EFFECTS OF HYDROGEN ON WELD MICROSTRUCTURE MECHANICAL PROPERTIES OF THE HIGH STRENGTH STEELS S 690Q and S 1100QL

The effects of hydrogen on the mechanical properties on weld microstructures of the high-strength steels S690Q and S1100QL were investigated by tensile testing of hydrogen-charged notched and unnotched specimens. It turned out that the mechanical influences of the constructional notch partly cover hydrogen related effects and thus, hydrogen degradation of the material properties is better detectable by using plane instead of notched specimens. In contrast to strength values, ductility parameters have been identified to be immensely significant for describing the impact of hydrogen on a particular microstructure. In particular, the true fracture strain reflects the effects of hydrogen very pronouncedly for the examined weld microstructures of high-strength structural steels. The slope of this characteristic value has been approximated by exponential functions of the hydrogen concentration, in order to provide a hydrogen dependent crack criterion for specific weld microstructures which can particularly be used for numeric simulations of hydrogen assisted cold cracking and for life time determinations of welded structural steel components.

I. INTRODUCTION

Due to economic and constructional reasons, light weight design of modern steel structures is marked by wall thickness reduction and thus, high-strength low-alloy structural steels with yield strengths up to 1100 MPa are increasingly applied in ship building, offshore industry and especially in crane fabrication [1].

However, recent spectacular failure cases showed that there is not always paid sufficient attention to the risk of hydrogen assisted cold cracking in welded joints of these materials during manufacturing and service operation of such components [2]. On one hand this has to attributed to the widely-held opinion that the difficulties with cold cracking during and after welding of structural steels have been solved. This is actually the case regarding structural steels of lower strength than 690 MPa. But, it is usually not considered that the respective standards have been elaborated based on the research results from steels of significant lower strength usually ranging at around 355 MPa. Within the design of modern steel constructions the fact that hydrogen uptake cannot always be avoided during fabrication welding vanishes increasingly into the background. As another reason for such increasing occurrence of cold cracking within the recent two years the technological test procedures used for the avoidance of hydrogen assisted cold cracking often do not reflect the real situation during the welding process. Primary objective of such tests is the specification of limits for the preheating temperatures, carbon equivalents, heat inputs and diffusible hydrogen concentrations, for instance. It can only be emphasized that non-satisfactory transferability of such laboratory test procedures is particularly accounted for by non-quantified restraint intensities of the real weld in a structural steel component [3, 4].

In order to gain more comprehensive understanding of the cold cracking risk of high strength structural steel welds at such an extremely high variation of the components, a more widespread viewpoint of hydrogen assisted cold cracking is required which should be based on the influencing local parameters microstructure, hydrogen concentration and mechanical load. As a first step towards this direction it is necessary to clarify the principal effects of hydrogen on the mechanical properties of the different weld microstructures of high-strength structural steels. For those reasons, tensile tests with pre-hydrogen charged notched and unnotched specimens have been performed. In order to cover the total strength range of modern structural steels, the hydrogen related mechanical properties of the steels S 690Q and S 1100QL have been investigated and compared to the results for supermartensitic stainless steels [5] and for an offshore steel of the category S 355 [6]. A special objective of the present investigations was to identify a simple parameter, in order to describe the impact of hydrogen on a specific microstructure dependent on the mechanical load which can be transferred to every type of weld at various components. For a first verification of such a parameter, the Implant-, the TEKKEN- and the CTS-Test have been considered.

II. EXPERIMENTAL

In order to the determine hydrogen-characteristic strength and strain values of the different weld microstructures of typical welded joints of the above mentioned materials, tensile specimens were manufactured from the base material (BM) as well as from the weld metal (WM) of real welded joints. The chemical compositions of the base and filler materials as well as the mechanical-technological properties of the different hydrogen-free microstructures determined at plane tensile specimens are summarized in Table 1. The weld metals needed for the tensile specimens were produced by the multi-run metal active gas welding according to EN 1597 with the welding parameters specified in Table 1.

In order to copy the notch effects arising at welded joints, stress concentration factors ranging between 2.91 and 3.45 were determined in cold cracking-critical weld regions of the TEKKEN- and CTS-specimens and were transferred to notched tensile specimens (Table 1).

As a specific disadvantage of the more technological TEKKEN Test and the CTS-Test the notch depth cannot be determined accurately. At least, the notch radius ranging between

0.07 mm and 0.12 mm in the TEKKEN Tests and the CTS tests, is similar to that of theImplant specimen having a value of 0.1 mm. In order to provide tensile specimens with weld microstructures of the coarse grained heat affected zone (CHAZ) and of the fine-grained heat affected zone (FHAZ), respective weld simulated microstructures have been produced. On the basis of the recorded temperature time processes within the respective range of the typical HAZ of real welds by means of thermocouples (Figure 1 and 2), parent metal specimens (11 mm�11mm�150 mm) were subjected to defined temperature cycles by means of conductive heating [7]. The respective material properties of the hydrogen-free microstructures are listed in Table 2.

Table 2: Mechnical properties of hydrogen-free S 690Q and S 1100 QL weld microstructures

As controlled by metallografic investigations, a homogeneous microstructure (Figure 3) has been realized over a length of about 25 mm in a SmithWeld simulator. Thus, the weld simulated microstructure covered the total gage length L0 =15 mm of the specimens.

The tensile specimens were cathodically charged with hydrogen at a current density of 10 mA/cm² in a 0.05 m sulphuric acid and 0.1 m sodium arsenite solution as a promotor for hydrogen uptake. By respective variation of the charging duration in a series of preceeding experiments it was confirmed that the tensile specimens were charged thoroughly in diameter.

All tensile specimens were examined by x-ray to be free of cracks and pores before charging and testing, all weld metal blanks were examined by x-ray. All tensile tests were performed according to DIN 50125 specimens with a gage diameter of 3 mm. The tensile tests were accomplished way-controlled at a crosshead velocity of 1mm/min and with automatic data registration. After rupture, one half of a sample was stored in liquid nitrogen again for subsequent determination of the hydrogen concentration. The other half was used for fracture analysis. In order to keep hydrogen effusion as small as possible, the whole testing time was limited to below 5 minutes.

used. By such less time consuming procedure, hydrogen has been extracted at 850 °C and thus, always the total hydrogen concentration including trapped residual portions which also might contribute to the material degradation has been determined.

III. RESULTS AND DISCUSSION

The fracture surfaces of the presaturated tensile specimens exhibited typical hydrogen related topographies. With increasing hydrogen concentration, the fracture mode changed from ductile dimples to brittle transgranular cleavage-like cracking.

Figure 4 shows a SEM micrograph of a clevage-like fracture surface of the weld simulated S 690Q heat affected zone charged with hydrogen. The topography in Figure 5 exhibits a typical fisheye on the fracture surface of a respective weld metal specimen.

Figure 4: SEM micrograph of cleavage-like cracking of Figure 5: SEM micrograph of a fisheye on a Union

III.1. Hydrogen dependent material properties of S 690 weld microstructures

The hydrogen dependent technical strength values, i. e. tensile strength and 0.2 % proof stress, of the four different S 690 weld microstructures have been assigned to Figure 6 dependent on hydrogen concentration. By comparison of the diagrams it becomes obvious that the base material can dissolve hydrogen at much lower concentrations of up to 3 Nml/100g than the weld metal and also than the two heat affected zone microstructures which can take up hydrogen concentrations of up to 6 ml/100g.

This behaviour differs significantly from that of supermartensitic stainless steels, for instance, which exhibited the lowest hydrogen solubility in the weld metal and in the HAZ, respectively. However, supermartensitic stainless steels have martensitic microstructure in the weld metal, the HAZ and the base material, and the different hydrogen solubility in the weld microstructure is caused by the different amounts of austenite and carbide precipitates. In contrast, as confirmed by respective light microscopy, the HAZ of the steel S 690 microstructures as well as the weld metal consisted of much more bainite and martensite caused by rapid cooling. It has to be anticipated that such bainitic and martensitic microstructures provide higher hydrogen solution capabilities than the base material by a larger number of dislocations and other lattice defects acting as hydrogen traps.

Similar to previous experiments with the steel S 355, all S 690 microstructures showed a small increase of the 0.2 % proof stress Rp0.2 at increasing hydrogen concentrations. Such effects have to be attributed to hydrogen dislocation interactions [8, 9] which are causing hardening effects at lower concentrations and a respective decrease in strength at higher concentrations. Similarly, for the as delivered material and the weld metal, but not for the weld simulated HAZ microstructures, a small increase of the tensile strength was registered. Again, such effects might be explained by hydrogen dislocation interactions.

It is doubtful, however, if the decrease in strength at high hydrogen concentrations represents a real softening with respect to ductility. As can be drawn from Figure 6, the tensile strength decreases to the level of the 0.2 % proof stress above a specific hydrogen concentration. This means that the ratio between the tensile strength and the 0.2 % proof stress diminishes to 1, and thus, such microstructures will have nearly no ductility if the hydrogen concentration increases to above a specific limit of about 2 to 3 ml/100g. The safety-relevance of this fact for practical design and fabrication welding of welded high strength steel structures can only be emphasized.

These results already highlight that hydrogen actually affects less the strength, but significantly reduces the ductility of high-strength structural steels and their weld microstructures. This effect becomes most obvious, if the effect of hydrogen on the technical stress and strain properties of a steel is considered in the way as it is schematically illustrated in Figure 7. As shown by such illustration, a hydrogen related reduction of the strength ratio to 1 will reduce the strength of a material only by 20 %, but will decrease the technical strain of a material by about 80 %.

Figure 7: Effect of hydrogen on the technical stress-strain-behavior of high strength structural steels

With the view to define a parameter to describe the hydrogen degradation of a specific microstructure most appropriately, the present contribution predominantly focuses on hydrogen dependent strain values. It turned out that true strain values, in particular the true fracture strain, represents such a parameter and, in addition, might represent a suitable crack criterion in terms of the limiting strain, a specific microstructure might withstand at a

Figure 8: True fracture strain depending on the hydrogen concentration (S 690Q) a) As-delivered

b) Weld simulated fine-grained HAZ

c) Weld simulated coarse-grained HAZ

d)Weld metal Union NiMoCr

specific hydrogen concentration. Vice versa, the critical hydrogen concentration a weld microstructure might tolerate at a specific local strain can be specified.

For those reasons, the true fracture strain has been determined from tensile testing of smooth specimens by using the formula

A₀

(eq. 1) = e with A0: initial cross section of the specimen ^Af

and Af: cross section of the specimen after fracture

which has to be applied, if necking still occurs which is the case for specimens which are not totally embrittled by hydrogen and fractured at plane stress. The true fracture strain dependent on hydrogen concentration has thus been assigned to Figure 8 for all weld microstructures of the steel S 690. As can be drawn from those diagrams, already at relatively small hydrogen concentrations of approximately 1Nml/100g a significant reduction the true fracture strain can be registered. The scatter of the fracture strain is quite lower than that of the strength values (Figure 6) and can thus be described by a lower and an upper envelope curve. By respective regression analyses it turned out that such envelope curves can most conveniently be described by exponential equations which are summarized in Table 3 for the four different microstructures of the S 690 welds. In addition, such equations represent a very suitable crack criterion. If the hydrogen concentration is known, the limiting strains can be calculated by these formulas and cracking occurs, if these values are exceeded. In this respect, the upper envelope curve represents the best case while the lower curve allows a worst case assessment.

The hydrogen dependent values of the notch tensile strength determined by respective testing of the notched samples have been assigned as smoothing curves to Figure 9 for the

Figure 9: Notch tensile strength depending on the hydrogen concentration (S 690Q) a) As-delivered

without notch notch 0,10 notch 0,15 notch 0,25	depth of the notch

without notch notch 0,10 notch 0,15 notch 0,25	depth of the notch

without notch notch 0,25 depth of the notch	notch 0,10 notch 0,15

notch 0,25	without notch notch 0,10 notch 0,15 depth of the notch

.

t

tensile srengthrespnotchtensile srengt

]MPa[h

t

b)Weld simulated fine-grained HAZ

tensil

.pserhtgnertse

] MPa[

notch tensile strength

c)Weld simulated coarse-grained HAZ

.

tensile strength resp

] MPa[

notch tensile strength

d) Weld metal Union NiMoCr

.

tensile strength resp

] MPa[

notch tensile strength

different weld microstructures of the steel S 690. For comparison, these diagrams also include the respective curves of the strength properties determined with smooth samples. As to be expected, the notch tensile strength increases with the sharpness of the notch at all four hydrogen-free weld microstructures, due to strain hardening. The weld simulated microstructures of the coarse-grained and fine grained HAZ exhibit a higher tension level than the parent metal and the weld metal which corresponds to the higher hardness and is consistent to the technical strength values shown in Figure 6. The effects of hydrogen are pronounced by a linear decline on the smoothened notch tensile strength curves. In contrast, increasing sharpness of the notches causes a parallel shift of this strength parameter towards higher values, at least for the as-delivered material, the weld metal and in the weld simulated fine grained HAZ. However, for the weld simulated coarse grained HAZ the gradient of the notch tensile strength curves with hydrogen significantly varies with the notch sharpness.

This particular diagram at least gives some evidence to the fact that the effects of hydrogen on the mechanical properties of a specific microstructure might be covered by mechanical notch effects, as for instance strain hardening ahead of the notch tip. In this respect, the use of notched and, in particular, pre-cracked specimens instead of smooth samples to identify and to quantify the effects of hydrogen on the mechanical properties of a weld microstructure adequately remains in question.

III.2. Hydrogen dependent material properties of S 1100 weld microstructures

In order to provide an insight into the hydrogen dependent mechanical properties of weld microstructures over the total high strength range, welds of the steel S 1100 have been investigated in addition to the lower strength steel S 690. Generally, the S 1100 base material dissolves more hydrogen than the as-delivered steel S 690. As one reason, a probably higher dislocation density in the higher strength material might be addressed. Additionally, it has to be considered that the higher amounts of Nickel and Carbon in the steel S 1100 cause a more bainitic to martensitic microstructure and might also provide easier conditions for the formation of retained austenite. However, the microstructures of the weld simulated HAZ and of the weld metal exhibited a similar hydrogen solubility to those of the steel S 690Q. Similar to the steel S 690, a slight increase of the 0.2 % proof stress appears at slightly higher hydrogen concentrations, indicating a hardening characteristic. Again, the typical reduction of mechanical-technological strength values can be observed with increasing hydrogen concentration in all four weld microstructures of the steel S 1100 (Figure 10).

Figure 10: Tensile strength and 0.2 % proof stress depending on the hydrogen concentration (S 11100) a) As-delivered

b)Weld simulated fine-grained HAZ

c)Weld simulated coarse-grained HAZ

d) Weld metal Union X96

In this context it has to be mentioned that the strength level of the weld metal is slightly lower than that of the base material exhibiting a minimum yield strength of 960 MPa, because a commercial matching filler material for the steel S 1100 was not available.

Figure 11: Critical hydrogen concentration for which the tensile strength ranges at the level of the 0.2 % proof stress dependent on the strength classification level of the structural steels S 355, S 690 and S 1100

As can be drawn from a comparison of Figure 10 with Figure 6, the specific hydrogen concentration at which the tensile strength coincides with the 0.2 % proof stress ranges at substantially lower values for all four weld microstructures of the steel S 1100 than for those of the steel S 690. The hydrogen concentration above which the tensile strength ranges at the same level as the 0.2 % proof stress represents a significant value for the effects of hydrogen in limiting the ductility of a material nearly to zero and has thus been assigned to the diagram presented in Figure 11 dependent on the strength classification level of structural steels. While such a limit has not been identified in previous investigations of the as delivered steel S 355, it could well be identified for the high strength base materials S 690 and S 1100. However, for the different weld microstructures such a limiting hydrogen concentration exists and decreases to very low values, if the strength level of the base material is increased. For the heat affected zone, the limiting hydrogen concentration ranges at a value of 2 ml/100g for the materials S 690 and the S 1100 , while the HAZ of the steel S 355 still maintains ductility up to hydrogen concentrations of about 8 ml/100g. Facing the fact that small amounts of hydrogen below 3 ml/100g can only be measured with high uncertainty, the safety limiting aspect of the above behaviour of the high strength HAZ with respect to practical welding can only be emphasized. Such low limits for hydrogen decreasing the ductility of high strength structural steel welds definitely mean that it has carefully to be avoided during welding that any hydrogen is introduced.

The values of the true fracture strain have been assigned to Figure 12 for all four S 1100QL microstructures dependent on the dissolved hydrogen concentration. The curves exhibit a sharp decrease of the ductility in these microstructures already with relatively small hydrogen

Figure 12: True fracture strain depending on the hydrogen concentration (S 1100QL) a) As-delivered

b)Weld simulated fine-grained HAZ

c)Weld simulated coarse-grained HAZ

d) Weld metal Union X96

concentrations. The most significant reduction of the true fracture strain was observed for the undermatching weld metal. This indicates that the strength level of a material might be less important for its sensitivity to hydrogen cracking as compared to the cooling or quenching process after welding.

The envelope curves of the hydrogen dependent true fracture strain of these microstructures have again been analysed and the respective exponential equations of first and second order are listed in Table 4.

Table 4: Exponential functions of the envelope curves for the true fracture strain as a function of the hydrogen concentration S 1100QL weld microstructures
	upper curve (best case)	lower curve (worst case)
as-delivered	0564 . 0 e1545 .1 2659 . 1 HD w +�=e -	0329.0 e3390.0 e8062.0 8324. 0 HD 7967. 0 HD w +�+�=e --
weld simulated fine-grained HAZ	0714. 0 e996. 0 058 .1 HD w +�=e -	023.0 e585.0 e467.0 873. 0 HD 8697.0 HD w +�+�=e --
weld simulated coarse-grained HAZ	1147. 0 e9135. 0 312 .1 HD w +�=e -	0134,0 e2456. 0 e7464.0 147.1 HD 15. 1 HD w +�+�=e --
weld metal Union NiMoCr	13. 0 e593 .0 3307. 0 HD w +�=e -	639. 3 HD 2296.0 HD w e093.0 e614.0 --�+�=e

The hydrogen dependent notch tensile strength of the S 1100QL weld microstructures has been assigned to the diagrams in Figure 13. Similar to the results obtained for the S 690Q weld microstructures, the notch tensile strength increases with the notch sharpness, due to strain hardening effects at the notch tip. In contrast to the steel S 690Q the strength level is nearly equal for the hydrogen-free base material and the weld simulated HAZ microstructures. Only the weld metal exhibits a 22 percent lower notch tensile strength, due to undermatching effects. With increasing hydrogen concentration, all S 1100QL weld microstructures show a linear decrease of the notch tensile strength. In contrast to the diagrams for the steel S 690Q, a parallel shift of these lines with increasing notch sharpness could only be observed for the weld simulated fine grained heat affected zone (Figure 13b). For the other S 1100QL weld microstructures the decrease of the notch tensile strength becomes more pronounced at sharper notches. This becomes most obvious for specimens with 0.15 and 0.20 mm deep notches exhibiting a decrease of the notch tensile strength below the level of the specimens with a 0.10 mm notch at higher hydrogen concentrations. However, it can also be observed that the decrease of the notch tensile strength for the specimens with a 0.10 mm deep notch is less pronounced than the decrease of the tensile strength of the smooth specimens. Interactions of hydrogen with an increased dislocation activity at the notch tip have to be anticipated as the reason for this behaviour.

Figure 13: Notch tensile strength depending on the hydrogen concentration a) As-delivered

without notch notch 0,10 notch 0,15 notch 0,25 depth of the notch

without notch notch 0,10 notch 0,15 notch 0,25	depth of the notch

without notch notch 0,15 notch 0,25 depth of the notch	notch 0,10

without notch notch 0,15 notch 0,25 depth of the notch	notch 0,10

.

t

tensile srengthrespnotchtensile srengt

]MPa[h

t

b)Weld simulated fine-grained HAZ

tensil

.pserhtgnertse

] MPa[

notch tensile strength

c)Weld simulated coarse-grained HAZ

.

tensile strength resp

] MPa[

notch tensile strength

d) Weld metal Union X96

.

t

tensile srengthrespnotchtensile srength

]MPa[

t

However, again the notch effects on the mechanical properties obviously cover the hydrogen related degradation of these microstructures.

From all above investigated hydrogen dependent mechanical properties it can be derived that the S 1100QL weld microstructures are more sensitive to hydrogen degradation than those of the steel S690Q. Such behaviour can most clearly and conveniently be evaluated by ductility related parameters, in particular by the true fracture strain.

IV. CONCLUSIONS

From the present investigations of the hydrogen dependent mechanical properties of high strength structural steels the following conclusions can be drawn:

1. As shown by the application of notched tensile specimens, notch related effects on the mechanical properties of a specific weld microstructure cannot clearly be separated from the influence by hydrogen. Additionally, hydrogen interacts with the strain field ahead of the notch tip, as for instance by trapping and enhancing effects alongside dislocations. For those reasons the application of notched or pre-cracked specimens for the evaluation of hydrogen related effects on a specific microstructure remains in question. Furthermore, the meaning of investigations with such specimens with respect to a transfer to real welded components appears low, if the notch or crack geometry cannot be determined.
2. Based on the present investigations and in view of a sufficient transferability to real welded components a mechanical parameter seems to be more practical to describe the effects of hydrogen on a specific weld microstructure. As a suitable parameter the true fracture strain has been identified which most significantly reflects hydrogen degradation in the investigated high strength structural steel welds.
3. The true fracture strain provides the capabilities for quantitative descriptions of hydrogen assisted cracking at real welded components by time strain fracture diagrams which have already successfully been used for the evaluation of the hydrogen assisted stress corrosion cracking risk at welded supermartensitic stainless steel pipelines [10]. In such diagrams the increasing local strains caused by welding and cooling can be plotted versus time and can be compared to the critical strain related to the hydrogen concentration introduced via the welding process. This is exemplified by the schematic illustration in Figure 14. The crack initiation time can exactly be identified for the point at which the local strain exceeds the critical strain for a specific microstructure.

[5]	Th. Boellinghaus, H. Hoffmeister, L. Reuter: Material Properties of As Delivered and Quenched
	Modified Martensitic Stainless Steels Dependent on Hydrogen Concentration,
	Proc. of the 1st Intern. Conf. on Supermartensitic Stainless Steels 99, Belgian Welding Institute,
	Bruxelles, 1999, 264-271
[6]	E. Ruyter: Development and Assessment of Welding Procedures
	for Avoiding Weld Joint Cracking in Highly Restrained Offshore Steel Structures,
	Dissertation, Hamburg, 1993
[7]	Thomich, W.: Kombinierte Analysenapparatur zur Bestimmung des diffundierbaren und
	gesamten Wasserstoffs, Stahl und Eisen, 1983, Nr.10, S.497-500
[8]	Oriani, R.A.:Berichte der Bunsen-Gesellschaft, 76, 1972
[9]	Nelson, H.G.:Hydrogen Embrittlement, Treatise on Materials Science and Technology,
	Vol.25, Briant, C.L., Banerji, S.K., Academic Press, New York, 1983, pp. 275-359
[10]	H. Hoffmeister, Th. Boellinghaus: Time-Strain-Fracture Diagrams for the Evaluation of
	Hydrogen Assisted Stress Corrosion Cracking in Supermartensitic Stainless Steels,
	Proc. EuroCorr 1999, p. 275
[11]	Th. Boellinghaus, H. Hoffmeister: Numerical Model for Hydrogen Assisted Cracking,
	Corrosion, Vol. 56, 2000, No. 6, pp. 611-622