Pulsed vs Continuous Laser: Which One for Rust Removal?

Pulsed vs Continuous Laser: Which One for Rust Removal?

1964: a beam that never switched off, and slid straight off rust. 1989: a beam that switched off 100,000 times a second, and cleaned a marble angel without touching the stone. Both machines are on the same price list today, and almost nobody selling them will tell you they are not competitors.

1964 - The beam that could not stop

Continuous wave (CW) came first, and it came from cutting. A CW laser emits without interruption: switch it on and energy pours out at a constant rate, forever, until you switch it off. For cutting and welding that is exactly what you want, because you are not removing a layer - you are opening a kerf, and steady energy plus assist gas does that beautifully.

Cleaning is a different problem. Rust is a 5-80 µm layer sitting on metal you must not damage. To remove the layer and spare the substrate, you need the energy to be absorbed in the layer and gone before it reaches the bulk. A continuous beam gives you no such control: it deposits heat into a spot for as long as the spot is there, and steel conducts that heat inward at roughly 4 × 10⁻⁶ m²/s. The rust and the plate end up at the same temperature, which is the definition of thermal damage.

There was a second, quieter reason CW lasers struggled on metal in that era: wavelength. The CO₂ laser works at 10.6 µm, where bare steel absorbs only 2-8% of the beam at room temperature. You can brute-force that with 4 kW and a gas jet. You cannot do it gently on a 1.2 mm panel.

1972-1989 - The museum detour that invented your machine

The first serious laser cleaning was not industrial. It was conservation: Venetian marble, Florentine bronze, limestone statuary eaten by sulphur dioxide. Conservators had a constraint no factory had ever faced - remove the black crust, keep the patina, do not touch the stone. Mechanical tools could not do it. Solvents could not do it. A brush took the crust off and 200 years of surface history with it.

What they found was that if you deliver energy in a short burst instead of a steady stream, a process window opens. The crust has a low ablation threshold - roughly 0.5-1.5 J/cm² for typical atmospheric soiling. The marble behind it needs 5-10 J/cm² to start breaking down. Choose a fluence between the two and the process is self-limiting: once the crust is gone there is nothing left to absorb enough energy to damage the stone. The operator's skill stops being the safety system. The physics becomes the safety system.

That is the entire intellectual foundation of every handheld cleaning laser sold today. It took twenty years to leave the museum and reach a shipyard, because the lasers were laboratory instruments with water cooling the size of a desk and a price in six figures.

2009 - The dial appears: MOPA

For most of the pulsed era you got the pulse your resonator gave you. Q-switched Nd:YAG delivered roughly 8-20 ns at whatever repetition rate it felt like, and peak power was locked to average power by the pulse width. If you needed more peak, you bought more average - which also meant more heat, which defeated the purpose.

MOPA (Master Oscillator Power Amplifier) fibre lasers broke that link. By 2009-2010, commercial units let you set pulse width anywhere from about 2 ns to 500 ns and repetition rate from 1 to 4,000 kHz, largely independently of average power. Peak power became a design choice rather than a side effect. That single change is why a 200W handheld unit today can clean a 0.8 mm car panel that a 2009 machine of the same average power would have heat-tinted.

Run the arithmetic on a typical setting: 200W average, 200 ns pulse, 50 kHz repetition. Pulse energy is 200 ÷ 50,000 = 4 mJ. Peak power is 4 mJ ÷ 200 ns = 20 kW - one hundred times the nameplate. The duty cycle is 1%. For 99% of the time the beam is off and the surface is cooling. That on/off rhythm is the machine.

2013-2026 - Two branches, two job sites

Once MOPA existed, the industry did something predictable: it split. Pulsed went where parts are thin, intricate, expensive or visible - automotive restoration, moulds, weld prep on stainless, aerospace skins, heritage. Continuous went where parts are thick, dirty and counted in hectares - ship plate, storage tanks, structural steel, mill scale on 12 mm sections, where a 3,000W CW head on a gantry eats 30+ m² an hour and distortion is irrelevant because the plate is 20 mm thick.

They did not merge, and the reason is engineering, not marketing. A pulsed cleaning head needs a short cavity, high peak damage threshold optics and a scanner synchronised to the pulse train. A multi-kW CW head needs serious thermal management on the delivery fibre and a water chiller that weighs more than the operator. Putting both in one box means compromising both. The 2023-onwards "dual-mode" units that do exist are real, but they are two sources sharing a housing and a control cabinet - not a single beam that behaves like both.

The two branches never competed for the same job. They were separated by part thickness, and that separation still holds in 2026.

The physics, in one number

Everything above reduces to a single comparison: how long the energy stays in one place versus how long heat takes to escape that place.

For steel, the thermal diffusion time across a distance d is roughly t ≈ d² / 4α, with α ≈ 4 × 10⁻⁶ m²/s. Across 10 µm - the thickness of a typical rust layer - that is about 6.3 µs. A 200 ns pulse is 31 times shorter than the time heat needs to cross the layer. The energy never gets out of the rust. The rust ablates; the plate below stays at ambient.

Now the continuous case. A 1 mm spot scanning at 2,000 mm/s dwells on any given point for 0.5 ms - about 80 times longer than the diffusion time. Heat diffuses roughly √(4αt) ≈ 89 µm deep in that single pass, and because adjacent tracks overlap 50-70%, the accumulated heat affected zone in production measures 0.3-1.5 mm. That is why a CW beam that cleans beautifully on 10 mm plate will bow a 1 mm panel and leave a straw-coloured tint on stainless.

Peak density figures assume a 0.1 mm spot for the pulsed unit and a 1 mm spot for the CW unit - the spot sizes each is normally operated at.

What that does to your parts

The HAZ number is abstract until it is your panel. The table below is measured bow on 500 mm coupons, single face cleaned, both sides unrestrained - the worst realistic case, and the one that decides whether a panel goes to paint or to the scrap bin.

Note the 2 mm row: this is where the two technologies overlap and where most buyers make an expensive guess. Below 2 mm there is no choice to make. Above 5 mm there is no penalty either way.

What it costs per square metre

Cost per m² is where pulsed looks bad and CW looks unbeatable - and where the comparison usually stops, one step too early. Figures below assume $18/h fully loaded labour, $0.14/kWh, medium rust on hot-rolled steel, one operator, consumables included.

Rates are for a single pass to bare metal on 3-8 mm hot-rolled plate. On thin sheet both laser rates drop 20-40% because power has to be pulled back.

Read that table and CW wins by 33% over the best pulsed option. Now add one column that the table cannot hold: rework. At 1 mm thickness, a 1500W CW unit puts 0.4-0.9 mm of bow into a 500 mm panel. Straightening that is 6-11 minutes per panel on a press, plus the inspection, plus the 3-8% you scrap outright. On a job that runs 4,000 m² a year of 1 mm work, that is roughly $8,000-11,000 a year - which erases the entire per-m² advantage and then some.

Where continuous still wins, honestly

We sell both, so there is no incentive here to pretend CW is obsolete. There are four situations where a pulsed machine is the wrong purchase, and they are not marginal.

And the mirror image: there are jobs where a CW unit is not "less suitable" but physically incapable. Under 2 mm, on aluminium, on cosmetic stainless, on anything with a machined or ground finish within 30 mm of the cleaned zone. No amount of parameter tuning fixes a heat affected zone that is 50× the layer you are removing.

Five years, three shops, one table

Purchase price is the least interesting number in this decision. Here is the five-year view for three profiles we quote constantly - machine, consumables, energy and labour, with rework included where it actually occurs.

Includes machine, lens and consumables, energy, labour at $18/h, and straightening/scrap cost for the sheet metal profile. Excludes financing and installation.

Two rows, opposite winners, same magnitude of difference - about $23,000 either way. That symmetry is the honest answer to this whole question: there is no better beam, only a beam matched to a thickness. The shop that buys the wrong one does not lose 10%. It loses the cost of the machine plus five years of the wrong labour rate.

If you are still unsure which branch you belong to, answer three questions and the table decides for you: what is your thinnest part, how many square metres did you clean last year, and is the head held by a person or a robot. Send us those three and we will send back Table 6 with your numbers in it - including the version where the machine you asked about is the wrong branch.

Pulsed vs continuousMOPA fibre laserHeat affected zoneRust removalLaser cleaning