Checking the cleanness of a sample of steel can be a time consuming business. The usual way of going about it is to use metallographic techniques to prepare a thin slice of the material being analysed. Several chemical or mechanical preparation steps can be required. The surface of the sample is then inspected optically or using particle interactions to determine the level of inclusions. A new technique promises to take some of the work out of this quality control process, reducing analysis time to just a couple of minutes. The technique is known as single spark evaluation spectroscopy.
How Does Single Spark Evaluation Work?
Single spark evaluation (SSE) involves the time resolved multispectral detection of individual emission intensities from the spark plasma. In other words, each individual spark is logged by the system to build up a histogram of the intensities of the emissions. This histogram can provide details about the distribution of an element in the sample, so picking out inclusions and measuring, for example, the levels of both soluble and insoluble aluminium in steel samples.
Determining Compositions of Metal Samples
Spark optical emission spectroscopy is normally used in an integrating mode for measuring line emission intensities. This type of analysis provides a value for the concentration of an element within a sample. However, this value is an average for the ‘burning spot’ - the area over which the hundreds of spark discharges have occurred.
Individual sparks only form craters of approximately 10-30µm in diameter, and so the emission line intensities represent a sample amount of only 30-500ng. By registering and evaluating the line intensities from each spark, SSE therefore gives access to the local concentration of elements within sample portions of about 20μm or 100ng in diameter. This makes SSE potentially very useful for measuring the size and distribution of non-homogeneities in a material, and for measuring the chemical constitution of inclusions.
Identifying Inhomogeneities in Metal Samples
The SSE method for detecting inhomogeneities in an element’s distribution relies on a simple approximation. The histogram of individual spark emission intensities from a homogeneously distributed element can be represented satisfactorily as a Gaussian curve. So if any sparks are picked up that do not fit into the Gaussian distribution, these must be the result of inhomogeneities. The relative amounts of the element in the homogeneous phase and inhomogeneous phase can be calculated by the number of sparks under and outside the Gaussian curve. SSE applied to a number of selected elemental emission lines allows the identification of those elements in an inclusion, and calibration graphs can be used to calculate the amounts of each element present.
Steps in a Single Spark Evaluation Process
The essential steps in the SSE process are as follows:
• Register all spark intensities for selected spectral lines
• Select the most suitable section of the sparking sequence to give an accurate representation of the material’s composition• calculate the best Gaussian fit for each spectral line and separate the non-Gaussian parts, known as ‘outliers’
• Calculate the intensity of the homogeneous and non-homogeneous parts of the elements being analysed, and calibrate for each element if reference samples are available
• Calculate the relative numbers and intensities of the `outliers' of a selected element to another element
• Use this data to identify the inclusions by their elemental composition.
Inclusions can then be characterised by calculating the number of atoms within an inclusion, assuming an average spark crater is represented by a hemisphere of 30μm diameter. Inclusion size can be worked out, and the overall analysis used to show the frequency of occurrence of each inclusion.
Determining Aluminium Content in Low Alloyed Steel
SSE has already been successfully applied to the measurement of the soluble and insoluble components of aluminium in low alloyed steel, and for the characterisation of inclusions in low alloyed steel. For determining insoluble aluminium, the first ‘diffuse discharges’ in the sparking series that contain very low iron intensities must be ignored. Once the level of iron has stabilised, the next 500 sparks are relevant for picking out the insoluble aluminium component.
The tests carried out by Spectro Analytical on 40mm diameter steel rods showed that the level of the insoluble aluminium varied considerably through the sample, by as much as 30%. Soluble aluminium levels varied by only a few percent. The technique proved successful, correlating well with the levels of aluminium in the sample and providing an improved analytical facility for distinguishing the soluble and insoluble phases.
Analysis of Inclusions in Low Alloyed Steels
SSE has also been used to analyse inclusions in rolled bars of low alloyed steel, with particular attention paid to certain element combinations that occur frequently. One search looked for any combination of aluminium, calcium, magnesium, oxygen and silicon. Another looked for any combination of calcium, manganese, oxygen, sulphur and zirconium, and a final search looked for combinations of aluminium, titanium and nitrogen.
Using correlation analysis of the ‘outliers’ found for each element, the SSE system was able to pick out the inclusion species and also show the size and frequency of occurrence of each inclusion. Frequency and size both varied considerably, but the system highlighted several relationships between the frequency of occurrence of certain species and the size of the inclusions. Sulphides, oxides and nitrides were all picked up easily, demonstrating that SSE could be used to perform a quick cleanness test in less than a minute on a sample of steel.
The levels of the dominant inclusion species found using SSE qualitatively agree with those determined using standard metallographic techniques. This augurs well - if further studies and agreement between SSE and metallographic techniques are shown, SSE will become a time-saving alternative to existing practices. Quality and process control will benefit greatly from the new technique.
In conclusion, inclusion characterisation is possible using correlation analysis to check which elements are present in relation to each other. These inclusions can be quantified by their size and frequency of occurrence. Non-homogeneities can be separated by SSE reproducibly. Further research into applications of the technique could open up many new possibilities for exploiting SSE in practice.