Reconsidering the Basicity of a FCAW Consumable — Part 1: Solidified Slag Composition of a FCAW Consumable As a Basicity Indicator

SUPPLEMENT TO THE WELDING JOURNAL, MARCH 2000 Sponsored by the American Welding Society and the Welding Research Council

Reconsidering the Basicity of a FCAW Consumable — Part 1: Solidified Slag Composition of a FCAW Consumable as a Basicity Indicator

A basicity index for a flux cored electrode was developed, taking into consideration the metal sheath, fill ingredients and weld metal composition

BY E. BAUNÉ, C. BONNET AND S. LIU

ABSTRACT. Based on an investigation expressing the flux/slag basicity. The slag detachability, the metallurgical performed using a set of five experimen- newly defined basicity index is found to properties of the final weldment, etc. tal FCAW electrodes, an improved ver- offer superior correlation with the weld Therefore, the multitude of flux ingredi- sion of the IIW basicity index formula is metal oxygen content, demonstrating ents used in a FCAW electrode, each fea- developed. This new methodology is de- the validity of the assumptions made in turing various functions, make the work scribed in two papers, titled Part 1: So- the present investigation. of formulators rather complex. For a par- lidified Slag Composition of a FCAW ticular FCAW electrode, with proper in- Consumable as a Basicity Indicator and Introduction formation on the chemical composition Part 2: Verification of the Flux/Slag of the deposited weld metal, the compo- Analysis Methodology for Weld Metal A flux cored arc welding (FCAW) elec- sition and nature of the core flux, a sim- Oxygen Control. To accomplish this pur- trode contains multiple powdered ingre- ple compositional relationship can be pose, the partition of the various ele- dients within a metal sheath. Moreover, obtained to describe the distribution of ments contained in the formulation of the variety of the ingredients that can be each metallic element present in the one FCAW electrode is studied and mod- used in FCAW is enormous. For these rea- electrode between the covering slag and eled in Part 1. Correspondingly, the com- sons, the intrinsic nature of welding the weld metal. position of the solidified slag is predicted fluxes is rather complex. Also, due to the In the present paper, for one particu- for this particular electrode. To verify the various chemical reactions involving lar experimental FCAW electrode, based model, the prediction of the slag chemi- these ingredients in the arc environment, on a mass balance considering the metal cal composition is compared with ex- it is not a simple task to understand how sheath, the electrode fill and the weld perimental measurements. Good accor- each constituent contributes to the gen- metal chemical composition, a simple dance is found, which shows the model eral behavior of the flux with regard to approach to predicting the composition is applicable. Also, a new way of defin- the electrode performance, i.e., the metal of the solidified slag is proposed. Hence, ing the basicity of a FCAW consumable transfer stability, the slag viscosity, the a methodology to determine the basicity based on the chemical composition of index of the slag is developed. the slag is derived. In Part 2, comparison of this innovative methodology with the FCAW Using CO2 as Shielding Gas IIW formula is achieved, as well as with other means reported in the literature for KEY WORDS A flux cored electrode is a composite, tubular electrode that consists of a metal FCAW Electrode sheath containing a core of flux (Ref. 1). E. BAUNÉ is R&D Engineer and C. BONNET Flux Cored The flux is made up of a mixture of pow- is Technical Manager with Air Liquide/Centre Basic dered ingredients, both metallic and Technique des Applications du Soudage Basicity Index nonmetallic. The main function of the (CTAS), Pontoise, France. S. LIU is Professor of Slag metallic materials is to alloy the weld Metallurgy, Center for Welding, Joining and Consumable metal, together with the alloying ele- RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT Coatings Research, Department of Metallurgi- ments contained in the metal sheath, in cal and Materials Engineering, Colorado order to increase the strength of the de- School of Mines, Golden, Colo. posited metal and deoxidize it. The non-

WELDING RESEARCH SUPPLEMENT | 57-s be determined as a function of ∆G0 , the

free energy of decomposition of CO2, Ρ CO2 and theΡCO ratio. This reads: 2  ρCO2   −⋅2 ∆G0  aOO==ρ ⋅ exp  22 ρ   RT   CO  (4)

Furthermore, from the various plasma- metal reactions occurring between the deoxidizing elements and oxygen, oxide compounds are formed, which tend to float to the molten weld pool surface. As solidification of the weld pool proceeds, these oxide compounds become part of the covering slag. A small amount of the oxide products may also be trapped in the solidified weld metal in the form of inclusions. In the present study, several assumptions are made to simulate the various reactions involving oxygen in welding experiments carried out using an experimental basic-type flux cored wire under 100% CO2 shielding. A descrip- tion of the reactions taking place in the Fig. 1 — X-ray diffraction pattern of slag sample collected after welding with the experimental arc column is then conceived. As a result, electrode. the partition of the various elements present in the constitution of the FCAW electrode between the deposited weld metal metallic ingredients are often slag form- dioxide can then further decompose into and the solidified slag can be deter- ers and arc stabilizers, i.e., helping to carbon and oxygen (O2), as indicated by mined, as well as the composition of the produce a smoother arc. They provide a the following reactions: solidified slag. From this knowledge, a secondary shielding action and help pu- 1 methodology to calculate the basicity CO (g) → CO (g) + ⁄2 O (g) (1) rify the weld metal. The nonmetallic in- 2 2 index of the FCAW electrode is derived. gredients also help in reducing weld 1 CO (g) → C (s) + ⁄2 O (g) (2) spatter and in controlling the melting 2 Basicity Index of a Flux/Slag System characteristics of the electrode. Proper selection of the core elements As a result, a certain amount of oxy- The concept of basicity was first is of prime importance for any welding gen is available in the arc atmosphere to adopted to answer the needs of the steel application of acceptable quality. When react with the various elements present in making industry. It was used to evaluate choosing the ingredients for a new flux the molten metal (metal droplets and the sulfur refinement ability of a slag in design, formulators need to carefully weld pool). Therefore, when CO2 is used steel ladle refining practices (Ref. 3). study the physical and chemical charac- as the shielding gas and dissociates dur- Later, the use of basicity index was teristics of the raw materials to utilize, ing welding, a strong oxidizing effect broadened to approximately measure the e.g., their particle shape and size distrib- takes place. That is partly why deoxidiz- flux oxidation capacity, and the same ution, their chemical composition and ers have to be added to the fill ingredients principles were applied to the welding purity, their hygroscopic tendency and of the electrode to compensate for the technology. Consequently, it has been their chemical stability. Also, the fill ratio various oxidation sources that include general practice to use the basicity index of the electrode, i.e., the ratio of the flux the oxidizing effects of CO2 and the O2- to characterize a welding flux with regard weight over the total electrode weight, containing constituents of the electrode. to its physical and chemical properties. has to be closely controlled in order to Also, in the presence of carbon in the The slag viscosity and the weldment obtain reproducible weld quality along molten weld pool, CO2 can react to form quality, as represented by the amount of the weld bead. CO according to the reaction below — oxygen pickup in the weld metal and the Carbon dioxide is certainly the most Boudouard equation (Ref. 2). weld metal notch toughness, can both be widely used shielding gas for FCAW steel estimated as a function of the basicity electrodes that require auxiliary gas CO2 (g) + C (s) → 2 CO (g) (3) index. Thus, a high basicity slag is con- shielding. Its selection is mainly due to sidered to be one that exhibits a high the possibility of high welding speeds, The simultaneous presence of CO and concentration (or a high activity) of free Ρ 2– good weld penetration and, most of all, CO2 oxide ions O . It would also tend to ex- hibit an increased breakdown of the low cost. At room temperature, carbon CO determines theΡCO ratio, which dioxide is a relatively inactive gas. When 2 three-dimensional silicate network struc- also controls Ρ , the partial pressure of RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT heated to high temperatures in the weld- O2 ture (Refs. 3, 4). Such a slag would there- ing arc, however, part of the CO2 disso- oxygen of the arc environment, as gov- fore be composed of a large quantity of ciates to form carbon monoxide (CO), erned by Equations 1 and 2. Thus, the ac- dispersed cations in the broken silicate which is more stable than CO2. Carbon tivity (or partial pressure) of oxygen can network, contributing to a more fluid slag

58-s | MARCH 2000 in the molten state. Defining the concept of basicity of slags by means of the O2– activity, however, is virtually not feasible, since activities or concentrations of free oxide ions are impossible to measure experimentally. In practice, weld metal mechanical properties and weld metal oxygen content are related to the welding fluxes by means of the basicity index, which is usually defined as the following: Σ Basic Oxides Basicity Index = Σ Acidic Oxides (5)

Equation 6 is the most widely used basicity index formula and is recognized by the International Institute of Welding Fig. 2 — Schematic representation of the multilayer deposit. (IIW). It is often referred to as Tuliani’s expression (Ref. 5), in which flux compo- nent contents are expressed in weight quantitatively describe the oxidizing ca- of the welding parameters. Finally, Inda- percent. pability of the welding fluxes. cochea, et al. (Ref. 23), studied various CaF2 ++CaO MgO +++ BaO SrO For many existing welding consum- manganese silicate fused fluxes in SAW. Na22 O+++×+ K O Li 2 O05. () MnO FeO ables, a general trend seems to indicate They came to the interesting conclusion BI..= the weld metal oxygen level decreases as that by holding the SiO content constant SiO22322+×05. () Al O + TiO + ZrO 2 (6) the basicity index increases. In particular, at 40%, the weld metal oxygen content the index is well suited for submerged arc increased as the FeO content in the flux Using the above expression, when the fluxes, which usually contain mostly sim- was increased. Their results brought basicity index is less than 1.0, the flux is ple oxides and calcium fluoride, along about another contradiction to the basic- regarded as acidic, between 1.0 and 1.2, with ferro-alloys for deoxidization and ity index theory wherein FeO appears in as neutral, and a flux with a B.I. greater alloying purposes. However, correlating the numerator of the majority of formulas than 1.2, as basic. the basicity index of a flux system based describing the basicity index. With time, dozens of empirical or sta- on the amount of all oxides present with Therefore, the amount of oxygen in tistical formulas have been developed for its oxygen potential can actually be far the weld metal is not strictly governed by the quantitative determination of the ba- from the real deoxidation mechanism the value of a basicity index based on the sicity of welding fluxes (Refs. 6–8). The that takes place in the weld pool. For ex- amount of oxides present in the original degree of successful application of each ample, manganese oxide and iron oxide flux. Defining such a basicity index ap- of these equations depended on the sam- both have been found to increase the pears irrelevant since it is always possible ple population and chemical reactions oxygen content in the deposited metal, to add strong deoxidizers in their metal- involved. Nonetheless, problems related although their presence increases the ba- lic form, e.g., Mn, Si, Mg, Al, Ti, ..., which to the evaluation of the basicity of weld- sicity of the flux/slag system. In an inves- are not included in the basicity index cal- ing fluxes/slags have always been a sub- tigation carried out by Gaspard-Angeli, et culation. Nevertheless, these com- ject of discussion (Refs. 6, 11–14). Thus, al. (Ref. 20), using various SAW fused pounds play a major role in the weld pool other concepts for expressing the flux ba- fluxes, the weld metal oxygen content deoxidization process. sicity have been investigated and re- was found to increase from 427 to 450 In FCAW, not only do flux formulators ported in the literature (Refs. 10, 15–18). ppm as the amount of ferrous oxide (FeO) use oxides but also ferro-metallic ingre- In particular, the Bz index, described by increased from 2.0% to 4.3%, all other dients, e.g., Fe-Mn, Fe-Si and multiple Zeke in 1980 (Ref. 16), as well as the op- welding parameters being identical. In other complex fluorides and minerals, tical basicity index, presented by Datta, addition, the same trend related to the ef- not always reported in the open litera- et al., in 1989 (Ref. 10), are two examples fect of magnetite (Fe3O4) on the transfer ture. Hence, because of the many com- of indexes that were developed to ex- of oxygen in the weld metal could be plicated and sometimes unknown reac- press the basicity of a welding flux. Fur- drawn from data in shielded metal arc tions taking place in the welding arc ther explanation of these two basicity welding (Ref. 21). The same conclusion environment, it cannot be justifiable to theories is provided in Part 2 of the paper, was reached for the influence of FeO flux correlate the composition and mechani- which starts on page 66-s of this issue of additions in a systematic study combin- cal properties of the weld metal with a the Welding Journal. ing several SAW fluxes, various solid basicity formula based only on the Furthermore, Pokhodnya, et al. (Ref. wires and two sets of welding parameters weight percentages of the oxides and cal- 19), evaluated the oxidizing capacity of (Ref. 22). Some SAW fluxes featuring cium fluoride contained in the welding submerged arc welding fluxes using a lower FeO contents, therefore, were flux. Conversely, the composition of the mass spectroscopy technique. As such, found to promote the formation of acic- solidified slag on top of a weld bead is a these researchers heated and melted typ- ular ferrite in the weld metal, favoring the good indicator of the extent of the ical SAW fluxes. Then, they measured the possibility for higher toughness. Also, slag/metal interactions during welding, quantity of oxygen, i.e., the oxygen po- some data in this paper showed the effect and provides substantial information tential of the gaseous phase generated. of MnO on the oxygen content was more about the various chemical reactions that RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT Eventually, the weld metal oxygen con- complex in that it could actually either took place in the weld pool. Therefore, tent could be predicted. This method, help decrease or, on the contrary, in- regarding the need for defining the ba- therefore, was used as a basicity index to crease the oxygen content, as a function sicity index of a FCAW electrode, it

WELDING RESEARCH SUPPLEMENT | 59-s network modifiers. The concentration of iron oxide, as well as the melt temperature and the oxygen partial pressure, were also found to affect the reduction of Fe3+ to Fe2+. For the sake of simplicity, iron was therefore supposed to be present in the form of ferric oxide (Fe2O3) in the slags investigated in the present paper. In this work, the basicity theory will be utilized to quantitatively characterize each FCAW steel consumable investigated with respect to its oxidation potential and metallurgical quality.

Assumptions Several assumptions were made in the present study. By doing so, the various reactions taking place during welding are simplified, which allows for the development of the methodology. 1) It is known that silicates and titanates are common in slags found in steel welding systems. They frequently form as solid solutions with variable ratios of Ca, Mn, Fig. 3 — Representation of the distribution of single elements in the solidified slag, as predicted Mg, Fe, etc. (Ref. 26). Larsenite from calculations, and as measured. –Ca2SiO4–, bustavite –(Ca,Mg)3Si3O9–, olivine –(Fe,Mn)2SiO4– and fayalite –Fe2SiO4– are examples of these compounds. Other constituents determined should be done based on the chemical Equation 7 is identical to the expres- using X-ray diffraction are K2Zr2O5, composition of the solidified slag col- sion proposed in 1969 by Tuliani, et al. hematite –Fe2O3– or CaF2. Besides these, lected after welding. As such, this ap- (Ref. 5), except the mole percentage of many phases found in a welding slag are proach is different from most investiga- each oxide is used instead of the mass amorphous, which cannot be determined tions done on slag-metal reactions since fraction. This indeed better accounts for by X-ray diffraction. they invariably predict the chemical the relative importance of the various ox- For simplicity’s sake, all elements that composition of the weld metal instead of ides, which form from the reactions be- transfer from the original flux to the slag that of the solidified slag (Refs. 6, 10, 13, tween the ingredients that dissociate to were assumed to preferentially form ox- 24, 25). As an example, Mitra and Eagar their metallic forms and the oxygen avail- ides in the covering slag. (Refs. 24, 25) developed a kinetic model able in the arc atmosphere. Also, FeO is 2) Every ingredient of the original core to describe the transfer of alloying ele- replaced by Fe2O3 in Tuliani’s formula flux was assumed to dissociate in the ments from various submerged arc fluxes since this latter oxide is presumably the welding arc. Recombination of the disso- to the weld pool. They established a iron oxide that forms in a welding slag, on ciated flux core elements with oxygen quantitative relationship between the the basis of their Gibbs energy of forma- will form the slag that covers the molten welding consumable composition and tion at approximately 2000°C, and con- weld pool. the resulting weldment chemical compo- firmed by X-ray diffraction (Ref. 26). Also, Fluorspar, CaF2, which also partici- sition. Despite the good agreement be- the fraction of iron oxide contained in a pates in the electrode flux formulation, is tween the model and experimental re- slag as given by X-ray fluorescence was treated differently in the analysis. CaF sults, no special consideration was given 2 expressed in the form of Fe2O3 in the pre- originating from the flux was believed to to the solidified slag, which would be in sent investigation. One reason for this transfer as a whole into the slag. Evidence intimate contact with the molten weld practice is the preparation technique em- of its presence as a major constituting pool until solidification. ployed, which consists of obtaining a phase in the solidified slag was indeed fused bead, therefore implying complete shown through X-ray analysis, in the case Proposed Methodology oxidation of the iron contained in the of the experimental electrode studied in slag. This actually opposes results from Part 1 — Fig. 1. In the present investigation, an inno- Medeiros, et al. (Ref. 27), who observed 3) In an article published by AWS in vative methodology for calculating the Fe2O3 initially present in a SMAW coat- 1979 (Ref. 29), data were reported on the basicity index of a FCAW steel electrode ing would tend to reduce to FeO in slags weight of fumes per weight of deposited is thus conceived by measuring the produced in underwater wet welding. metal for various flux cored steel elec- chemical composition of the solidified Furthermore, Pargamin, et al. (Ref. 28), trodes, including for two E70T-5 elec- slag and applying the expression given used Mössbauer spectroscopy to deter- trodes, i.e., for electrodes somewhat sim- by Equation 7. mine the coordination of Fe2+ and Fe3+ ilar to those investigated in the present ions in quenched silicate melts. They work. These two sets of data indicated the CaF22++ CaO MgO ++++ BaO SrO Na O RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT found that at low concentrations of iron weight of fumes to constitute exactly KO++× LiO 0.5 MnO + Fe O B.I. = 22() 23 oxides in alcaline silicate slags, both Fe2+ 1.79% and 2.36%, respectively, of the SiO+× 0.5 Al O + TiO + ZrO 22322()and Fe3+ were actually octahedrally co- weight of weld metal deposit for these (7) ordinated with oxygen, thus behaving as two electrodes. On the basis of these

60-s | MARCH 2000 numbers published by AWS, the amount nE = Total number of moles of element E minerals that contained Na, K, Mg, Li, Al of material carried away in fumes during in the slag; and Zr. Table 2 lists the ingredients used welding was neglected. No volatile nExOy = Total number of moles of oxide for manufacturing the electrode. species was considered in the present ExOy in the slag; Multipass welds were prepared on work. It is assumed, therefore, that when ME = Molecular mass of E; ASTM Type A633 steel coupons using the electrode is heated in the arc atmos- MExOy = Molecular mass of ExOy; 100% CO2 shielding gas. These chemical phere, only carbon dioxide evolves from E%slag = Total concentration of element composition pads were prepared using a the decomposition of the carbonates pre- E transferred into the solidified covering method equivalent to that specified by sent in the core flux. The equation below slag, expressed as a fraction of the total AWS A5.29-80 specifications — Fig. 2. describes the decomposition of carbon- weight of the slag; As such, the dilution effect caused by the ates into an oxide and carbon dioxide: ExOy%slag = Total concentration of oxide base material could be avoided. The top ExOy transferred into the solidified cover- bead of the multilayer weld pad was Ex(CO3)y (s) → ExOy (s) + y CO2 (g) (8) ing slag, expressed as a fraction of the ground flat and then analyzed using op- total weight of the slag. tical emission spectroscopy. Three runs where Ex(CO3)y (s) represents the car- By applying Equation 11 to all ele- were performed at different locations on bonate of element E (e.g., Ca or Mg, etc.) ments E of the solidified slag, the slag the same weld pad and the average value and ExOy (s), the oxide of E that becomes composition can be determined in terms was reported. part of the covering slag. of stoichiometric oxides and calcium flu- The experimental basic-type FCAW 4) The amount of material lost in the oride only. Then, the basicity index of the consumable was used at the welding form of spatter is neglected. welding consumable can be computed condition of 260 A, 31 V and a travel 5) The collected slags were assumed using the different fractions of oxides cal- speed of 7 mm/s (16.5 in./min), which is to be homogeneous in composition at culated from Equation 11. It is notewor- the best operating performance for this any location along the weld bead. Also, thy to point out Equation 12 actually re- electrode. The contact tip-to-work dis- no nonmetallic inclusions are assumed sults from the expression given in tance was set and carefully controlled at 3 to form in the weld metal, which is hy- Equation 7. 20 mm ( ⁄4 in.). A Miller Maxtron 450 con- pothesized to be fully deoxidized. This stant voltage power supply was used. Ac- assumption implies the resulting oxide CaF2 %slag + ∑ () Exy O %slag + cording to the manufacturers’ recom- E Ca,Mg,Ba,Sr,Na,K,Li compounds float to the surface of the = mended practices, the electrode was molten weld pool, where they become 0.5 × ∑ ()Exy O %slag welded using direct current electrode E=Mn,Fe part of the solidifying slag. B.I. = negative (DCEN) polarity. Finally, espe- ∑ ()EO%xy slag +× 0.5∑ () EO%xy slag Based on the above assumptions, the ESi= E=Al,Ti,Zr cially due to the nature of the shielding overall slag composition, therefore, was gas, a globular, repulsive type of transfer expressed in terms of fixed, stoichiomet- 12) was observed in the conditions of the ex- ric basic and acidic oxides and calcium What follows in the present paper will periments. fluoride only. describe a series of experiments that For the flux cored electrode of Table were performed to predict the covering 2, a method was proposed for the calcu- Solidified Slag Composition slag composition from the knowledge of lation of the composition of the solidified the composition and nature of the flux- slag using a mass balance technique con- From the knowledge of the solidified cored electrode and the chemical com- sidering the FCAW fill, sheath and weld slag elemental composition, the typical position of the deposited weld metal. In metal chemical analyses. Then, a oxidation reaction involving a single addition, comparison with the actual methodology was developed to account for the basicity of the electrode, based on metallic element E and its oxide ExOy has composition of the collected slag was to be considered, according to the previ- made. the slag composition. For mass balance, ously stated assumptions and as repre- steel coupons used for welding were sented by Equation 9. Experimental Procedure weighed before and after welding to de- y termine the weight of the deposited weld xE() s + Og↔ EOsxy() The experimental 1.2-mm (0.045-in.) metal, as well as that of the solidified cov- 2 2 () (9) diameter basic-type FCAW (E70T-5 AWS ering slag. Also, the length of the elec- which yields grade) electrode investigated in this paper trode consumed during a weld test was xn×= n ExOy e (10) consisted of a low-carbon, low-alloy steel measured for the calculations of the wire and sheath, the composition of which is given efficiency, i.e., the weight of deposited in Table 1, and a core flux containing 14 metal for 100 g of consumed wire. M EOxy

metallic and nonmetallic ingredients, in- Furthermore, solidified slag samples RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT EO%xy slag + E%slag xM× E (11) cluding iron, ferro-silicon, ferro-man- were collected from weld beads de- ganese powders, SiO2, TiO2, CaCO3, posited using the experimental flux cored In Equations 9, 10 and 11, CaF2, ZrO2 and various other oxides and electrode for subsequent analyses. Ele-

WELDING RESEARCH SUPPLEMENT | 61-s ments such as Si, Mn, Ti, Mg, Fe, Zr, Ca, that w.e. was equal to 79%, and s.e., to spatter produced — Equation 17. Al and P present in the slag sample were 7.7%. The welding experiments exhib- (E%total x 100%) = (E%weld metal x w.e.) + measured by fusing a bead in a platinum ited good reproducibility for this con- (E%slag x s.e.) + (E%fume x f.e.) + crucible in a propane/oxygen atmos- sumable. As for f.e. and sp.e., they were (E%spatter x sc.e.) (17) phere. The sample was then analyzed not calculated since neglected, accord- using quantitative X-ray fluorescence. El- ing to the earlier stated assumptions. No- with ements such as Na, K and Li were dis- tice also the sum of w.e. and s.e. does not E%total = total concentration of a sin- solved in an acidic solution and analyzed equal 100%, clearly indicating the gle element E contained in the flux cored using inductance coupled plasma spec- weight of the consumed cored wire does electrode, expressed as a fraction of the troscopy. Fluorine was analyzed in solu- not simply distribute between the weld total weight of the electrode; tion after pyrohydrolysis using a specific metal and the slag. E%weld metal = total concentration of E electrode that featured a monocrystalline First of all, due to its basic nature, the deposited in the weld metal, expressed as membrane (PF4-L type electrode). The welding operating characteristics of the a fraction of the total weight of the de- activity of free F– anions was measured FCAW electrode utilized were rather posited weld metal; by immersing the electrode in the solu- poor, with a harsh arc and significant E%slag = total concentration of E trans- tion. The amount of fluorine in slag sam- spatter levels. The weight of the de- ferred into the solidified covering slag, ples was thus measured as a potential dif- posited weld metal and that of the col- expressed as a fraction of the total weight ference, on the basis of Nernst’s law. lected solidified slag did not include the of the slag; Finally, the sulfur content was estimated droplets ejected from the welding arc. E%fumes = total concentration of E lost using semi-quantitative X-ray spectrome- Note the use of 100% CO2 shielding gas in the welding fumes, expressed as a frac- try. This method was selected instead of was deliberate in the present investigation of the total weight of the fumes pro- a combustion technique because of the tion, as part of a strategy selected for de- duced during welding; rather large quantity of fluorine con- veloping a viable 100% CO2-shielded E%spatter = total concentration of E tained in the slag samples. Fluorine evo- basic-type FCAW consumable. One can transferred into the solidified covering lution would actually damage the mea- suppose using a more “forgiving” shield- slag, expressed as a fraction of the total surement cells in a combustion ing gas, such as a 80%Ar/20%CO2 or a weight of the spatter ejected during technique such as that employed in a 98%Ar/2%O2 shielding gas mixture, welding. Leco combustion analyzer. would have improved the metal transfer Next, since losses in the form of va- characteristics with a considerable re- pors or spatter are ignored according to Results and Discussion duction in the amount of spatter during the earlier stated assumptions, Equation welding. These two gases were not used 17 transforms to Equation 18. The equa- Prediction of the Solidified Slag in this program. However, although fea- tion of conservation reads as follows: Composition turing a rather high spatter level, the experimental electrode exhibited a high (E% x 100%) = (E% x w.e.) + From the knowledge of the exact for- total weld metal metal transfer stability, as demonstrated (E%slag x s.e.) (18) mulation of the flux cored electrode, i.e., by arc voltage and current signals gath- the chemical analysis of the metal sheath ered using a high-speed data acquisition An expression for calculating the and the exact proportion of the ingredi- system (Ref. 30). Consequently, due to amount of each element that transfers ents composing the flux, a method for the very stable metal transfer characteris- from the original core flux to the solidi- predicting the composition of the solidi- tics of the electrode, it was concluded the fied slag can thus be derived, as indicated fied slag was derived. weld deposit oxygen content would not below. First of all, when a weld bead is de- be greatly affected by the spatter level ob- posited on a steel plate, it is possible to served in the present investigation. Nei- E% % ()total ×100 − define the arc welding deposition effi- ther abrupt changes in the metal transfer ciency, as represented by w.e., which E% w.e mode, nor chaotic transfer events, be- ()weld metal × . corresponds to the actual weight of the E% = lieved to promote an increase in the oxy- slag s.e. (19) weld metal for 100 g of the consumed gen transfer to the weld pool across the cored wire. Similarly, s.e., as defined as arc, was observed.    the “slag efficiency,” as well as f.e., the  ()f.r.×+ E%flux  Secondly, no volatile species are con-   ×100% −  1 − f.r.× E%  “fume efficiency,” and sp.e., the “spatter sidered in the calculation, which may ()metal sheath   efficiency,” can be determined. The fol- constitute another source of error. Finally, ()E%weld metal × w.e. lowing equations define w.e., s.e., f.e., E% = regarding the slag that results from the slag s.e. and sp.e., respectively. multiple reactions that involve gaseous species with the dissociated flux ingredi- (20) weight of deposited weld metal w.e. = × 100 ents in the arc column, the quantity of the where f.r. was the fill ratio of the flux weight of consumed cored wire (13) shielding gas that reacts with the elec- cored wire, i.e., ratio of the weight of weight of the solidified weld metal s.e. = × 100 trode is difficult to evaluate. Conse- core flux fill over the total weight of the weight of consumed cored wire quently, for the above reasons, adding up flux cored electrode fill, per unit length. (14) w.e. and s.e. would not necessarily give In addition, E%flux was the total concen- weight of the fumes produced a total of 100%. tration of element E contained in the core f.e. = × 100 weight of consumed cored wire (15) For the calculation, the concept of flux, expressed as a fraction of the total and mass conservation of a particular ele- weight of the flux, and E%metal sheath, the ment present in the welding system is re- total concentration of E contained in the

RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT weight of the spatter produced sp.e. =×100 quired. This condition implies the mass metal sheath, expressed as a fraction of weight of consumed cored wire (16) of an element E present in the original the total sheath weight. It was experimentally found for the electrode must also be present in the The complete chemical analysis of the experimental flux cored wire of Part 1 weld metal, slag, smoke and, eventually, weld metal deposited using the experi-

62-s | MARCH 2000 mental electrode was measured. As such, Analysis of the Solidified Slag Composition the values for E%weld metal are reported in Table 3. Slag chips were collected from the By knowing the exact constitution welds produced for complete analysis. and formulation of the flux cored elec- The analysis allowed for the determina- trode, and by substituting the values for tion of the slag composition in terms of E%weld metal from Table 3 into Equation single metallic elements. Also, Equation 20, an elemental composition of the so- 11 was applied to this analysis in order to lidified slag could be calculated. Addi- get the composition of the slag in terms tionally, Equation 11 was used to give the of stoichiometric oxides. The results of composition of the slag as simple oxides. the analysis are compiled in Table 5. The chemical composition of the solidified slag, expressed both in terms of sin- Results from Actual Slag Chemical Analysis gle elements and oxides, is reported in Table 4. Table 5 compares well with Table 4. Additionally, Fig. 3 presents both sets of Predictions from Calculations data that describe the composition of the covering slag, expressed as a function of From Table 4, it is found the main con- the single elements present in it. stituents of the solidified slag are CaF2 First of all, it is important to observe (37.4%), SiO2 (21.0%) and TiO2 (16.7%). that the total given by the slag composi- This agrees well with the fact that most of tion expressed in terms of oxides is very the matter constituting the slag consists of close to 100%, as indicated in Table 5. various silicates, titanates, and calcium This shows that the assumption that all el- fluoride. One can also see from the re- ements transferring from the original flux sults in Table 4 the total given by the sum to the slag form oxides and CaF2 in the of all oxides is equal to 108.8%, which is covering slag was reasonable. This as- greater than 100%. This discrepancy cer- sumption allowed for good predictions of tainly arises from shortcomings in the as- the overall slag composition. Besides, it sumptions, particularly the negligence of is apparent from these results that a rather the amount of material carried away in good agreement is obtained between the the form of spatter and welding fume. calculated and the measured slag com- Also, it was assumed that CaF2 compositions. The five main elements consti- pletely transferred from the core flux to tuting the slag, i.e., Ca, F, Mn, Si and Ti, the solidified slag. Even though X-ray were predicted with a fairly acceptable analysis showed evidence of the pres- match. Therefore, the methodology pro- ence of CaF2 in the solidified slag of the posed for predicting the composition of electrode, it is probable this ingredient the covering slag was reasonable. Fur- reacts differently when transferred across thermore, from the constitution of the the welding arc. Finally, in order to ex- slag expressed in terms of stoichiometric plain the discrepancy of Table 4 regard- oxides and CaF2, the basicity was calcu- ing the total chemical composition of lated in each case. As shown in Tables 4 108.8%, assuming the slag is made up and 5, the experimental FCAW steel elec- only of simple oxides is less than rigor- trode was assigned a basicity index ap- ous. For example, when a compound proximately equal to 1.9. Both calcu- such as olivine –(Fe,Mn)2SiO4– forms in lated and measured slag compositions the slag, fewer oxygen equivalents are led to this basicity index value with an needed than what is required for the for- excellent agreement. values might be attributed to the losses in mation of Fe2O3, 2MnO and SiO2 (4 O Figure 3 also shows all of the pre- vapors of the corresponding compound equiv. vs. 7). Likewise, assuming the for- dicted concentrations are larger than of the element present in the flux at the mation of 3CaO, 3MgO and 3SiO2 in- those given by the slag analysis — except high temperatures in the welding arc. In stead of bustavite –(Ca,Mg)3Si3O9– re- for zirconium, aluminum and iron. One particular, as seen in Fig. 3, only about quires three extra oxygen equivalents (12 of the reasons for this may be the vapor- half of the predicted Li, Na and K were vs. 9). Therefore, the discrepancy of Table actually measured in the slag. These re-

ization of flux ingredients and reaction RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT 4 is most likely due to the assumption the products during welding. For each ele- sults are not surprising considering these slag forms stoichiometric oxides instead ment, the magnitude of the difference be- elements feature high volatility. Conse- of complex oxides. tween the predicted and the measured quently, a large proportion of these ele-

WELDING RESEARCH SUPPLEMENT | 63-s ments would be concentrated in the the weld bead for a complete analysis, welding. Welding Journal 57(2): 76-s to 80-s. fumes, which were neglected according calculating for each metallic element 8. Potapov, N. N., and Kurlanov, S. A. to the assumptions. Notice here, if Li, Na present in the slag the number of oxygen 1978. A quantitative evaluation of the basicity and K constituted a large percentage of equivalents consumed for the formation of welding fluxes. Welding Production 25(9): the slag-forming ingredients of the elec- of its most stable stoichiometric oxide, 39–42. trode, as would be the case in many expressing the overall composition of the 9. Iwamoto, N. 1974. Structure of slags metal-cored electrodes, then the pre- solidified slag in terms of molar fractions (Review I) — Basicity of slag. Transactions of dicted slag composition, assuming no of the constituting oxides together with JWRI 3(2): 89–98. fume losses, would be considerably dif- CaF2– and calculating the basicity index 10. Datta, I., and Parekh, M. 1989. Filler ferent from the actual slag composition. of the slag using a derivative of Tuliani’s metal flux basicity determination using the Notice, however, in the case of the pre- formula from the molar fraction of each optical basicity index. Welding Journal 68(2): sent electrode, Li, Na and K represented oxide and CaF2 present in the slag. 68-s to 74-s. only 3.2% of the total flux weight. Thus, 3) One of the main advantages of this 11. Palm, J. H. 1972. How fluxes deter- vaporization losses of these elements cer- innovative basicity theory is that the slag mine the metallurgical properties of sub- tainly had a minor effect on the final out- basicity index is always predictable from merged arc welds, Welding Journal 51(7): come of the analysis. Finally, as to the a mass balance considering the exact for- 358-s to 360-s. presence of Al and Zr in the solidified mulation of the FCAW consumable. Prior 12. North, T. H., Bell, H. B., Nowicki, A., slag, both of them are believed to origi- to welding, the slag basicity index can be and Craig, I. 1978. Slag/metal interaction, nate from “impurities” contained in the estimated from calculations. oxygen and toughness in submerged arc weld- core flux — in particular, from CaCO3 ing. Welding Journal 57(3): 63-s to 75-s. and CaF2 — since no compound of these Acknowledgments 13. Wegrzyn, J. 1985. Oxygen in the sub- two elements was actually present in the merged arc welding process. Metal Construc- core flux of the experimental electrode. E. Bauné wishes to express his sincere tion 17(11): 759R–764R. From these results, with a silicon con- gratitude to Air Liquide for the financial 14. Olson, D. L., Liu, S., and Fleming, D. tent equal to 0.23% and a manganese support in the development of his re- A. 1993. Welding flux: nature and behavior. content equal to 1.26% in the weld metal search work and studies at the Colorado Report MT-CWR-093-001, Golden, Colo., deposited using the electrode of Part 1, it School of Mines. The authors gratefully Colorado School of Mines. was found 80% of the total amount of sil- thank the personnel working in the Weld- 15. Wagner, C. 1975. The concept of the icon originally contained in the flux cored ing Consumables and Laboratory De- basicity of slags. Metallurgical Transactions B electrode was transferred into the solidi- partments at CTAS and in particular, P. Le 6B(9): 405–409. fied slag, and only 44% of the manganese Seigneur, F. Richard and B. Leduey, who 16. Zeke, J. 1980. Recommendations for did. As for the constituents of the core flux also contributed to the investigation pre- expressing the flux basicity index by means of other than those containing Si and Mn, sented herein. Finally, thanks are also ex- the oxygen anion ionic fraction. Zvaranaie i.e., elements such as Ti, Ca, Zr, Li, Mg, tended to Prof. D. L. Williamson for as- 29(7): 193–204. Na, K and Al, they were virtually not found sistance with the X-ray diffraction 17. Mills, K. C. 1993. The influence of in the weld metal. They indeed exhibited analyses conducted at the Colorado structure on the physico-chemical properties only minor or null effect on the final School of Mines. of slags. ISIJ International 33(1): 148–155. chemical composition of the weld metal. 18. Frohberg, M. G., and Kapoor, M. L. Finally, one of the drawbacks of the References 1971. The application of a new basicity index present model is that no Fe can be pre- to the metallurgical reactions. Stahl und Eisen dicted in the covering slag since the 1. Smith, D. C. 1970. Flux-cored elec- 91(4): 182–188. method employed consists in a mass bal- trodes — Their composition and use. Welding 19. Pokhodnya, I. K., Golovko, V. V., Kush- ance in which iron is the base element. Journal 49(7): 535–547. nerev, D. M., and Shvachko, V. I. 1990. Eval- However, the analysis showed 2.7% Fe 2. Liu, S., Olson, D. L., Ibarra, S., and Run- uation of the oxidizing capability of ceramic was actually present in the covering slag. nerstam, O. 1992. Oxygen as a welding para- fluxes. Avtom. Svarka, No. 2: 45–48. meter: the role of light metallography. Metal- 20. Gaspard-Angeli, A., and Bonnet, C. Conclusions lographic Characterization of Metals after 1986. How to optimize the properties of the lon- Welding, Processing, and Service, Microstruc- gitudinal welds of pipes using a fused flux. Proc. The achievements of the investiga- tural Science 20, ASM International: 31–44. 3rd Int. Conf. of Welding and Performances of tions carried out in Part 1 of the present 3. Turkdogan, E. T. 1996. Fundamentals of Pipelines. Paper 42. London, England. paper can be summarized as follows: Steelmaking. The Institute of Materials, Lon- 21. Leduey, B. 1997. Hydrogéne diffusible 1) Knowing the constitution of an ex- don, England. et humidité ambiante. Rapport Technique no. perimental basic-type FCAW electrode 4. Podgaetskii, V. V., and Galinich, V. I. 1324., Pontoise, France, L’Air Liquide C.T.A.S. and the chemical analysis of the weld 1981. The structures of molten welding slags 22. Bonnet, C., and Charpentier, J. P. metal deposited with it, a simple method (Review). Avt. Svarka 7: 36–45. 1983. Effect of deoxidization residues in wire for predicting the composition of the so- 5. Tuliani, S. S., Boniszewski, T., and and of some particular oxides in CS fused lidified slag was developed. Rather good Eaton, N. F. 1969. Notch toughness of com- fluxes on the microstructure of submerged arc agreement was obtained with the com- mercial submerged arc weld metal. Welding weld metals. The Effects of Residual, Impurity, position obtained from a slag analysis. and Metal Fabrication 37(8): 327–339. and Micro-Alloying Elements on Weldability 2) A conceivable approach to quanti- 6. Almqvist, G., Polgary, C. S., Rosendhal, and Weld Properties Int. Conf. Paper 8. Lon- tatively characterize FCAW electrodes by C. H., and Valland, G. 1972. Some basic fac- don, England. means of a basicity index based on the tors controlling the properties of weld metal. 23. Idacochea, J. E., Blander, M., Chris- chemical analysis of their collected slag Proc. Conf. on Welding Research Relating to tensen, N., and Olson, D. L. 1985. Chemical RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT was developed using a derivative of the Power Plant, Central Electricity Generating reactions during submerged arc welding with well-known Tuliani’s formula. The differ- Board, Leatherhead: p. 204. FeO-MnO-SiO2 fluxes. Metallurgical Transac- ent steps of this methodology consisted 7. Eagar, T. W. 1978. Sources of weld metal tions B 16B(2): 237–245. of collecting the solidified slag on top of oxygen contamination during submerged arc 24. Mitra, U., and Eagar, T. W. 1984. Slag-

64-s | MARCH 2000 metal reactions during submerged arc welding 27. Medeiros, R. C., and Liu, S. 1998. A ronment, A Research Report on Fumes and of alloy steels. Metallurgical Transactions A predictive electrochemical model for weld Gases Generated during Welding Operations. 15A(1): 217–227. metal hydrogen pickup in underwater wet 1979. American Welding Society, Miami, Fla. 25. Mitra, U., and Eagar, T. W. 1991. Slag- welds. Proc. 17th Int. Offshore Mechanics and 30. Bauné, E., Bonnet, C., and Liu, S. metal reactions during welding. part I: evalu- Arctic Engineering Conf. — Material Sympo- 1999. A methodology for assessing the metal ation and reassessment of existing theorie — sium, ASME/OMAE. Paper 98-2211. Lisbon, transfer stability and spatter severity in flux part II: theory — part III: verification of theory. Portugal. cored arc welding. Paper C, Session 4. 80th Metallurgical Transactions B 22B(1): 65–71, 28. Pargamin, L., Lupis, H. P., and Flinn, P. Annual AWS Convention and International 73–81, 83–100. A. 1972. Mössbauer analysis of the distribu- Welding and Fabrication Exposition, St. 26. Private correspondence with R. Hamil- tion of iron cations in silicate slags. Metallur- Louis, Mo. ton, with Mountain Technical Center, J. gical Transactions 3: 2093–2099. Manville, Denver, Colo., Nov. 1997 29. Fumes and Gases in the Welding Envi-

Preparation of Manuscripts for Submission to the Welding Journal Research Supplement

All authors should address themselves to the Keep in mind that proper use of terms, following questions when writing papers for submission abbreviations and symbols are important to the Welding Research Supplement: considerations in processing a manuscript for ◆ Why was the work done? publication. For welding terminology, the Welding ◆ What was done? Journal adheres to ANSI/AWS A3.0-94, Standard ◆ What was found? Welding Terms and Definitions. ◆ What is the significance of your results? Papers submitted for consideration in the Welding Research Supplement are required to undergo Peer ◆ What are your most important conclusions? Review before acceptance for publication. Submit an With those questions in mind, most authors can original and one copy (double-spaced, with 1-in. logically organize their material along the following 1 margins on 8 ⁄2 x 11-in. or A4 paper) of the manuscript. lines, using suitable headings and subheadings to Submit the abstract only on a computer disk. The divide the paper. preferred format is from any Macintosh® word 1) Abstract. A concise summary of the major processor on a 3.5-in. double- or high-density disk. elements of the presentation, not exceeding 200 words, Other acceptable formats include ASCII text, to help the reader decide if the information is for him or Windows™ or DOS. A manuscript submission form her. should accompany the manuscript. 2) Introduction. A short statement giving relevant Tables and figures should be separate from the background, purpose and scope to help orient the manuscript copy and only high-quality figures will be reader. Do not duplicate the abstract. published. Figures should be original line art or glossy 3) Experimental Procedure, Materials, photos. Special instructions are required if figures are Equipment. submitted by electronic means. To receive complete 4) Results, Discussion. The facts or data instructions and the manuscript submission form, obtained and their evaluation. please contact the Peer Review Coordinator, Doreen 5) Conclusion. An evaluation and interpretation of Kubish, at (305) 443-9353, ext. 275; FAX 305-443- your results. Most often, this is what the readers 7404; or write to the American Welding Society, 550 remember. NW LeJeune Rd., Miami, FL 33126. 6) Acknowledgment, References and Appendix. RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT

WELDING RESEARCH SUPPLEMENT | 65-s