A technique for determining austenite to ferrite ratios in welded duplex stainless steels

TWI Bulletin, November 1985

Brian Ginn is Group Leader (Experimental) in the Stainless and Corrosion Section of the Materials Department of The Welding Institute.

In duplex ferritic/austenitic stainless steels, it is important to know the proportions of the phases present in welded joints, but identification and measurement can be difficult and laborious. The simple method of colour contrasting the phases described in this article makes possible rapid measurement using an image analyser.

Ferritic austenitic duplex stainless steels are used increasingly in oil and gas production systems to reduce the risk of stress corrosion cracking in chloride containing environments or where exposure to sour gas (hydrogen sulphide) is likely.

The two phase duplex structure of austenite and delta ferrite is determined by the chemical composition of the steel and the thermal cycle it has experienced. Rolled plate is normally solution treated at 1050°C to give an austenite to ferrite ratio of approximately 50:50. This parent structure may be affected, however, by additional heating and cooling cycles, such as in welding, where the rapid cooling rate suppresses austenite precipitation and increases ferrite levels, resulting in a reduction in material properties. Similarly, weld metal properties may decline, necessitating use of a filler containing an austenitiser if high joint toughness and corrosion resistance are required.

Determination of actual austenite or ferrite level therefore becomes of prime importance, and in attempts to measure it a variety of methods has been developed. These are of two types: those relying on magnetic effects, such as the attraction of a magnet to the ferrite phase, or the magnetic permeability of the sample; or those using metallographic techniques such as point counting or surface area measurements. All the methods have limitations either in the range of measurements they cover or in accuracy. Magnetic methods, for instance, may give low readings compared with point counting if ferrite levels are low and grain size is small, and correspondingly high levels if grain size is large and ferrite levels are high. ^[1]

Metallographic analysis using either linear (point counting, etc) or area determinations is generally accepted as reliable, but is dependent on etching and staining techniques to assist in distinguishing between the ferrite and austenite phases. However, point counting is slow and tedious, particularly so if measurements are required from specific areas, such as a coarse heat affected zone region of a single weld pass.

The Welding Research Council (WRC) advisory sub-committee on 'Welding stainless steels' accepted the limitations on currently available techniques for measuring ferrite levels and decided to use determination of the tear off force of a magnet as a standard method. The method is defined in AWS Standard A4 2-74, ^[2] and uses a proprietary instrument, the Magne-Gage, as the primary measuring instrument. A similar method in line with the AWS standard has since been developed by the International Institute of Welding (IIW) and forms the basis of the IIW standard. ^[3] Both AWS and IIW standards recognise that the ferrite measurements obtained are arbitrary values rather than true volume percentages and they define these as ferrite numbers (FN). The Magne-Gage however has a working range of only 0-28FN, well below the ferrite levels normally experienced in duplex stainless steels. A method of extending this working range has been developed and ferrite levels determined by it are defined as extended ferrite numbers (EFN). ^[4] Difficulties may be experienced when attempting to make measurements using a Magne-Gage in narrow regions of a welded sample because the magnet is influenced by surrounding regions which may have higher or lower ferrite levels than the area of interest. ^[5]

Technique development

In view of the limitations of available techniques and because it is important to be able to measure pre-determined regions in a welded joint, a fresh assessment of metallographic techniques has been made. The most popular techniques, point counting and area measurement, both require development of microstructural features and this is usually carried out by chemical etching or staining. For duplex stainless steels, fine microstructural variations may be hard to distinguish visually following etching ( Fig.1 and 2) and are difficult to determine accurately either by an optical point counting technique or use of an image analyser. For speed and convenience of surface area measurement an image analyser has many advantages, but simple laboratory instruments scanning only black and white have difficulty differentiating between the various colour shades produced between ferrite and austenite by conventional etching and staining methods.

Fig.1. Duplex stainless weld metal etched electrolytically in 20% H ₂ SO ₄

Fig.2. Heat affected zone in duplex stainless steel weld etched electrolytically in 20% H ₂ SO ₄

After some experimentation, it was found that when an as-polished duplex stainless sample was coated with a ferro-fluid, in this case an iron colloid solution comprising iron oxide (Fe ₃ O ₄ ) in paraffin, the iron oxide adhered to the magnetic ferrite phase only. It was not necessary to magnetise the sample, its residual magnetism was sufficient to attract the iron oxide. The ferrite phase is coloured and appears in shades of dark blue and brown by this method, while the austenite phase remains white, see front cover. These marked colour differences assist in visual point counting, and when viewed in black and white contrast are sufficient to allow a simple image analyser to differentiate between the two phases ( Fig.3 and 4.).

Fig.3. Duplex stainless weld metal - as polished and coated with iron oxide - white phase is austenite

Fig.4. Heat affected zone in duplex stainless steel weld - as polished and coated with iron oxide

To determine the reliability of the iron powder/image analyser technique, a series of measurements was made on a set of parent plate and weld metal samples with ferrite levels ranging from 2-74%, as previously determined by either point counting or Magne-Gage measurement. Ten measurements each were made on randomly located areas by two operators. Microscope magnification was X500 and area measurements were made using a Metaserv GM 7000 image analyser. This magnification gave the best overall results for this series of specimens, but measurements have also been made successfully at X1000 when resolving fine weld metal structures. The Table shows the range and averages obtained using the image analyser and compares them with measurements made by point counting and using a Magne-Gage. Point counts were made at 500 intersections except for two samples where 2000 were counted.

The results in the Table indicate a good correlation between the two operators using the iron powder/image analyser method. In addition, generally good agreement is evident between image analyser results and those obtained by either point counting or using a Magne-Gage. However, a wide range of ferrite levels is obtained by each operator during the series of ten area measurements, especially at higher ferrite levels. As a result, some differences between point counting and image analyser results are seen. The image analyser data are, however, in line with earlier work by Gunia and Ratz which indicated that, because of ferrite variability in weld metals and cast structures, measurements made metallographically or magnetically may vary by as much as +- 17% at the 50% ferrite level. ^[6] .

Summary

A technique has been developed for shade contrasting the ferrite phase of a duplex stainless steel with the austenitic phase using residual magnetism of the ferrite to deposit iron oxide selectively. Sufficiently good contrast has been obtained on a variety of parent plate and weld metal samples, with ferrite levels ranging from 2-75%, to enable area measurements to be made using an image analyser. The values of percentage ferrite determined using this technique show reasonable agreement with those determined by point counting or the use of a Magne-Gage.

Results obtained with Image analyser compared with measurements made by point counting and Magne-Gage

References