Progress on hydrogen analysis standardisation

TWI Bulletin, November/December 1990

Norman Jenkins is Head of the Chemical Laboratory with responsibility for providing an analytical service to TWI and its members. The laboratory is widely recognised for its work on welding fume and the measurement of hydrogen in weld metals. Norman is Chairman of BSI and ISO Committees concerned with the development of standards on both these topics.

Norman Jenkins explains the background to International Standard 3690, and to other Standards on hydrogen analysis, and suggests further improvements which might be incorporated in a revised ISO Standard.

The International Standard for measuring diffusible hydrogen in test welds is ISO 3690: 1977. ^[1] This Standard has been a major force in achieving reliable hydrogen data, which in turn has made possible the continued development of procedures for reducing the hydrogen content of welds, so reducing the risk of hydrogen-induced cracking. Since the publication of ISO 3690, there has been much progress in analytical instrumentation and an increased appreciation of the factors affecting weld hydrogen levels which should be controlled and monitored in the production of the test weld - see Figure 1.

Hydrogen determination

The determination of hydrogen in steel is, in principle, a simple procedure because heating the sample is in itself sufficient to separate the hydrogen from its matrix, after which a suitable collection or detection device will enable a rapid and precise measurement of its volume to be made. However, while hydrogen, because of its adverse effect upon the properties of steel, is one of the most important impurities present, steelmakers have no standard method of direct hydrogen content measurement.

This notable omission in the range of analytical techniques is due almost entirely to the fact that hydrogen is highly mobile in steel at room temperature - which makes it difficult to define a suitable sampling procedure. This difficulty may be largely overcome by maintaining the sample, prior to analysis, at a low temperature using solid carbon dioxide or liquid nitrogen. Even so, it is recognised that some losses of hydrogen do occur, and the original hydrogen content of the steel will be underestimated, albeit by a small amount. The only way of obtaining a reproducible sample suitable for hydrogen analysis is to specify rigorously the procedures which shall be used to (a) take a representative sample, (b) store the sample, and (c) prepare the sample for analysis.

When the objective is to determine hydrogen in weld metal, the problems of sampling solid and molten steel are exacerbated by the additional variables introduced by the welding process. It is known that welding parameters have an influence upon the weld hydrogen content, and hence the validity of a hydrogen result is dependent not only upon the accuracy of the analytical method, but also upon the accuracy with which the welding parameters have been controlled, measured and reported.

Welding Standards committees (ISO, CEN BSI) have long recognised that, even with the most rigorously specified and controlled sampling procedure, a weld hydrogen measurement does not produce an accurate statement on the actual hydrogen content of a given real weld. Nevertheless, by standardising sampling and analysis procedures, it has proved possible to correlate test weld hydrogen data with the cracking behaviour of steels, so that welding procedures can be controlled to avoid hydrogen cracking. ^[2] Prevention of hydrogen cracking therefore requires a knowledge of the hydrogen potential of the welding consumable, evaluated by analysing test weld samples. This enables weld hydrogen to be controlled during fabrication by adopting a suitable procedure for drying consumables and, if necessary, by using preheat or a post-weld hydrogen removal heat treatment, based upon the hydrogen content.

Weld hydrogen measurements differ from all other chemical analyses where a representative sample of the material is taken and the total amount of the element in question is measured. The figures reported for hydrogen content depend critically on the conventions of the method adopted and, if these are changed, the hydrogen content changes. Internationally standardised methods of measurement are vital for sensible discussion of hydrogen control.

Approach to analysis

The measurement of hydrogen in weld metal differs from other methods of analysis in two other very important aspects. First, the hydrogen normally measured and reported is diffusible hydrogen - ie that released at room temperature - and is not necessarily the total amount present. The reason for this is the traditional belief that hydrogen cracking is caused by diffusible hydrogen rather than by that proportion of the hydrogen 'trapped' within the weld metal at sites such as inclusions. In principle, an obvious advantage of room temperature measurement is that it simplifies the analysis procedure, ie the physical measurement of the length and diameter of a gas column collected from a test weld bead ( Figure 2) in a precision bore tube over mercury ( Figure 3).

This technique is described as a primary method, because it uses primary units of length; it is not dependent upon secondary calibration methods, and has therefore been adopted in ISO 3690:1977.

The second important difference is that diffusible hydrogen is reported as the concentration at STP of hydrogen in ml per 100g of deposited weld metal. This is based upon an accurate measurement of the deposited weld weight, and these units have been used in the development of weld cracking avoidance procedures and hence are the basis of current consumable classification systems. It may be argued that fused metal hydrogen results would provide a clearer view of consumable hydrogen potential, but it is a difficult and lengthy task to derive a reliable estimate of fused metal weight.

Although hydrogen measurements are reported in ml per 100g of deposited metal, it is pointed out that this represents a very low concentration which may be converted to parts per million (ppm) by multiplying by 0.892. There have been critical comments on the accuracy of hydrogen analyses but, in terms of precision, the primary method compares well with methods for measuring other elements in steel at the ppm level. In general, a reproducibility of ± 10% at the 10 ml/100g level may be expected for the mean of triplicate results, and this is as good as can be obtained for most other methods of 'trace' element analysis.

The main advantage of the mercury collection method - the simplicity and accuracy of the technique - leads to its status as a primary method for measuring difiusible hydrogen in weld metal. However, the method has a major disadvantage in that the time for complete hydrogen removal ( ie the analysis time) is governed by the diffusion rate of hydrogen in the weld metal, and this may be fairly slow. For the size of weld bead required by ISO 3690:1977, analysis times are typically 10 to 14 days, but for some steel compositions may be as much as 40 days. Such times are clearly unacceptable for production quality control, and in recent years industry has investigated use of more rapid methods of analysis based upon heating samples to temperatures as high as 4000, thus achieving analysis times of about 30 minutes - see Figure 4.

As noted above, the primary method of analysis measures room-temperature diffusible hydrogen. Extracting the hydrogen at temperatures such as 400C can also release residual hydrogen, adding a positive bias to the result and indicating that it might be more logical to report total hydrogen concentration rather than what is, in effect, a halfway house.

This is not a new concept, as a suitable analytical technique was developed by Coe and Jenkins as long ago as 1960. ^[3] They proposed using carrier gas to extract the hydrogen at 650C, followed by its measurement using a thermal conductivity detector and integrator. This technique provides a total hydrogen measurement and indeed the principle forms the basis of current instrumental methods.

TWI continues to play a leading role in developing and using analytical procedures for measuring diffusible hydrogen, and a recent Group Sponsored Project has examined the application and implications of extraction at temperatures above normal ambient. While extraction at, say, 400C is certainly feasible, some differences in behaviour have been noted between various consumable types, and current activity includes a new Group Sponsored Project to study the release of residual hydrogen at 400C using a wider range of consumables. ^[4]

Standards

Routine testing has inevitably moved away from the primary method involving collection over mercury - which is too slow. The International Institute of Welding (IIW), Commission II, recognised this trend and in 1986 published a draft Standard ^[5] which used collection over mercury as the primary, or reference, method, but which also admitted more rapid instrumental methods, provided that they were correlated against the primary method.

This draft also incorporated many practical and instructional improvements compared with the existing ISO 3690 Standard, which it was intended to replace. Concurrently with the preparation of the IIW draft of Committee WEE 39/6, the British Standards Institution was revising the British Standard method for diffusible hydrogen found in Appendix C to BS 639:1976 ^[6]

The aim was to develop a British Standard fully compatible with the IIW proposals which would eventually also be compatible with the new ISO Standard. The new British Standard ^[7] was published in 1988 as BS 6693, Parts 3, 4 and 5, pertaining to manual metal arc, submerged-arc and gas-shielded welding respectively. Issuing BS 6693 as a separate Standard gave the measurement of diffusible hydrogen a status independent of the demands of welding-consumable hydrogen classification.

Moreover, revisions of the consumables Standards, as and when they occurred, provided an opportunity for defining specific welding parameters to produce test welds appropriate in size for instrumental analysis, or for analysis by the primary method. A more detailed discussion of the practical aspects of the ISO/IIW and BSI Standard methods and reliability is given by Coe. ^[8]

The IIW draft Standard was published in Welding in the World, ^[5] with the intention that it be submitted to ISO as a full international Standard. It was expected that the various national standardising authorities would, as occasion demanded, revise their individual Standards in line with the IIW draft. Unfortunately, with a few exceptions, this has not occurred - indeed, the American Welding Society has published a Standard wherein the sample size is incompatible with either the primary method as described by IIW, or the majority of instrumental techniques, ^[9] although the AWS procedure is published with a correlation factor which relates its results to those obtained by the IIW method. Obviously, it cannot be mandatory for different countries to follow the IIW Standard, even if it achieves ISO status, but a more widespread acceptance must help to achieve international uniformity of hydrogen analysis.

The current situation is that a draft international Standard, prepared by a recognised standardising authority - the IIW - has been in existence for over four years, but has not yet been adopted by the ISO. The reason for this is unclear but, because of the delay, it is suggested that the draft should be re-edited to take advantage of two years' experience of BS 6693. For example, advice can now be given on handling manual metal arc electrodes and submerged-arc flux during the period between removal from the drying oven and use.

Other additions might cover the measurement and use of relative humidity data during welding, electrode extension during welding with a continuous electrode, and drying shielding gas. Cautionary advice might also be given on the possibility of releasing residual hydrogen when samples are analysed by heating to 400C - which may occur if porosity is present in the weld, or with weld types having a high proportion of residual hydrogen. The question of rounding off the analytical result also requires consideration.

Recent years have seen significant improvements in hydrogen-controlled welding consumables and it is now realistic to aim at hydrogen levels below 3 ml/100g. While the precision of the hydrogen measurement technique is more than adequate at such levels, there are few data on the reproducibility of the results when analysing actual test welds. An evaluation of method performance is required in order to establish the confidence with which lower levels of hydrogen classification may be introduced. As with Bs 6693 Part 3, a statistical evaluation of the method should be an intrinsic part of an analytical Standard.

Conclusions

To achieve uniformity of procedure it is important that the revision of ISO 3690, as prepared by the IIW in the early 1980s, be re-examined to ensure that it reflects state-ofthe-art technology and is appropriate for present-day industrial requirements. On behalf of Commission II of the IIW, TWI is revising the original IIW draft Standard and it is expected that the new draft will be available for submission to ISO some time during 1991.

References