Spark emission spectrometry, know for decades, is widely used for the chemical analysis of metals and alloys. In terms of overall performance for bulk analysis, GD and Spark are very comparable and therefore Spark is more widely used due to lower price and simplest operation mode.
However, GD is the necessary choice when one wants to perform depth profile analysis (which Spark cannot do) together with bulk. GD has also some benefits in some rare cases, due to its specific characteristic low Matrix Effect and linearity allowing easier extrapolations. GD is therefore used for precious metals or nodular cast iron (with Mg).
In Spark, the atomization and excitation steps occur in the same step. The sample surface is locally melted and the sample material vaporized. The electric arcs of a Spark analysis can also be affected by microstructures and inhomogeneities in the material. Finally the emission lines in Spark are subject to strong self-absorption leading to second order calibration curves, the need to select several lines for the same element to cover the required concentration range, and great uncertainties when extrapolating results. Spark is finally very sensitive to surface oxidation (requiring sample polishing just before analysis).
GD separates the atomization and excitation processes. In GD, sample atoms are released through sputtering and further excited in the gas phase away from the sample surface. Therefore strict matrix-matching is not necessary in GD which is an advantage for complex alloys for which no Certified Reference Materials (CRM) are available.
In addition, GD is a low thermal process and offers linear calibration curves over orders of magnitude (generally from DL to 100%), therefore one line per element is the rule of thumb in GD. Finally GD averages signals over the entire sputtered area and is therefore less sensitive to local inhomogeneities.
GD is of interest when thin foils (less than 2 mm) have to be measured. The photo below is an example. The thermal effects in Spark are too strong to measure these types of samples accurately, which cannot be easily polished prior to measurement.
There is not much difference in the overall speed of a complete analysis between Spark and GD: an individual Spark measurement is faster than a GD one (15 s vs 1 mn typically) but, in order to obtain representative information, several spots must be done at different locations requiring a manipulation of the sample between two analyses (and repolishing if the sample is small). In GD, the pre-integration time notably is longer than in Spark but successive measurements can then be done at the same location as the process continuously penetrates the sample and the previously puttered material is continuously taken away.
The two techniques are very different and rarely have to be compared. X-Ray techniques have limitations for light elements, which is not the case for GD.
Growing fields of interest in GD relate to Li batteries or H storage materials; H and Li cannot be measured with X-rays. For common elements, pulsed RF GDOES offers sensitivity superior to EDX but comparable to WDX.
X-ray instruments are sometimes also proposed for depth profile analysis. Depending on the incident energy the X-rays can penetrate more or less deeply in a material, and if the layer sequence is known, information on thickness can be derived. This might work for plated materials if the same element is not present in two layers (the sequence Au/Ni/Cu/glass can be measured where Au/Ni/Cu/Ni/glass will lead to errors). With X-rays, no sputtering is needed but low concentrations in the layers and potential contaminants at interfaces cannot be seen.
Prior to the European project coordinated by HORIBA Scientific (https://cordis.europa.eu/project/id/32202/reporting/de) that lead to the development of a brand new instrumentation type called Plasma Profiling TOFMS, the acronym “GDMS” was only used for a few instruments worldwide dedicated to ultra trace analysis of metals.
These instruments feature a magnetic sector sequential by design (they are therefore not adapted to depth profile analysis of thin layers) and DC sources limit them to conductive materials only. They can achieve outstanding Detection Limits (down to the sub ppb range) and have found niche applications for high purity material assessment.