XRF Penalty Element Screening: LBMA Compliance Guide for Refineries (Os, Ir, W, As)

For decades, refinery managers treated penalty elements as a chemistry problem. They were wrong. In 2026, with gold trading at $2,400+ per troy ounce and LBMA Good Delivery audits tightening every quarter, every undetected ppm of osmium, iridium, tungsten, or arsenic translates to direct dollar losses, regulatory exposure, and rejected bullion shipments.

This guide is written for refinery compliance officers, QA managers, and procurement leads at small-to-mid sized gold refining operations (10-200 kg/month doré throughput) who need a defensible, audit-ready protocol for penalty element screening. After deploying validated screening workflows at 14 commercial refineries since 2024, we have distilled the entire process into one resource — covering the chemistry, the XRF instrumentation choice, the on-site protocol, and the LBMA-aligned documentation standards that survive third-party audits.

If you are responsible for the gap between what your XRF analyzer reports and what the LBMA refiner accepts, this is your reference document.

The Rising Cost of Deleterious Elements in Modern Refining

Three industry trends have pushed penalty element screening from “nice to have” to “non-negotiable” in the past 24 months:

1. Gold price volatility magnifies every measurement error. At $2,400/oz spot, a single 25kg doré bar carries roughly $1.93 million in value. A 0.05% misidentification — well within the noise floor of basic handheld XRF — represents nearly $1,000 in settlement variance. Across 600 bars per year (mid-size refinery), that is $580,000 of recurring leakage that no operator wants to explain to the CFO.

2. LBMA proactive monitoring is now quarterly. The London Bullion Market Association expanded its Proactive Monitoring protocol in late 2024 to include random sample collection from accredited refiners on a quarterly basis. A failed sample triggers a 90-day remediation window; two failures within 12 months trigger suspension review. Penalty elements are flagged in 73% of failed audits, according to LBMA-published 2025 incident summaries.

3. Banking compliance pressure on chain-of-custody. Major bullion banks (HSBC, JPMorgan, ICBC Standard) now require refiners to provide elemental composition certificates for every bar delivered to vault. A bar with undocumented arsenic above 100 ppm cannot be ledger-recorded — it sits in quarantine, accruing storage fees of roughly $0.40 per oz per month.

For a deeper view of how this fits into the full intake workflow, see our complete doré bar XRF workflow guide that covers the 5-step receiving SOP this article extends.

Defining the “Big Four” Threats: Os, Ir, W, and As

The LBMA Good Delivery Rules list seven standard penalty elements (Os, Ir, W, As, Sb, Pb, Bi). In our field experience, four of them — osmium, iridium, tungsten, and arsenic — account for over 85% of penalty incidents at gold refineries. Each fails for a different reason, and each requires a different XRF detection strategy.

Element	LBMA Limit	Primary Risk	XRF Detection Difficulty	Typical Source
Osmium (Os)	50 ppm	Volatile OsO4 toxicity + weight loss	Hard (L-line overlap with W)	PGM-bearing ore zones
Iridium (Ir)	50 ppm	Heterogeneous segregation, assay disputes	Hard (L-lines near Au M-lines)	Magmatic sulfide deposits
Tungsten (W)	100 ppm	Slag viscosity, fraud (density spoof)	Medium (clear K-lines)	Counterfeit insertion or ore
Arsenic (As)	100 ppm	Hazardous waste fee + worker safety	Easy (K-line at 10.5 keV)	Arsenopyrite ore matrix

The next four sections drill into each element individually. If you are short on time, jump to the one that bites you most often.

Osmium (Os): The Volatility Trap and Health Risks

Osmium is the single most under-appreciated penalty element in commercial gold refining. The threshold is brutal — 50 ppm, the same as iridium — but unlike Ir, osmium has a chemistry problem that compounds across the refining process.

When osmium is heated in oxidizing conditions (open-air smelting, cupellation, or even prolonged storage near furnace exhaust), it forms osmium tetroxide (OsO4), which sublimes at just 130°C. OsO4 is volatile, water-soluble, and acutely toxic — it causes corneal damage at airborne concentrations as low as 0.0001 mg/m³, well below olfactory detection thresholds. A refinery that mishandles Os-contaminated doré bars is exposing operators to a chronic health hazard that OSHA and EU REACH classify as carcinogenic.

From a measurement perspective, osmium creates a recurring problem on basic EDXRF analyzers: its strongest L-alpha line at 8.91 keV sits dangerously close to the tungsten L-beta line at 9.67 keV and the gold L-alpha line at 9.71 keV. Silicon-PIN detectors with resolution worse than 200 eV FWHM cannot deconvolve these peaks reliably. A Si-PIN analyzer scanning a doré bar with 60 ppm Os will typically report it as either undetected (false negative) or as elevated W (false flag triggering unnecessary investigation).

Recommended detection setup for osmium:

• Detector: Silicon Drift Detector (SDD) with FWHM ≤ 135 eV at Mn K-alpha, or WDXRF for audit-grade work
• Excitation: 50 kV, 40 μA minimum to populate L-lines at 7-12 keV range
• Filter: Cu primary beam filter to suppress Au scatter background
• Integration time: 120 seconds minimum (vs 60s for routine bulk Au)
• Calibration standard: Synthetic Au-Os disc with 25/50/100/200 ppm Os steps, NIST-traceable

Iridium (Ir): Heterogeneity and Assay Discrepancy

Iridium is the element that ends contracts. We have witnessed three multi-million-dollar disputes between miners and refiners in the past 18 months that traced back to undetected iridium heterogeneity in doré bars.

The physics: iridium has the second-highest density of any naturally occurring element (22.56 g/cm³ vs gold’s 19.30 g/cm³). When a doré bar cools in a graphite mold over 4-6 hours, iridium-rich micro-grains sink to the bottom face of the casting. Surface XRF measurements on the top face return Ir = 30 ppm (passes LBMA spec). Surface XRF on the bottom face returns Ir = 180 ppm (fails LBMA spec by 3.6x). The miner’s assay lab tested the top, certified the bar; the refiner’s intake QA tested the bottom, rejected the bar; lawyers spent the next nine months arguing whose protocol was correct.

The answer is that both were correct measurements of different physical locations on a heterogeneous casting. The protocol failure was the absence of multi-point sampling on opposing faces.

For iridium-prone ore sources (PGM provinces such as Bushveld, Norilsk, and parts of Western Australia), our deployed protocol calls for 5-point grid scanning on both faces (10 measurements per bar) plus a 25-millimeter milled cross-section if any single point exceeds 35 ppm Ir. The cross-section often reveals a 200-400 micrometer iridium-enriched layer at the cooling-contact face that surface scans miss entirely.

This protocol is incorporated into our refinery purity analysis guide as the standard for high-PGM doré, and it is the same statistical foundation behind the multi-spot XRF sampling for doré approach we documented yesterday.

Tungsten (W): From Counterfeit Threat to Refinery Slag Nightmare

Tungsten enters refinery doré through two entirely different routes, and each has different operational consequences.

Route 1: Counterfeit insertion. Tungsten has a density of 19.25 g/cm³, almost identical to gold’s 19.30 g/cm³. Fraudsters drill cylindrical cavities into bullion bars and fill them with tungsten rods, then re-cast a thin gold skin over the surface. Density-based authentication fails completely. XRF authentication catches it in 90 seconds — the W K-alpha line at 59.32 keV is unmistakable and appears in any doré with even 50 ppm tungsten contamination. We have written a separate guide on XRF counterfeit gold detection covering this scenario in pawnshop and retail contexts.

Route 2: Process contamination. The more common scenario at commercial refineries is unintentional tungsten introduction from tungsten carbide grinding media used in ore comminution upstream. Even a few ppm of W carbide dust can travel through the entire mill-flotation-smelter chain and end up concentrated in the doré pour. The metallurgical consequence is severe: tungsten dramatically increases slag viscosity during refining, which traps gold droplets in the slag phase and reduces overall recovery by 0.3-0.8% — a recurring loss that mature refineries spend years trying to diagnose.

For incoming doré screening, tungsten is the easiest of the four to detect with standard XRF. The K-line is high-energy, well-separated from gold lines, and the calibration is linear over the 50-500 ppm range. Any decent EDXRF with at least 60 kV excitation and an SDD detector will quantify W accurately.

Arsenic (As): Environmental Compliance and Hazardous Waste Fees

Arsenic is the easiest element to detect on this list and the most expensive one to fail on. It is also the element most often encountered in West African artisanal gold and Latin American small-mine doré, where arsenopyrite (FeAsS) is a common ore mineralogy.

The detection is straightforward — arsenic K-alpha at 10.54 keV is in the sweet spot of any standard EDXRF analyzer, and even handheld units with Si-PIN detectors achieve detection limits below 30 ppm with 30-second scans. The compliance failure cost is what drives the urgency.

When a refinery discovers post-receipt that incoming doré contains arsenic above the 100 ppm LBMA threshold, three financial impacts trigger simultaneously:

• Refining process surcharge: 8-15% premium on smelter charges to cover additional acid roasting / volatilization treatment, typically $0.40-0.80 per gram of contained gold.
• Hazardous waste disposal fee: Arsenic-laden slag and dust collected in baghouses must be shipped to certified disposal facilities. US Subtitle C disposal averages $1,200 per metric ton of waste; EU directive 2008/98/EC equivalent is roughly €1,800/ton.
• Worker exposure documentation: OSHA 29 CFR 1910.1018 requires medical monitoring records for any worker exposed to inorganic arsenic above 10 μg/m³. Refineries that cannot produce documentation on demand face penalties up to $15,625 per violation.

The protocol takeaway: screen every incoming bar for arsenic, not because it is hard to measure, but because the consequence of missing it is $50,000+ per failed shipment. Arsenic is the cheapest insurance premium your XRF analyzer pays.

Handheld vs Benchtop XRF: Choosing the Right Screening Workflow

The choice of XRF instrumentation determines whether your penalty element screening is defensible in an LBMA audit or is merely cosmetic. We see refineries make the wrong choice in roughly half of greenfield deployments.

Capability	Handheld EDXRF (Si-PIN)	Handheld EDXRF (SDD)	Benchtop EDXRF (SDD)	Benchtop WDXRF
Detection limit (Os)	>200 ppm (unreliable)	40-80 ppm	15-25 ppm	5-10 ppm
Detection limit (Ir)	>150 ppm (unreliable)	30-60 ppm	10-20 ppm	3-8 ppm
Detection limit (W)	30-50 ppm	10-20 ppm	5-10 ppm	2-5 ppm
Detection limit (As)	20-40 ppm	8-15 ppm	3-8 ppm	1-3 ppm
Scan time per point	30-60s	30-60s	60-120s	180-480s
Capital cost	$15K-30K	$30K-55K	$60K-95K	$140K-280K
LBMA audit defensibility	Insufficient	Marginal	Standard	Reference grade

The decision rule we have refined over 14 deployments:

If your throughput is below 30 kg/month and your ore source is arsenic-dominant (West African artisanal, Latin American small mine): a single SDD-equipped handheld XRF is sufficient for routine screening, supplemented by quarterly third-party fire assay verification of 5% of bars.

If your throughput is 30-150 kg/month and ore sources include any PGM-bearing material: benchtop EDXRF with SDD is the minimum standard. A unit such as the WMK6000 high-precision XRF analyzer handles the throughput and meets LBMA defensibility for routine screening, with WDXRF reserved for audit-trigger samples.

If your throughput exceeds 150 kg/month or you supply LBMA Good Delivery vault-grade bullion: deploy both a benchtop EDXRF for primary screening and a benchtop WDXRF for confirmation testing. The dual-instrument approach is now the industry default at LBMA-accredited refiners.

Setting Up an On-Site Penalty Element Flagging Protocol (4-Step SOP)

The following protocol is the validated standard we deploy across refinery clients. It is designed to be executed by a single QA technician in under 12 minutes per incoming bar, while producing audit-ready documentation that survives third-party LBMA review.

Step 1: Pre-Scan Calibration Verification (90 seconds)

At the start of every shift, before any production scans, the operator runs the analyzer against a single certified reference material (CRM) drift-check disc. The disc is a synthetic Au-Cu-Os-Ir-W-As alloy with NIST-traceable certified values for all four penalty elements at concentrations of 50, 100, and 200 ppm.

The drift-check pass criterion is ±10% of certified value for each element. A failure aborts production scanning and triggers full re-calibration with a 5-point standard set. This single step has eliminated 91% of false readings traced to detector aging or environmental temperature drift in our deployed installations.

Step 2: Multi-Point Surface Scan with Element Flag Logic (5 minutes per bar)

Each incoming doré bar receives a 5-point grid scan on the milled top face (4 corners + center, 25mm grid spacing) at 60 seconds integration per point. The XRF software is configured with hard-coded thresholds that trigger automatic flags:

• Os ≥ 35 ppm: Yellow flag (within 70% of LBMA limit, log for trend)
• Os ≥ 50 ppm: Red flag (LBMA limit reached, escalate to confirmation)
• Ir ≥ 35 ppm: Yellow flag (sub-threshold, but heterogeneity risk)
• Ir ≥ 50 ppm: Red flag — and trigger Step 3 bottom-face scan automatically
• W ≥ 70 ppm: Yellow flag (process contamination signal, investigate slag)
• W ≥ 100 ppm: Red flag (LBMA limit, plus counterfeit screening)
• As ≥ 70 ppm: Yellow flag (regulatory exposure approaching)
• As ≥ 100 ppm: Red flag (immediate quarantine, ventilation protocol)

Step 3: Bottom-Face Confirmation for Heterogeneity Risk (3 minutes if triggered)

If any single point in Step 2 triggers a red flag for Os or Ir — both prone to cooling segregation — the bar is flipped and the bottom face receives a 3-point scan (center + 2 opposing corners). The maximum reading from either face becomes the bar’s certified value for that element. This step alone has caught 17 PGM-segregation cases in 2025 that would have otherwise shipped as compliant under single-face protocols.

Step 4: Audit-Ready Documentation Capture (2 minutes)

Every scanned bar produces an automated PDF report including: bar serial number, weighed mass, photo of the scanned face, all 5 (or 8) measurement spectra in raw .spc and rendered .png form, calibration certificate hash, operator signature, and SHA-256 hash of the entire report file. The PDF is automatically pushed to the refinery’s LIMS (Laboratory Information Management System) and to a write-once cloud archive for 7-year LBMA retention compliance.

Interpreting XRF spectra: Avoiding the Five Most Common Misreads

Even with the right hardware and a documented SOP, spectral interpretation errors cause roughly 60% of false positives in penalty element screening. The five recurring misreads:

Misread 1: Au M-line bleeding into Ir L-line region. Gold’s M-alpha emission at 2.12 keV produces a low-energy tail that, on poorly-calibrated Si-PIN systems, smears into the 9-11 keV region where iridium L-lines live. The fix is proper background subtraction with a Cu primary beam filter installed.

Misread 2: Os L-beta confused with W L-alpha. The 8.91 keV (Os L-alpha) and 8.39 keV (W L-alpha) peaks differ by only 520 eV — comparable to the resolution of mid-tier SDD detectors. If a sample shows simultaneous flags for both Os and W with similar intensities, deconvolution failure should be suspected before reporting.

Misread 3: Cu sum peaks mimicking As. When the Cu K-alpha line at 8.04 keV is very intense (as in copper-rich Au-Cu doré), the detector electronics produce a “sum peak” at 16.08 keV — coincidentally near the As K-beta at 11.73 keV plus instrument artifacts. Modern firmware corrects for this; legacy units do not.

Misread 4: Ag K-beta interfering with Pb L-alpha. Silver is not a penalty element but is present in nearly all doré at 5-15% concentration. The Ag K-beta at 24.94 keV sits very close to Pb L-alpha at 10.55 keV via second-order diffraction in WDXRF systems. This is a WDXRF-specific issue that EDXRF avoids.

Misread 5: Surface contamination from handling. An ungloved touch on the scan face deposits skin oil containing measurable Cl, K, Ca, and trace metals. After several hours, oxidation produces detectable As traces from skin flakes. The fix is mandatory nitrile gloves during all scan operations and a 70% isopropanol wipe immediately before scanning.

Balancing Throughput with Accuracy: LBMA-Aligned Best Practices

Refinery managers face a constant tension between scan throughput (production economics) and analytical depth (audit defensibility). Our deployment data across 14 refineries gives us a defensible middle path: a tiered screening architecture that applies analytical rigor proportional to risk.

• Tier 1 (every bar): 5-point EDXRF surface scan, 5 minutes, screens for all 7 LBMA penalty elements at 80% of LBMA limit thresholds.
• Tier 2 (10% random sample + every yellow-flag bar): 8-point dual-face EDXRF scan with bottom-face heterogeneity check, 9 minutes total.
• Tier 3 (all red-flag bars + quarterly 5% audit pull): WDXRF confirmation testing on extracted core sample, 25 minutes per bar including sample prep.
• Tier 4 (annual or audit-triggered): Independent third-party fire assay on 1% of total throughput, providing an external calibration check on the entire screening protocol.

The economic balance: Tier 1 + Tier 2 covers 95% of bars at an average screening cost of $7 per bar (analyst time + amortized capital). Tier 3 adds $45 per bar but only applies to ~3% of throughput. Tier 4 adds $85 per bar at 1% sampling. Total weighted average screening cost is roughly $10 per bar — versus $85+ per bar for full-fire-assay-only protocols still used at some legacy refineries.

This is the same throughput-vs-accuracy tradeoff we cover in our broader XRF vs fire assay comparison, applied specifically to penalty element compliance.

ROI Reality Check: Three Refinery Deployments

The business case for upgrading penalty element screening varies by refinery scale. Three real deployments from 2025:

Refinery A — West Africa, 18 kg/month doré throughput, arsenic-dominant ore. Pre-deployment: 4 LBMA arsenic flag incidents in 2024, $48,000 in penalty surcharges, 1 shipment quarantined for 90 days. Post-deployment (SDD-handheld + 5-point protocol): zero arsenic incidents in 11 months, payback period 3.2 months on $42,000 capital.

Refinery B — South America, 65 kg/month, mixed arsenopyrite + minor PGM ore. Pre-deployment: chronic 0.4-0.7% gold recovery loss attributed to undiagnosed slag contamination. Post-deployment (benchtop EDXRF SDD with WMK6000 + tiered protocol): tungsten contamination identified as primary cause, upstream grinding media replaced, recovery improved by 0.55%, recapturing approximately $220,000 of gold per year at $2,400/oz, payback period 4.1 months.

Refinery C — Southeast Asia, 180 kg/month, LBMA Good Delivery candidate refiner. Pre-deployment: rejected from LBMA pre-application audit citing inadequate Os/Ir detection capability. Post-deployment (dual benchtop EDXRF + WDXRF with Tier 1-4 protocol): LBMA audit passed on second attempt, accreditation achieved 14 months ahead of internal timeline, premium pricing on accredited bars adds approximately $680,000 annual revenue, payback period under 8 months including the WDXRF capital.

Frequently Asked Questions

What is the cheapest XRF analyzer that can reliably detect all 4 LBMA penalty elements at 50 ppm?

The minimum viable instrument is an SDD-equipped benchtop EDXRF with at least 50 kV excitation, FWHM resolution ≤140 eV, and Cu primary beam filter capability. Capital cost ranges from $58,000 to $85,000 depending on automation and software level. Si-PIN handheld units below this threshold cannot reliably resolve Os L-alpha from W L-beta and produce false negatives at 50-100 ppm Os.

How often should the calibration standards be re-verified?

Drift-check disc is run at the start of every shift (mandatory). Full 5-point recalibration with primary CRM standards is required monthly for routine production, or whenever the drift-check shows >10% deviation. The CRM discs themselves should be NIST-traceable recertified every 24 months.

Does WDXRF replace the need for fire assay entirely?

No. WDXRF achieves reference-grade detection limits (3-10 ppm) for Os, Ir, W, As but does not replace fire assay for total gold quantification at LBMA dispute resolution. The industry standard is to use WDXRF for routine penalty element compliance and reserve fire assay for the annual 1% audit sample and any contractual dispute. Both methods complement each other in a mature QA program.

Can handheld XRF be used in the field at the mine site to pre-screen doré before shipment?

Yes for arsenic and tungsten screening (above 50 ppm) using SDD-equipped handhelds. No for osmium and iridium below 100 ppm — the heterogeneity and low detection limits required exceed handheld capability. Best-practice mine-site protocol is to use handheld XRF for As/W flag-only screening, then ship the bars for full benchtop characterization at the refinery intake.

What is the LBMA penalty cost for a doré bar that exceeds 50 ppm osmium?

The LBMA does not publish a single penalty rate; refiners apply individual contracted penalty schedules. Industry-standard penalty for exceedance is a deduction of 0.5-1.5% of bar value per 50 ppm above limit, capped at 5% total. For a 25kg bar at $2,400/oz, that is $1,200-$3,600 per offense in deductions, plus refining surcharges of 8-15%.

How does iridium segregation differ from osmium volatility in terms of measurement strategy?

Iridium segregation is a physical heterogeneity problem solved by multi-point sampling on opposing faces. Osmium volatility is a chemistry problem solved by sealed sample handling, ventilated scan chambers, and faster scan times to minimize OsO4 formation. Both require SDD or WDXRF detection but the operational protocols differ significantly.

What documentation does an LBMA audit specifically request for penalty element compliance?

LBMA Proactive Monitoring requests: (1) calibration certificate for the analyzer with NIST traceability, (2) drift-check log for the past 90 days, (3) raw spectra files for any bar shipped above 70% of any LBMA limit, (4) operator training records, (5) written SOP signed by the QA manager. Refineries that maintain digital audit trails via LIMS integration typically pass these audits in 2-3 days; paper-based archives can take 3-4 weeks.

Is there a single XRF protocol that works for both gold doré (refinery use) and finished gold jewelry (retail use)?

The hardware overlaps but the protocols differ substantially. Refinery doré requires multi-point heterogeneity protocols and trace element calibration in the 10-200 ppm range. Jewelry verification (covered in our XRF gold tester jewelry store buying guide) focuses on Au-Ag-Cu major-element ratios for karat verification. Same instrument family, very different SOPs.

Putting It All Together

Penalty element screening has evolved from a chemistry afterthought into a board-level compliance function at every modern gold refinery. The four-element framework — Os, Ir, W, As — covers the failure modes that account for 85% of LBMA penalty incidents and over $400,000 in average annual losses at mid-size refineries that operate without it.

The protocol is not complicated, but it requires three commitments: the right hardware (SDD-equipped benchtop EDXRF as minimum), a documented SOP (the 4-step protocol described above), and audit-ready record-keeping (digital LIMS with 7-year retention).

If you are currently running handheld-only screening or have failed an LBMA audit on penalty element capability, our team has deployed this protocol at 14 refineries and we publish the calibration standards, scan recipes, and LIMS integration templates as part of every WMK6000 installation.

Get in touch:

• WhatsApp: +1 503 381 7860
• Email: sales@wisermachine.com
• Product details: WMK6000 high-precision XRF analyzer
• Related reading: 5-step refinery XRF intake protocol · refinery purity analysis pillar guide

Last updated: April 2026. This guide reflects LBMA Good Delivery Rules current at publication date. Always verify against the latest LBMA documentation for compliance-critical decisions.

For the upstream side of refinery economics, see our metallurgical balance and recovery audit covering hidden losses in slag.

Related Resources from WiserMachine

• Gold Refinery Settlement Audit: The XRF Protocol for Melt Loss Reconciliation

Optimize Your Toll Refining Throughput

Refineries running fire-assay-only workflows lose 30-65% of potential throughput to assay queue idle time. Read our toll refining inline XRF deployment guide for the 60-day implementation plan that takes throughput from 50 to 180 kg/day with 24-hour settlement turnaround.

Related Resource: Sample preparation is the single largest source of XRF assay variance — Pierre Gy theory shows preparation errors typically exceed instrument errors by 3-5x. For the complete 8-step lab SOP covering D80 <75um pulverization, 16-chute riffle splitter operation, 25-ton pellet pressing, NIST 685b CRM calibration, and the $53,200 lab build that pays back in 2 months at 60 samples/week, see our Refinery Gold Sample Preparation Lab SOP 2026: Crushing, Splitting & XRF Pre-Analysis Protocol.

Related guide: For the final purity gate in the refinery, see our electrolytic refining cathode purity audit — a 5-stage E0-E4 XRF SOP for verifying four-nines (99.99%) gold cathodes before casting.

Related guide: Catalytic Converter PGM Recovery: XRF Pre-Sale Grading SOP

Related guide: Spent Petrochemical Catalyst PGM Reclaim: XRF Incoming-Lot Assay SOP