Alloy identification, which technique to use in each case

Alloy identification means determining which elements a metal is made of and in what proportion, with enough precision to decide.

Between X-ray fluorescence, spark spectrometry and ICP there are three orders of magnitude of difference in sensitivity, and one difference that is not of degree but of nature: two of the three cannot see carbon. The industrial engineering laboratory settles that choice before measuring, because once the sample is dissolved there is no going back.

The opening question is not what instrument you have to hand, it is what answer you need and what you are willing to sacrifice of the part. It is the first step of any chemical analysis of metals.

What alloy identification means and what each case asks for

Alloy identification admits two different answers, and confusing them is what leads to picking the wrong technique. One is knowing which family and which grade it is, the other is knowing the exact concentration of one specific element.

What result you need, the family or the exact grade

The first answer is classification: whether this part is an austenitic stainless or a duplex, a 6000 series aluminium or a 7000, a brass or a bronze. For that it is enough to see the major alloying elements and compare them with the family table.

The second answer is certification: whether this steel is a 304 or a 304L, whether this aluminium meets the magnesium range of its grade, whether the impurity you are worried about sits below the limit. There the result has to carry its uncertainty and the method has to work in the concentration range you are deciding on. A 304 and a 304L are indistinguishable by their metallic elements, because the difference is the maximum carbon (0.08% against 0.03% by mass).

The distinction between alloy families and their uses is developed in analysis of alloys and their industrial utility, and the route from the analytical result to the equivalent standardised grade, in material composition analysis to reproduce a part.

Before choosing the technique you have to write down the number that is missing. If what you need is to separate a 316 from a 304, any instrument that sees molybdenum will do; if you need the maximum carbon of a stainless, two of the three techniques are out from the start.

What the part itself imposes, size, shape and acceptable damage

The part narrows the list of candidates before any analytical criterion, and when you have to identify a material with no technical documentation it is the only thing that narrows it. Three properties decide: whether it tolerates having material removed, whether it has a flat area to rest an instrument on, and whether what you are measuring sits inside or at the surface.

Acceptable damage is the harshest filter. A unique part, one that is assembled or one that has to go back only tolerates non-destructive measurement; a part you can sacrifice a piece of opens the other two routes. Size rules next: spark needs a flat area of 10 to 20 millimetres and ICP between 0.1 and 1 gram for the digestion.

The third constraint is where the data sits. If the part carries a coating, zinc plating, paint, oxide or a decarburised skin, any surface measurement reports on the layer and not on the base material. Preparation is not a formality, and alloy analysis on competitor parts for performance and cost optimisation always starts from a surface prepared down to bare metal.

Which elements each technique sees and which it does not, starting with carbon

Carbon separates the three techniques better than any other variable. Spark spectrometry measures it, X-ray fluorescence cannot see it at all and ICP does not determine it as an alloying element.

Carbon, the element that decides which technique works

X-ray fluorescence detects elements from magnesium upwards and is blind to carbon, nitrogen, oxygen and boron, because the energy of their characteristic radiation is too low for the detector. It is not a calibration limitation, it is the physics of the method.

In steels that rules the technique out for anything that depends on carbon, which is nearly everything mechanical. A C22 and a C45 have the same metallic elements and differ in carbon (0.17 to 0.24% against 0.42 to 0.50% by mass), which is what changes the hardness reachable on quenching and the weldability.

Spark spectrometry does measure carbon, and it is the reference by which steel grades are certified in production. When the part cannot take the spark burn there is a third route, combustion with infrared detection, which determines carbon and sulphur on a chip of a few hundred milligrams.

Light elements, trace levels and minor alloying elements

Below carbon there is a second group of elements that marks differences between techniques. The light ones (lithium, beryllium, boron) fall outside the reach of X-ray fluorescence, and magnesium and aluminium sit at the edge of its range, where accuracy drops: in a 5000 or 6000 series aluminium alloy, where magnesium defines the grade.

Trace levels are ICP territory. ICP-OES reaches the milligram per kilogram level and ICP-MS goes down to micrograms per kilogram, that is parts per billion, three orders below. That margin is what allows you to show a residual element sits under a limit.

The concentration level of the deciding element orders the choice better than the name of the technique. Above 0.1% by mass spectrometry rules; below 0.01% ICP rules, and between the two there is an overlap band where it is worth measuring by both routes.

Achievable precision, destructive character and sample preparation

The three techniques line up on a scale of trade-off: the more precise and more sensitive the method, the more material it consumes and the more preparation it demands.

What precision each technique gives and what it is enough for

Handheld X-ray fluorescence gives the alloying elements with a relative error of a few percent and detection limits of the order of tens of parts per million for transition metals in an iron matrix. That confirms a grade and separates alloys from one another, and it does not certify compliance with a narrow range.

Spark spectrometry is the one that gives the precision useful for certifying. Its carbon repeatability in low ranges sits of the order of ±0.005 to ±0.01% by mass, which is the resolution needed to tell a 304 from a 304L. The instrument works with calibration programmes by base matrix (iron, aluminium, copper, nickel), so you have to know which family the material belongs to before measuring it. If the family is the unknown, a classification measurement comes first and the quantification follows with the right programme.

ICP is the most sensitive and the one that needs the most preparation. Its precision on minor and trace elements is beyond the other two, and it is not the most convenient route for quantifying the major element, because the dilution needed to get into the instrument’s range degrades precision on that 70% or 80% of matrix.

What you do to the part, from non-destructive to full dissolution

X-ray fluorescence consumes no sample and leaves no mark, and it can be applied to the part in place wherever there is access. It is the first option when the object is unique or is the evidence in a dispute.

Spark spectrometry is slightly destructive. It leaves a burn mark 5 to 10 millimetres across and removes a surface layer, so the part comes back marked at the measuring point.

ICP is fully destructive. The sample is dissolved in acids and disappears. That irreversibility forces you to think about order: first the non-destructive, then the slightly destructive, and dissolution last and only on material you can afford to lose. Metallographic testing fits into that order before dissolution, because the prepared section shows the microstructure and settles where to measure.

Sample preparation, the step that decides reliability

Preparation is where most results are lost, and it is proportional to the sensitivity of the method. For X-ray fluorescence a clean, flat surface of 3 to 8 millimetres is enough, because its analysis depth runs from a few micrometres for light elements to tens of micrometres for heavy ones: a 10 micrometre zinc plating is enough to skew the reading completely.

Spark demands reaching the base metal by machining or grinding, with a flat area and conductive material, and repeating the measurement at three different points to rule out local heterogeneity. ICP demands weighing the sample, digesting it with the right acid mixture, preparing standards matched to the matrix and running a method blank. Refractory elements (niobium, tantalum, titanium, zirconium, silicon) need hydrofluoric acid or prior fusion, and if they do not fully dissolve the result comes out low with nothing to flag it.

Technique	What it measures and what it misses	Order of precision	Damage and preparation	Reference or scope
X-ray fluorescence	Magnesium to uranium; blind to carbon, nitrogen, oxygen and boron	A few percent relative; detection in tens of ppm	Non-destructive; clean flat surface of 3 to 8 mm	Scope: classification and grade confirmation by metallic elements
Spark spectrometry	Alloying elements including carbon, plus sulphur and phosphorus	Carbon with repeatability of ±0.005 to ±0.01% by mass	Burn mark of 5 to 10 mm; machined flat area of 10 to 20 mm	Scope: grade quantification with calibration by base matrix
ICP-OES	Metals and metalloids in solution; does not give carbon	Minor and trace elements at the mg/kg level	Destructive; dissolves 0.1 to 1 g	Scope: minor and trace elements with matrix-matched standards
ICP-MS	The same with far greater sensitivity; no carbon either	Traces at µg/kg, parts per billion (ppb)	Destructive; same digestion and blanks	Scope: impurities below the ICP-OES range
Combustion with infrared detection	Carbon and sulphur only	Carbon in the same order as spark	Destructive; a chip of a few hundred milligrams	Scope: carbon and sulphur when the part cannot take a spark

Where each technique fails

Each technique has cases where it gives a credible and wrong number, which is worse than giving none.

Where the handheld measurement fails without showing it

X-ray fluorescence fails by three routes, and all three return a normal looking result. The first is the coating or surface layer: zinc, chrome, paint, oxide or decarburisation, and what it measures is the layer. The second is geometry, because the instrument integrates an area of several millimetres and needs a flat rest.

The third is the matrix effect. The signal of one element depends on the other elements present, and a calibration built for steels applied to a nickel alloy or a brass returns shifted percentages with nothing flagged by the instrument. With a handheld XRF gun for elemental analysis the rule for deciding is short: if the element your case turns on is light, sits below 0.1% or sits under a layer, the handheld measurement does not close it.

Where spark and ICP fail

Spark fails when the part does not meet its physical conditions, and that happens more often than it seems: parts too small or too thin to rest and to survive the discharge, curved surfaces, non-conductive material, and matrices the laboratory has no calibration programme for. There is also a silent failure mode, measuring on an area affected by heat treatment or by welding, where the local composition does not represent the material.

ICP fails through preparation. An incomplete digestion leaves part of the element undissolved and the result comes out low; a contaminated blank (the acid, the vessel material, the water) artificially raises a trace; and a badly chosen dilution takes the element out of the calibration range. The method blank and matrix-matched standards are not optional.

All three techniques can give a reasonable and false number. The difference is the failure mode: the handheld fails on what sits above the surface, spark on where you rest it and ICP on what did not dissolve. A cross-checked result between two techniques removes all three.

Measuring the same point with two techniques of different principle turns a number into data, and it is the procedure of testing of metallic materials and alloys when the material goes into a formal record or a qualification.

Choose by the figure you need, not by the instrument on hand

The choice between X-ray fluorescence, spark spectrometry and ICP resolves with three questions in this order. First, does carbon or a light element decide? If so, spark or combustion with infrared, and the handheld is out. Second, can you sacrifice material? If not, only the non-destructive route remains and you have to accept its scope. Third, is the element that matters above or below 0.01%? Above, spectrometry; below, dissolution and ICP.

---

---

Frequently asked questions