Metal AM powder is qualified before a build by stacking five measurement families: particle size distribution (typically 20 to 63 micrometres for laser powder bed fusion), particle shape and morphology, apparent density and flow, bulk/true density, and chemistry plus crystalline phase [S3][S4]. PSD narrows, sphericity, and oxygen/nitrogen control predictably separate a build-ready lot from one that will throw defects.
For laser powder bed fusion (LPBF) and electron beam powder bed fusion (EBM) the typical PSD window sits in the 20 to 80 micrometre range, with sub-20 micrometre fines removed because they ignite, agglomerate, and disrupt recoater flow [S3]. Binder jetting uses a coarser 20 to 80 micrometre window too but tolerates more shape variation, while directed energy deposition with blown powder accepts a wider 45 to 150 micrometre band. Metal injection moulding is a different process entirely and uses 1 to 10 micrometre powder [S3].
Particle size distribution: laser diffraction versus dynamic image analysis
Laser diffraction is the workhorse for PSD in AM powder QA, reporting D10, D50, and D90 from a representative dry or wet dispersion; this aligns with ASTM B822 and the ISO/ASTM 52907 powder feedstock guidance, and gives the cumulative volume distribution that printer profiles are tuned against [S5][S7].
Dynamic image analysis is the necessary complement because it captures shape at the same time: a CAMSIZER X2 with a 0.8 micrometre to 8 millimetre range and over 300 frames per second can flag satellites, fused particles, and elongated grains that laser diffraction averages out [S3]. Image-based shape descriptors (aspect ratio, circularity, convexity) correlate with packing density and recoater drag; powders that are too irregular leave streak defects in LPBF. The standard sieve method is too coarse for the AM window, but sieving is still used in AFS number calculations where ASTM mesh stacks overlap the relevant size classes [S3].
Flow, apparent density, and Hausner ratio
Flow behaviour is not a single number; it is the integrated outcome of PSD, shape, surface texture, and inter-particle friction, and it is checked with a calibrated funnel (Hall flow per ASTM B213), a Carney funnel, and increasingly a FT4 powder rheometer that measures basic flowability energy, specific energy, and compressibility under consolidation stresses that simulate the recoater [S5].
Apparent density (ASTM B212) and tap density (ASTM B527) feed the Hausner ratio (tap/apparent); a ratio above roughly 1.25 signals poor packing and is a common release-rejection criterion in aerospace powder specifications. The FT4 also reports a stability index and a variable flow rate index, both used in process qualification to detect lot-to-lot drift even when Hall flow time still passes [S5].
Chemistry: alloy composition, interstitials, and surface oxides
Bulk alloy chemistry is verified with X-ray fluorescence for elements from sodium to uranium, plus ICP-OES or ICP-MS for trace and interstitial content, and LECO combustion analysis for carbon, sulfur, oxygen, and nitrogen [S5]. These interstitials are the silent killers: titanium alloy powder (Ti-6Al-4V Grade 5 ELI) is typically released with oxygen below 0.13 wt% (worse in recycled stock), while 316L stainless is held to nitrogen under 0.10 wt% and oxygen often under 0.05 wt% in aerospace grades.
Surface oxides are particularly important for reactive alloys. X-ray photoelectron spectroscopy quantifies the oxide film thickness on each particle surface, and is the technique the NIST round robin used to show how a single DMLS build cycle measurably changes the surface chemistry of recycled stainless and cobalt-chrome powder [S1][S2]. This kind of recycling drift, together with picked-up argon or nitrogen entrapment in EBM, is why powder suppliers specify a maximum number of reuse cycles (commonly 5 to 30 depending on alloy) and require re-sieving and blending on each cycle.
Phase and microstructure: XRD, pycnometry, and SEM
X-ray diffraction identifies the phases actually present in the powder (for example alpha versus beta phase balance in Ti-6Al-4V, or retained austenite versus martensite in 17-4 PH) and is the only routine technique that flags a wrong crystal structure before the build, where a non-equilibrium phase can drive distortion or cracking during cooling [S1][S5].
True density comes from helium pycnometry (ASTM B555 or B923), apparent density from a Carney or Hall cup, and bulk density from a Scott volumeter, and the three together bound the packing fraction the printer will actually achieve [S1]. Scanning electron microscopy plus energy-dispersive X-ray analysis (SEM-EDX) is used at qualification time and on failure investigations to inspect single-particle morphology, satellites, hollow particles, and surface contamination; X-ray computed tomography on powder beds adds 3D packing geometry that the other techniques cannot see [S1][S2]. For a broader look at how metal powder is graded and traded across industrial buyers, the feedstock standards and sieving practices overlap heavily with those used in additive manufacturing material qualification.
Recycling, lot-to-lot drift, and process linkages
Recycled powder in LPBF changes measurably after each cycle: PSD broadens as fines stick to spatter particles, oxygen and nitrogen pick up, satellites grow, and Hausner ratio drifts upward [S1][S4]. The NIST DMLS study on stainless steel and cobalt-chrome documented the chemistry and morphology shifts using laser diffraction, X-ray CT, optical and SEM, XRD, EDX, and XPS, and showed that recycled lots can no longer be treated as equivalent to virgin material [S1][S2].
Production-grade programs therefore tie every powder lot to a release certificate covering PSD, flow, apparent/tap density, full alloy chemistry including interstitials, moisture per Karl Fischer, and visual inspection under stereomicroscope, with re-test cadence on a defined fraction of the recycled stock, and re-blending with virgin powder at a controlled ratio, typically 50/50 in critical aerospace work [S4][S6]. ISO/ASTM 52907 is the central powder-feedstock standard cited by testing labs, while ISO/ASTM 52900 covers the process taxonomy (PBF-LB/M, PBF-EB/M, DED, binder jetting) that decides which subset of the test stack applies [S7].
Comparison of characterization techniques for AM powder
Five test families dominate the field, each with a different role: laser diffraction PSD for the volume distribution in the 20 to 80 micrometre window, dynamic image analysis for shape and satellites down to about 0.8 micrometre, Hall flow plus FT4 rheometry for the integrated flow response, ICP-OES plus LECO plus XRF for alloy and interstitials, and XRD plus SEM-EDX plus XPS for phase, morphology, and surface chemistry [S3][S4][S5].
On the four criteria that drive release decisions (cost per test, sensitivity to surface oxides, sensitivity to particle shape, ability to detect a bad lot before a build), laser diffraction is cheap and PSD-strong but blind to shape; dynamic image analysis catches shape defects that laser diffraction misses; FT4 rheometry catches lots that pass Hall flow but fail under recoater shear; and XPS plus LECO catch the surface oxide and interstitial pickup that no imaging test will ever see [S3][S5]. A specification that omits any one of these will leak defects; a specification that uses all five with documented limits is what aerospace Tier-1 buyers and most medical device programs require [S1][S6][S7]. Related production-side controls, including traceability of every lot into a manufacturing execution system, are detailed in coverage of MES for additive lines, which complements the powder QA stack described here.
Limitations, failure modes, and what the tests still miss
No single test predicts print behaviour: even a full PSD, flow, chemistry, and phase release can pass while a still-acceptable satellite population drags surface finish out of spec, or while a moisture spike above roughly 0.05 wt% on titanium alloy powder causes porosity that none of the dry-powder tests see [S1][S4].
Common failure modes the standard test stack can miss include argon entrapment in EBM powder (only visible in CT or build-section metallography), static-charged fines bridging in the recoater, and a recycled lot whose D50 has not shifted but whose D90 has crept upward by 5 to 8 micrometres, enough to change layer thickness on a 30 micrometre build [S1][S4]. A program-grade spec adds Karl Fischer moisture, particle counting by image analysis on a minimum number of particles, and a coupon build per lot to catch what the powder tests cannot. The background on metal material properties links back to the same composition and phase considerations that drive the QA limits.
Watch on every inbound CofA: D10, D50, D90 versus the printer profile (any drift over 3 to 5 micrometres is a flag), oxygen and nitrogen trend across the last five lots, and FT4 basic flowability energy as a leading indicator ahead of Hall flow time. These three numbers together catch most of the recycling-related drift modes the NIST round-robin work documented, and they are the cheapest signals to track on a continuous basis [S1][S5][S6].