratio for hydrogen in steel is around 0.4. Read those numbers again: when aluminum freezes, essentially all of the dissolved hydrogen it was carrying has to come out, and it has to come out in the narrow window between liquid and solid.

On top of that, aluminum rebuilds its oxide film within minutes of cleaning, that film is hydrated — it carries water — and the puddle temperature is far lower than steel’s, so the escape window is short. The practical result: a slightly dirty brush, a humid afternoon, or a joint that sat uncovered for an hour will make aluminum porous in ways mild steel quietly tolerates. If you want the metal-side details of that surface prep, our guide on what gas do you use to weld aluminum walks through shielding choices alongside it, and what color tungsten for aluminum covers the GTAW-side setup errors that compound the same problem.

Four Ways a Bubble Gets Trapped

Knowing that bubbles form is only half the picture. The mechanisms that trap them differ, and each one points at a different cause, so it is worth separating them before moving on to sources.

• Keyhole collapse. In high-energy welding — deep penetration MIG on aluminum, any process where a keyhole forms — the keyhole is a cavity of vapor surrounded by liquid metal. If it shuts while gas is still inside, that gas is sealed into the finished bead. The rapidly advancing solidification front closes faster than the vapor can climb out. This is the dominant bubble-initiation mechanism in keyhole welding.
• Surface tension pinning. A bubble that reaches the pool surface normally breaks free and is done. In a small, fast-moving puddle, or when gas is generating faster than it can leave, the bubble never gets clear of the surface film — it is held, recirculated by the flow patterns in the puddle, and frozen wherever the front catches it.
• Rapid solidification. The simplest mechanism and the one you control most directly. A pool that freezes quickly gives gas less time to escape, whatever the source. This is why short-circuit transfer is more porosity-prone than spray transfer, and why cold plate, cold wire, and hurried travel speeds all make things worse.
• Geometric traps. Crevices and blind pockets hold gas by shape alone. The gap under the vertical leg of a T-joint welded from both sides is the classic example: gas generated in that corner has no upward path, and the second pass seals it in permanently. Poor joint design can build porosity into a weld before anyone strikes an arc.

Quick Note: If a joint design has a pocket that cannot vent upward, no shielding gas setting will save that weld. Fix the joint — add a vent or change the weld sequence — before you touch the flowmeter.

The Three Gases Behind Almost Every Pore

Of every gas you might encounter in a welding environment, only three dissolve in steel in meaningful amounts: nitrogen, hydrogen, and oxygen. Learning that list is genuinely useful because it eliminates most of the advice you will find online. If someone blames argon for porosity, or helium, or carbon dioxide “dissolving” into the puddle, they have the chemistry backwards — and the fix they recommend will be wrong too.

Nitrogen: Air That Sneaked Past the Shield

Nitrogen from the atmosphere is the leading cause of porosity in gas-shielded processes — MIG and TIG above all. The shielding gas exists for one job: to push air away from the puddle and keep it there while the metal is liquid. Anything that lets air back in gives nitrogen a direct route into solution, and nitrogen that dissolves in the puddle has no useful escape once freezing begins.

What surprises people is the number of routes. A pinhole in a hose, a regulator fitting that weeps only when the bottle is cold, a loose gun connection, an extraction arm parked in the wrong place, a fan pointed across the bay, a cylinder that ran empty mid-shift — all of them deliver the same result: air where argon was supposed to be. Excessive stickout does it too, because the arc stretches beyond the reach of the nozzle’s shield and welds in ambient air. So does an extreme work angle or heavy weaving, which folds air into the trailing edge of the puddle as the pool rolls.

The oddity, the one worth tattooing inside your eyelids, is that too much shielding gas causes porosity just as surely as too little. That deserves its own section.

Hydrogen: Moisture Wearing a Disguise

Hydrogen is the gas to worry about with coated electrodes, damp surfaces, and aluminum. Its source is always water or a water-bearing compound, and water finds its way into welding work in more forms than people expect: electrodes left out of the rod oven, plate stored under a tarp, condensation on cold stock brought into a warm shop, mill scale and rust, grease and cutting fluids, drawing lubricants left on wire, and the hydrated oxide film aluminum rebuilds within minutes of brushing.

Electrode coatings are where hydrogen enters the conversation with numbers. Cellulosic rods such as E6010 and E6011 were designed to burn hot and dig, and a side effect is a hydrogen-rich atmosphere around the arc — on the order of 30 to 45 milliliters of hydrogen per 100 grams of deposited weld metal. A basic low-hydrogen electrode like E7018 deposits less than 5 under the same accounting, roughly a factor of ten, and that difference is baked into the classification for a reason. For the rod-side background, our piece on what 7018 rods are used for covers storage expectations alongside applications.

Important: A low-hydrogen classification describes the electrode as it leaves the sealed container at the manufacturer’s drying level — not the electrode after a night in an open job box, a humid truck bed, or a damp corner of the shop. Once the rod takes on moisture, the number on the classification no longer describes what you are running.

Oxygen and Carbon Monoxide: Gas Made Inside the Puddle

Oxygen has a route to trouble that does not involve dissolving at all. It reacts inside the pool. Iron oxide in a contaminated puddle reacts with hydrogen and produces water vapor; carbon in the steel reacts with oxygen and produces carbon monoxide. Both reactions generate gas in the middle of the molten metal, which means the weld can manufacture its own porosity from contamination you put there — mill scale, rust, paint residue — even when the shielding envelope was perfect. This is why a weld can be porosity-free one day on clean plate and full of holes the next on the same settings.

Zinc is the standout case. Arc welding galvanized steel vaporizes the coating and floods the weld zone with zinc oxide, giving you porosity in the bead and a serious fume exposure for the operator at the same time. Paint and thick primer have their own signature: large elongated cavities known as wormholes are the classic marker of gross organic contamination, and they almost always mean somebody welded over a surface that was never prepped. Dig into the coating side with our guides on welding galvanized steel with MIG and flux-core work on galvanized stock.

The Gases That Cannot Cause Porosity

Argon and helium are noble gases. They are essentially insoluble in molten steel, so they pass through the puddle without dissolving and cannot form pores as dissolved species. Carbon dioxide is likewise negligible as a direct pore-former — carbon monoxide earns its place on the problem list only through the in-puddle reactions described above, not through solubility.

This is worth holding onto because it resolves entire internet arguments. Argon does not cause porosity. It can contribute indirectly — flowing too fast and stirring air into the jet, or not flowing at all and leaving the weld uncovered — but as a dissolved species it is inert to the problem. The same goes for helium. When someone reports “argon porosity,” the real story is almost always turbulence, a leak, a draft, or a dirty surface, and the gas bottle is a bystander.

Quick Note: Half the contradictory advice online disappears once you know the cast of characters. Only hydrogen, nitrogen, and oxygen dissolve in steel. Everything else is either innocent or guilty of a different crime — like creating fume, or destabilizing the arc.

Where the Gas Actually Comes From on the Job

Theory says nitrogen, hydrogen, oxygen. The shop floor says: the flowmeter, the door, the rod box, and the plate that came in from outside. This section translates the chemistry into the five places gas physically enters your work.

The Gas Flow Reflex, and Why It Backfires

Turn the shielding gas up and the porosity goes away — that is the reflex, and nearly every welder develops it at some point. It works exactly often enough to become habit, which is what makes it dangerous. Shielding gas protects the weld by displacing air with a calm, laminar blanket. Push the flow past the nozzle manufacturer’s specification and that blanket stops behaving. The stream goes turbulent, and a turbulent jet does not exclude air — it entrains it, sucking air in along the edges of the jet and delivering it straight to the puddle, inside the very shield that was supposed to keep it out.

Research on the effect has put numbers on it: air entrainment of roughly one percent of the shielding gas flow produces distributed porosity, and above about one and a half percent you start seeing gross surface pores. So the welder who opened the flowmeter to solve a porosity problem can easily have become the cause of it — until someone turns the flow back to specification and the porosity vanishes for reasons nobody understands, which is how the myth survives another decade. For the parameter side, our MIG gas setting guide lays out practical ranges, and C25 versus C100 explains how gas composition changes your tolerance for exactly this kind of entrainment.

Pro Tip: Set flow from the machine and nozzle manufacturer’s specification, not from memory. If you suspect you have been running rich, put it back to spec for one shift before changing anything else — you may solve the whole problem without touching a single fitting.

Leaks, Empty Bottles, Long Stickout, and Open Doors

Once the flow setting is eliminated as a suspect, walk the gas path physically. Every connection between the cylinder valve and the contact tip is a potential air inlet: regulator face fittings, hose crimps, the gas solenoid, the liner-to-gun connection, the nozzle threads. A soapy-water check on every joint takes two minutes and finds leaks that no amount of reasoning will. This is the cheapest diagnostic in welding and the one most often skipped.

The empty cylinder deserves its own mention because it fails silently. A bottle on fumes does not announce itself; it just delivers whatever pressure remains, thinning out gradually over the shift. The porosity that results appears slowly, so it gets misread as technique drift — “I must be moving too fast today” — when in fact the shield simply died. Weigh the bottle or read the gauge honestly at the start of the shift, not when the weld goes bad.

Drafts are the other classic. A fan across the bay, a roller door left open, an extraction arm relocated for another job — none of these look like welding variables, and all of them move air across the arc at speeds far below what you would guess matters. Wind shielding for outdoor or doorway work is cheap and pays for itself on the first saved reject. The same applies to stickout: a long stickout places the arc beyond the nozzle’s effective reach, so the shield arrives late and diluted. Excessive weaving and extreme work angles belong here too — both rock the puddle in a way that folds the trailing edge under, pulling air into the last liquid metal to freeze.

Contamination on the Plate

Rust, mill scale, paint, primer, galvanizing, oil, grease, cutting fluid, grinding dust, and yesterday’s spatter all carry oxygen or hydrogen into the puddle through the in-puddle reactions described earlier. The rule that works is simple: clean the joint right before welding, not an hour before. Surfaces re-contaminate — cold stock sweats, oily fingerprints transfer, ground surfaces oxidize, and shop air deposits dust.

Sequence matters as much as effort. Degrease first, then remove the oxide or coating layer, then weld; do it in reverse and you grind the oil into the clean surface. On aluminum the routine is specific: wipe with acetone, scrub with a dedicated stainless wire brush that has never touched steel, then weld within a few minutes before the oxide film rebuilds. On galvanized stock, strip the coating back several inches from the joint. On rusty salvage plate, grind back to bright metal along the full length of the intended bead — spot cleaning gives the puddle a clean inch and a dirty foot, and porosity follows the dirty foot.

Moisture in Consumables

Electrodes and fluxes are packaged, handled, and stored as moisture-control devices, which is a useful way to think about them. A stick electrode’s coating is a chemistry experiment that includes hydrogen management; a flux-cored wire’s tubular fill includes deoxidizers tuned for a dry, clean joint. Expose either to humidity and you are running a different product than the one specified on the box.

Low-hydrogen rods belong in a rod oven held at the manufacturer’s stated temperature and soak time. We have covered that from the DIY angle with a homemade welding rod oven build, and from the practical angle with can you dry welding rods in the oven — including what happens when you over-bake a coating and destroy the electrode instead of saving it. The discipline that matters: if a low-hydrogen rod has been out of its sealed container or out of the oven overnight, treat it as damp. The classification on the box no longer applies to it.

Wire has a quieter version of the same problem. Small-diameter wire has a high surface-to-volume ratio, so more of every pound is exposed to handling, oxidation, and moisture than on larger wire. Drawing lubricants left on the surface add hydrogen. Wire stored uncovered, on the floor, or on a spool that got rained on brings all of it into the puddle with no chance for the shielding gas to help. Nothing dramatic happens — you just get slightly worse welds than the same wire stored in a cabinet, and you blame your hands. We covered the storage side in how to keep MIG wire from rusting.

What the Machine Itself Can Cause

Machines do produce porosity, just far less often than environments do. An inverter with a gas-flow fault code or an underflow sensor is telling you something specific, and it is worth believing. A failed solenoid, a cracked gas line inside the wire drive, a porous liner that lets gas escape before it reaches the gun — all are real, and all are found with the same soap-and-water walk-through described above. Erratic feed also matters indirectly: stuttering feed makes a stuttering arc, and an unstable puddle folds gas in.

Settings come last on this list deliberately. Voltage and wire speed that worked this morning have not changed on their own. When porosity appears mid-job, the machine is the ninth suspect, not the first — and the troubleshooting order later in this article makes that sequence explicit.

What the Location of the Pores Tells You

Reading porosity placement is the single highest-value skill in this whole subject, and it costs nothing but attention. Where the holes sit is a clue about where the gas came from, and a welder who can read that clue turns a two-hour guessing session into a five-minute check.

Notice what is missing from that table: a row for “the settings were wrong.” Incorrect parameters do cause defects — a puddle too cold freezes too fast and traps more gas — but pure parameter error almost never produces these classic shapes and placements. If the porosity has a pattern that makes sense, the cause is environmental or contaminant-related, and no amount of dial-turning will fix it. That table is worth copying onto a card for the toolbox; we expand the same diagnostic approach in flux core welding troubleshooting for the defect family beyond gas holes.

Short Note: Pattern first, always. Spend thirty seconds deciding whether you are looking at scatter, clusters, root porosity, or wormholes before you lay a hand on the regulator.

Porosity by Welding Process

Every arc process has its own relationship with gas entrapment, driven by transfer mode, shielding arrangement, and the chemistry of the consumable. The core physics never changes, but the usual suspects shift from process to process.

GMAW (MIG)

Transfer mode matters here more than most welders expect. Spray transfer, available on argon-rich mixtures at adequate current, is the clean one: metal crosses the gap as fine droplets, the puddle stays open longer, and gas has time to rise and leave. Short-circuit transfer at lower current freezes the pool dozens of times per second and traps gas far more readily. If you have been chasing porosity on thin sheet running short circuit, part of the answer lives in the transfer mode, not only in the flowmeter. Our wire speed and voltage chart shows where the modes sit relative to each other.

Gas selection changes your tolerance for contamination in ways that surprise people. One hundred percent CO2 cannot produce true spray transfer — it runs globular or repelled, throws more spatter, and lays a wider, flatter puddle. But because it is oxidizing, it cleans the weld surface and tolerates entrained nitrogen and oxygen better than you would guess. An 80/20 argon-CO2 blend sits between the extremes with good tolerance for entrained gas. The practical translation: if you are getting porosity on pure CO2 thin-sheet work, an argon blend is worth a trial; if you are already on a blend and getting porosity, you need the right gas and an undisturbed shield, not more flow. Related comparisons live in FCAW versus MIG and our core piece on whether MIG welds are strong.

Quick Note: On aluminum MIG specifically, the oxide film is the usual culprit rather than the gas. Acetone, dedicated stainless brush, weld within minutes — and keep the stickout short, because aluminum wire feeds long and soft through the liner and arc instability follows.

FCAW (Flux-Cored)

Flux-cored welding is more forgiving of shielding problems than most people expect when they switch over from solid wire. The slag layer blankets the puddle, the flux carries its own deoxidizers, and those deoxidizers actively resist nitrogen absorption. On dual-shield wire, the external CO2 is often an asset rather than a liability. For the fundamentals of how the tubular wire works, our what is flux core welding primer covers it from the wire up.

When flux-cored welds do come out porous, the suspect list narrows fast: contaminated wire stored badly, a dirty gas line, or a wet bottle — rarely the parameters, because the flux chemistry tolerates a reasonable settings range. If you are comparing the two processes, flux core versus solid wire and dual shield welding go deeper on the shielding arrangements that make the difference.

SMAW (Stick)

Moisture in the coating is the main risk in stick welding, and cellulosic coatings generate hydrogen by design. Porosity concentrated near the start of a bead or at a restart is common, because the operator rebuilds the puddle from nothing and whatever moisture sits at that spot gets its moment of heat. Rod-oven discipline earns its keep here: a restart on a properly stored 7018 should be unremarkable, while a restart on a rod that slept in a damp toolbox often is not. Polarity mistakes contribute as well. For decoding electrodes, our electrode chart for arc welding and the walkthrough in welding rod numbers explained give you the tools.

GTAW (TIG)

TIG is gas-shielded through and through and shares nitrogen exposure with MIG, but the failure points are almost always hardware or environment rather than consumable. The torch has no wire to disrupt the flow, so a long stickout, a dirty or dented ceramic, a cracked gas hose, or argon that is not flowing cleanly through the torch body is usually the whole story — leaks at the rear of the torch dilute the argon before it ever reaches the puddle.

On aluminum, TIG is the process most likely to produce keyhole porosity when coverage is poor, which is why a full argon purge on the back side matters on any joint where the root will see atmosphere. Achieve full purge flow before striking the arc, not during the first inch of the weld. TIG-specific habits live in our TIG tips and tricks collection.

How Porosity Is Actually Found

Four inspection methods dominate real work, and each one has a blind spot that gets people hurt. Knowing what each method cannot see is as important as knowing what it can.

Visual Testing

Every weld gets looked at, and visual testing is a required step under major construction codes. Two technique points separate useful visual inspection from a glance. First, clean the weld before you look: slag, spatter, smoke deposits, and oxide hide exactly the features you are hunting. Second, use raking light — a lamp held at a low angle across the bead — rather than flat overhead illumination. Raking light throws shadows into pores and makes them pop; flat light falls into a hole and returns nothing.

Appearance tells you whether porosity exists and roughly how bad it looks, but it cannot quantify acceptance and cannot find anything buried inside the weld. That ceiling is where the other methods take over.

Radiography

X-ray and gamma-ray inspection is the strongest volumetric tool for porosity. It images internal voids directly, distinguishes porosity from slag by shape, and produces a permanent record with indications you can measure against acceptance charts — which is exactly why it settles arguments that visual inspection starts.

The limits are real. Small pores are hard to resolve in thick sections, and a planar defect like lack of fusion can slip past if it is angled away from the beam, because the rays travel through it edge-on and leave no shadow. Radiography is also the slowest and most expensive of the common methods, so it gets used where the code calls for it. Reading the film is a certification-level skill in its own right.

Ultrasonic Testing

Ultrasonics shines on planar defects and thick material — lack of fusion, lack of penetration, crack-like indications. For scattered porosity it is less dependable: small round reflectors are difficult to size consistently, and coarse-grained austenitic stainless welds degrade performance badly because grain structure scatters the sound before it reaches the target. The practical division of labor holds steady: porosity-heavy joints go to radiography, fusion-suspect joints go to ultrasonics, and mixed cases get both if the application justifies it.

Liquid Penetrant Testing, and the False Reassurance It Gives

Here is the correction to make if you picked up the wrong idea somewhere, because this mistake is common and it breeds dangerous confidence. Liquid penetrant testing does not find subsurface porosity. It cannot. Penetrant works purely by capillary action, drawing dye into discontinuities that break the surface. A pore sealed inside the weld metal is invisible to it, and so is embedded slag and internal lack of fusion.

Run a dye penetrant test on a weld full of internal porosity and the result comes back clean. That is a false pass, and it sounds exactly like a real one. Penetrant does one job well: finding surface-breaking cracks and surface-connected defects on non-porous materials. Use it for that. Use radiography or ultrasonics when the question is what is inside the bead.

Pro Tip: On a smooth-looking bead with a suspected internal problem, raking-light visual plus penetrant will both tell you “fine” while the interior is full of voids. If the application is pressure-retaining or fatigue-critical, insist on a volumetric method before signing anything off.

What the Codes Really Allow

This surprises everyone the first time: some porosity is acceptable, and a weld with visible pores can still be a fully code-compliant weld. Acceptance is not zero — it is a table.

AWS D1.1, the structural steel welding code used across most American construction and fabrication, publishes porosity limits in its visual inspection acceptance tables, and the limits change with joint type and loading direction. A complete-penetration groove weld in a butt joint carrying stress transverse to the weld is the most restricted case in the book, where visible piping porosity is simply not permitted. Static fillet welds and other statically loaded groove welds get far more permissive treatment, allowing scattered pores within stated size and density limits. The logic is mechanical, and the section on weld properties below explains why the code draws the line where it does. Our comparison of full penetration versus fillet welds gives context on why those joint categories carry different stakes.

Two practical warnings attach to any code conversation. First, clause numbering trips people up: the porosity numbers live in the visual inspection acceptance tables, not where many people go looking, and the edition matters because tables get revised. Second, pressure and piping work runs on different criteria entirely. The honest advice is invariable: open the current edition of the code your job is built to and read the table that applies to your joint. Do not rely on a number quoted in a blog post or a forum reply, including this one. The American Welding Society publishes the standard itself through the AWS site, and pressure-work criteria come from ASME.

Important: Before you pick up a grinder because a weld “looks porous,” check the acceptance table for that joint under that loading. Excavating a compliant weld is pure waste — and every gouge is a new chance to undercut or leave a profile defect where none existed before.

Can You Weld Porosity Out?

No — and it is worth understanding why, because the instinct to reheat the bad area is strong and it reliably makes things worse. Adding heat does not drive gas out of solid metal. What it does is speed up the chemical reactions that generate gas from oxides and moisture, and slow the cooling of the puddle. A second pass laid over a porous area produces a worse weld than the first one, essentially every time: the puddle you are melting already contains the voids, and remelting traps fresh gas around them.

The correct repair sequence, in order:

1. Find and correct the cause first. Repair before fixing the cause and you will be repeating the repair tomorrow.
2. Excavate. Gouge or grind the porous metal out until you reach sound, dense weld metal or clean base metal. Carbon arc gouging works for heavy sections; angle grinding suits smaller repairs. See our guide to gouging for the technique side.
3. Clean the excavation. You have just exposed fresh metal and possibly left carbonized or contaminated surfaces behind. Wire-brush and inspect before any deposit.
4. Reweld to a qualified procedure — correct preheat where applicable, correct consumable, correct parameters.
5. Re-inspect with a method that can actually see the repair, matched to the depth of the deposit.

Short Note: Grinding a visible pore flush does not remove it — it usually just opens the top of a cavity and smears metal across the rest. If the porosity is subsurface, the only legitimate repair is full excavation to sound metal.

Preventing Porosity: Where the Control Lives

Prevention organizes naturally by who controls each variable. Gas setup belongs to whoever prepares the machine. Material cleanliness belongs to the handler. Consumable storage is a shop discipline. Technique belongs to the welder. Machine condition belongs to maintenance — equipment makers such as Lincoln Electric publish nozzle and flow specifications worth following to the letter. Assigning the categories that way stops the circular argument where everybody assumes somebody else handled it.

On the Shielding Gas

• Set the flow from the manufacturer’s specification for your nozzle and process, then leave it alone. Resist adding “just a little more.”
• Soap-test every hose, fitting, regulator connection, and gun joint before blaming the arc. Two minutes, no parts ordered.
• Confirm the cylinder is actually full and the regulator is not iced up — a starved bottle mimics a technique problem every time.
• Keep the nozzle clear of spatter; a partially blocked nozzle distorts the flow pattern unpredictably.
• Match the flowmeter to the gas. An argon scale on a CO2 flowmeter reads lies, and the number you are trusting is fiction.
• Shield the work from drafts. A cheap windscreen solves a problem that otherwise costs real money in rejects.

On the Material

• Clean the joint immediately before welding, not earlier in the day.
• Remove rust, mill scale, paint, primer, grease, and oil along the full length of the intended bead plus a margin.
• Strip galvanizing from the weld zone, or choose a process and fume controls built for it.
• On aluminum: acetone wipe, dedicated stainless brush, weld within minutes before the oxide film returns.
• Store base metal indoors and off the floor — cold stock brought into a warm shop sweats condensation.

On Consumables

• Hold low-hydrogen electrodes in a rod oven at the specified temperature for the specified soak time.
• Never use a rod that has been out of its container overnight as if it were still low-hydrogen.
• Store wire clean, dry, and covered — spools off the floor, bags resealed after use.
• Check the gas bottle before the shift starts, not after the weld goes bad.
• When you switch to a new batch of anything, run a scrap bead first. If the scrap is porous, the batch is suspect and production stays untouched.

Pro Tip: Tape a small card inside the welder door listing the flow specification for each nozzle in the shop. Habit is the enemy of specification, and a visible number beats a remembered one every time.

On Technique

• Pre-flow gas and establish the puddle quickly instead of creeping into it.
• Fill the crater before breaking the arc — crater-fill feature on wire machines, or step the electrode back into the crater on stick.
• Avoid heavy weaving; hold a short arc, a steady work angle, and let travel speed distribute the bead.
• Watch stickout constantly. The shield does not reach the end of a long stickout, no matter what the flowmeter says.
• Go easier on cold material and cold wire — both raise the freezing rate and shrink the escape window for bubbles.
• On aluminum, achieve full purge flow before striking, and back-purge any root that will be exposed.

On the Machine

On an inverter with fault codes, read the display before you touch anything else — a gas-flow fault or underflow condition is a specific message about a specific system, and believing it saves an hour. Beyond electronics: check the liner and drive rolls when feed is erratic, because unstable feed makes an unstable arc and an unstable puddle swallows air. Wire-feed consistency also governs your ability to hold parameters, which is why maintenance items like drive rolls for flux core wire are porosity-relevant and not just throughput trivia.

These five control points map directly onto the everyday defects catalogued in common flux core welding mistakes — the same categories, different wire.

What Porosity Does to a Finished Weld

Inspection debates get easier once you understand what is mechanically at stake. Codes restrict porosity in some joints and tolerate it in others for four distinct reasons, and the four are not equally important in every application.

Read those four together and the code logic explains itself. Scattered fine porosity at low volume barely moves tensile properties, because the remaining cross-section carries the load and the specimen usually fails in the base metal rather than the weld. As void fraction climbs, the failure migrates into the weld and turns brittle — the physical reason acceptance tables tighten as loading moves from static to transverse tension. Static structural joints get permissive limits because the remaining section carries the load and there is no leak path and no cycle. Fatigue-critical details and pressure boundaries get strict limits because one of these mechanisms is live in them.

Fatigue deserves special emphasis because it is where porosity earns its reputation. Smooth, isolated, subsurface pores are far less damaging than surface-breaking clusters at the weld toe — and the toe already carries the highest geometric stress concentration in most joints. A component in cyclic service therefore cares far more about porosity sitting at the toe than about visually identical porosity mid-bead, even though a visual inspection would score them the same. If you want the broader picture of joint quality versus load path, our piece on how strong a weld actually is covers the strength side of the equation.

Short Note: When someone says “a little porosity never hurt anybody,” ask which of the four mechanisms they have ruled out. For a static fillet on a handrail, often none apply. For a pressure boundary or a fatigue detail, the first sentence of the argument is already wrong.

A Troubleshooting Order Worth Taping Inside the Door

When porosity shows up on your own work, the order of checks matters as much as the checks. Random diagnosis — opening the flow, then changing voltage, then swapping wire — burns material and teaches nothing, because you have changed three variables at once. This sequence runs from cheapest and most likely to most involved, and it never starts at the machine settings.

1. Read the pattern first. Scattered? Clustered at stops? Root only? Wormholes? The location narrows the cause before anything is touched, using the diagnostic table earlier in this article.
2. Soap-test the gas path. Regulator connection, hose fittings, gun connection, cylinder valve. Two minutes, no parts, and a leak is the most common environmental cause after drafts.
3. Check the cylinder. Weigh it or read the gauge honestly. A nearly empty bottle shields with what is left, which is often nothing, and it fails gradually — which is exactly why it gets misread as technique drift.
4. Measure flow at the nozzle with a flowmeter — actual flow at the gun, not the setting on the dial. If you increased it recently, put it back to specification and retest.
5. Check the room. Door open? Fan running? Extraction moved? A draft that did not exist yesterday will reproduce the sudden-onset scattered pattern exactly.
6. Check the surface. Was the joint cleaned right before welding — in the last half hour, or yesterday? Any grease from handling since?
7. Check consumables. Rods out of the oven or sealed box? Wire uncovered overnight? A new batch of anything is suspect until proven clean.
8. Review technique changes. New joint geometry with a crevice? Longer stickout than usual? Faster travel because the schedule tightened?
9. Only now consider the machine. Fault codes, flow sensors, solenoid behavior, and — on wire machines — drive and liner condition if feed has been erratic.

The discipline is in the order, not the individual steps. Steps one through eight cost nothing but attention and resolve the overwhelming majority of field cases. A welder who starts at step nine will eventually stumble onto a setting that makes the porosity vanish by accident, learn nothing about why, and meet the same defect again next week under slightly different conditions.

Quick Note: Write the sequence on a card and keep it with your settings notebook. On a bad day — late shift, ugly film, pressure on — nobody improvises well, and the cheap checks are the first things skipped. The sequence exists so the diagnosis runs in the right order when you are least inclined to follow it.

Porosity and Welding Fume Safety

Porosity is a quality problem, but it lives inside a bigger occupational health conversation, and the two share sources often enough that it pays to address both at once. Welding fume is metal oxide particulate plus gases, and its composition depends entirely on what is being welded. Mild steel fume carries iron, manganese, and silicates. Stainless adds chromium and nickel, including hexavalent chromium. Aluminum fume can carry fluorine and copper. Zinc from galvanized or brass work produces zinc oxide — the same zinc that puts wormholes in your bead also puts a respiratory irritant in your air.

One regulatory point gets repeated incorrectly all the time. OSHA does not currently enforce a general permissible exposure limit for “welding fume” as a single substance. A 5 mg/m³ welding fume limit appeared in an older rule that was remanded and is not in force; compliance runs through the constituent limits instead — manganese at a 5 mg/m³ ceiling, iron oxide fume at 10 mg/m³ as an eight-hour time-weighted average, hexavalent chromium at 5 µg/m³. NIOSH takes a harder line on manganese, recommending 1 mg/m³ as an eight-hour average and 3 mg/m³ short-term, and it treats welding fume as carcinogenic. You can read the agency positions at OSHA’s welding and cutting page and NIOSH’s welding topic page, with the ventilation requirement text at 1910.252 on eCFR.

The practical controls do not change with the regulatory fine print. Local exhaust ventilation at the source — a fume extractor arm parked roughly a foot from the arc — outperforms opening a door across the shop by a wide margin. Respirators belong where local exhaust is not feasible, and only as part of a written program with fit testing. We have written the detail pieces separately: are welding fumes toxic covers what is in the smoke, respirator selection covers the PPE side, and welding safety tips covers the wider shop context. Environmental context on air standards sits at EPA.

Important: Note the link back to the defect itself: porosity and fume share sources. Wet plate, dirty coatings, and zinc produce both the defect and the exposure, so the surface-preparation discipline that saves the weld also protects the welder.

Three Situations You Will Actually Run Into

Theory gets you halfway; these three cover most of what comes through the shop door, and recognizing which one you are in finds the cause faster than any amount of setting-chasing.

Situation One: It Started Suddenly on a Job That Was Fine

Nothing in the procedure changed, so the change is in the environment. This is the pattern from the rinse tank story at the top of this article, and it is the most common surprise visit in fabrication. Work the gas system first, in this order: the bottle and regulator — a cylinder that quietly ran down mid-shift starves the shield until the weld gives up; then the hose and fittings, because a connection that behaved on Monday can crack on Tuesday; then a cup of soapy water on every joint in the path, watching for bubbles.

After the gas path, look at the room. Someone opened a door on a windy day, a pedestal fan got switched on, the extraction arm moved for another job. A cross-draft pulls air into the arc across a wide area and does not need to be strong to do it. Finally, the consumables: a fresh box of electrodes that rode in the truck through a humid night, or wire that has sat uncovered since the last shift. The tell for this scenario is distribution — the porosity is fine and evenly scattered rather than clumped, because a general air supply produces general defects.

Situation Two: Clusters at Every Start and Every Stop

This points at arc initiation and arc termination, and usually at consumable moisture alongside them. Every strike builds a puddle from nothing and the first gas released has nowhere to go; every break leaves an open crater that shrinks as it cools and can suck in atmosphere while it does. Both moments are vulnerable, and both get worse on a slow, cold start.

Fix the start by pre-flowing gas before striking and establishing the puddle quickly instead of creeping into it — a confident start is a shielding technique. Fix the stop by filling the crater before breaking the arc: crater-fill on a wire machine, or the stick technique of stepping the electrode back into the crater before lifting away. If the clusters are severe or keep returning after you fix the technique, go straight to the rods. Start-and-stop clustering is the signature of damp electrodes, and no amount of arc skill outruns a wet coating. The 7018 rod guide walks through what happens to that coating once storage discipline slips.

Short Note: Clustering at stops with a hollow crater center is often shrinkage rather than gas — same neighborhood, different cause. Crater-fill technique solves both, but knowing which one you have tells you whether to look at the machine’s crater setting or at the gas supply.

Situation Three: Only the First Weld of the Day Is Bad

A cold-start problem, and almost always one of two things. Moisture first: cold plate carries condensation overnight, that water boils off during the first pass or two, and the problem appears to fix itself — which convinces everyone it was technique when it was actually a dehumidifying puddle. Stickout second: on a cold machine the wire is cold and the contact tip has not thermally settled, so the arc stretches out past the end of the gas nozzle where the shield cannot reach.

The fix is preparation: dry the plate and consumables, and bring the gun and wire to temperature before the production weld begins. Running a short weld on scrap as a warm-up sounds like folklore until you know the reason — it drives condensation off the work area and normalizes the equipment. Shops that do it deliberately see fewer mystery first-piece rejects.

Four Porosity Myths That Refuse to Die

Certain beliefs circulate in workshops with the persistence of folklore. Each one contains a fragment of truth, which is exactly why they survive.

• “Turn the gas up.” The fragment: inadequate shielding does cause porosity, and a welder who has never verified a flowmeter might genuinely be running low. The myth: that flow is a volume knob where more is always safer. Turbulence past the specified range entrains air into the shield meant to exclude it, so the fix for low flow and the fix for high flow look identical on the valve and produce opposite results. Set to specification and find the actual leak or draft.
• “A little porosity is always acceptable.” The fragment: codes do permit limited porosity in statically loaded joints, so treating any pore as an automatic reject is genuinely wrong. The myth: that the allowance generalizes. In a transverse-loaded complete-penetration butt weld, in a pressure boundary, or in a fatigue-critical detail, porosity that would pass a static fillet check is a defect. Whoever quotes the permissive limit without quoting the joint condition has given you half an answer — the table is joint-specific and loading-specific.
• “It’s the machine.” The fragment: machines do cause porosity — a gas fault code, a failed flow sensor, a stuck solenoid. The myth: frequency. Machine failures are rare and usually visible, while environmental causes — the door, the draft, the damp rod box, the cleaned-yesterday joint — recur constantly and quietly. When porosity appears on a job that ran clean all week, the procedure did not change and the machine did not change; the room did.
• “The welder’s hands are to blame.” The fragment: technique contributes — stickout, weave, crater fill, travel speed all belong to the operator. The myth: that the operator should be the default suspect. In the patterns that actually walk through the shop door, environment and consumables outweigh technique every time. The welder is usually the person who can see all the cheap causes fastest, which makes them the investigator, not the defendant.

Pro Tip: When a fix sounds like a single knob — more flow, more heat, more amps — treat it skeptically. Porosity comes from systems: gas path, room, materials, consumables. Knob-turning addresses systems about as well as a horn fix addresses a flat tire.

Frequently Asked Questions

What causes porosity in welding?

The weld pool traps gas that fails to escape before the metal freezes. In practice that gas is nitrogen from air getting into the shield, hydrogen from moisture, or oxygen and carbon monoxide generated when oxides and coatings break down in the arc. The most frequent individual culprits are an incorrect gas flow rate — often too high — a leaking or depleted gas supply, a cross-draft, dirty base metal, damp electrodes, and welding over paint or galvanizing.

Is porosity always a serious defect?

No. Acceptance depends on joint type and how it is loaded. Structural codes permit limited porosity in some statically loaded fillet and groove welds while restricting it heavily in complete-penetration butt welds loaded transverse to the weld, and pressure work runs on its own criteria. Read the acceptance table for your joint in the current code edition before deciding anything needs repair.

Can you burn porosity out of a weld?

No. Extra heat accelerates the reactions that generate gas and slows puddle solidification, so it makes the problem worse. The only legitimate repair is to excavate the porous metal back to sound material, correct the underlying cause, and reweld to a qualified procedure — in that order.

Why am I getting porosity with 7018 electrodes?

Three usual suspects. Electrode moisture: if the rods have been out of the oven or out of their sealed container, they have absorbed hydrogen and the low-hydrogen classification no longer describes them. Plate contamination: rust, oil, paint, or mill scale releasing gas in the puddle. Shielding problems: a leak or draft pulling nitrogen in even though the electrode itself is fine. The rod’s usage guide covers polarity and settings alongside storage.

Does liquid penetrant testing find porosity?

Only porosity that breaks the surface. Penetrant enters discontinuities by capillary action through an open path, so voids sealed inside the weld are invisible to it. Internal porosity needs radiography or ultrasonic testing; a clean penetrant result says nothing about what is inside the bead.

Does argon cause porosity?

Not directly. Argon does not dissolve in steel and cannot form bubbles as a dissolved species. It contributes only indirectly — flowing too fast and entraining air into the jet, or not flowing at all and leaving the weld uncovered. Helium is in the same position: innocent as a solute, potentially guilty as a delivery system for air.

Why is my MIG weld porous on aluminum?

Almost always one of three things: the oxide film was not removed properly — acetone, then a dedicated stainless brush, then weld within minutes; the gas coverage cannot keep up with travel speed or the outdoors is blowing it away; or the wire is feeding a long, soft path through the liner, introducing arc instability that shows up as porosity at the puddle. Aluminum’s extreme hydrogen solubility swing means it reacts faster to all three than steel does.

How much gas flow is too much?

Follow the machine and nozzle manufacturer’s specification — commonly around 20 to 25 cubic feet per hour for standard MIG nozzles, but nozzle size, gas type, and process all shift the number. The reliable test is behavioral: if increasing flow makes porosity worse, you have crossed into turbulence-driven entrainment. Return to specification and fix the actual cause instead.

Can I weld over galvanized steel without worrying about porosity?

You can weld it, but the zinc is contamination in the puddle and a fume exposure in your lungs at the same time. Grind or cut the coating back several inches from the joint, use local exhaust ventilation, and follow your respiratory protection program. The porosity side and the health side share a root cause, so the same surface-prep discipline solves both. Our guides on MIG on galvanized and flux core on galvanized cover the technique adjustments.

Final Thoughts

Porosity is not random and it is not mysterious. Gas enters the weld from one of three places — the air, moisture, or oxides — and once you can name the source, the cure is usually cheap. The film that came back with dots on it was never an accusation; it was a map, and now you can read it.

Two habits will serve you better than anything else on this page. The first is reading where the holes sit: fine scatter means air, wormholes mean contamination, root porosity means the back of the joint was never protected, and clusters at the ends mean technique or damp consumables. That single skill compresses troubleshooting more than any tool in the box. The second is respecting the gas flow specification — turning the flow up to fix a shielding problem is the reflex that makes porosity worse, and it is a reflex worth breaking early in your career.

When verification beats guessing, remember the division of labor: visual testing finds surface indications, radiography and ultrasonics find internal ones, and penetrant only finds what breaks the surface. If the porosity is internal, use a method that can see it. If the code says the level in front of you is acceptable, put the grinder down and sign the report honestly — knowing what to leave alone is part of the job. For the wider ecosystem around this defect, our related guides on flux core settings and dual shield welding pick up where this article leaves off.