Probing Element Transfer Behavior during the Submerged Arc Welding Process for CaF₂-SiO₂-Na₂O-Cr₂O₃ Agglomerated Fluxes: A Thermodynamic Approach

Abstract

Submerged arc welding joins metal by the heating of the electrode, base metal, and flux in the arc plasma, while the weld pool is protected under the granular flux and molten slag. Due to complex chemical reactions occurring between the arc plasma, weld pool, and molten slag (flux), flux essentially affects the weld metal composition, which, in turn, dictates the mechanical properties of the weldment. Therefore, fine-tuning the weld metal composition is essential to ensure a sound weld, and efforts worldwide have been focused on the control mechanism of flux on the weld metal composition. Recently, agglomerated fluxes have been widely applied due to low energy consumption during manufacture. The Cr₂O₃-bearing agglomerated flux is one of the most commonly used flux types in fields of heavy industrial applications. However, few works concern the element transfer behavior when Cr₂O₃-bearing agglomerated fluxes are used. Within this framework, typical agglomerated CaF₂-SiO₂-Na₂O-Cr₂O₃ fluxes with varying Cr₂O₃ content from 10 to 50 wt.% are designed and applied to Q345A steel. The influence of Cr₂O₃ content upon the transfer behaviors of essential elements, including O, Cr, and Mn, is quantified and interpreted from the point of thermodynamics. By incorporating a gas-slag-metal equilibrium consideration, the assumptions made in previous studies are justified. Additionally, evidence regarding the loss of Cr and Mn to the arc plasma is provided, and a possible thermodynamic approach to predict element transfer levels is proposed. It is revealed that the gas-slag-metal equilibrium consideration is able to qualitatively analyze the transfer behaviors involved in the submerged arc welding system, even under high temperatures. Based on the quantitative data, the practical implications as well as limitations of the gas-slag-metal equilibrium model are proposed.

1. Introduction

Submerged arc welding (SAW) is an extremely versatile automatic arc welding method widely applied in fields of heavy industrial applications, owing to its high deposition rate and excellent reliability [1]. Since the control of WM compositions is important for producing the weldment of sound quality, an understanding of the mechanisms that alter the weld metal (WM) compositions would be a primary aid in such control [2,3,4,5,6].

Flux plays a complex role during the SAW process. In addition to stabilizing arc plasma, providing slag, and promoting slag detachability, flux essentially affects WM compositions due to the chemical reactions occurring between the flux (slag), weld pool, and arc plasma. Therefore, to better control WM compositions, an in-depth understanding of the element transfer behavior in SAW is necessary [7,8]. In SAW engineering, a Δ value is adopted to quantify the element transfer between flux and WM [9,10]. The Δ value may be positive or negative depending upon the element transfer direction; that is, a positive Δ value indicates an elemental gain from the flux, whereas a negative Δ value indicates an elemental loss from the WM to the slag [11,12,13,14].

Understanding the mechanisms in terms of the element transfer between flux and WM can be attained by investigating the influence of a flux chemical additive on the quantified element transfer, viz. the Δ value. For instance, Zhang et al. [11,12,14] designed a series of CaF₂-based fused fluxes with varying oxide addition levels and evaluated the impact of SiO₂, MnO, and TiO₂ on the element transfer between flux and WM. Dallam et al. [7] documented the influence of CaF₂, CaO, and FeO addition on the Δ values of O, Si, and Mn when manganese-silicate fluxes were used. Kanjilal et al. [15], in contrast, studied the effect of CaF₂ and NiO addition on the ΔMn value for CaF₂-CaO-SiO-Al₂O₃-based agglomerated fluxes.

However, the mechanisms responsible for the element transfer behavior in SAW due to Cr₂O₃ addition into fluxes have not been clearly described. One typical thermodynamic investigation was conducted by Mitra et al. [16]; nonetheless, within the above work, the levels of Cr₂O₃ in fluxes were confined in a narrow range from 10 to 18 wt.%.

Cr₂O₃-bearing fluxes are widely applied when heat-resistant steel is submerged-arc-welded. The C₂O₃ in the flux exerts a significant impact on the contents of O and Cr of the submerged-arc-welded metal, which, in turn, dictates the mechanical properties of the weld. Therefore, a deep understanding of the transfer behavior in SAW when Cr₂O₃-bearing fluxes are applied is essential [17,18,19,20].

The objective of this study is to investigate the influence of Cr₂O₃ on the element transfer behavior from the point of thermodynamics over a wide compositional range of Cr₂O₃ in fluxes. By using agglomerated fluxes, the role of Cr₂O₃ in the determination of WM final compositions is qualified. By incorporating a gas-slag-metal equilibrium consideration, the quantified Δ values are interpreted and several assumptions regarding the transfer of elements are justified thermodynamically. Additionally, evidence regarding the loss of Cr and Mn to the arc plasma is provided by using measured data, and a possible thermodynamic approach to predict the Δ values for Cr₂O₃-bearing fluxes is suggested.

2. Materials and Methods

2.1. Flux Preparation

For each flux recipe, 1 kg of reagent-grade powders were weighted according to the formulas given in

Table 1. Formulas of initial fluxes (wt.%).

Flux	Cr2O3	CaF2
F-1	10	90
F-2	20	80
F-3	30	70
F-4	40	60
F-5	50	50

. In addition to Cr₂O₃, CaF₂ (the non-oxide with no O potential) was added as a diluent to lower the melting temperature of the flux [21]. The powders were mixed in a V blender at 0.5 Hz for an hour. Then, the powders were bonded by a 150 g sodium-silicate solution; as such, SiO₂ and Na₂O are incorporated since they help improve the slag detachability and arc stability [10,14]. Subsequently, the bonded mixtures were pelletized and dried in a muffle furnace at 973 K for 3 hours. At last, the mixtures were broken up and screened to a 14 to 100 mesh [9]. In this study, “flux” means the starting material before SAW, while “slag” implies the molten or solidified flux during or after SAW [8,22].

2.2. Welding Experiment

A typical low alloy grade steel, Q345A, was selected as the base metal (BM). Bead-on-plate double-electrode single-pass SAW (Lincoln Electric Power Wave AC/DC 1000 SD, Lincoln Electric, Cleveland, OH, USA) was performed at a heat input of 60 kJ/cm (DC-850 A/32 V for electrode forward, AC-625 A/36 V for electrode backward, 500 mm/min).

2.3. Chemical Composition Analysis

X-ray fluorescence (XRF, model S4 Explorer, Bruker, Germany) was used to determine the compositions of fluxes and slags. The content of F was determined by a titration method. The analytical compositions of fluxes and slags are given in

Table 2. Measured compositions of fluxes (wt.%).

Flux	Cr2O3	SiO2	Na2O	CaF2	BI
F-1	12.58	7.99	0.52	78.91	5.56
F-2	22.89	7.56	0.48	69.07	3.66
F-3	32.88	7.63	0.55	58.94	2.47
F-4	41.99	7.88	0.61	49.52	1.74
F-5	53.19	7.66	0.49	38.66	1.14

and

Table 3. Measured compositions of slags (wt.%).

Flux	Cr2O3	SiO2	Na2O	FeO	MnO	Al2O3	TiO2	CaF2
F-1	5.63	8.1	0.25	4.65	3.54	0.18	0.11	76.75
F-2	11.25	8.52	0.21	7.26	4.25	0.17	0.10	67.41
F-3	18.11	8.22	0.28	10.37	5.50	0.20	0.10	56.46
F-4	24.72	8.58	0.21	12.41	5.30	0.21	0.13	47.41
F-5	32.45	10.11	0.14	13.91	6.66	0.2	0.11	35.63

Only components > 0.1 wt.% were taken into account.

, where the basicity index (BI) of each flux is determined by Equation (1) [23].

$$BI=\frac{{\mathrm{CaO}+\mathrm{CaF}}_2{+\mathrm{MgO}+\mathrm{Na}}_2{\mathrm{O}+\mathrm{K}}_2\mathrm{O}+0.5\times (\mathrm{MnO}+\mathrm{FeO})}{{\mathrm{SiO}}_2{+1/2(\mathrm{Al}}_2{\mathrm{O}}_3{+\mathrm{Cr}}_2{\mathrm{O}}_3{+\mathrm{TiO}}_2{+\mathrm{ZrO}}_2)}$$

(1)

Inductively coupled plasma optical emission spectrometry (ICP-OES, Perkin Elmer) was used to determine the compositions of metallic elements, while the LECO analyzer was used to determine the contents of O. The compositions of the electrode and BM are given in

Table 4. Measured chemical compositions of the BM and electrode (wt.%).

	C	Si	Mn	Ti	Cr	O
Q345A	0.112	0.142	1.540	0.015	0.018	0.003
Electrode	0.127	0.049	1.650	0.015	0.015	0.003

.

The Δ values for O, Cr, and Mn are quantified from Equation (2).

$$\Delta =M_{\mathrm{WM}}-M_{\mathrm{N}}$$

(2)

In Equation (2), M_WM is the measured WM composition, while M_N is the nominal composition (the composition considering only the dilution effect of the BM and electrode) [9,12,14]. The M_N value is determined from Equation (3), where M_BM represents the measured composition of the BM, M_el represents the measured composition of electrode, and d represents the dilution value of the BM.

$$M_{\mathrm{N}}=M_{\mathrm{BM}}\times d+M_{\mathrm{el}}\times (1-d)$$

(3)

To determine the value of d, the weldments were cross-sectioned, polished, and etched by a 4 wt.% nital solution. Then, the value of d was calculated from Equation (4).

$$d=\frac{\mathrm{Area} \mathrm{of} \mathrm{fused} \mathrm{base} \mathrm{metal}}{\mathrm{Area} \mathrm{of} \mathrm{weld} \mathrm{metal}}$$

(4)

The data of nominal compositions, measured WM compositions, and Δ values for O, Cr, and Mn are summarized in

Table 5. Nominal compositions, measured WM compositions, and quantified Δ values (wt.%).

Weld Metal	WM-1	WM-2	WM-3	WM-4	WM-5
Flux	F-1	F-2	F-3	F-4	F-5
(O)A	0.049	0.074	0.118	0.147	0.162
(O)N	0.003	0.003	0.003	0.003	0.003
ΔO	0.046	0.071	0.115	0.144	0.159
(Cr)A	1.050	1.620	1.800	2.080	2.620
(Cr)N	0.016	0.017	0.017	0.017	0.016
ΔCr	1.034	1.603	1.783	2.063	2.604
(Mn)A	0.710	0.510	0.310	0.260	0.190
(Mn)N	1.601	1.593	1.586	1.591	1.616
ΔMn	−0.891	−1.083	−1.276	−1.331	−1.426

.

2.4. Thermodynamic Calculation

Due to the limited understanding of the SAW process, researchers often used the “effective equilibrium temperature” to perform a thermodynamic equilibrium calculation [22,24,25]. The effective equilibrium temperature does not imply the measured equilibrium temperature of SAW but indicates the temperature at which the experimental mass action index equals the equilibrium constant. Within this framework, the effective equilibrium temperature was set to 1973 and 2273 K, which was concluded by Mitra et al. [16] when Cr₂O₃-bearing fluxes were applied.

Recently, interactions at the gas-slag-metal interface in SAW have been documented [1,26]. Major phases and reaction interfaces associated with gas phases are given in Figure 1a; the blue point in Figure 1c indicates the plasm(gas)-slag-gas interface, while the green point in Figure 1b indicates the bubble(gas)-slag-gas interface [26]. As was concluded in our previous study, the nucleation and release of bubbles are closely related to complex factors, although the mechanisms are not fully understood; nonetheless, the existence of arc plasma guarantees the existence of a bubble(gas)-slag-gas interface [26].

There are two issues complicating the investigation over the development of the compositional prediction model for SAW [27]:

• It is impossible to capture the gases in the arc plasma and to sample the molten slag for analytical purposes since the arc plasma, molten slag, and weld pool are shielded under the flux during the SAW process.
• The effective equilibrium temperature of the SAW chemical system is as high as 2000 °C, under which the thermodynamic information remains scarce.

Hopefully, a number of thermodynamic databases for gases, oxides, and alloy systems have been developed by the Calphad technique in recent decades [28,29]. Additionally, with the help of the applicable thermodynamic models (such as the cell model and the modified regular solution model), the thermodynamic data could be extended to a SAW temperature as high as 2273 K [30].

In this regard, gas-slag-metal equilibrium calculations were performed at temperatures of 1973 and 2273 K to aid in the discussion on the quantified element transfer data. This approach was used in our previous study to illustrate the transfer of Ti and O from TiO₂-bearing basic-fluoride fluxes to WMs subject to varying TiO₂ levels [31]. The details of the thermodynamic calculation are given in Appendix A.

3. Results and Discussion

3.1. Transfer of O

O is one of the most essential elements for submerged-arc-welded metal that must be carefully controlled [32]. It is accepted that excessive O in the WM tends to cause several problems, such as promoting porosity, reducing toughness, and decreasing hardenability [10,25]. However, a WM with too low O levels shows poor impact toughness since there are insufficient inclusions to promote the formation of acicular ferrite (AF) [33,34].

In comparison to other arc welding methods, one salient feature of the SAW process is the significant O uptake from the flux [10,35]. Chai et al. [21] assumed that under the presence of an arc plasma of high temperature, oxide in the flux tends to decompose into suboxides, releases O₂ gas in the arc cavity, and transfers O to the weld pool. Based on an observation from

Table A2. The summary of metal nominal compositions (wt.%).

	WM-1	WM-2	WM-3	WM-4	WM-5
(C)N	0.120	0.119	0.118	0.119	0.121
(Si)N	0.091	0.098	0.103	0.099	0.088
(Mn)N	1.600	1.592	1.586	1.591	1.604
(Ti)N	0.015	0.015	0.015	0.015	0.015
(Cr)N	0.016	0.017	0.017	0.017	0.016
(O)N	0.003	0.003	0.003	0.003	0.003

and

Table A3. Parts of the gas-slag-metal equilibrium calculation outputs under 1973 K for 100 g WM.

Weld Metal		WM-1	WM-2	WM-3	WM-4	WM-5
Flux		F-1	F-2	F-3	F-4	F-5
CrF3	Vol.%	3.824	4.800	5.354	6.322	7.620
MnF2		1.366	1.042	0.930	0.684	0.415
O2	10−10 atm.	4.691	7.198	7.230	7.248	7.562
CrO	Weight (g)	0.215	0.302	0.276	0.322	0.411
Cr2O3		0.018	0.032	0.027	0.025	0.026
FeO		0.266	0.320	0.298	0.329	0.373

, Cr₂O₃ and CrO are the primary Cr oxides in the slag. Therefore, based on the assumption made by Chai et al. [21], it is speculated that Cr₂O₃ tends to decompose into CrO and O₂ gas via Reaction (5) at the flux(slag)-plasma interface.

$${2(\mathrm{Cr}}_2{\mathrm{O}}_3{)=4(\mathrm{CrO})+\mathrm{O}}_2(\mathrm{g})$$

(5)

An empirical concept of the flux basicity index (BI) has been adopted to identify the flux O potential (the driving force for O transfer from the flux to the WM); generally, a higher BI value means a lower flux O potential, and, thus, a lower O content of the submerged-arc-welded metal [9,10]. Based on the experimental data, a general relationship between the weld metal O content and the flux BI is summarized in Figure A1, based on which the O content can be predicted from the flux BI [23].

Another prediction approach was developed by Zhang et al. [31], in which the O level was predicted from the gas-slag-metal equilibrium calculations; this approach was based on the assumed local attained gas-slag-metal thermodynamic equilibrium involved in the SAW process. The predicted data from the BI model and the gas-slag-metal equilibrium model for O contents are given in Figure 2.

Based on an observation from Figure 2, both the flux BI and gas-slag-metal equilibrium models are capable of predicting the upward changing trend of ΔO values with a higher addition level of Cr₂O₃ in the flux (see the green and blue dot-dash lines in Figure 2). Specially, as shown by the blue-shaded area in Figure 2, the measured ΔO values lay in the ranges calculated by the gas-slag-metal equilibrium calculations at 1973 and 2273 K. Therefore, the consideration of the gas-slag-metal equilibrium is able to place limits on the transfer of O between flux and WM.

Additionally, as was assumed by Lau et al. [36] and Mitra et al. [37], the O uptake from the flux is governed by the level of O₂ gas pressure in the arc plasma. To verify this assumption, the gas-slag-metal equilibrium O₂ gas pressure was calculated and is provided in Table A3 and

Table A4. Parts of the gas-slag-metal equilibrium calculation outputs under 2273 K for WM of 100 g.

Weld Metal		WM-1	WM-2	WM-3	WM-4	WM-5
Flux		F-1	F-2	F-3	F-4	F-5
CrF3	Vol.%	8.269	11.990	11.883	11.710	10.479
MnF2		5.770	4.291	3.309	2.342	1.489
O2	10−8 atm.	3.520	8.240	10.420	11.400	12.280
CrO	Weight (g)	0.155	0.548	0.996	1.947	1.010
Cr2O3		0.008	0.065	0.165	0.374	0.233
FeO		0.301	0.679	1.070	1.837	0.967

, from which it is seen that the gas-slag-metal equilibrium O₂ gas pressure generally increases with a higher level of Cr₂O₃ addition in the flux.

Another parameter to identify the flux O potential is the FeO uptake in the slag [16,37,38]. Considering Reaction (6) occurring at the slag-metal interface, Zhang et al. [11,12,14] assumed that the FeO level in the slag is proportional to the WM O content since a higher flux O potential tends to drive Reaction (6) to the right side, resulting in a higher level of FeO uptake in the slag. This assumption is confirmed by the increasing FeO content in the slag with Cr₂O₃ addition (corresponding to a higher flux O potential), as shown by the red dash line in Figure 3.

$$\left[\mathrm{Fe}\right]+\left[\mathrm{O}\right]=\left(\mathrm{FeO}\right)$$

(6)

However, it is noted that the measured FeO level was higher than the gas-slag-metal equilibrium ones, as shown in Figure 3. This phenomenon can be explained by the assumption proposed by Mitra et al. [38], that is, Reaction (6) only proceeds forward, and the equilibrium of Reaction (6) is not attained. The measured FeO level in this work may provide the evidence proposed by Mitra et al. [38] regarding the transfer of Fe in previous studies.

3.2. Transfer of Cr

Figure 4 illustrates the levels of the actual and predicted ΔCr values as a function of the Cr₂O₃ addition level in the flux. As shown by the blue-shaded area in Figure 4, the incumbent approach is capable of placing a limit on the values of ΔCr, except at 12.48 wt.% Cr₂O₃. Mitra et al. [16] assumed that the transfer of Cr between Cr₂O₃-bearing fluxes and submerged arc welded metals governed by Reaction (7) at the slag-metal interface; to check whether such assumption is feasible for this study, the activities of Cr oxides calculated from the gas-slag-metal equilibrium model are plotted in Figure 4 to aid in the analysis [31,39].

$${(\mathrm{Cr}}_2{\mathrm{O}}_3)=2[\mathrm{Cr}]+3[\mathrm{O}]$$

(7)

It seems that the sole consideration of Reaction (7) is insufficient to explain the transfer behavior of Cr for this case study. As was demonstrated in our previous study, the transfer between the suboxide and the weld pool, such as Reaction (8) at the slag-metal interface, should be considered in terms of transfer for an alloy element [31].

$$\left(\mathrm{CrO}\right)=\left[\mathrm{Cr}\right]+\left[\mathrm{O}\right]$$

(8)

As shown in Figure 5b, the activities of CrO in the molten slag were much higher than those of Cr₂O₃. Considering the conclusion that the level of element transfer is essentially controlled by the oxide activity in the slag, it is speculated that Reaction (8) should be considered regarding the transfer of Cr, which is rather remarkable and different from the previous assumption raised by Mitra et al. [14,16,31]. As such, it is speculated that the improvement of both Cr₂O₃ and CrO activity drives Reaction (8) to the right side, leading to a higher level of ΔCr with Cr₂O₃ addition to the flux, even at a higher flux O potential.

Additionally, it is well known that the loss of Cr tends to occur in SAW. Such loss is confirmed by the mass balance calculation given in Appendix B. Glasser et al. [40] assumed that Cr is lost from the silicate mixtures in the form of trivalent Cr. Based on an observation from Table A3 and Table A4, CrF₃ is the primary Cr-contained gas generated in the SAW process. Therefore, the gas-slag-metal equilibrium model may justify the assumption raised by Glasser et al. [40].

3.3. Transfer of Mn

The levels of ΔMn calculated from the gas-slag-metal equilibrium model are plotted in Figure 6. It is seen that the gas-slag-metal equilibrium model is able to predict the transfer direction of Mn between the flux and the submerged-arc-welded metal, viz. the negative ΔMn values (see the blue circles and triangles in Figure 6). Especially, the ΔMn values are constrained by using the gas-slag-metal equilibrium model, as shown by the red squares and the blue-shaded area in Figure 6. Similar to the case of Cr transfer, Mn tends to be lost from the SAW system via gas formation, which is confirmed by the mass balance calculation in Appendix B.

Mn is an essential element for a WM. Mn is an AF-promoting agent, and its level, ranging from 0.6 to 1.8 wt.%, could increase the AF fraction and depress the formation of polygonal and side plate ferrites [26]. Therefore, an electrode and/or a BM with higher Mn levels is recommended to match the Cr₂O₃-bearing flux, thereby compensating the possible Mn loss incurred by oxidation reactions.

4. Conclusions

Within this framework, the transfer behaviors of O, Cr, and Mn involved in SAW were quantified and evaluated from a thermodynamic perspective when CaF₂-SiO₂-Na₂O-Cr₂O₃ system fluxes were used. The following conclusions can be drawn:

• By performing a gas-slag-metal thermodynamic equilibrium calculation, the transfer direction of O, Cr, and Mn can be detected and the level of element transfer (ΔO, ΔCr, and ΔMn) can be constrained, which may pave a viable way for the prediction of element transfer behaviors when Cr₂O₃-bearing fluxes are applied.
• The measured slag compositions, coupled with thermodynamic data, demonstrated that the thermodynamic equilibrium for Fe transfer is not achieved.
• The loss of Mn from the weld pool is enhanced due to a higher level of Cr₂O₃ addition to the flux. The evidence regarding the loss of Cr and Mn from the SAW system to the gas phase was provided; such loss is predictable by using a gas-slag-metal equilibrium model.
• An electrode and/or a BM with higher Mn levels is recommended to match the Cr₂O₃-bearing flux, thereby compensating the possible Mn loss incurred by oxidation reactions.
• The content of Cr₂O₃ in the flux should be controlled under 30 wt.% since a WM with an O level higher than 1000 ppm may incur unexpected issues, such as enhanced porosity, reduced toughness, and depreciated hardenability.
• With a higher level of Cr₂O₃ addition to the flux, the Cr contents in the electrode and the BM should be restricted to avoid redundant Cr uptake from the flux.

Nonetheless, the measurement of thermodynamic data at ultra-high temperature is still a technical problem. Therefore, the author suggests that the limitations of this work are that:

• Since the high-temperature thermodynamic data are extended from the model, there must be an error with the real and predicted data.
• FactSage only considers the assumed thermodynamic equilibrium involved in SAW. However, kinetic factors should also be considered to improve the overall accuracy.
• More work on SAW experiments is required to determine the optimal calculation temperature.