Manufacturing
Plasma Cutting
A constricted electric arc turns gas into a metal-melting jet
Plasma cutting forces an electric arc through a narrow nozzle, ionizing a gas into a 20,000–25,000 °C jet that melts metal and blows the molten metal away. Conventional, dual-gas, and high-definition variants trade speed, edge squareness, and dross against cost. Found in CNC cutting tables, shipyards, structural-steel fabrication, and ductwork.
- MechanismConstricted transferred arc
- Jet temperature20,000 to 25,000 °C
- CutsAny conductive metal
- Practical thickness0.5 to ~50 mm (160 mm sever on big systems)
- Kerf width1 to 4 mm (HD lower)
- Failure modeDross, bevel, consumable burn-out
Interactive visualization
Press play, or step through manually. The visualization is yours to drive — try it before reading on.
Watch the 60-second explainer
A condensed visual walkthrough — narrated, captioned, under a minute.
How plasma cutting works
Start with the fourth state of matter. Heat a gas hot enough and its atoms shed electrons; the gas becomes a soup of free electrons and positive ions called plasma. That soup is electrically conductive and ferociously hot. Plasma cutting is the trick of generating a small, controlled, knife-thin column of this plasma and dragging it through metal.
The torch does four things in sequence. First a pilot arc jumps from a central electrode to the surrounding copper nozzle (or is struck by a high-frequency spark or a "blowback" contact-start), ionizing the gas inside the torch. Second, the now-conductive gas lets the main current "transfer" to the workpiece — the part itself becomes part of the circuit. Third, the gas is forced through a tiny constricting orifice in the nozzle, which squeezes the arc into a narrow, high-velocity jet: constriction is the whole game, because it raises both the temperature and the velocity of the plasma. Fourth, that jet melts a spot of metal and its own momentum, helped by the gas flow, blows the molten metal out the bottom of the cut.
The key insight is that a free-burning, unconstricted arc (like a welding arc) spreads out and runs "cool" — a few thousand degrees over a wide, soft cone. Force the same arc through a 1–3 mm orifice and you concentrate the current into a tiny cross-section. Current density skyrockets, the gas is heated far hotter, and it exits at speeds that can approach the local sound speed. The same electrical power becomes a cutting tool instead of a welding tool.
The four functional pieces:
Electrode (cathode) → hafnium-tipped copper; emits electrons
Constricting nozzle → copper orifice that squeezes the arc
Plasma / cut gas → air, O2, N2, or Ar-H2 (ionized to the jet)
Shield gas + cap → surrounds the jet, sharpens the cut edge
Circuit (transferred arc):
power supply (+) → workpiece → plasma column → electrode (−) → supply
The governing physics: power, current density, kerf
Plasma cutting is, at heart, an energy-balance problem. The plasma delivers power; the metal needs a certain amount of energy per unit volume to be heated to melting and ejected; cut speed is set by how those two balance.
The electrical power dumped into the arc is simply:
P = V × I
V = arc voltage (≈ 100–200 V across the constricted arc)
I = arc current (20 A handheld → 400+ A industrial)
Example: a 130 A machine at ~140 V → P ≈ 18 kW into the arc.
Not all of that reaches the cut — a chunk is lost to the nozzle, radiation, and the kerf walls. The fraction that actually does cutting work, the cutting efficiency, is typically 40–60%. What matters at the metal is the energy needed to remove material:
Energy to melt & eject a volume of steel:
E/volume = ρ × [ c_p × (T_melt − T_0) + L_f ]
ρ ≈ 7,850 kg/m³ (density of steel)
c_p ≈ 0.49 kJ/(kg·K) (specific heat)
T_melt ≈ 1,510 °C, T_0 ≈ 20 °C
L_f ≈ 247 kJ/kg (latent heat of fusion)
E/volume ≈ 7,850 × (0.49×1490 + 247) ≈ 7.7 GJ/m³ ≈ 7.7 J/mm³
The material removal rate (kerf cross-section × cut speed):
MRR = w × t × v
w = kerf width, t = plate thickness, v = cut speed
So: v ≈ (η · P) / (E/volume · w · t)
Plug in numbers. A 130 A torch (P ≈ 18 kW) with η ≈ 0.5 puts ~9 kW into a 10 mm steel plate. With a 2 mm kerf, the steel removal demands ~7.7 J/mm³ × 2 mm × 10 mm = 154 J per mm of travel; 9,000 W ÷ 154 J/mm ≈ 58 mm/s, or about 3.5 m/min. Real machine charts for 130 A on 10 mm mild steel land near 2–3 m/min once you account for the melt-ejection inefficiency — a good sanity check on the simplified model.
Two more design levers fall out of this. Kerf width scales with the orifice diameter: a wider orifice cuts a wider, more wasteful kerf but tolerates more current. And the arc's swirl — the gas is injected tangentially to spin around the electrode — stabilizes the column and centers the hot cathode spot, at the cost of depositing slightly more heat on one side of the kerf (the source of the characteristic bevel).
Variants: conventional, dual-gas, water-injection, HD
- Conventional (single-gas) plasma. One gas does everything — usually shop air. Cheapest, simplest, handheld and light-industrial. Air plasma cuts mild steel fast but nitrides the cut edge (air is ~78% nitrogen), which can cause porosity if you weld over it without grinding.
- Dual-gas plasma. A separate plasma gas and shield gas. Oxygen plasma on mild steel adds an exothermic oxidation boost (like oxy-fuel) for faster, cleaner cuts; nitrogen or argon-hydrogen on stainless and aluminum avoids oxidation and improves edge color. The shield gas tunes the kerf and reduces top-edge dross.
- Water-injection plasma. Radial water injected at the nozzle constricts the arc further (the water vaporizes and pinches the column), giving a squarer cut and cooling the nozzle for longer consumable life. The underside of the table is often a water table that catches fume and dross and dampens noise.
- High-definition / high-precision (HD) plasma. Tighter orifice geometry, vortex shield gas, and tightly regulated standoff produce kerfs near 1 mm and bevel under 1–2° — approaching laser quality on medium plate. Hypertherm's HPR/XPR and Kjellberg's HiFocus are the canonical product families. The trade is more expensive consumables and a narrower process window.
- Underwater / submerged-arc plasma. Cutting with the plate submerged 50–75 mm under water drastically cuts noise, UV, and fume — common on large CNC tables for stainless and aluminum.
Plasma vs other thermal & abrasive cutting
| Plasma | Oxy-fuel | Laser (fiber) | Waterjet | Wire EDM | |
|---|---|---|---|---|---|
| Cutting mechanism | Melt + blow (electric arc) | Oxidation + melt (flame) | Melt/vaporize (light) | Erosion (abrasive + water) | Spark erosion |
| Materials | Any conductive metal | Ferrous (steel) only | Metals + some non-metals | Anything | Conductive only |
| Sweet-spot thickness | 6 to 50 mm | 10 to 300+ mm | 0.5 to 25 mm | 1 to 200 mm | 0.1 to 300 mm |
| Kerf width | 1 to 4 mm | 1 to 5 mm | 0.1 to 0.4 mm | 0.5 to 1.5 mm | 0.02 to 0.4 mm |
| Edge squareness | 3 to 5° bevel (HD <2°) | ~2° bevel | Near square | Slight taper | Excellent |
| Cut speed (10 mm steel) | ~2 to 3 m/min | ~0.4 to 0.6 m/min | ~1.5 to 3 m/min | ~0.3 to 0.8 m/min | ~2 to 5 mm/min |
| Heat-affected zone | Moderate | Large | Small | None (cold cut) | Tiny recast layer |
| Capital cost | Low to medium | Lowest | High | High | Medium to high |
Real-world specs and consumable economics
| System class | Current | Pierce / cut capacity (steel) | Typical use |
|---|---|---|---|
| Handheld / hobby air plasma | 20 to 45 A | Cut ≤12 mm, sever ≤16 mm | Repair, farm, light fabrication |
| Light industrial | 60 to 105 A | Cut ≤25 mm, pierce ≤16 mm | Job shops, ductwork, art tables |
| CNC mid-range | 130 to 200 A | Cut ≤38 mm, pierce ≤25 mm | Structural steel, general fab |
| Heavy industrial | 300 to 400 A | Cut ≤50 mm, pierce ≤38 mm, sever ≤80 mm | Shipyards, heavy plate, pressure vessels |
| Special high-current | 600 to 1000 A | Sever 150 to 160 mm | Scrap reclamation, mill gouging |
The hidden running cost is consumables. The hafnium- or zirconium-tipped electrode and the copper nozzle erode every time the arc starts and stops. The hafnium emitter recedes a fraction of a millimetre per pierce; once the pit exceeds ~1.5 mm depth the electrode is scrap. A mid-range CNC torch consumes an electrode-and-nozzle set roughly every few hours of arc-on time, and each set can cost a few to several dollars. Because most wear happens at arc start and stop, a nesting program that minimizes pierces (chaining contours, using common cut lines) directly lengthens consumable life — a real-money optimization, not a nicety. Gas is the other line item: shop air is nearly free, but oxygen, nitrogen, and argon-hydrogen mixes add cost and need clean, dry, regulated supply.
Standoff height is tightly controlled in CNC: the torch rides at a fixed height (often via an arc-voltage height control, since arc voltage rises as the gap grows) typically 1.5–4 mm above the plate. Too low and the nozzle double-arcs and dies; too high and the cut bevels and loses energy.
Where plasma cutting is used
- CNC cutting tables. The dominant use. A gantry drives the torch over a slat or water table, cutting nested parts from sheet and plate at production speed. Pairs with arc-voltage height control and often a marking/etching mode.
- Shipbuilding and heavy structural steel. 300–400 A systems slice 25–50 mm plate for hulls, bridges, and pressure-vessel blanks faster than oxy-fuel and on stainless where oxy-fuel can't go.
- HVAC and ductwork. Fast cutting of thin galvanized and stainless sheet; air plasma's speed on thin material is its strong suit.
- Gouging. A flatter torch angle and a gouging nozzle let plasma carve out weld defects, remove old welds, or back-gouge a root pass — replacing carbon-arc gouging with less smoke and noise.
- Demolition, scrap, and field repair. Portable inverter units cut rebar, beams, and machinery on site. High-current systems chop scrap for recycling.
- Robotic 3D plasma. A torch on a 6-axis robot bevel-cuts pipe ends and structural members for weld prep, where the controlled bevel that's a defect on flat plate becomes a feature.
Common failure modes and pitfalls
- Dross on the underside. The most common complaint. Low-speed dross (rounded, easily knocked off) means cutting too slowly; high-speed dross (thin, hard bead on the top edge) means too fast. The fix is matching cut speed, current, and standoff to the thickness chart — there is a sweet-speed window for each combination.
- Excessive bevel. Some bevel is inherent to the swirling arc (the "square" side is to the right of cut direction). Excess bevel signals wrong standoff, a worn nozzle, or cutting too fast/slow. HD plasma and proper consumable replacement bring it back in spec.
- Double-arcing. If the torch dips too close to the plate or the nozzle is worn, the current finds a second path — electrode-to-nozzle and nozzle-to-work — instead of one clean transferred arc. It instantly melts the nozzle bore. Arc-voltage height control and not letting the torch crash prevent it.
- Consumable burn-out from bad start/stop. The hafnium emitter erodes mostly at arc ignition and termination. A worn-electrode "ramp-down" gas post-flow protects the emitter as the arc dies; skipping it, or hot-starting on a worn electrode, blows the consumable early.
- Fume, UV, and noise. Plasma generates intense UV (welding-shade eye protection required), ozone, metal-oxide fume (hexavalent chromium when cutting stainless — a regulated carcinogen), and noise above 100 dB at high current. Water tables, downdraft extraction, and shade-rated PPE are not optional in production.
- Trying to cut non-conductors. Plasma needs the workpiece in the transferred-arc circuit. Aim it at glass, stone, or plastic and there's no return path — no cut. (Non-transferred plasma exists for spraying, but not for cutting these.)
Frequently asked questions
How hot is a plasma cutting arc?
The constricted plasma jet runs roughly 20,000 to 25,000 °C (about 36,000 to 45,000 °F) in the core — several times hotter than an oxy-fuel flame's ~3,100 °C and far above the melting point of any common metal. Steel melts at ~1,500 °C and aluminum at ~660 °C, so the plasma doesn't gently warm the metal: it instantaneously melts a path while the high-velocity gas blows the molten remainder out the bottom of the cut.
Can a plasma cutter cut any metal?
It can cut any electrically conductive metal — mild steel, stainless steel, aluminum, copper, brass, titanium, cast iron. That conductivity is mandatory: the workpiece is one terminal of the transferred-arc circuit, so the current must flow through it back to the power supply. Plasma cannot cut wood, plastic, glass, concrete, or other non-conductors. Oxy-fuel, by contrast, only cuts ferrous (iron-bearing) metals because it relies on iron oxidizing exothermically.
What is the difference between a transferred and non-transferred plasma arc?
In a transferred arc the current flows from the electrode, through the plasma, into the conductive workpiece and back to the supply — the part is part of the circuit. This puts maximum energy into the metal and is what every cutting torch uses. In a non-transferred arc the current loops between the electrode and the nozzle itself; the part isn't in the circuit. Non-transferred arcs are used for plasma spraying and for cutting non-conductors, but deliver far less cutting power because the workpiece never carries current.
What is dross in plasma cutting and how do you avoid it?
Dross is the re-solidified molten metal that clings to the bottom edge of the cut. Low-speed dross (thick, rounded, on the underside) means you're cutting too slowly; high-speed dross (thin, hard, on the top edge) means too fast. The cure is dialing in the correct cut speed, arc current, and torch standoff height for the material thickness — there's a sweet-speed window for each. Correct consumables (a worn nozzle widens and destabilizes the jet) and the right gas also matter. A clean cut on thin steel leaves essentially dross-free edges.
Why does a plasma cut edge come out beveled instead of square?
The plasma jet swirls (the gas is given a vortex to stabilize the arc), and the swirl deposits more energy on one side of the kerf than the other, producing a few degrees of bevel — typically 3 to 5° on conventional plasma, with the "good" square side on the right of the cut direction. High-definition plasma constricts the arc tighter and tunes the gas swirl to bring bevel down to under 1 to 2°. For perfectly square edges in thick plate, fabricators still reach for laser, waterjet, or a secondary machining pass.
Plasma cutting vs laser cutting — when do you pick which?
Plasma wins on thick conductive metal, capital cost, and cutting speed in the 6 to 50 mm range — a 400 A system cuts 50 mm steel that most fiber lasers struggle with, at a fraction of the machine price. Laser wins on thin sheet (under ~6 mm), edge squareness, kerf width (~0.2 mm vs plasma's 1 to 4 mm), fine detail, and the ability to cut non-metals. A common shop split: laser the thin precision work, plasma the heavy plate.