Proceedings of International Workshop Conference on Hydrcgen Management for Applications, Ottawa. Ontario. Canada: 6-8 Oct. 1998. J.E.M.Braid,C.V. Hyan. D.L.OIson, G.N.Vigilante,eds. CANMET (Canada Centre for Mineral and Energy Technology). Ottawa. Ontario. 105-110.1998

MECHANISTIC UNDERSTANDING OF IN STEEL WELDS

D. Suh and T.W. Eagar Dept. of Materials Science and Engineering Massachusetts Institute of Technology

ABSTRACT

Hydrogen behavior in steel welds is reviewed from the viewpoint of a new absorption model based on both monatomic and diatomic hydrogen. According to the model the monatomic hydrogen is absorbed much more readily than the diatomic hydrogen. Under this model, the major zone of absorption should be around the cooler outer edge, rather than the high- temperature central region, of the weld pool.

INTRODUCTION

The principal sources of hydrogen in welding consumables are from moisture in the or the shielding gases, water in the form of hydrated oxides, hydrocarbons, oil, din. and the like [I-31. Whatever the source, hydrogen is present in the are plasma as both monatomic and diatomic hydrogen. It is absorbed at the interface between the plasma and the weld pool, and is transported by convection in the weld pool [2 J.

It would be desirable to have a reaction model at each step in order tofigure out the behavior of hydrogen during welding. Most researchers who have studied hydrogen in welding have taken a simple model based on Sicven's law assuming that the reaction between diatomic hydrogen in the gas and absorbed hydrogen in the melt is in equilibrium as given by: 112 Ha(@ = H (ppm) in liquid metal

Combining the above equation with the equilibrium constant, Sievert's law states that absorbed hydrogen is proportional to the root of the partial pressure of diatomic hydrogen, and increases exponentially as the reaction temperature increases as shown by

where Ci and C2are constants. Qualitatively. Sieven's lawseems to explain well theeflfoct of diatomic hydrogen partial pressure because the amount of absorbed hydrogen increases monotonically with partial pressure. Many researchers simply fined their data points with Sievenoslaw. A representative example for steel is shown in Fig. 1. D. Suh and T.W. Engar

¥ However, quantitatively, the reaction temperature calculated basedon Sievert's law for steel is unreasonable. The calculated values are much higher than 2500°Cthe maximum temperature obainable in steel as shown by Block-Bolton and Eagar [4]. Even if Terasaki's theoretical analysis is applied to convert measured diffusible hydrogen to that which was initially absorbed, one cannot quantitatively reconcile the Sievert's law model with the experimental data [S]. Researchers who used Sievert's law were unable to explain their results in terms of realistic reaction temperatures [6,7]. Gedeon and &gar [8]proposed a new model that involved both monatomic and diatomic species. This model, which has been explained further by DebRoy for absorption from the plasma, produces more realistic reaction temperatures.

NEW MECHANISM OF HYDROGEN ABSORPTION To understand the model, the behavior of hydrogen at the interface between the plasma and the weld pool must be considered. The temperature of the plasma in arc welding is greater than 8WK,which is sufficient to dissociate diatomic hydrogen into monatomic hydrogen. According to Dinulescu and Mender's study [9],the anode and cathode boundary layer spatially defines an energy exchange free path, X.E. between the plasma and the weld pool surface where ionized monatomic hydrogen will arrive at the cathode without recombination with an electron. Thus, monatomic hydrogen will have a substantially higher temperature than the weld pool surface, which means that the reaction temperature of monatomic hydrogen will be higher than that of diatomic hydrogen. The new model consists of two steps: dissociation of diatomic hydrogen in the bulk plasma and absorption of monaiomic hydrogen at the liquid interface.

The dissociation of hydrogen can be presented with the following two equilibrium reactions

By combining these equations, the absorption of monatomic hydrogen can be expressed in one reaction equation 1101.

Thus one can obtain thauunount of absorbed hydrogen as a function of the partial pressure of monatomic hydrogen for various reaction temperatures as shown in Fig. 3 [8].

In the case of monatomic hydrogen (Fig. 2) the trend of absorption is clearly different from that of diatomic hydrogen @g. 1). Firstly, the amount of absorbed hydrogen decreases with temperature rather than increases, which means that this model not only predicts a very different behavior of hydrogen absorption than Sievert's law but can explain experimental results without exceeding the maximum temperature of the weld pool. This new model shows that the solubility of active gases increases with decreasing temperature in high temperature systems [l I]. Secondly, the gradient of hydrogen absorption increases as the reaction temperature decreases, which means that the major absorption will take place around the outer edge of the weld pool as shown in Fig. 3. This directly contradicts postulates based on Sicvert's law, which leads to the result that the maximum absorption of hydrogen occurs in the high-temperature central region of the weld pool. Mechanistic Understanding of Hydrogen In- WeM*

e In order for the model to be effective, sufficient amounts of hydrogen continuously dissociate and arrive at the weld pool surface. According to the calculation by Gedeon [12], the percent dissociation increases significantly with even minute increases in reaction temperature as shown in Fig. 4. Another important point from Fig. 4 is that most of the diatomic hydrogen dissociates into monatomic hydrogen at just above the maximum temperature of the weld pool. This implies that most of the hydrogen in the cathode boundary layer is monatomic hydrogen providing the dissociation model with a rational basis. According to the dissociation model, the hydrogen absorption during welding is governed by the degree of dissociation and the reaction temperature

that is determined by the temperature distribution at the cathode boundary layer...... ,.

AFTERMATH OF ABSORPTION It is widely accepted that absorbed hydmgen can be uniformly distributed into the weld pool due to vigorous convection. In addition, the cold part of the weld pool is adjacent to the ambient gas phase. not the plasma. Under this circumstance, the behavhdab-. . divided into either,desorption from the cooler part of the weld pool or trapping and diffusion away from the solidification front. Hydrogen in solidified weld metal and the HAZ diffuses away to reach an equilibrium state in terms of chemical potential of the hydrogen before it is trapped in the iron lattice at temperatures below 200° [13]. The lattice diffusion of hydrogen is aided by tensile stress fields that developed at (lie toe and root of the weld, particularly at notches or inclusions and other discontinuities and by thermal contraction of the cooling weld. As a result, the hydrogen solubility around defects in the HAZ is higher, and eventually hydrogen . cracking occurs. The various conditions for cracking have been described by others [1.14].

SUGGESTIONS FORWJTURE RESEARCH

The mechanisms of hydrogen absorption, diffusion and crack nucleation in steel andsteel weldments have been sniffied extensively for half a century. In the area of absorption, the past ten years has produced anew theory based on monatomic hydrogen as the dominant species in the arc, leading to hydmgen in the weld pool. This theory has several predictions diametrically opposed to the previous theory based on Sicvat's law, e.g., with the monatomic absorption theory, cooler weld poQs absorb more hydrogen, whereas in the diatomic theory, hotter weld pools absorb more hydrogen. With such vast diff~~cnoesin these theories, it should be possible to design experiments capable of distinguishing between the theories. Such basic research can be justified since acorrect theory can lead to process changes designed to minimize the amount of hydrogen entering the weld pool, with subsequent reductions in the prevalence of cracking.

There has been lienew insight into hydrogen diffusion in recent years, in part because of the transient nature of the hydrogen - it diffuses out of the sample spontaneously at room temperature - plus the fact that no hydrogen microprobe is available to measure concentration variations over small distances. Finally, near ambient temperature, there are many trapping sites that can cause two to three orders of magnitude variation in the diffusivity.

With regard to cracking, there is hope that some of the new ab initiocalculation methods may produce new insight into how hydrogen promotes cracking in metals. REFERENCES

1. Bailey. N., Cue. F.R., Gooch T.G.. Han. P.H.M.. Jenkins. N. and Pargeter. R.Jà "We- steels withom hydrogen cracking", Abingion Publishing and ASM International, ch. 1.1993. 2. National Materials Advisory Board, "Effective use of weld metal yield strength for HY-iiteek", Report No. NMAB-380. Contract MDA-903-82-C-0434. National Academy Press. 1983. 3. Grong. 0.. "Metallurgical modelling of welding". The Institute of Materials, ch. 2. 1997. 4. Block-Bolten. A. and Eagar. T.W., *'Metal vaporization from weld pools", Met. TmtSB(9). 441-469.1984. 5. TerdiT., Akiyama, T., Hamashima, S. and Kishikawa, K.. "An analysis on specimen size for deteraiinaiion of diffusible hydrogen content in weld metal", Trails. Japan Welding Society 17(1), 93-101.1986. 6. Salter, G.R. and Milner. D.R., ¥'Gaabsorption from an: atmosphere", British Welding Journal, 89-10. Feb. 1960. 7. White, D.R., "Process measurement of hydrogen in welding", Ph.D. dissertation, Champaim, HI.. University of Illinois. 1986. 8. Gedeon. S.A. and Eagar, T.W. ¥fhennocbemicaanalysis of hydrogen absorption in welding". Welding Research Suppkmem, 264-271, July 1990. 9. Dinulescu, H.A. and Pfendcr. E.. "Analysis of die anode boundary layer of high intensity arcs". Journal of Applied Physics 5 l(6). 3 149-3157.1980. 10. Elliot, J.F., Gleiter, M. and Raniakrishna. V.. ¥Thermoche&tr for sittlmiking", VoL Addison-Wesley Publishing Co.. Reading. Massachusetts, 1963. 1 1. Flengas. S.N. and Block-Bolien, A. "Solubilities of reactive cases in molten salts". Advance in Molten Salt Chemistry, eds. EJ. Braun-itein G. Manontov mid G. Smith, Plenum Press. New York, N.Y.. 27-81.1973. 12. Gedmn, S.A., -Hydrogen-assisted cracking of high-strength steel welds". PkD. dissertation, Qnnbridge, Mass., Massachusetts Institute of Technology, 19R7. . . 13. Bailey. N.. Coe. F.R. Gooch T.G.. Han. P.m.. Jenkins, N. and Pargcta. R.J., "Welding steels without hyitrogcn cracking". Abington Publishing and ASM International, ch. 5.1993. 14. Grong, 0..Metallurgical modelling of welding". The Institute of Materials, cb. 7,1997. Mwtenlfc Undçrstendlnof Hydrogen hi BtMl WeMs

0 1, tl:l~~4la#l~ 0 2 4 6 8 10 1C, OIATOMIC HYDROGEN IN ATMOSPHERE

Fig. 1. Equilibrium hydrogen solubiltiy as a function of diatomic hydrogen partial pressure.

I MONATOMIC HYDROGEN IN ATMOSPHERE

Fig.2. ~~uilibriumhydrogen solubility as a function of the partial pressure of monatomic hydrogen gas for various assumed absorption temperature D. Suh and T.W. EMMr

Fig.3. Fraction of hydrogen dissociated as a function of both partial pressure of diatomic hydrogen and reaction temperature of monatomic hydrogen. Curves are 1,0.1,0.01, 0~001,~0.0001,and 0.00001 atrn from righi.

o MOMATOM IC ABSORPTION

is I

Fig. 4. Theoretical hydrogen absorption due to both rnonaiornic and diatomic hydrogen as a function of weld pool location. The calculated points assume a dissociation temperature of 2500¡C0.01 atrn hydrogen added to the shielding gas. and an absorption temperature as given by Krause (Ref. 19) for the surface temperature of the molten weld pool. .