Metal sorting - on-site methods for chemical analysis of steels

TWI Bulletin, May 1986

Sheila Stevens, LRSC, is a Senior Research Chemist in the Materials Department.

To achieve maximum economic and technical advantage from chemical analysis, the reasons for analysis, the sampling possibilities and the analytical techniques available must be fully understood. This article examines on-site analysis, with a description of the methods available, including laboratory methods for instances where only small amounts of materials may be removed for subsequent analysis.

From the moment of manufacture a steel product undergoes numerous processing and handling operations on a number of different sites, before fabrication into a finished product. It is unfortunate that, even with the best quality control systems, this multiplicity of handling provides opportunities for mislabelling and mix-up of materials. Even when a batch of material has been shown by analysis to be within specification, there is the possibility of variation in composition between particular pieces of material. To confirm or eliminate the occurrence of a mix-up, or to confirm on-site compliance with a material specification in a critical application, a range of test methods has been developed. These vary from the simple inexpensive spot test, using chemicals, to expensive portable instrumental methods which provide the rapid analyses sometimes necessary for quality control.

The range of applications of site analysis is equally wide, from simple metal sorting to full chemical analysis where a certificate of composition is required. On-site analysis techniques are particularly useful for carrying out a co-ordinated inspection from the fabricator's stock yards to onsite assembly and welding. A further application of on-site analysis is in preparation for repair welding, either during fabrication or as a result of a failure during operation. In these circumstances the chemical composition of the parent material must be ascertained, to make it possible to select the most appropriate welding process and consumables for the repair.

It is sometimes only possible to remove small amounts of material for analysis in the laboratory, and in such instances it has been necessary to develop new techniques of sample preparation to enable accurate instrumental analysis to be performed with only a limited sample size. The sample requirements are dictated by the method of analysis, but on-site sampling limitations do not always enable the optimum sample to be obtained.

On-site methods

Spot testing

Chemical methods for spot testing ^[1] may consist of tests with specific acids to show whether or not attack occurs; for instance nitric acid can be used to distinguish between carbon steels and stainless steels as the former produce brown fumes of nitrogen dioxide, while the latter do not react. Alternatively, reagents may be used to show the presence or absence of individual alloying elements, such as molybdenum (to differentiate between 316 and other 18-8 grades of stainless steel). Tests normally involve taking the steel into solution, usually with a small amount of acid, and testing the resulting solution with specific reagents, or transferring the solution to a filter paper which may be impregnated with an appropriate chemical.

An alternative dissolution technique is electrographic sampling. ^[2] A filter paper impregnated with a suitable solution to act as the electrolyte is placed on the steel, which is made the anode of an electrical circuit. The wet area of the filter is connected to the cathode and a current is applied, producing cations from the electrolytic dissolution of the anode, which are transferred to the filter paper for subsequent analysis.

Spark testing

Some metals, when finely divided, oxidise rapidly on heating, e.g. iron. In spark testing ^[2,3] a high speed grinding wheel is used to abrade small particles of steel; these oxidise as they are thrown off as a series of sparks, which can easily be followed against a dark background. The test is mainly used as an indication of carbon content as this element has a characteristic effect on the appearance of the sparks. An experienced tester may be able to distinguish differences of 0.05% carbon in the range 0-0.4%, and differences of over 0.1% above this. ^[3] If alloying elements are present, they can produce characteristic effects on the sparks produced by a low carbon steel and it is said to be possible to recognise silicon above 0.5%, manganese above 0.1%, nickel above 1.0% and chromium at 0%, 1.5% and 3.0%. ^[3] However, the sparks obtained on the sample being tested should always be compared with those produced by standard materials.

Physical testing methods

These are based on physical properties and, as these can be affected by sample composition, sample form, metallurgical history and sample condition, they are only capable of indicating chemical composition indirectly.

Magnetic

It is possible, using a permanent magnet, to distinguish between ferromagnetic materials such as iron, mild and low alloy steels, and non-magnetic austenitic steels. However, some caution is necessary since the presence of delta ferrite causes some of the more highly alloyed austenitic steels to be slightly magnetic, while high nickel alloys (with more than 35% nickel) can also be ferromagnetic. ^[1] Moreover, it is possible for steels of differing compositions to have the same magnetic properties, while the reverse is also true, as steels from the same heat may have varying magnetic properties because of factors such as different cooling rates after rolling or variations in the degree of cold working. ^[2,4] Thus the approach is at best limited, and is most commonly applied only to confirm that a particular steel is austenitic rather than ferritic or martensitic.

Electromagnetic

Use of metal sorters based on electromagnetic properties is, for the reasons mentioned above, limited to comparison of a reference steel with that of an unknown sample to see whether they are similar or different. Using a bridge circuit, samples are placed in two matched coils, and their differential hysteresis loops are compared.

Electric - eddy current

Eddy current testing is based on the principle that conductivity is affected by chemical composition. ^[2] Devices available incorporate an AC coil to generate a magnetic field which induces eddy currents in conductive materials, and a meter to measure changes in the magnetic field caused by these currents. In simple instruments, the voltage across the coil is used as a measure of the eddy current effect, while in others the changes in electrical conductivity are measured.

Thermoelectric

Thermoelectric methods utilise the 'Seebeck effect', ^[2] namely that when two dissimilar metals are in contact, a heated junction between the two produces a potential difference, the magnitude of which is dependent on the junction temperature and the chemical composition of the materials in contact. ^[3] Measuring devices consist of two probes, one of which is heated, connected by a voltmeter. By varying the probe materials, a wide range of compositions may be sorted.

Spectrometric methods

Optical emission

The principle of optical emission spectrometry (OES) is that atoms in a solid sample are excited by the absorption of energy from an electrical discharge and emit characteristic electromagnetic radiation in the form of light. The light is dispersed by the spectrometer into component wavelengths. Each element gives rise to lines which occur at definite positions in the spectrum and can be used for identification purposes and, when their intensities are measured, for determination of the concentration of a particular element. The technique has been refined and used to produce advanced computer controlled direct reading optical emission spectrometers which are used in the laboratory. However, hand held spectroscopes allowing direct vision of the spectrum, and portable spectrometers incorporating some of the features of the laboratory instruments are available for on-site testing as follows.

1. Direct-vision spectroscopes: these devices incorporate an eyepiece for viewing the spectrum resulting from an electrical discharge between the surface of the testpiece and a suitable counter electrode. The spectrum ( Fig.1) is compared with spectra obtained from standard materials. Elements which can be detected, and typical manufacturer's detection limits, are given in Table 1. It is claimed that a trained operator can use the spectroscope semi-quantitatively by comparing the intensity of a line with that of an adjacent element line or with a known standard.
   
   
2. Portable spectrometers: in these instruments, the electrical discharge is produced using a hand-held pistol containing the counter electrode. The sample surface should ideally be prepared by removing scale, although a DC arc can burn through thin scale. The light emitted by the sample is transmitted along a flexible quartz light guide, normally about 10m long, to the spectrometer, where it is separated into component wavelengths and the intensity of each is measured. Data processing is by a microprocessor or computer.

The sensitivity and spectral resolution are not as good as for the laboratory instruments, and the accuracy is reduced because the normal mode of operation for metal sorting is unlikely to take into account inter-element effects arising from overlapping spectral lines. Samples should therefore be analysed on calibrations produced using standards of a similar composition. This should not be a problem for 'mix-up' analysis or alloy sorting where there is knowledge of the expected composition, but may cause difficulties with unknown samples or where no similar standards exist ( e.g. for weld metal analysis).

The elements which may be analysed, and typical detection limits, are given in Table 1. Determination of carbon has been investigated by several manufacturers, and it may be feasible to measure carbon to an accuracy of ± 0.05%.

Table 1 Typical detection limits for OES methods

Isotope analysis

Portable X-ray fluorescence (XRF) instruments are available which incorporate a radioactive isotope source mounted in a hand-held test probe which is touched to the surface of the sample (surface preparation is often not necessary). The radiation causes the atoms of some of the elements in the sample to fluoresce and emit X-rays of a particular energy, which are then separated into discrete energy regions whose intensities are measured. The intensities are converted into relevant concentrations by a microprocessor. The elements which can be determined depend on the particular source employed; for example, Cd ¹⁰⁹ is suitable for Fe, Mn, Ni, Cr, Mo, Cu, Nb, Co, and W, while Fe ⁵⁵ is usually used for lighter elements such as S, P, Si, V, Ti, and Ca.

A feature of XRF analysis is that accuracy can often be improved by increasing the analysis time. The quantitative analysis modes of the isotope analysers incorporate matrix corrections, which may require previous identification of alloy type, but which further improve accuracy.

Small sample laboratory methods

Instrumental carbon and sulphur analysis

Using a commercially available carbon/sulphur analyser it is possible to obtain accurate analyses on as little as 0.1g of material. Tests on one such instrument indicate that an accuracy of ±0.01% is achievable for carbon levels up to at least 0.3%. ^[5] The sample is usually combusted in oxygen to produce carbon monoxide, sulphur dioxide, and water, which pass through a moisture trap and into an infrared cell for sulphur determination, and then through a catalyst which converts carbon monoxide to carbon dioxide, and sulphur dioxide to sulphur trioxide. The latter gas is trapped and the carbon dioxide passes through a second infrared cell where the carbon is measured.

Dilution with pure iron

In this method a measured amount of the sample is mixed with a known amount of pure iron, or of a certified standard material, and remelted in an argon arc remelter to provide a button for either XRF ( Fig.2) or OES ( Fig.3) analysis. If the former technique is used, a separate carbon determination is necessary. The accuracy decreases with increasing dilution, but the technique has been successfully demonstrated for stainless steels using as little as 2g of sample. These tests are described in Ref ^[6] .

Mounting technique

This method is described in full in ref. ^[7] and involves XRF analysis after mounting the sample in a resin containing a convenient amount of a metallic element: a copper containing resin for example can be used to analyse a steel sample. The principle of the technique isthat the presence of a sample causes a decrease in intensity of the metallic element by comparison with the intensity for the metallic element in the resin alone. A factor can be derived, based on the intensities obtained, whichdepends on the surface area of the sample, and this can be used to convert the intensities of the various elements in the sample to the values which would have been obtained from a full size sample. These corrected intensities can beconverted into concentrations using normal calibration graphs. Separate carbon, and also sulphur and phosphorus, determinations are necessary.

Rubbing technique

This technique also utilises XRF analysis and is almost completely non-destructive. It involves removing a sample from the test material by rubbing with an abrasive paper and presenting this to the instrument. The effect is that ofanalysing a large area of a thin sample, rather than a small area of a thicker sample, as in the mounting technique. The use of silicon carbide paper precludes use of the technique for silicon, but satisfactory results have beenobtained on low alloy ^[5] and stainless steels ^[7] for Mn, Ni, Cr, Mo, V, Cu, Nb, Ti, Co and W. Again, separate carbon, sulphur and phosphorus determinations are necessary.

Discussion

A variety of methods is available for on-site metal sorting and identification. In general, instrumental methods give results that cannot be achieved by field tests necessitating only simple equipment, but the former aresubstantially more expensive and it may be difficult to justify the investment involved. On-site analytical services are offered by subcontractors, however. Different techniques for site analysis may find particular use for specificapplications but may be inappropriate for others. The major characteristics of the methods described here and their advantages and disadvantages are summarised in Table 2.

Table 2 Advantages, disadvantages and applicability of the methods available for site analysis

In recent years, chemical spot tests and spark testing have become less widely used as they have been replaced by modern methods which are less subjective and often capable of providing more information. Nonetheless, they should not be overlooked because they have the advantage of simplicity. This is true also for techniques based on physical properties, although these are normally restricted to routine sorting of batches of samples having the same metallurgical history and, sometimes, the same physical shape. Often, knowledge of the expected composition is desirable.

Analytical techniques which are most commonly used include spectroscopes, portable spectrometers and isotope analysers. The spectroscope, in the hands of a trained operator, is a useful, relatively inexpensive tool capable of alloy sorting and identification. There is also the possibility of obtaining semi-quantitative results. The more expensive portable spectrometers and isotope analysers are becoming more widely employed as they can provide rapid analyses and are easy to use. These instruments are eminently suitable for alloy sorting and identification and are capable of providing quantitative analyses for a large number of elements. They cannot match the accuracy and sensitivity of the more expensive laboratory instruments, however, and are not able, for instance, to distinguish between different grades of a carbon-manganese steel. Their major drawback is the inability to determine carbon accurately. In the welding industry, accurate carbon analysis is particularly important to determine preheat and post weld heat treatment procedures to avoid hydrogen cracking in steel weldments.

If the opportunity exists for removal of small amounts of material, only a few grams, laboratory techniques can provide satisfactory analyses. In this way it is possible to obtain accurate carbon and sulphur analyses on very small samples. The range of elements determined may be greater than for the on-site techniques, but the laboratory methods are more time-consuming.

Summary

The choice of an on-site analysis technique for a particular application must ultimately be based on the quality of results required, but will be influenced also by speed of analysis, cost, and availability of personnel.

Where small amounts of material may be removed, laboratory techniques are available which have the advantage over on-site methods of being able to provide accurate carbon analyses. Such laboratory techniques are available at The Welding Institute and Members interested in using them are invited to contact Graham Carter.

References