Electropolishing stainless steel parts: mechanism, process control, and tolerances

• 1
  What electropolishing stainless steel parts actually does
• 2
  Bath chemistry and the operating window
• 3
  Metal removal, surface finish, and dimensional effect
• 4
  Edges, burrs, and internal surfaces
• 5
  Fixturing, racking, and masking
• 6
  Hydrogen embrittlement, fatigue, and metallurgical cautions
• 7
  Specifying the finish and inspecting it
• 8
  Electropolishing versus passivation versus mechanical finishing

Electropolishing is anodic dissolution. The part hangs on the positive side of a DC circuit in an acid bath, and metal leaves the surface ion by ion instead of being cut or abraded. No tool touches the work. The result is a surface that is dissolved, not smeared.

The leveling comes from a film effect. In phosphoric and sulfuric acid blends, a viscous oxide-rich layer builds on the part. It is thinner over micro-peaks, where current density is highest, and thicker in micro-valleys. Peaks therefore dissolve faster than valleys, and the surface flattens at a scale below about 0.5 mm.

That mechanism sets the boundary. Electropolishing removes the smear layer left by machining or grinding, and it dissolves the free iron and sulfide inclusions that seed pitting. It does not fill scratches, weld spatter, or deep tool marks. Whatever the bath cannot reach, it cannot fix.

Practically, this is why the process suits parts that are already machined to a decent finish. A turned 316L flange at Ra 1.6 μm comes out bright and uniform. A hand-ground surface with 0.1 mm gouges comes out bright and still gouged. The gloss changes; the geometry underneath does not.

Most stainless baths are concentrated phosphoric and sulfuric acid blends, often 50–75% phosphoric acid by volume. Sulfuric acid carries the conductivity and water controls viscosity. The mix is not a fixed recipe; it drifts as parts are processed and metal accumulates.

Typical running conditions are 50–90 °C at 5–20 V, with current density in the 10–50 A/dm² band. Alloy and finish target move you within that band. A 316L part aiming for a bright cosmetic surface runs differently from a 17-4PH part where the target is deburring and passivation.

Bath analysis matters more than the setpoint on the panel. Acid ratio, water content, dissolved metal, and temperature all shift over a production week. Shops that check titration and specific gravity on a schedule hold repeatable results. Shops that only watch the voltmeter chase problems.

Viscosity is the lever most people ignore. A warmer, thinner bath diffuses faster and levels less. A cooler, thicker bath polishes brighter at lower removal rates and takes longer. When finish and cycle time pull against each other, temperature is where the trade is made.

Removal is generally 5–25 μm per surface, and 10–15 μm is a common default for general parts. The figure is per surface, so a Ø20 mm shaft loses roughly twice that on diameter while a bore grows by the same amount on radius.

Tolerance planning follows directly. A feature held to ±0.005 mm should not go into the bath without pre-compensation on the drawing. Either the machinist leaves stock, or the feature is masked. Threads tighter than class 6g and press fits are the usual candidates for masking.

Finish gains are real but bounded. A part starting near Ra 0.8 μm typically lands around Ra 0.2–0.4 μm. A 0.4 μm start may reach 0.1–0.2 μm. Going below about Ra 0.05 μm is not a routine outcome; it usually needs mechanical polishing first, with electropolishing as the final clean-up.

Time does not buy unlimited gloss. Past a point, extra minutes remove more metal without improving Ra much, and they start to widen existing pits. The process is a finisher, not a grinder.

Current density concentrates at high points, so edges and burrs dissolve first. A 0.1 mm burr on a laser-cut or machined edge usually clears in one standard cycle. Sharp edges round to roughly 0.05–0.2 mm radius, depending on how much metal you remove.

That rounding is a benefit for cleanability and a hazard for function. Seal grooves, press fits, and knife edges lose their sharpness. If a face must stay flat to a gasket, mask it or accept a radius and design around it.

Internal surfaces polish only where current can reach. A through-hole at Ø3 mm and 20 mm deep polishes well; a blind pocket with a 1 mm corner radius does not. Deep blind holes get bright at the mouth and dull at the bottom, and no amount of cycle time changes that.

Electropolishing also tends to open the surface slightly at inclusion sites. On free-machining grades such as 303, sulfide strings can leave a fine etched texture. That is often acceptable on hidden parts and not acceptable on a visible panel, so grade choice belongs in the same conversation as finish.

Parts sit on titanium or copper fixtures, and the contact has to carry the full current without heating or arcing. A weak contact leaves a dull patch and can etch a contact mark deep enough to fail cosmetic inspection. Contact area is a real design decision, not an afterthought.

Masking with acid-resistant tape or lacquer protects threads, bearing seats, seal grooves, and any surface that must stay matte. Mask lines are sharp, so plan them where a line will not be seen or where a step is acceptable.

Racking marks are the most common cosmetic complaint. They are avoidable in most cases. Tell the shop which faces are visible so the fixture can grip a hidden surface, and put that note on the drawing rather than in an email.

For parts that combine a polished exterior with a tight bore, the usual answer is to mask the bore, polish the outside, then run a light second operation. It costs a step, and it keeps both features inside tolerance.

Electropolishing charges the part with hydrogen. Austenitic 304 and 316L are generally low risk at normal strength levels. Martensitic and precipitation-hardening grades above roughly 1000 MPa tensile strength are the ones to watch, including 420, 440C, and 17-4PH in the higher condition.

For those grades, specify a post-bake. A typical window is 2–4 hours at 190–220 °C, and it should be mandatory rather than optional on the purchase order. Bake timing relative to the bath matters; a bake that happens days later is not the same process.

Fatigue life is a separate question. A polished surface removes machining marks that act as stress raisers, which helps. But a part that is already thin at a fillet loses another 10–25 μm, and that is where the cycle life is decided. Check the wall before and after, not just the surface.

Spring temper sheet and fine wire are also worth flagging. Both are sensitive to hydrogen and to handling, and both are usually better served by a shorter cycle at lower current density, or by passivation alone if cleanability is the only goal.

A useful drawing callout names the standard, the removal allowance, and the masked features. Something like: electropolish per ASTM B912, 10–15 μm per surface, mask M8 threads and the Ø12 H7 bore. That sentence prevents most of the arguments that follow.

Ra alone is a weak acceptance criterion. Two parts can read Ra 0.3 μm on the same profilometer and look completely different, because Ra hides pits and waviness. Add a visual comparator or a defined gloss level for cosmetic surfaces, and a pit count for wetted surfaces.

Inspection is usually visual plus dimensional. Measure critical features before and after to confirm the removal allowance, check masked zones for stray attack, and look at edges under low magnification. Reports are available on request.

Cost is driven by removal depth, racking density, masking complexity, and bath volume per part. Small parts that pack densely on a rack are cheap per piece. Large weldments with lots of blind geometry are not, because most of the current goes to the outside.