Hydrocarbon Contamination and Diffusible Hydrogen Levels in Shielded Metal Arc Weld Deposits

Patchett and Yarmuch layout:Layout 1 11/10/10 2:45 PM Page 262

Oil contamination of basic H4 and H4R SMAW electrodes removes low-hydrogen characteristics on contact, and conventional baking treatments for humidity and water exposure restore electrodes to approximately the H8 designation

BY B. M. PATCHETT AND M. A. R. YARMUCH

ABSTRACT ence, for example 10 mL/100 g equating to 11 ppm. This difference is often within the typical variability of results in electrode Published literature suggest the avoidance of hydrocarbon contamination (oil or testing. grease) of shielded metal arc welding (SMAW) electrodes to prevent diffusible hy- The International Institute of Welding WELDING RESEARCH drogen cracking in weld zones, but there are no published data on the contamination (IIW) designation system for hydrogen mechanisms of exposure to hydrocarbons on the diffusible hydrogen level. This paper potential of welding consumables are explores hydrocarbon contamination of low-hydrogen basic SMAW electrodes. Con- “very low” for up to 5 mL/100 g; “low” for tact with oil causes instant and gradually increasing contamination and diffusible hy- 5–10 mL/100 g; “medium” for 10–15 drogen levels with time, and the contamination is greater with lower oil viscosity. mL/100 g; and “high” for more than 15 Standard reconditioning heat treatment for water contamination lowers the dif- mL/100 g of weld metal deposited. The fusible hydrogen content to about H8 levels for the electrodes and oils investigated. American Welding Society assesses electrodes via a logarithmic scale for diffusible hydrogen levels in a weld deposit. H16 is Introduction to weld deposits by various welding for 16 mL/100 g of weld metal (17.6 ppm), processes was hampered by imprecision in H8 is for an electrode producing less than The presence of hydrogen in ferritic the measurement of the diffusible hydro- 8 mL/100 g (8.8 ppm), the common upper steels welded by the shielded metal arc gen in the deposited weld metal. This was limit for “low hydrogen,” and H4 is for less welding (SMAW), submerged arc welding traced to the varying solubility of molecu- than 4 mL/100 g or 4.4 ppm. Commercial (SAW), flux cored arc welding (FCAW), lar hydrogen in the liquid media used to consumables that are able to reduce the and other flux-bearing processes has been collect the hydrogen expelled by a weld diffusible content further down the loga- studied for many years. The studies have sample (Ref. 3). Standardized tests based rithmic scale (2 or 1 mL/100 g) are not re- concentrated on humidity and moisture on the use of a liquid without measurable liable at present for arc welding processes effects on absorbed hydrogen levels. Hy- solubility (mercury) or vacuum extraction involving fluxes (SMAW, FCAW, and drogen-assisted cracking (HAC) in hard- produced accurate and reproducible re- SAW). AWS A5.1, Specification for Car- enable steel weld zones is controlled by sults (Refs. 4, 5). In these extraction tests, bon Steel Electrodes for Shielded Metal Arc several methods — electrode flux chem- results are conventionally reported as Welding, was revised in 2004 to reflect this istry and conditioning, procedure control “mL/100 g deposited weld metal,” which is new optional (voluntary) designation sys- (a combination of suitable preheat and a characteristic of the hydrogen collection tem. The specification also permits an op- heat input) in steels of relatively low hard- method, not of atomic hydrogen levels in tional supplemental “R” suffix designator enability, and the addition of postweld the actual weld deposit. This paper uses a for electrode coverings that satisfy ab- heat treatment (PWHT) on steels of high dual reporting system including parts per sorbed moisture limitations. Note that the hardenability (carbon equivalent) (Ref. million (ppm), a more relevant number H16, H8, and H4 designations should not 1). The strength level and microstructure for atomic hydrogen in solid solution. be confused with the H1 (extra-low hy- of low-carbon steels has emerged as an- There is only approximately a 10% differ- drogen ≤ 5.5 ppm or 5 mL/100 g), H2 (low- other criterion (in addition to hardness) hydrogen ≤ 11 ppm or 10 mL/100 g), and governing the susceptibility to HAC (Ref. H3 (hydrogen not controlled) designa- 2). In both cases, the amount of diffusible tions in AWS D1.1 Annex XI for assess- hydrogen imparted to the weld zone is of KEYWORDS ment of hydrogen cracking susceptibility importance in determining suitable proce- via the Pcm method. dural parameters. Initial assessment of the Hydrocarbon Contamination Attempts to connect weld metal dif- amount of diffusible hydrogen imparted Shielded Metal Arc Welding fusible hydrogen to moisture in the flux (SMAW) (Ref. 6) and weight gains during exposure Diffusible and Low-Hydrogen to humidity met with higher variability in B. M. PATCHETT ([email protected]) is H8 Levels professor emeritus of welding engineering at the the results. This is due to the fact that University of Alberta, Canada, and M. A. R. Low Moisture Pickup (LMP) weight gain is partially due to carbon diox- YARMUCH is program leader — welding engi- ide absorption (Ref. 7). Water exists in neering at the Alberta Research Council, Edmon- most low-hydrogen fluxes in the following ton, Alb., Canada.

262-s DECEMBER 2010, VOL. 89 Patchett and Yarmuch layout:Layout 1 11/10/10 2:46 PM Page 263

AB AB

Fig. 1 — Flux structure of standard basic low-hydrogen electrodes. A — Fig. 2 — Flux structure of low moisture pickup basic low-hydrogen elec- Cross section; B — surface. trodes. A — Cross section; B — surface.

5 two forms: water of crystallization in where HD = IIW diffusible hydrogen in type, 4 mm ( ⁄32 in.) in diameter. Both stan- binders or agglomeration stabilizers and mL/100 g; a1 = as-baked coating moisture dard and moisture-resistant (low moisture adsorbed water via hygroscopic compo- %; a2 = adsorbed moisture %; and b = at- pickup (LMP)) types were assessed. All nents in the flux. The former is “perma- mospheric humidity in mm Hg. electrodes were conditioned at 375°C nent,” in the sense that total removal This equation strictly applies only to (707°F) for 1 h before use. Cooled and destabilizes the mechanical integrity of the electrodes tested (manufactured in weighed electrodes were then immersed the flux, while the latter is transitory and Japan) and must be used with circumspec- in a graduated cylinder to 25 mm (1 in.) can be removed without destabilizing the tion if differing flux chemistries from from the top of the flux coating in two sin- flux. This dual behavior limits the temper- other manufacturing sites are used. gle viscosity grade mineral lubricating oils ature of baking to approximately These numerous investigations into of differing viscosity, a 10-W low-viscosity 400°–425°C. Published information on the the effects of humidity, adsorbed water, grade and a 30-W medium-viscosity grade. subject of water content concerns the re- and water of crystallization are not Multigrade oils were avoided to isolate lationship between water uptake and ex- matched by investigations into the effects any effect of viscosity. After various times posure conditions, including steps to fol- of hydrocarbon contamination, although of immersion, excess oil was stripped from 1 low to minimize or reduce the net amount the deleterious effects of oil and grease the flux covering with a 1.5-mm- ( ⁄16-in.-) of deposit diffusible hydrogen. Since the contamination on diffusible hydrogen thick flexible rubber grommet squeegee water from exposure appears to be ad- content are generally assumed. This paper containing a hole slightly smaller in diam- sorbed (surface) rather than absorbed is intended to provide an initial assess- eter than the electrode coating. The slight (bulk) by the flux, part of the water is dis- ment of the effects of hydrocarbon con- pressure ensured a consistent removal of

persed into the atmosphere by resistance tamination on diffusible hydrogen only. surface oil. Immediate weighing deter- WELDING RESEARCH heating of the SMAW electrode during Complex hydrocarbons may contain other mined the weight gain caused by oil im- welding, and adsorbed water adds less dif- elements of interest, such as sulfur, but mersion. Since the “dry” electrode weight fusible hydrogen than does bound water further investigation is necessary to study varied, weight gain was defined as a per- (Ref. 8). The substantial efforts of several other possible contaminants. centage gain for each electrode, rather investigators, over many years, has shown than an absolute weight gain. Welding be- that the diffusible hydrogen content of Experimental Program havior/diffusible hydrogen measurements SMAW deposits is related strongly to followed within 5 min. Procedural condi- water content of the flux up to approxi- The electrodes used in the experimen- tions were 24 V, 190 A on electrode posi- mately 0.3% water, but the relationship tal program were of the E4918 (E7018) tive polarity and a welding speed of 200 becomes more scattered at higher water contents (Refs. 9, 10). Diffusible hydrogen Table 1 — Chemical Composition of Electrode Fluxes is also affected by atmospheric humidity at the point of welding, which is more no- Element Percentage Percentage Percentage Change— ticeable in low-hydrogen electrode depo- Standard Electrode Low Moisture Standard to Low sition than it is in electrodes producing Pickup Electrode Moisture Pickup higher hydrogen levels (Ref. 9). An equation (Ref. 9) is available that relates dif- Silicon 17.8 14.4 –23 fusible hydrogen to atmospheric humidity Potassium 15.5 10.9 –42 and flux water content, which is consid- Calcium 57.0 54.0 –6 ered accurate up to about 0.3% adsorbed Titanium 0.0 12.7 + X and crystalline water: Manganese 3.5 3.8 +9 Iron 5.9 3.7 –55 1⁄ HD = [260a1 + 30a2 + 0.9b – 10] 2 Total >99 >99

WELDING JOURNAL 263-s Patchett and Yarmuch layout:Layout 1 11/10/10 2:46 PM Page 264

Fig. 3 — Weight gain vs. exposure time during electrode immersion. Fig. 4 — Weld deposit diffusible hydrogen content vs. hydrocarbon exposure time.

mm/min, producing a heat input of 1.4 played discrete mineral particles of diam- of moisture (Ref. 9).

WELDING RESEARCH kJ/mm. Bead-on-plate (BOP) welds were eters ranging from about 50 to 300 mi- The results show that the adsorption of produced on ASTM A36 steel pads 10 mm crometers in a fine matrix. The LMP flux oil by the flux coating causes a very rapid 13 ( ⁄32 in.) thick, 75 mm (3 in.) wide, and 300 was similar but had smaller average parti- initial weight gain by the electrode coat- mm (12 in.) long with the SMAW process cle size and many small cracks in the ma- ing. An increase in viscosity produces a to assess electrode usability characteristics trix. A fractured surface of each flux lower weight gain for a given time of im- after oil contamination. Weld metal dif- showed that each contained porosity be- mersion — Fig. 3. Long immersion times fusible hydrogen levels were determined tween the various mineral particles. The of up to 24 h slightly increase weight gain. on IIW specimens under mercury accord- surface features and internal structures of This is probably an indication of satura- ing to ISO 3690. Bakeout after hydrocar- the two fluxes are shown — Figs. 1, 2. The tion or a stable balance between oil pene- bon contamination was identical to the results show that the flux coatings are tration and entrapped air in the surface conditioning procedure, 375°C for 1 h. small-scale analogs of porous rock and subsurface pores in the extruded flux formations. coating. Results and Discussion The chemistry of each flux type was as- Studies of crude oil movements sessed in the SEM at the same time using through porous rock have shown that per- The diffusible hydrogen contents were energy-dispersive X-ray (EDX) analysis meation depends inversely on viscosity determined for each electrode type after (Table 1). (Ref. 13) and that surface-active compo- conditioning to establish a base line for The chemistry of the fluxes is not a nents dominate adsorption, with the comparison with oil contamination levels. complete and detailed analysis (e.g., no higher molecular weight fractions adsorb- The standard electrodes produced an av- determination of oxygen was made), but ing preferentially (Ref. 14). Mineral oils erage diffusible hydrogen level (3 deter- there are significant differences, particu- used for lubrication are usually paraffinic minations) of 5.9 ± 1.1 ppm or 5.4 mL/100 larly the titanium level. This suggests that oils, which have long chain molecular g and the LMP types an average of 4.3 ± the rutile level is increased in the LMP structures. Higher-viscosity oils have 0.7 ppm or 3.9 mL/100 g. flux. The mineral constituents are there- longer molecular chains on average. In the Before assessing the oil adsorption fore somewhat different, as might be ex- interaction between porous rock and oil characteristics, the surfaces of the elec- pected. This strongly suggests that behav- alone, the adsorption behavior is polar trode fluxes were inspected in a scanning ior during hydrocarbon exposure may vary and the more surface-active elements of electron microscope (SEM) to assess the from one manufacturer’s flux to another, the oil are adsorbed (Ref. 12). The mole- particle size and structure of the flux coat- or from country to country, as demon- cules are typically in the C20 to C70 range ings. The standard electrode flux dis- strated previously in assessing the effects (number of carbon atoms in a molecule), and the boiling point is in excess of 370°C. Molecules in the C15 to C50 range (diesel Table 2 — Bakeout Effects on Diffusible Hydrogen Levels fuel) boil at temperatures between 300° and 600°C (Ref. 15), indicating that the Electrode Oil Type Exposure Time Diffusible Hydrogen Diffusible Hydrogen higher average molecular weight mole- from Contact after Baking cules in lubricating oils would have an mL/100 g ppm mL/100 g ppm even wider range of boiling temperature. Therefore, the normal bakeout tempera- Standard 10 W 0.1 min 17.3 19.2 5.2 5.8 ture range for basic low-hydrogen electrodes (350°–400°C) may cause boiling of Low Moisture Pickup 10 W 0.1 min 21.3 23.7 4.8 5.3 the lower molecular weight fractions, but Low Moisture Pickup 10 W 24 h 40.7 45.2 5.7 6.3 not all of the higher molecular weight Low Moisture Pickup 30 W 24 h — 6.9 7.7 fractions. There is an instant increase in dif-

264-s DECEMBER 2010, VOL. 89 Patchett and Yarmuch layout:Layout 1 11/10/10 2:53 PM Page 265

fusible hydrogen levels after contact be- “very low-hydrogen” behavior. In terms of drogen content of welds. Metal Construction tween the oil bath and flux coatings. Short- the logarithmic AWS scale, baking causes 7(10): 508–511. term immersion results show that the dif- a reduction in diffusible hydrogen to ap- 6. Boniszewski, T. 1979. Manual metal arc fusible hydrogen level jumps to about 20 proximately the H8 level. welding — Old process, new developments. Metallurgist & Materials Technologist ppm or 18 mL/100 g after only a few sec- 6. The inability of baking to completely 11(10): 567–572; 11(11):640–643; and 12(12): 697–704. onds of contact with the oil — Fig. 4. reverse oil adsorption is likely due to a 7. Siewert, T. A. 1985. Moisture in welding Long-term immersion increased the combination of molecular fraction boiling filler materials. Welding Journal 64(2): 32–41. weight gain incrementally, and an appar- temperatures and bonding of higher mo- 8. Coe, F. R. 1986. Hydrogen measurement ent saturation level was reached. The dif- lecular weight fractions to pore surfaces in — Current trends versus forgotten facts. Metal fusible hydrogen results reflect this, with the flux during adsorption. Construction 18(1): 20–25. the maximum of about 40–50 ppm (36–45 9. Kotecki, D. J. 1992. Hydrogen reconsid- mL/100 g) appearing for oil exposure time Acknowledgments ered. Welding Journal 71(8): 35–43. up to 24 h. This is similar to diffusible hy- 10. Kiefer, J. H. 1996. Effects of moisture drogen results for Exx10-type cellulosic The authors wish to thank Clark Bick- contamination and welding parameters on diffusible hydrogen. Welding Journal 75(5): 155-s electrodes. nell of the welding laboratory at the Uni- to 161-s. Baking reduces the diffusible hydrogen versity of Alberta for valuable technical as- 11. Hart, P. H. M. 1986. Resistance to hy- in the weld deposits substantially, as sistance and Barry Bilida for assistance in drogen cracking in steel weld metals. Welding shown in Table 2, to the IIW “low-hydro- the execution of the experimental program. Journal 65(1): 14-s to 22-s. gen” (5.5–11 ppm) diffusible hydrogen 12. Buckley, J. S., Liu, Y., and Monsterleet, level but cannot restore full IIW “very-low References S. 1998. Mechanism of wetting alteration by hydrogen” behavior. On the AWS loga- crude oils. SPE Journal 3: 54–61. rithmic scale, the results are at or below, 1. Coe, F. R. 1973. Welding steels without 13. Dullien, F. A. L. 1992. Porous Media — the H8 designation. The diffusible hydro- hydrogen cracking. The Welding Institute. Fluid Transport and Pore Structure, 2nd ed. San 2. Widgery, D. J. 2002. Welding high Diego: Academic Press. gen content after baking for the specific 14. Akhlaq, M. S., Kessel, D., and Dorrow, electrodes assessed is similar for both strength pipelines. 4th International Pipeline Conference. Calgary, Alberta: ASME. W. 1996. Separation and chemical characteriza- electrode coating types, both oil viscosi- 3. Smith, D. C., Rinehart, W. G., and Jo- tion of wetting crude oil compounds. SPE Jour- ties, and is also similar for any contamina- hannes, K. P. 1956. Effect of moisture in the nal 180(2): 309–314. tion time from 0.1 min to 24 h. coatings of low-hydrogen iron-powder elec- 15. Dalbey, W. E., and Biles, R. W. 2002. Re- The postbaking SMAW process behav- trodes. Welding Journal 35(7): 313-s to 322-s. cent respiratory toxicology investigations of ior during manual operation was accept- 4. Smith, D. C. 1959. Development, proper- mineral oils: Post-1990. Symposium on Mineral able for the welder and did not produce ties, and usability of low-hydrogen electrodes. Oils and Metal Processing Oils. Cincinnati, any visual flaws in BOP welds, e.g., pin- Welding Journal 38(9): 377-s to 392-s. Ohio: ACGIH. holes or cracks, which were the expected 5. Evans, G. M., and Baach, H. 1975. Hy- results from the H8 levels of diffusible hydrogen. No formal weldability tests were conducted due to the potentially large size of a comprehensive alloy assessment program.

Conclusions An Important Event

1. Contact between basic low-hydrogen on Its Way? SMAW electrodes and oil causes instant and incrementally increasing contamina- Send information on upcoming events to the Welding Journal Dept., 550 WELDING RESEARCH tion by adsorption, leading to significantly NW LeJeune Rd., Miami, FL 33126. Items can also be sent via FAX to increased diffusible hydrogen levels in de- (305) 443-7404 or by e-mail to [email protected]. posited weld metals. 2. Diffusible hydrogen levels from contaminated electrodes are above about 18 mL/100 g or 20 ppm for any time of contact and for both electrodes and both oil viscosities used in this investigation. 3. Weight gain from contact with hydrocarbons is reduced by higher oil viscos- REPRINTS REPRINTS ity, which shows that larger molecules penetrate less rapidly. Oil thus appears to To order custom reprints be adsorbed by the electrode flux. of 100 or more of articles in 4. The low-moisture-pickup flux, intended to resist adsorption of water, also Welding Journal, reduced the rate at which oil is adsorbed, call FosteReprints at possibly by minimizing flux porosity and (219) 879-8366 or flow passage access. 5. Baking electrodes contaminated by (800) 382-0808. oil at a typical temperature and time rec- Request for quotes can be faxed to (219) 874-2849. ommended for adsorbed water removal You can e-mail FosteReprints at reduces the net diffusible hydrogen levels [email protected]. to the 5–10 ppm range, corresponding to IIW “low-hydrogen” rather than IIW

WELDING JOURNAL 265-s