PMI Testing for Pipe Fittings: XRF vs OES, What Each Method Can and Cannot Detect
Positive Material Identification (PMI) is the in-service or in-shop verification that a pipe fitting is actually made from the alloy stated on the tag or certificate. Material mix-ups — where a lower-grade fitting is incorrectly identified as a higher-grade alloy — are a real occurrence in the pipe fitting supply chain, and PMI is the safeguard against them reaching a pressure system. The two techniques used for PMI in practice are X-ray fluorescence (XRF) and optical emission spectrometry (OES), and they have very different capabilities and limitations.
XRF: Principles and Capabilities
XRF instruments (handheld "alloy analysers") work by irradiating the sample with X-rays from an internal source (typically an X-ray tube or radioactive source). The X-rays excite electrons in the sample atoms, causing them to emit characteristic fluorescent X-rays at energies specific to each element. The detector measures these energies and intensities to calculate elemental composition. XRF is excellent at detecting: chromium, nickel, molybdenum, manganese, copper, niobium, titanium, vanadium, tungsten, cobalt, and most other metallic alloying elements present at concentrations above approximately 0.05–0.1 wt%. A modern handheld XRF instrument can reliably distinguish P22 (2.25Cr-1Mo) from P91 (9Cr-1Mo-V-Nb), distinguish 316L from 304L (by the molybdenum content), and identify duplex 2205 from super duplex 2507 (by Mo and Ni differences). The critical limitation of XRF is that it cannot detect carbon, sulphur, phosphorus, or nitrogen — elements that are too light for practical X-ray fluorescence analysis in field conditions. This means XRF cannot distinguish carbon steel from stainless steel (both are iron-based, and the difference is primarily carbon and chromium — XRF will detect the chromium correctly but the carbon cannot be confirmed).
OES: The Laboratory Standard
Optical emission spectrometry (OES) — also called spark emission spectrometry — vaporises a small spot of the sample surface using an electric spark and analyses the light emitted by the excited atoms. Each element emits light at characteristic wavelengths, which are separated by a diffraction grating and detected by photomultipliers or CCD arrays. OES detects all elements including carbon, sulphur, and phosphorus, which XRF cannot measure. OES is therefore the definitive method for distinguishing: WPB (carbon steel, C ~0.30%) from P11 (alloy steel, same carbon level but 1.25Cr-0.5Mo); or 304L from 316L when molybdenum readings are ambiguous (OES gives more precise Mo measurement); or P91 from P92 (P92 has tungsten ~1.5% and slightly different Cr/Mo — OES distinguishes these more reliably). OES requires a flat, clean surface for the spark, produces a small burn mark (~3 mm), and is typically a bench instrument rather than truly handheld — though portable OES units exist for field use. Measurement uncertainty for carbon by OES is approximately ±0.01% absolute, which is sufficient to confirm the difference between carbon steel (0.25–0.30% C) and alloy steel (0.10–0.15% C).
The Carbon Blind Spot in XRF
The inability of XRF to detect carbon has practical consequences for pipe fitting PMI. The distinction between carbon steel (WPB) and low-alloy steel (P11) cannot be made by XRF if the chromium content is similar — P11 has 1.25% Cr, which XRF will detect, but if a WPB fitting is incorrectly tagged as P11, the XRF will show "no chromium" and correctly identify it as carbon steel. The problem is the reverse: a stainless steel fitting (316L, ~16% Cr) is easily distinguished from carbon steel by XRF. But a mix-up between two CrMo grades with similar Cr and Mo but different carbon levels — for example, a grade misidentified in the foundry — cannot be detected by XRF alone. For P91 specifically, where the carbon content window (0.08–0.12%) is critical for microstructure stability, OES is required for complete grade verification. This is why NACE RP0582 and many owner specifications require OES (or combustion carbon analysis) in addition to XRF for P91 and P92 fittings.
Specifying PMI in the Purchase Order
PMI requirements must be specified in the purchase order — they are not included in standard ASTM A234 or EN 10253-2 supply unless explicitly required. The specification should state: which fittings are subject to PMI (typically 100% of alloy steel and stainless/nickel alloy fittings; carbon steel WPB is often excluded or sampled); the PMI method (XRF is standard; OES additionally required for P91/P92); the elements to be reported and acceptance criteria (comparison to the chemistry ranges in the material specification); and the point at which PMI is performed (pre-shipment at the supplier's works is standard; field PMI on receipt provides a second check). The PMI report should record: instrument type and serial number; calibration date; fitting identification (heat number, piece mark); elements measured; measured values; and pass/fail verdict. Supplier PMI reports should be retained as part of the material documentation package — they are evidence that the correct alloy was shipped.
Surface Preparation and Measurement Accuracy
Both XRF and OES require the measurement surface to be clean metal — paint, scale, heavy oxide, and weld spatter must be removed before measurement. For XRF, a light grind or wire brush to bright metal is usually sufficient. For OES, the surface must be flat and smooth — a 120-grit grind is standard. Measuring through paint or heavy scale will produce erroneous results. Thin coatings (electroless nickel, surface treatments) can also confuse XRF readings — the instrument averages the composition over a shallow depth (0.1–1 mm depending on element and energy), so a nickel-plated carbon steel fitting would show high nickel by XRF despite being carbon steel beneath. For painted or coated fittings, the coating must be removed at the measurement point before PMI is valid.