Detecting Tungsten-Filled Gold Bars with XRF Analyzers

Tungsten-cored gold bars are the one counterfeit that defeats the two checks most buyers rely on. The density matches. The surface assays as pure gold. A bar can pass both tests and still be mostly tungsten.

This tutorial explains precisely why XRF cannot see a buried core, what it can see, and how to build a layered verification sequence that closes the gap. The honest starting position is that no single instrument solves this problem — anyone claiming otherwise is selling something.

Why tungsten specifically

Counterfeiters converged on tungsten for one reason: density.

• Gold: 19.32 g/cm³
• Tungsten: 19.25 g/cm³

That is a difference of 0.36 percent. No other affordable metal comes close. Lead is 11.34, brass around 8.5, and both are exposed instantly by a hand scale and a ruler. Tungsten is not.

The economics are compelling from the fraudster’s side. Tungsten trades in the region of tens of dollars per kilogram against gold in the tens of thousands per kilogram. A one-kilogram bar with a substantial tungsten core carries a fraction of the metal value while presenting the correct weight and dimensions.

Where tungsten fails as a substitute

Tungsten is not a perfect impostor. Its melting point is 3,422°C against gold’s 1,064°C, so the two cannot simply be cast together. Tungsten is hard and brittle where gold is soft and malleable. Its thermal and electrical conductivity differ substantially from gold’s.

Every practical detection method exploits one of these differences rather than density, because density is the one property the counterfeiter matched deliberately.

What XRF can and cannot see

This section matters more than any other, because misplaced confidence in XRF is how these bars get accepted.

The depth limit

XRF measures the composition of a thin surface layer. For gold L lines in a gold matrix, useful signal originates within roughly 20 to 60 microns of the surface — a few hundredths of a millimetre.

A tungsten core is typically covered by one to several millimetres of gold. That is ten to a hundred times deeper than XRF can penetrate. The instrument is not being fooled; it is accurately reporting a surface that genuinely is pure gold.

No detector upgrade changes this. No extended counting time changes this. The fluorescent photons from the tungsten are absorbed within the gold and never reach the detector. The information does not exist in the measurement.

What XRF does catch

XRF remains valuable against several real fraud variants:

• Thin plating over tungsten. Where the gold layer is under a few microns, tungsten Lα at 8.40 keV appears in the spectrum. Cheaper fakes fall here.
• Surface tungsten contamination from manufacturing, exposed at edges, corners, or a drilled and refilled plug.
• Non-tungsten fakes entirely — brass, lead, and tungsten-alloy composites are all identified immediately.
• Alloy inconsistency where a bar claiming 9999 fineness shows trace elements inconsistent with refined gold.

The resolution factor

Tungsten Lα at 8.40 keV sits 1.31 keV below gold Lα at 9.71 keV. Whether a small tungsten contribution is separable from the skirt of a large gold peak depends on detector energy resolution.

A silicon drift detector at 125 to 140 eV FWHM resolves the two cleanly. A Si-PIN at 180 to 220 eV produces broader peaks whose tails overlap, and a modest tungsten signal can hide inside the gold peak’s shoulder. This is one of the clearest practical arguments for detector quality, developed further in our detector comparison.

There is also a complication worth knowing: instruments with tungsten X-ray tube targets scatter their own tungsten emission off the sample. Distinguishing tube-scattered tungsten from sample tungsten requires the software to account for it, and it is a reason rhodium-target instruments are often preferred for this application.

Density: necessary but insufficient

Archimedes displacement remains a core check despite tungsten’s near-match, because it catches everything else and narrows the field.

The measurement

Weigh the bar in air, then suspended in water. Density equals dry weight divided by the weight difference, adjusted for water temperature.

Done carefully with a balance reading to 0.01 g, this achieves roughly 0.1 to 0.2 percent precision on a substantial bar. Against tungsten’s 0.36 percent density difference, that is marginal but not useless.

Why it is marginal in practice

Several factors erode the theoretical precision:

• Air bubbles clinging to the surface displace water and bias the result. A wetting agent and careful immersion technique are required.
• Water temperature changes density measurably and must be compensated.
• Surface texture and cavities trap air in ways that are hard to control.
• Composite cores. A sophisticated counterfeit adjusts the tungsten-to-gold ratio and may incorporate a small void or a lighter inclusion to bring measured density to exactly 19.32.

The practical position is that density reliably exposes brass, lead, and steel, and provides a weak signal against tungsten. Treat a density pass as ruling out the crude fakes, not as clearing the bar.

Ultrasonic testing: the method that actually works

Ultrasonic velocity measurement is the most effective non-destructive test for this specific fraud, because it exploits a property the counterfeiter cannot match.

The physical basis

Sound travels at characteristic speeds through different materials:

• Gold: approximately 3,240 m/s
• Tungsten: approximately 5,180 m/s

That is a 60 percent difference — enormous compared to density’s 0.36 percent. Tungsten’s stiffness relative to its density makes it acoustically nothing like gold, and no adjustment of the core composition can bring the two into agreement while preserving the density match.

How the test is performed

An ultrasonic thickness gauge sends a pulse into the bar and times the echo from the far face. Knowing the bar’s physical thickness, the instrument computes the velocity through the material.

A solid gold bar returns a velocity near 3,240 m/s and a single clean echo. A tungsten-cored bar returns either an anomalous velocity or, more tellingly, multiple echoes from the internal gold-tungsten interface. That interface reflection is direct evidence of a layered structure and is difficult to disguise.

Practical requirements

The method needs a reasonably flat surface, couplant gel between probe and bar, and an operator who understands what a clean echo looks like. It works best on bars of substantial thickness and is less straightforward on thin wafers or coins.

Equipment cost is modest relative to the value being protected, and for any operation regularly handling investment bars it is the highest-return addition to the verification toolkit.

How these bars are actually made

Understanding the manufacturing constraints tells you where to look, because every construction method leaves evidence somewhere.

Machined core with a pressed shell

A tungsten blank is machined to the required dimensions, then gold sheet is wrapped, pressed, and the seams worked closed and polished. This is the most common approach because it avoids any need to melt tungsten.

Evidence to look for: seam lines along edges, slightly rounded corners where a genuine pressed bar is crisp, and thickness variation in the gold shell detectable by ultrasonic scanning at multiple points. Corners are the weak spot, since wrapping sheet around a right angle thins it there.

Drilled and plugged

A genuine bar is drilled, gold removed, tungsten rod inserted, and the hole plugged and refinished. Often the drilling is done from the underside or through the stamped face where surface disruption is easier to disguise.

Evidence: a plug boundary sometimes visible under raking light or magnification, slight surface texture differences over the plug, and XRF readings that vary between the plug area and the surrounding metal if the plug material differs at all. This variant is the one most likely to be caught by XRF, because a poorly executed plug can expose tungsten at the surface.

Cast around a core

Molten gold cast around a preformed tungsten insert. Tungsten’s melting point far exceeds gold’s, so the insert survives casting intact.

Evidence: casting porosity, less crisp detail than a struck or pressed bar, and often a density slightly off because casting shrinkage creates voids. Ultrasonic testing reveals the interface clearly in this variant.

Layered composite

Alternating layers of gold and tungsten, engineered to tune overall density precisely. This is the most sophisticated construction and the hardest to detect by density alone.

Evidence: ultrasonic scanning produces multiple interface echoes at different depths, which is a signature no genuine bar produces. This is exactly the case where ultrasonic testing is not merely helpful but essential.

Coins versus bars

The two categories present different risk profiles and call for different emphasis.

Why coins are somewhat harder to fake

Bullion coins carry design detail, edge reeding, and precise dimensional tolerances that are difficult to reproduce while accommodating a core. Their smaller mass also limits the absolute gain from coring.

More usefully, coins have tightly specified diameter, thickness, and weight. A counterfeit that matches weight by using a tungsten core generally cannot also match all three dimensions, because tungsten’s density is close to gold’s but not identical. Precision calipers and a scale catch a meaningful proportion of coin fakes on specification alone.

Where coins remain vulnerable

Thicker coins and larger denominations offer more room. Coins in sealed assay cards cannot be measured dimensionally without breaking the seal, which creates a tension between verification and preserving numismatic packaging value.

Ultrasonic testing on coins is also more difficult, since thin sections produce closely spaced echoes that are harder to interpret than the clean returns from a thick bar.

The specific gravity approach for coins

For coins, careful specific gravity measurement combined with dimensional verification is often more practical than ultrasonics. The published specifications for common bullion coins are precise enough that a genuine coin should match all parameters simultaneously, and a core forces at least one to deviate.

Building policy rather than reacting case by case

Operations that handle bars regularly are better served by a written policy than by ad hoc judgment, because judgment degrades under time pressure and commercial incentive.

Define testing tiers by value and counterparty

Set explicit thresholds. For example: visual and dimensional on everything; XRF on everything; density above a stated value; ultrasonic above a higher value or for any unfamiliar counterparty. Writing the thresholds down means they get applied when the transaction is attractive and the seller is impatient.

Record what each test established

A log entry reading “XRF verified” is actively misleading on a bar that was never ultrasonically tested, because it implies a level of assurance that was not obtained. Record which tests were run and which were not.

Maintain reference material

Keep known genuine bars from the brands you handle most. Comparative inspection against a real example catches subtle stamping and finish differences that are invisible in isolation. This costs nothing beyond holding inventory you would hold anyway.

Train on the failure mode, not the equipment

Staff need to understand why a clean XRF reading does not clear a bar. An operator who believes the analyzer settles the question will not escalate, no matter what the policy says. The conceptual point — that XRF measures the surface, and the fraud is beneath it — is the training content that matters.

Supporting methods

Several additional checks add independent evidence, each exploiting a different property.

Electrical conductivity

Gold conducts at roughly 44 percent IACS; tungsten at about 31 percent. Eddy current conductivity meters measure this quickly and non-destructively.

The limitation is penetration depth. Eddy currents concentrate near the surface, so a thick gold layer over tungsten may read as gold. Conductivity is a useful supporting signal on thinner items and a weak one on heavily clad bars.

Thermal conductivity

Gold conducts heat at approximately 318 W/m·K, tungsten at about 173. Handheld thermal probes exploit this, and the difference is large enough to be meaningful.

The same depth caveat applies. Surface thermal response is dominated by the surface layer, so this method suits thin plating rather than deep cores.

Magnetic response

Both gold and tungsten are essentially non-magnetic, so a simple magnet test does not distinguish them. It remains worth performing because it instantly exposes steel and nickel-based fakes, and it costs nothing.

A magnetic slide test — observing how a strong magnet decelerates when sliding along the bar due to eddy current braking — is more informative than a static magnet, and reflects conductivity differences.

Dimensional and visual inspection

Genuine bars from recognized refiners have precise dimensions, consistent finish, sharp stamping, and correct serial and assay markings. Counterfeits frequently show:

• Dimensions marginally off standard, since matching both weight and volume with a tungsten core forces geometric compromises
• Soft or shallow stamping where a genuine bar has crisp impressions
• Seam lines, plugged holes, or surface irregularities on faces or edges
• Packaging and assay certificate inconsistencies

Visual inspection catches a meaningful share of fakes before any instrument is used, and it costs only attention.

The drill and the cut

Drilling into the bar and testing the recovered material, or sectioning it outright, is definitive. It is also destructive and reduces the bar to scrap value for resale purposes.

This is the correct action when non-destructive methods leave genuine doubt on a high-value item. The cost of destroying one bar is far below the cost of accepting a counterfeit kilogram.

A layered verification sequence

No single test is sufficient. The following sequence orders checks by cost and speed, escalating only when required.

Level one: free and immediate

Visual inspection against known genuine examples. Dimensional check with calipers against published specifications. Magnet test. Weight check against stated weight.

Any failure here ends the process. A bar that is dimensionally wrong or magnetically responsive does not need further testing.

Level two: XRF surface assay

Test multiple faces, including edges and corners where a core is more likely to approach the surface. Look specifically for tungsten Lα at 8.40 keV, and for any trace elements inconsistent with the claimed fineness.

This clears thin-plated fakes and all non-tungsten substitutes. It does not clear deep tungsten, and the record should state that explicitly rather than logging the bar as “XRF verified.”

Level three: density

Archimedes displacement with careful technique. This confirms nothing about tungsten but definitively excludes lead, brass, steel, and hollow construction.

Level four: ultrasonic velocity

The decisive test. Look for velocity near 3,240 m/s and, critically, for the absence of internal interface echoes. This is where a tungsten core is actually caught.

Level five: destructive confirmation

Drill or section where doubt remains and value justifies it. Reserved for cases where levels one to four disagree or where the transaction size makes certainty essential.

Applying the sequence proportionately

Most operations do not run all five levels on every bar. A reasonable policy runs levels one and two universally, adds level three above a value threshold, and adds level four for investment-grade bars, unfamiliar counterparties, or any bar where earlier levels raised a question. The threshold structure used for incoming bar verification is set out in our bullion bar verification SOP.

Risk factors worth weighting

Verification intensity should scale with risk, and certain signals justify escalation regardless of value.

• Unknown counterparty with no transaction history
• Price below market — the most reliable single indicator of a problem
• Missing or inconsistent documentation, particularly assay certificates that do not match serial numbers
• Larger bars, since the economics of coring improve with size and kilogram bars are the primary target
• Pressure to complete quickly, which is a behavioural signal rather than a technical one but correlates strongly
• Bars outside recognized refiner branding, or brands that are unfamiliar and hard to verify

Small bars and coins are less commonly cored simply because there is less room for a core and the margin is thinner. This is a tendency rather than a rule.

Frequently asked questions

Can XRF detect a tungsten core in a gold bar?

Not when the core is buried under more than a few tens of microns of gold. XRF measures only the surface layer, and a typical core sits one to several millimetres deep. It does detect thin gold plating over tungsten and all non-tungsten substitutes.

Why does density testing not catch tungsten?

Tungsten’s density of 19.25 g/cm³ is within 0.36 percent of gold’s 19.32. Practical displacement measurement achieves 0.1 to 0.2 percent precision at best, leaving little margin, and sophisticated fakes tune the core ratio to land on the correct figure.

What is the most reliable non-destructive test?

Ultrasonic velocity measurement. Sound travels about 60 percent faster through tungsten than gold, and an internal interface produces a characteristic secondary echo. Unlike density, this is a property the counterfeiter cannot match while preserving the density match.

Will a magnet detect a tungsten core?

No. Both gold and tungsten are essentially non-magnetic. A magnet test is still worth performing because it immediately exposes steel and nickel-based fakes, but it says nothing about tungsten.

Are smaller bars safer?

Somewhat. Coring a small bar leaves little room for the core and yields a smaller absolute gain, so kilogram bars are the primary target. This reduces risk rather than eliminating it, and small bars have been cored.

Does buying from a recognized refiner eliminate the risk?

It reduces it substantially, since counterfeits typically enter through secondary markets rather than direct from the refiner. Verify the chain of custody as well as the branding, since counterfeit bars are frequently made to imitate exactly the brands buyers trust.

Can a bar pass XRF and still be fake?

Yes, and this is the central point. A tungsten-cored bar with a genuine gold shell passes XRF because the surface genuinely is gold. Passing XRF means the surface is what it claims to be, not that the interior is.

Is drilling worth the damage on a suspect bar?

When non-destructive methods leave genuine doubt on a high-value bar, yes. The loss from reducing one bar to scrap value is far smaller than the loss from accepting a counterfeit kilogram, and a genuine bar can still be sold on metal content.

What to take away

XRF is a necessary part of bar verification and an insufficient one. It reliably confirms surface composition and reliably exposes plated and non-tungsten fakes. It cannot see through millimetres of gold, and treating a clean XRF result as clearance is precisely the assumption this fraud is designed to exploit.

Density excludes the crude substitutes and provides only weak evidence against tungsten. Ultrasonic velocity is the test that actually resolves the question, because acoustic behaviour is the property tungsten cannot fake while holding its density match.

The defensible position is a layered protocol with thresholds agreed in advance, applied consistently, and documented in a way that records what each test did and did not establish.

For the depth limitation in quantitative detail, see plating thickness measurement limits. For broader counterfeit patterns beyond tungsten, see our counterfeit gold detection guide.

What to do when a bar fails

Detection is only useful if the response is decided in advance. A failed verification in progress, with a seller present, is the worst moment to improvise.

Do not accuse

The person presenting the bar is frequently not the person who made it. Counterfeit bullion moves through several hands, and an intermediate holder may have bought it in good faith and been defrauded themselves.

State the finding factually: the ultrasonic reading is inconsistent with solid gold, and the transaction cannot proceed on that basis. Avoid characterizing intent.

Do not return it without a record

If the bar leaves without documentation, it will be presented elsewhere. Photograph it including serial numbers and stamps, record the test results, and note the presenting party’s details to the extent your jurisdiction permits and requires.

Understand your reporting obligations

Requirements vary considerably by jurisdiction. Dealers in precious metals are subject to anti-money-laundering reporting in many countries, and suspected counterfeit bullion may trigger a specific obligation. Knowing what your obligation is before you encounter a fake is part of the preparation.

Preserve the evidence chain

If a matter proceeds formally, contemporaneous records carry weight that recollection does not. The test results, photographs, and timestamped notes made at the counter are the record that matters.

Cost of verification against cost of failure

Operations sometimes resist adding ultrasonic capability on cost grounds. The arithmetic does not support that position for anyone handling investment bars.

A single accepted kilogram bar with a substantial tungsten core represents a loss in the tens of thousands. Ultrasonic thickness gauges suitable for this application cost a small fraction of that, and the equipment does not expire.

The time cost is likewise modest. An ultrasonic check on a flat bar adds a minute or two to a transaction that already involves weighing, XRF, and paperwork. Against transactions of this value, that is not a meaningful friction.

The real cost is organizational: writing the policy, training staff, and maintaining the discipline to apply it when a large attractive transaction is sitting on the counter and the seller is impatient. That discipline is what the equipment enables rather than replaces.

Where verification cost genuinely does not justify itself

Small scrap gold, jewelry lots, and low-value transactions do not warrant ultrasonic testing. The economics of coring do not work at small scale, and XRF plus density is proportionate.

The threshold worth setting is the value at which a single undetected fake would materially hurt the business. Below it, screen efficiently; above it, verify properly. Applying investment-bar protocols to every jewelry transaction wastes effort that should be spent where the exposure actually sits.