Spark optical emission spectrometry (OES) – method, application and LIMS integration in the metals industry and foundries

What is spark OES – and why does it dominate metals analysis?

In foundries, steelworks, aluminium and non-ferrous metal processing, practically every batch is checked today by spark optical emission spectrometry. The reason is simple: no other analytical technique delivers the chemical composition of a metallic sample as fast, as precisely and as reproducibly as OES.

The key properties that have made the method an industry standard:

• Speed – a complete elemental analysis takes a few seconds, with minimal sample preparation
• Wide elemental range – practically all relevant alloying elements, from lithium to uranium
• High precision – even low concentrations are detected reliably
• Economic efficiency – significantly cheaper to run than wet-chemical methods
• Reproducibility – measured values are comparable across time, shifts and sites

In everyday laboratory work this means: a batch comes out of the melt, a sample is poured or turned, briefly ground, placed into the spectrometer – seconds later the analysis is available. Production does not wait; it can react immediately to the OES values.

The method step by step

OES sounds complex – but at its core, the method is elegant and intuitive. What happens physically between the sample and the detector:

1. 1
   Spark discharge in an argon atmosphere Between the metal sample and a counter-electrode (typically tungsten or silver), a high-energy spark discharge is ignited under an argon shielding gas. Argon prevents atmospheric oxygen from distorting the measurement.
2. 2
   Plasma & light emission The discharge produces a plasma several thousand degrees Celsius hot. Atoms from the sample are excited into the plasma state and emit light as they fall back to the ground state – every element on its characteristic wavelengths.
3. 3
   Spectral decomposition The emitted light is fed through optical fibres into the optical system and there decomposed spectrally by a grating – similar to a prism, but with much higher resolution. Wavelength range: 130 to 800 nanometres (UV + visible light).
4. 4
   Detection & evaluation Photomultipliers or CCD detectors measure the intensity of each relevant wavelength. The instrument software (e.g. Bruker QMatrix, Spectro SPARK ANALYZER) calculates the concentrations of the individual elements from the intensities – calibrated against reference materials.
5. 5
   Evaluation against the specification The result is compared with the specification of the grade to be tested: do all elements lie within tolerance? The answer decides whether the batch is released or reworked.

A variant gaining in importance: ICP-OES (inductively coupled plasma OES). Instead of a spark discharge, the plasma is generated by a high-frequency electromagnetic field in an argon flow. The advantage: even lower detection limits, also suitable for liquids and very fine trace analysis.

Application areas of OES

OES is used wherever metallic materials have to be tested, classified or released. The most important industries:

OES instrument manufacturers at a glance

The market for OES spectrometers is dominated by a handful of manufacturers. Anyone equipping an industrial laboratory typically deals with the following vendors:

In practice, many laboratories operate instruments from different manufacturers in parallel – e.g. a Bruker Q4 Tasman in the main laboratory and a Hitachi FOUNDRY-MASTER mobile on the shop floor. This is exactly where a central requirement of the laboratory’s IT lies: all instruments have to feed their data into a single system – otherwise isolated solutions emerge that no one can evaluate centrally any more.

The underestimated weak spot: what happens after the measurement

The OES measurement itself takes seconds. The spectrometer delivers precise, reproducible values. Yet this is exactly where the strength of many industrial laboratories ends – and the problem begins.

In laboratories without end-to-end data integration, practice often looks like this:

• Measured values are transferred manually from the instrument software into an Excel file (typing error rate: 1–3%)
• The Excel file sits locally on the instrument PC, not accessible centrally
• Shift supervisors collect the Excel files by email at the end of the shift
• In the office, the values are copied yet again into another tool for reporting
• Batch-to-sample linkage is done by hand – if at all
• At audits, values have to be reconstructed from several sources

The result: a high-precision instrument produces reliable data which then loses quality through manual transfer. In the worst case, release decisions are made on the basis of values that were typed in (and potentially incorrect).

This is exactly the point where a Laboratory Information Management System comes in.

LIMS as the data backbone of OES analytics

A modern LIMS such as [FP]-LIMS connects OES spectrometers directly to the central laboratory database – vendor-independent. Concretely, that means:

[FP]-LIMS supports all common OES manufacturers – more than 100 pre-configured instrument interfaces, including Bruker, Spectro Ametek, Hitachi, Thermo Fisher, Elementar and others. For the Bruker QMatrix integration there is a dedicated detailed article.

Practice: OES + LIMS in foundry and steel industry

Two practical examples make tangible what the OES + LIMS combination delivers in everyday industrial work – both with a time horizon that speaks for itself:

In both cases, OES is the measurement backbone, and the LIMS is the organisational one. Only together do they deliver what industrial operations need: reliable values, immediately available, completely documented, without manual transfer effort.

Comparable constellations can be found at AGOSI (precious metal processing, ISO/IEC 17025-accredited since 2012), COMPO EXPERT (fertilisers, around 700 employees) and many other long-standing customers in the metals and chemicals sectors.

Typical pitfalls in OES analytics without a LIMS

From over 30 years of practice with industrial laboratories – the classics where OES workflows keep getting stuck:

• Values are typed in by hand – typing error rate 1–3%. In safety-critical applications (automotive, pressure vessels) an open flank
• Spectrometer databases are isolated solutions – values from the Q4 Tasman in the main laboratory are not readily comparable with values from the Hitachi FOUNDRY-MASTER on the shop floor
• Specifications are maintained manually at the instrument – time is lost when switching grades, errors creep in
• Batch-to-sample linkage missing – at the audit it is no longer traceable which measurement belongs to which batch
• Mill test certificates are assembled by hand – from Excel, PDF and handwritten notes. Risk: complaints over incorrect evidence
• Measurement uncertainty missing in the test report – an audit finding under ISO 17025. More on this in the measurement uncertainty article.
• No trend analysis possible – drift on individual instruments only becomes apparent when complaints arrive, instead of being caught as an early indicator

Frequently asked questions on spark OES