Published Aug 30, 2026 · Last updated Nov 16, 2026 · 6 min · IndieRF

ENIG vs Bare Copper: Why Nickel Adds Microstrip Loss

On 50 Ω RO4350B microstrip at 10 GHz, ENIG adds about 0.17 dB/in over bare copper. The gold is a flash. The nickel is thicker than a skin depth. Change the finish in Transmission Line Calculator and read the loss.

ENIG is the usual outer-layer finish on RF boards because it solders and does not tarnish. It is also a nickel barrier with a gold flash. At microwave frequencies that nickel sits where the current lives.

On the stackup below (50 Ω microstrip, 6.6 mil Rogers RO4350B, ½ oz reverse-treated foil, relieved mask, 10 GHz), Transmission Line Calculator reports 0.56 dB/in with ENIG and 0.40 dB/in with bare copper. The extra 0.17 dB/in is the plating. On a 4 inch feed that's 0.66 dB, and only the finish or the length can buy it back.

Blame the nickel. The gold flash is a sixteenth of a skin depth, so the current goes straight through it.

Current hugs the surface

Skin depth is the distance the current penetrates a metal before it has mostly died away:

δ=1πfμ0μrσ\delta = \frac{1}{\sqrt{\pi f \mu_0 \mu_r \sigma}}

At 10 GHz in copper, δ≈0.66 μm\delta \approx 0.66\,\mu\mathrm{m}. The rule to carry: copper skin depth is about 66/f66/\sqrt{f} µm with ff in MHz, so 2.1 µm at 1 GHz. Half-ounce foil is 17.5 μm17.5\,\mu\mathrm{m} thick, so the RF current is a thin sheet on the surface. Coat that surface and the current flows in the coating.

Nickel is thick compared with a skin depth

IPC-4552B ENIG is a thin gold flash on electroless nickel. Transmission Line Calculator uses 0.05 μm0.05\,\mu\mathrm{m} Au on 4.5 μm4.5\,\mu\mathrm{m} Ni.

MetalRoleThicknessSkin depth at 10 GHz
Goldoxidation barrier0.05 μm0.05\,\mu\mathrm{m}0.79 μm0.79\,\mu\mathrm{m}
Nickelsolder-diffusion barrier4.5 μm4.5\,\mu\mathrm{m}0.72 μm0.72\,\mu\mathrm{m}
Copperthe foil you paid for17.5 μm17.5\,\mu\mathrm{m}0.66 μm0.66\,\mu\mathrm{m}

The gold flash is a small fraction of a skin depth, so it is almost transparent. The nickel is about six skin depths. Current on a plated face dies in the nickel and never reaches the copper.

Nickel also conducts worse than copper (σ≈1.4×107\sigma \approx 1.4\times10^7 versus 5.8×107 S/m5.8\times10^7\,\mathrm{S/m}) and it is ferromagnetic. This model uses μr=3.5\mu_r = 3.5 for plated nickel. Magnetism shrinks skin depth, which raises surface resistance further. In the skin inspector, about 94% of the plated-face current is in the nickel.

Most microstrip loss is on the unplated faces

ENIG coats the top and the sidewalls. The bottom face sits on the laminate and stays copper, and so does the ground plane under it. Microstrip carries most of its loss on those two faces, against the dielectric.

Transmission Line Calculator works out the plated share from the line's geometry, by Wheeler's rule applied to the strip's top and sides alone: 36% of the conductor loss on this line. A 2-D boundary-element solve of the same 50 Ω line gives 36% too. The nickel's surface impedance applies to that share, and the foil's roughness to the rest. That is why ENIG raises this line by about 40% instead of tripling it.

The share follows the width. On a 20 mil core the same solve puts 62% of the loss on the top and sides of a 6 mil strip and 22% on a 98 mil one, so a narrow ENIG line pays a larger fraction.

Grounded CPW is less forgiving. More current rides the plated gaps, so the same nickel stack costs more. Switch the topology in the calculator if that is your launch.

ENIG is an outer-layer finish. Inner-layer stripline is bare copper on most builds, so it skips this tax.

Same board, two finishes

Hold the laminate, thickness, foil, mask, and frequency fixed. Only the finish changes. Loss is locked to 50 Ω, so the width is the same in both widgets.

IndieRF Transmission Line Calculator

50 Ω microstrip, 6.6 mil RO4350B, ENIG (IPC-4552B), 10 GHz

Open in calculator
Length units

Pick a JLCPCB, PCBWay or OSH Park stackup to fill height, εr, and copper. Editing those fields, or the mask, switches this back to Custom.

(mil)
(mil)
(mil)

Lengths are shown to 0.01 mil. That is the drawing resolution, not a fabrication tolerance. Ask your fabricator for the etch allowance.

(GHz)
W = 13.69 milrelief = 8.00 milt = 0.69 milmask 0.79 milh = 6.60 milGround planeDielectric: Rogers RO4350B · εr 3.66 · tanδ 0.00375 mil
TraceFinishSolder maskDielectricGroundDrawn to scale.

Impedance Z0

50.00Ω

Insertion loss

0.574dB/in

0.0226 dB/mm

Closed form, field solve pending

Effective permittivity ε_eff

2.884

Static ε_eff 2.856

Where the loss comes from

29% of the loss is the ENIG plating (0.166 dB/in of 0.574 dB/in). At 10.00 GHz the nickel skin depth is 0.7 µm, smaller than the 4.5 µm nickel layer, so RF current crowds into the nickel instead of the copper under it. The largest share is still the copper under the finish, at 30%.

  • Copper30.4%
  • Plating28.8%
  • Roughness17.0%
  • Dielectric22.8%
  • Mask1.0%

Skin depth and plating

At 10.00 GHz, 99.8% of the current is in the plating and 0.2% is in the core. The nickel skin depth (0.7 µm) is smaller than the nickel thickness (4.5 µm).

Foil: Reverse treated (RTF). Rq = 1.2 µm. A larger value is a rougher surface and more conductor loss.

  • Goldt 0.05 µm · δ 0.79 µm · 6.2%

    Conductivity 4.10e7 S/m

  • Nickelt 4.50 µm · δ 0.72 µm · 93.7%

    Conductivity 1.40e7 S/m · relative permeability 3.5

  • Copper coret 17.50 µm · δ 0.66 µm · 0.2%

    Conductivity 5.80e7 S/m

Advanced roughness parameters

Huray treats the foil teeth as snowballs. a is the sphere radius and As/Af is the sphere area divided by the flat area. A rougher foil has a larger As/Af and more conductor loss. These are the numbers stored for this foil type, not a measurement of your board.

a = 0.7 µm · As/Af = 1.4

IndieRF Transmission Line Calculator

Same stackup, bare copper

Open in calculator
Length units

Pick a JLCPCB, PCBWay or OSH Park stackup to fill height, εr, and copper. Editing those fields, or the mask, switches this back to Custom.

(mil)
(mil)
(mil)

Lengths are shown to 0.01 mil. That is the drawing resolution, not a fabrication tolerance. Ask your fabricator for the etch allowance.

(GHz)
W = 13.69 milrelief = 8.00 milt = 0.69 milmask 0.79 milh = 6.60 milGround planeDielectric: Rogers RO4350B · εr 3.66 · tanδ 0.00375 mil
TraceSolder maskDielectricGroundDrawn to scale.

Impedance Z0

50.00Ω

Insertion loss

0.408dB/in

0.0161 dB/mm

Closed form, field solve pending

Effective permittivity ε_eff

2.884

Static ε_eff 2.856

Where the loss comes from

43% of the loss is the copper under the finish (0.174 dB/in of 0.408 dB/in).

  • Copper42.7%
  • Plating0.0%
  • Roughness23.9%
  • Dielectric32.0%
  • Mask1.4%

Skin depth and plating

At 10.00 GHz, 0.0% of the current is in the plating and 100.0% is in the core.

Foil: Reverse treated (RTF). Rq = 1.2 µm. A larger value is a rougher surface and more conductor loss.

  • Copper coret 17.50 µm · δ 0.66 µm · 100.0%

    Conductivity 5.80e7 S/m

Advanced roughness parameters

Huray treats the foil teeth as snowballs. a is the sphere radius and As/Af is the sphere area divided by the flat area. A rougher foil has a larger As/Af and more conductor loss. These are the numbers stored for this foil type, not a measurement of your board.

a = 0.7 µm · As/Af = 1.4

On ENIG, plating is about 30% of the 10 GHz loss. On bare copper it is zero. Dielectric and mask barely move. Roughness is still there on both. That loss lives in the foil's teeth against the laminate, under any finish.

Loss breakdown in the Transmission Line Calculator for this note's ENIG stackup at 10 GHz: plating is 29.4% of 0.563 dB/in, copper 31.0%, roughness 17.3%, dielectric 22.2%, and 93.7% of the current flows in the nickel
Loss breakdown for the ENIG microstrip on RO4350B at 10 GHz. The plating is 0.166 of the 0.563 dB/in (29.4%), and the nickel carries 93.7% of the current. Open this stackup.

Put your numbers in

The widgets are live. Change height to your core, frequency to your band, and finish to whatever is on the traveler. Leave lock-to-50 Ω on so the width follows.

A few finishes worth trying on this same line at 10 GHz:

FinishWhat the current seesLoss on this stackup
Bare copperCopper0.40 dB/in0.40\,\mathrm{dB/in}
OSPCopper (the organic is electrically transparent)0.40 dB/in0.40\,\mathrm{dB/in}
Immersion silverThin Ag, then copper0.40 dB/in0.40\,\mathrm{dB/in}
ENIGNi, then almost no copper0.56 dB/in0.56\,\mathrm{dB/in}
ENEPIGPd + Ni; nickel still dominates0.56 dB/in0.56\,\mathrm{dB/in}

Immersion silver is a better conductor than copper and it is thin, so the current still reaches the foil. It reads 0.001 dB/in under bare copper here, and that is the silver's own conductivity on the plated faces. The model holds the top and sides smooth with or without a finish, because Horn and coauthors (DesignCon 2016) measured untreated foil faces adding little loss, so the foil's teeth against the laminate are the only roughness either line carries. ENEPIG still has 4.5 μm4.5\,\mu\mathrm{m} of nickel, so it tracks ENIG. Hard gold is a connector stack (thicker Au-Co on nickel), so it inherits the nickel loss.

For a sweep to 20 GHz and side-by-side captures, open Transmission Line Calculator. Save the ENIG line to memory, switch to bare copper or OSP, and read Δα\Delta\alpha at 10 GHz and 28 GHz. In this model the nickel leaves Z0 alone. Width shifts from copper weight and solder mask are in how copper and solder mask change Z0. The same RO4350B line from 1 to 10 GHz, with Rogers' printed ED foil roughness split from the nickel, is in skin depth at 10 GHz in copper and nickel.

When ENIG is still the right call

The extra 0.17 dB/in is a tax, not a veto. I still spec ENIG when the board has to be assembled later, sit in a stockroom, or mix RF with parts that want a gold pad. OSP fingerprints and has a short shelf life. Immersion silver has its own tarnish and creep rules.

The engineering move is to spend the tax where it is cheap. Keep long RF runs on inner-layer copper. Keep ENIG on short launches, connectors, and mixed-signal outer layers. If the outer layer is the RF path, ask the fab whether OSP or immersion silver is allowed on that construction.

A 0.17 dB/in hit on a 0.4 inch launch is noise. The same hit on a 6 inch 10 GHz feed is 1.0 dB in the cascade before the first amplifier.

If 0.1 dB/in is your whole margin, measure a coupon on the actual stackup. The nickel μr=3.5\mu_r = 3.5 used here is a typical plated-Ni value. Electroless Ni-P is less magnetic than bulk nickel, and fab thickness wanders around the IPC-4552B window.

Questions? Contact

Discussion

No comments yet. Start the thread with a measurement, a correction, or a worked example.