Sham 8-14_Layout 1 7/15/14 1:22 PM Page 271

WELDING RESEARCH
SUPPLEMENT TO THE WELDING JOURNAL, AUGUST 2014
Sponsored by the American Welding Society and the Welding Research Council

Flux­Coating Development for SMAW Consumable

Electrode of High­Nickel Alloys
Extrudability, recovery analysis, and welding characteristics were incorporated into a
shielded metal arc welding electrode for welding high­nickel alloys

BY K. SHAM AND S. LIU

coating include providing a slag layer

ABSTRACT for a smoother weld contour and bead
shape, and surface cleanliness of the
There is interest in improved high­nickel alloy electrodes for shielded metal arc weld metal. Additionally, the flux coat-
welding (SMAW) to replace gas tungsten arc welding (GTAW) with cold wire feed for ing should be formulated with ingredi-
pressure vessel fabrication/repair in the power­generation industry. Without altering ents that produce a low-density slag,
the GTAW wire composition, a proper flux coating was designed to produce a SMAW
provide proper viscosity for out-of-
electrode to give an enhanced weld composition. SMAW electrodes were designed by
developing proper flux coatings for extrusion, welding characteristics, and weld position welding, promote slag detach-
integrity. The flux formulation design began with an equal distribution of three primary ability, and reduce spatter and fume.
ingredients: cryolite, rutile, and calcium carbonate to clean the weld, create a slag, and Fluxes can be separated into acidic,
provide shielding gas, respectively. The system was then optimized to enhance extrud­ basic, and neutral depending on which
ability, slag detachability, electrode weldability, alloying recovery, arc stability, and a oxides are dominant as flux ingredi-
number of weld properties. Detailed mass balance calculations were performed to de­ ents. The combination of these oxides
termine alloying element transfer across the arc and to understand weld metal recov­ determines the slag properties such as
ery from the flux coating, filler metal wire, and base metal. For chromium and melting temperature, thermal expan-
manganese, an average of 95 and 60% weld recovery was confirmed and determined, sion, density, electrical conductivity,
respectively. The elemental recovery data determined were important for calculating
and viscosity. Acidic oxides such as sili-
the proper additions of these elements into the flux formulation. Comparisons revealed
comparable welding characteristics for the in­house manufactured and commercial cates are primarily slag or network for-
SMAW electrodes. Arc voltage analysis was also performed to determine arc stability of mers. Basic oxides (CaO, MgO, Na2O
these electrodes. Results indicated differences in metal transfer modes between the and K2O) decrease viscosity since they
different flux coating iterations, and the final electrodes exhibited more consistent volt­ break down silica networks. Fluorides
age­time traces. This work demonstrates that even though it is common practice to use and chlorides are also added to se-
a GTAW wire to build a SMAW electrode, the flux coating can be designed to lected flux compositions as aggressive
beneficially alter the weld composition. chemicals that assist in removing
strong oxides from the weld joints. Ad-
ditions of fluorides decrease the vis-
KEYWORDS cosity of the slag, depending on the
•Covered Electrode • SMAW Consumable • Nickel Alloys • Flux Coating
type of fluoride. Slipping agents and
•Extrusion • Electrode Manufacture • Composition
binding agents are added to improve
the flow of the wet flux mixture dur-
predefined target weld metal composi- ing extrusion and bind the flux ingre-
Introduction dients to each other and to the core
tion. This work presents a design
methodology for developing a flux rod to form a solid coating. Sodium sil-
With the development of a weld coating formulation to convert a icate, potassium silicate, or mixtures
composition that demonstrates abili- GTAW wire into a SMAW consumable of the two are common binders used
ties to mitigate ductility dip cracking electrode for high-nickel alloys. to manufacture electrodes.
(DDC) and hot cracking in high-nickel The primary functions of the flux Welding arc stability is important
alloys while maintaining their high coating of an SMAW consumable elec- to the weldability of the electrode and
temperature and corrosion resistance, trode are to provide arc stability, pro- the integrity of the weld properties.
an existing GTAW wire was used as an tect the molten metal from the Some ingredients used to enhance arc
SMAW core rod with a flux coating de- atmosphere, and refine the weld pool stability are potassium silicate, sodium
signed specifically to meet the new (Refs. 1, 2). Other functions of a flux silicate, rutile (TiO2), and potassium

K. SHAM and S. LIU are with Colorado School of Mines Golden, Colo.

AUGUST 2014 / WELDING JOURNAL 271-s

Sham 8-14_Layout 1 7/14/14 4:39 PM Page 272

WELDING RESEARCH

Fig. 2 — Experimental GTAW wire de­coiled, straightened, and

cut into 356­mm (14­in.) lengths.

Fig. 1 — Experimental GTAW wire in coil form.

the flux coating con-

Table 1 — Commercial High­Nickel Core Rod tains fluorides. Slags
Compositions that contain the
spinel (AO.B2O3)
Element E­1 (wt­%) E­2 (wt­%) where A and B repre-
Carbon 0.006 0.028 sent cations such as
Silicon 0.05 0.16 Mg and Al and
Chromium 16.8 27.3 cordierite
Manganese 2.5 0.37 (Mg2Al4Si5O18) glass
Niobium 0.06 0.02
Iron 8.75 10
structure have also Fig. 3 — Schematic illustration of the extruder, top view (Ref. 14).
Nickel 71.3 60.4 been deemed diffi-
cult to detach from Another important aspect of the
the weld metal. An increase in alumina flux coating of the SMAW consumable
titanate (Ref. 3). Even though potas-
content in the flux composition has electrode is that it can be utilized as a
sium silicate acts predominately as the
shown improved slag detachability. conduit for alloying the weld metal
binder in the flux coating, it plays a
Another important factor to consider composition. The flux coating ratio de-
critical role in the stabilization of the
for slag detachability is the difference fined as the diameter of the core rod
arc in AC applications because potas-
in coefficient of thermal expansion divided by the diameter of the SMAW
sium ionizes at a lower energy, 4.3
(CTE) between the slag and the weld electrode affects the amount of alloy-
eV/atom. Sodium ionizes at 5.1
metal deposit (Refs. 5–7). It is desir- ing elements recovered in the weld
eV/atom and is therefore more readily
able to have a slag with a CTE different metal (Refs. 9–11). There are three
used as the binder for arc stabilizer of
from that of the weld metal. Too early main components that contribute to
DC applications where the ability to
slag detachment, however, may pres- the end weld composition: the core
reestablish the welding arc is not as
ent undesired heavy temper colors rod, the base metal, and the flux coat-
vital (Ref. 4). The addition of fluorides
(Ref. 1). To further explain the ther- ing. Thus, it is important to obtain
typically produces a more erratic arc,
mal expansion mismatch, Equation 1 chemical compositions for all three
despite its ability to remove surface
describes the strain that is produced components in order to calculate re-
oxides from the joint region.
due to a temperature change at the covery directly from the flux coating.
Slag detachability is of great con-
bonding interface between the slag Delta quantity is described as the
cern because the presence of slag on
and the weld metal. amount of a specific element that is
the weld surface will influence the in-
tegrity of multipass weldments and gained or lost during the welding
S = Δα(ΔT) (1)
affect process economics. Rutile-con- process. The delta quantity is repre-
taining fluxes generally produce slags S is defined as the strain, the Δα is the sented by the following mathematical
that exhibit good detachability. On the difference between the thermal expan- equation:
other hand, basic flux systems typi- sion of the slag and the weld metal, and
cally exhibit less than desirable wet- (ΔQuantity)i = (Analytical
ΔT, known as the differential contrac- Content)i – (Estimated Content)i
ting. Consequently, complete slag tion, is defined as the temperature dif-
removal may be more difficult in these (2)
ference at the bonding interface during The analytical content of element i is
systems. Slag removal has also been the welding thermal cycle (Ref. 8).
documented as more difficult when determined by chemical analysis of the

272-s WELDING JOURNAL / AUGUST 2014, VOL. 93

Sham 8-14_Layout 1 7/14/14 4:39 PM Page 273

WELDING RESEARCH

Fig. 4 — Pictorial depiction of di­ Fig. 5 — Schematic of SMAW. Fig. 6 — A relationship between the correct
ameter of filler metal rod and di­ amount of binder added and the percentage
ameter of electrode. of relative humidity in the working area.

weld metal, whereas the estimated lowing techniques were all used to- (E-2) used for comparison in this re-
content is determined by adding the gether: X-ray diffraction (XRD), energy search. Due to the sensitive nature of
contributions from the core rod and dispersive X-ray spectroscopy (EDS), this work, the exact composition of
base metal. A negative delta quantity wavelength-dispersive X-ray spec- the experimental wire is not revealed.
gives a loss of the specific element ei- troscopy (WDS), atomic absorption The experimental flux matrix pre-
ther to the slag or in a gaseous form. A (AA), and inductively coupled plasma sented in Table 2 displays all the major
positive delta quantity implies flux optimal emission spectroscopy (ICP- heats and changes to reach a final flux
contribution of a specific element, OES). The combination of chemical composition titled, “CSM Final.” An
having been transferred to the weld analysis techniques are used together to initial formulation of a 1:1:1 mass
metal to achieve a certain target weld determine both elemental composition ratio of cryolite (Na2AlF6), rutile, and
composition (Refs. 12, 13). The delta and crystalline phases to determine calcium carbonate (CaCO3) was taken
quantity concept plays an important possible mineral ingredients used to for- as a starting point in the formulation
part in understanding the role of the mulate the flux coatings. process. The descriptors “decrease”
flux ingredients in their contribution Experimental wire of 3.2-mm (1⁄8- and “increase” represent decreasing or
to the final weld composition. in.) diameter was procured in the form increasing quantities of an ingredient
of a large coil (Fig. 1) that had to be from the previous heat to promote no-
Experimental Procedures decoiled, straightened, and cut into ticeable variations, in particular, slag
356-mm (14-in.) lengths — Fig. 2. properties or weld behavior. The ma-
Flux coating chemistries for com- Table 1 gives the core rod composi- trix was designed to address the fol-
mercial consumables were determined tions of the commercial SMAW elec- lowing properties: extrudability, flux
using a variety of techniques. The fol- trodes ENiCrFe-3 (E-1) and ENiCrFe-7 coating finish, slag viscosity, slag de-

Table 2 — Experimental Flux Matrix for CSM High­Nickel SMAW Electrode Development

Flux Heat 1 Heat 1A Heat 2 Heat 3 Heat 4 Heat 5 Heat 6 CSM Final
Ingredient (wt­%)
Cryolite 21.6 Decrease Increase Decrease Increase Increase – –
Rutile 21.6 Decrease Increase – – – – Increase
Calcium Carbonate 21.6 Increase Decrease – – – – –
Silicon Dioxide n/a – – – – Increase – –
Talc n/a – Increase – Decrease Decrease – –
Kaolin Clay n/a Increase – Increase – Dercrease – –
Bentonite Clay n/a – – – – – Increase Decrease
Cellulose n/a – – – – – – Increase
Alginate n/a – – – – – – Increase
Ferroniobium n/a – – – – – Increase –
Manganese n/a – – – – – Increase –
Chromium n/a – – – – – Increase –
Binder Bal. Bal. Bal. Bal. Bal. Bal. Bal. Bal.

AUGUST 2014 / WELDING JOURNAL 273-s

Sham 8-14_Layout 1 7/14/14 4:39 PM Page 274

WELDING RESEARCH

Fig. 8 — Bead­on­plate weld produced with Heat 2 flux composition. Part of

the slag was retained for illustration.

Fig. 9 — Bead­on­plate weld produced with Heat 3 flux formulation with

Fig. 7 — Particle size distribution for coarse, fine, and
decreased cryolite content.
50% coarse and 50% fine combined for rutile.

an average flux coating ratio of 0.42.

Table 3 — Welding Parameters for E­1, E­2, and CSM Final Calipers and image analysis were used
to verify results. After drying in air for
Electrode E­1 E­2 CSM Final 24 h following extrusion, the elec-
Electrode Feed Rate (mm/s) 4.9 4.9 5.1 trodes were then baked in an oven at
Travel Speed (mm/s) 2.1 2.1 1.7 175°C (350°F) for one h, 260°C
Deposition Rate (g/s) 0.57 0.62 0.45 (500°F) for three h, and then 175°C
Welding Time (s) 36 43 42 (350°F) for one h. The drying sequence
was consistent with industrial practice
and selected after input from commer-
tachability, effect of cryolite on weld particle size distribution for the dry cial electrode manufacturers (Ref. 11).
cleanliness, welding characteristics, mix. One major manufacturing aspect
arc stability, alloying additions, and of concern is electrode eccentricity, Welding Process
SMAW electrode mass production. which affects the weldability of the
SMAW consumable electrode. Elec- Using ASTM A36 base plates and
Manufacturing SMAW trode eccentricity is measured by the Inconel® 690 base plates for the final
Consumable Electrodes relative position of the core rod with welds, bead-on-plate welds were de-
respect to the layer of flux coating. An posited using a Hobart TIG-300 con-
eccentric electrode with uneven thick- stant current power supply. Using
For the SMAW consumable elec- ness coating surrounding the core rod
trode extruder, a total of 2 to 3 kg (ap- direct current electrode positive
will produce an erratic arc with uneven (DCEP) configuration, an average lin-
proximately 4 to 6 lb) of the wet mix is melting of the core rod and preferen-
needed to make a successful extrusion ear heat input of 12 kJ/mm (0.5
tial flaking of the flux coating. kJ/in.) was obtained. Based on the
run. The extruder used for the experi- Another important aspect of extru-
mental small batch production for this manufacturer-recommended welding
sion is determining the appropriate ranges for E-1 (90–110 A) and E-2
work is capable of pressures ranging thickness of the flux coating, which is
from 500 to 1000 lb/in.2 (3.5 to 6.9 (75–100 A), the respective averages of
adjustable during manufacturing. In- 100 A and 88 A were adopted for weld-
MPa) as compared to a commercial ex-
appropriate flux coating thickness can ing for the 3.2-mm- (1⁄8-in.-) diameter
truder averaging around 5000 to 6000
affect the integrity of the weld and the commercial electrodes. The optimal
lb/in.2 (35 to 40 MPa). Figure 3 gives a
composition of the weld metal. The welding current for the in-house man-
detailed top-view schematic of how
thickness of the flux coating is meas- ufactured electrode was subsequently
the extruder functions during extru-
ured as a flux coating ratio defined as determined to be 85 to 95 A. Table 3
sion. A successful extrusion is depend-
the ratio between the electrode diame- gives the welding parameters for E-1,
ent on many factors such as
ter and the diameter of the core rod — E-2, and the “CSM Final.” Arc signals
experience with the extruder, success-
Fig. 4. Commercial electrodes showed were acquired during the welding
ful adjustment of the eccentricity of
an average flux coating ratio of 0.4. process for the bead-on-plate samples
the electrodes, proper preparation of
Thus, a die size comparable to 0.4 was for both the commercial and in-house
the core rods, adding the right amount
used for the manufacturing of the in- manufactured electrodes. These volt-
of binder and slipping agent, adequate
house manufactured electrodes giving age signals were obtained using a data
flux coating ratio, and having a good

274-s WELDING JOURNAL / AUGUST 2014, VOL. 93

Sham 8-14_Layout 1 7/14/14 4:39 PM Page 275

WELDING RESEARCH

Fig. 10 — Bead­on­plate produced with CSM Final flux formula­

tion with cryolite adjusted to 24 wt­%.

acquisition system with a sampling quence. Figure 5

rate of 1000 Hz and LabView® 7.1. displays all as-
These digital voltage signals were pects of SMAW
recorded for the duration of every to assist in the
weld, which ranged from 35 to 50 s. understanding
of the recovery
⎛ ⎛ RE ( in ) ⎞ ⎞ calculations
AFMC ( G ) = ⎜ 1 − ⎜ ⎟ ⎟ × AFM ( g ) that follow. Fig. 11 — Ternary phase diagram of CaO­TiO2­SiO2 (Ref. 15) with
⎝ ⎝ IE ( in ) ⎠ ⎠ compositions marked.
(3)
⎛ Zw ⎞
FM ( g ) = AFMC ( g ) WMRi = ⎜ ⎟ × 100%
Chemical Compositions and ⎝ f
Z + Z r + Z bm ⎠ i
⎛ OES − FM ( wt%) ⎞ (10)
Recovery Calculations ×⎜ ⎟⎠
⎝ 100 where WMR is weld metal recovery, Z
(4)
The chemical compositions of the represents the concentration of the el-
base metal, all the filler metal wires, ement (i) in discussion, w is the weld
and welds were analyzed using OES. AFCC ( g ) = metal, f is the flux coating, r is the
Filler metal wires were also melted ⎛ ⎛ RFC ( in ) ⎞ ⎞ filler metal core rod, and bm is the
into buttons to simulate the chemical ⎜ 1 − ⎜ IFC ( in ) ⎟ ⎟ × AFC ( g ) base metal.
⎝ ⎝ ⎠⎠ (5)
recovery of arc welding in an inert at-
mosphere. Alloying element recovery Results and Discussion
calculations were done with the meas- FC ( g ) = AFCC ( g )
ured compositions. Dilution calcula- ⎛ FC ( wt%) ⎞ The flux coatings of the two com-
tions were performed on all ×⎜
bead-on-plate welds in order to calcu- ⎝ 100 ⎟⎠ mercial nickel-alloy electrodes, E-1 and
(6) E-2, were examined as references for
late alloying element recovery in the comparison with the flux formulation
weld metal. Dilution percentage of a
bead-on-plate weld was calculated as
BMP ( g ) = AP cm 2( ) effort in this investigation. Table 4 pres-
ents the compositions of the E-1 and E-
37.1%. With these known contribu- 7.8 g
tions from the base metal and the × LW ( cm ) × 2 flux coatings. As can be seen, the
cc (7) three principal ingredients in the flux
filler metal, the weld composition can coating of commercial high-nickel
be determined numerically. Weld SMAW electrodes are cryolite, rutile,
metal recovery was determined for BM ( g ) = BMP ( g ) and calcite. E-1 contained a fourth
each of the major alloying elements: major ingredient, feldspar, which is gen-
niobium, manganese, and chromium. ⎛ OES − BM ( wt%) ⎞
×⎜ ⎟⎠
erally treated as filler material, added
A mass balance calculation was ⎝ 100 mainly for slag volume. A combination
conducted to determine the actual (8) of strontium and calcium carbonate was
amounts of the electrode consumed used in the commercial flux coatings be-
(both weight and length) to form the WM ( g ) = ( BMP ( g ) + WMD ( g )) cause strontium carbonate decomposes
weld. The amount of weld deposit at a higher temperature, 1100°C
(area of deposit and area of penetra- ⎛ OES − WM ( wt%) ⎞ (2012°F), than calcium carbonate and
×⎜ ⎟⎠
tion) was also determined using Im- ⎝ 100 (9) provides better coverage to the weld
ageJ® for the mass balance pool. An additional advantage of stron-
calculations. ImageJ is a public do- After completing calculations using tium carbonate is that it is not as hygro-
main image processing software de- Equations 3–9 for each element of in- scopic as calcium carbonate. Picking up
veloped at NIH (National Institute of terest, the following Equation 10 is less moisture, the electrode flux coating
Health). With all the data necessary, a used to determine the actual overall is expected to produce a more stable arc.
series of calculations using the equa- weld recovery: The E-2 flux coating contained a small
tions below was performed in se- amount of silica and some other alloy-

AUGUST 2014 / WELDING JOURNAL 275-s

Sham 8-14_Layout 1 7/14/14 4:39 PM Page 276

WELDING RESEARCH

Table 4 — E­1 and E­2 Flux Coating Composition

Mineral E­1 (wt­%) E­2 (wt­%)

Cryolite (Na3AIF6) 21 25
Rutile (TiO2) 14 17
Strontium Carbonate
(SrCO3) and Calcium 24 28
Carbonate (CaCO3)
Silicon Dioxide (SiO2) n/a 7
Feldspar
(K2O.Al2O3.6SiO2) 23 n/a
Fig. 12 — Self releasing slag systems, E­1 (top) and CSM Final (bottom). Manganese 13 12
Nickel 1 2
Iron n/a 3
Chromium n/a 3

between 34 and 55% with an average

relative humidity at 42%. The average
amount of binder added to the dry mix
to form the wet mix was 30 wt-% with
a standard deviation of less than or
equal to 2.5 wt-% depending on the
humidity. When the relative humidity
Fig. 13 — Heat 2 bead­on­plate exhibiting poor slag detachability. was above the average relative humid-
ity of 42%, the amount of binder was
less than 30 wt-% but within the devi-
ing elements for weld composition ad- were the amount of binder, the mixing ation of 2.5 wt-%. Figure 6 gives a
justment such as manganese, nickel, technique, the slipping agents, and the graphical representation of the rela-
and chromium. Due to the inert nature particle size distribution of the dry mix. tionship between binder amount and
of nickel, the value and importance of a The extrusion environment is also criti- relative humidity in the working area.
chemically aggressive flux ingredient cal. For example, relative humidity and The point with the highest relative hu-
cannot be overemphasized. Together temperature affect significantly the midity of 95% was data recorded dur-
with fluorides that are chemically ag- flow characteristics of the wet mix dur- ing the mass production of the
gressive for cleaning the weld joint, cry- ing extrusion. An extrudable flux can electrodes at a commercial electrode
olite prevents joint surfaces from quickly lose moisture and become diffi- manufacturer in Houston, Tex. It can
oxidation and improves the cleanliness cult to extrude during the time of man- be seen that the amount of binder
of the weld metal surface. Rutile prima- ufacturing one batch of electrodes. needed in making the wet mix de-
rily functions as a slag former and arc Therefore, this process is subjective and creases as the relative humidity of the
stabilizer. Calcium carbonate decom- based on experience in observing the work area increases with 23.5 wt-% of
poses around 890°C (1634°F) into CaO consistency of the wet mix, the fluidity binder being the lowest amount of
and CO2. CaO serves as a slag former by of the wet mix, and the adhesion be- binder needed.
reacting with TiO2 and SiO2. CO2 is the tween the core rod and the flux The poor extrudability of Heat 1 led
shielding gas that displaces the air from coating. to a decrease in binder and the addi-
the weld joint and blankets the entire Extrusion was performed on each tion of a slipping agent, kaolin
weld region. The performance of each of the experimental flux formulations, [Al2Si2O5(OH)4], to the flux paste to
flux formulation determined the including the base formula, to verify improve electrode extrusion. Heat 1A
amounts of addition or subtraction of the extrudability of the flux mixtures. had improved extrudability with too
each of the ingredients for the subse- As expected, the first flux formula- much slipping agent added; the flux
quent formulation. tion, Heat 1, was not extrudable be- paste flowed faster than the rod inter-
cause an excessive amount of the mittently even though the velocity of
Evaluation of Electrode binder (at ~35 wt-%) was utilized. the compressor was decreased to a
Extrudability Since the binder is typically 50% sili- minimum. For Heat 2, a combination
cate and 50% water, some data were of kaolin and talc, Mg3Si4O10(OH)2,
collected to determine if the amount was used in a decreased amount from
Extrusion of SMAW electrodes is not of binder added is dependent on the
a process that can be easily quantified. Heat 1A. These two slipping agents
humidity of the manufacturing envi- greatly improved extrudability of the
It depends on a number of variables in- ronment. The correct amount of
cluding the type and source of the min- base flux coating. For the latter extru-
binder is dependent on the relative hu- sion runs of the nickel SMAW consum-
eral, the appropriate amount of water, midity of the working area at the time
the particle size distribution, among able electrodes, including the larger
of mixing and extrusion. The relative batches, bentonite clay,
others. The extrusion variables studied humidity in the laboratory fluctuated [Al2O3(4SiO2H2O)] was used instead

276-s WELDING JOURNAL / AUGUST 2014, VOL. 93

Sham 8-14_Layout 1 7/14/14 4:39 PM Page 277

WELDING RESEARCH

Fig. 14 — Voltage data for E­1. Fig. 15 — Voltage data for E­2.

Fig. 17 — Voltage data for CSM Final.

technique, the correct

amount of binder, and an im- Table 5 — Flux Ingredients Broken Down into
Fig. 16 — Voltage data for Heat 2. the Slag Form
proved particle size distribu-
tion, “CSM Final” gave a
of kaolin clay due to what was avail- Flux Ingredients Primary Oxides
much improved extrusion experience. in Slags
able. Other than the difference in The dry mix of the CSM Final was opti-
chemical composition, both bentonite mized by using 50 wt-% fine grades and Na3AIF6 Na20, Al2O3

⇒
clay and kaolin clay, and talc function 50 wt-% coarse grades for the calcium TiO2 TiO2
as slipping agents to ease extrudabil- CaCO3 CaO
carbonate and rutile. Figure 7 depicts an Al2Si2O5(OH)4 Al2O3, SiO2
ity. Availability, proper particle size example of how the combination of the Mg3Si4O10(OH)2 MgO, SiO2
distribution, and desired weld compo- fine and coarse grades of rutile affected SiO2 SiO2
sition may be the deciding factors for the overall particle distribution. As K2SiO3.nH2O K2O, SiO2
selecting one and not the others. Fi- compared to Heat 1, which consisted of
nally, a small quantity of cellulose was larger quantities of 325 and 400 mesh
added to enhance homogeneous mix- size, “CSM Final” had lesser quantities cryolite, which produced weldments
ing of the wet mix during the commer- of 325 and 400 mesh size and displayed with good surface cleanliness. Surface
cial mass production. To maintain a more normal distribution around 200 cleanliness was inspected visually with a
extrudability and minimize water ab- mesh, which is the practice of industry shiny, silvery weld having good surface
sorption, a small amount of alginate formulators. The CSM Final electrodes cleanliness. Figure 8 depicts a bead-on-
was added as a slipping agent. have good extrudability and their flux plate weld produced with Heat 2 from
In Heat 5, the amount of slipping coatings have a smooth appearance. the experimental flux coating design
agent was further reduced and a better matrix displaying good surface cleanli-
particle size distribution for the dry mix ness. To test the effectiveness and opti-
was used to maintain extrudability. In
Weld Cleanliness mal amount of cryolite as a chemically
addition to the extrusion aspect, parti- aggressive flux ingredient to enhance
The first experimental Heat 1 of the the cleanliness of the weld metal, the
cle size distribution of the dry mix also
flux formulation used an even distribu- amount of cryolite was decreased from
affects the flux coating finish of a
tion of cryolite, rutile, and calcium car- 19 wt-% in Heat 2 to 13 wt-% in Heat 3.
SMAW electrode. Using the combina-
bonate as the starting point. Heat 1A As a result, the bead-on-plate weld from
tion of slipping agents, proper mixing
and Heat 2 contained 18–19 wt-% of

AUGUST 2014 / WELDING JOURNAL 277-s

Sham 8-14_Layout 1 7/14/14 4:39 PM Page 278

WELDING RESEARCH

Table 6 — Room Temperature CTE Values of Table 7 — Weld Recovery Results (wt­%)
Two Glass Phases and Inconel 600 and 690
Alloy (Refs. 15–18) Elements Heat 2 (%) Alloy 1 Heat 2 (%) Alloy 2 E­2 (%) E­1 (%)

Materials CTE at RT (10–6/°C) Chromium 100 100 100 88

Manganese 54 65 47 51
Sphene (CaTiSiO5) 6 Niobium 89 100 n/a* n/a*
Iron 100x 100 94 95
Perovskite (CaTiO3) 15.8
Nickel 83 99 100 100
Inconel 690 alloy 10.4
*Note — Niobium composition was not determined.

position of the flux coating. These val-

Table 8 — Known Alloy Recovery from SMAW Flux Coating (wt­%) (Ref. 5) ues are shown in Table 6. With the
compositions of the E-1 and E-2 flux
Alloy Element Form of Material in Electrode Covering Approximate Recovery of coatings, calculations were made to de-
Element, wt­% termine the respective crystalline
Aluminum Ferroaluminum 20 phases that could form in the resulting
Boron Ferroboron 2 slags. As can be seen in Fig. 10, the
Carbon Graphite 75 two commercial slags were determined
Chromium Ferrochromium 95
Niobium Ferroniobium 70 to produce the perovskite phase,
Copper Copper Metal 100 which has a larger CTE, 15.8 ×
Manganese Ferromanganese 75 10–6/°C, than that of the weld metal,
Molybdenum Ferromolybdenum 97 10.4 × 10–6/°C (Refs. 15, 18). The CTE
Nickel Electrolytic nickel 100
Silicon Ferrosilicon 45
of Inconel® 690 alloy is used for the
Titanium Ferrotitanium 5 weld metal since it is the most similar
Vanadium Ferrovanadium 80 to the composition of the experimen-
tal weld wire used.
Figure 11 shows the ternary system
Heat 3 produced a dull, oxidized weld pact productivity and weld integrity. with the respective phases that form
metal surface finish — Fig. 9. In Heat 4, The differences in the CTE between for Heats 2 through “CSM Final,” and
cryolite was increased back to the 19 the slag and the weld metal play a the commercial electrodes. Examina-
wt-% used in Heat 2 to restore weld vital role in the ability for the slag to tion of this figure shows that the com-
cleanliness. In Heat 5, cryolite was fur- detach easily. mercial electrode compositions fell
ther increased to optimize weld cleanli- Using a phase diagram of the CaO- into the perovskite phase. Heat 2 pro-
ness. The “CSM Final” flux formulation TiO2-SiO2 ternary system, shown in duced a dominant sphere phase in the
optimized cryolite to 24 wt-% and the Fig. 11, mass balance equations were slag with a CTE of 6 × 10–6/°C, which
respective bead-on-plate weld can be used to estimate the glass phases pro- is smaller than the weld metal, 10.4 ×
seen in Fig. 10. The “CSM Final” bead- duced in the slag. To perform these 10–6/°C, and displayed poor slag de-
on-plate weld displays the lustrous, sil- mass balance equations, the flux in- tachability. Heat 3 also produced a
very metallic appearance of nickel that gredients were first broken down to sphene phase with poor slag detacha-
Heat 3 (Fig. 9) does not with the de- their respective slag form (expressed bility (Ref. 15). Furthermore, the slag
creased amount of cryolite. Cryolite in the form of primary oxides)—Table developed a porous layer that adhered
plays a critical role in the cleanliness of 5). Similar oxides were then grouped tenaciously to the weld metal, which is
the weld metal surface. The amount together such that the ternary system an indication that a thin spinel phase
must be optimized for weld cleanliness diagram of CaO-TiO2-SiO2 can be is likely to be present on the surface of
without affecting the viscosity of the used. Na2O, MgO, and K2O were the weld bead. The occurrence of
slag flow since fluorides act as network grouped together with CaO, and Al2O3 spinel on top of a weld bead is usually
breakers and decrease the viscosity of was grouped with TiO2. This approxi- in the form of a thin layer on the order
the slag. Part of the design methodol- mation does not give an exact compo- of several atomic layers thick. The
ogy for flux coatings in a SMAW con- sition of the slag, but the thermal strain associated with con-
sumable electrode is to balance the methodology allows for the estimation traction and expansion of this layer is
multiple functions of each flux ingredi- of the predominant glass phase in the typically small. As such, the spinel
ent to optimize extrudability, weldabil- slag. The determination of the domi- layer is expected to have great adhe-
ity, and weld integrity. nant glass phase allows the estimation sion with the weld bead surface.
of the differences in the CTE values Heat 4 electrodes produced a domi-
Slag Detachability between the slag and weld metal, nant perovskite phase in the slag,
which helps predict slag detachability. which exhibited much better slag de-
Slag detachability is defined as the Values of linear CTEs were collected tachability due to the higher CTE val-
ease with which the solidified protec- from literature for the weld metal and ues. With a CTE value that is higher
tive slag layer can be removed. Poor the various crystalline phases that than that of the weld metal of
slag detachability can negatively im- could be formed based upon the com- Inconel® 690 alloy, the perovskite

278-s WELDING JOURNAL / AUGUST 2014, VOL. 93

Sham 8-14_Layout 1 7/14/14 4:39 PM Page 279

WELDING RESEARCH

Table 9 — Recovery Results from the Button Table 10 — Comparison of Core Rod and No­Dilution Weld Composition Compared to a Predefined
Melts with Respect to the Core Rods (wt­%) Target Weld Composition (wt­%)

Elements Alloy 1 (wt­%) Alloy 2 (wt­%) Elements Target Weld Experimental Heat 5 Heat 6
Composition Core Rod
Chromium 99 100 (HD 52)
Manganese 95 89
Niobium 100 89 Cr 27 26.8 24.7 27.2
Iron 62 60 Fe 2.5 2.6 2.91 4.9
Nickel 100 100 C 0.035 0.012 0.009 0.02
Ti 0.3 0.15 0.03 0.05
Nb 2.5 1.82 1.11 2.58
phase will contract much more than Mn 3 3.02 2.12 2.97
the weld metal upon solidification. Ni Bal Bal Bal Bal
The greater tensile stress induced in S <0.0015 0.001 0.001 <0.001
the slag phase will promote better slag P <0.005 0.001 0.001 0.005
Si 0.05 0.08 0.58 0.60
detachability. In order to maintain slag Mo 0.01 0.12 0.11 0.10
detachability for CSM Final flux for- Cu 0.1 <0.1 0.0015 0.05
mulation, rutile was increased to
enhance slag formation. The observa-
tions described above help to explain ured values in this research gave iron once the base flux formula was com-
the slag detachability behavior ob- recovery at 97%. Lastly, average recov- pleted. The recovery rate used for
served in the experiments. CTE differ- ery value of nickel was also calculated to chromium additions in the flux was
ences between the dominant slag be about 95%. These recovery values are 95%. A value of 60% was used for the
phase and weld metal can be used as a important for the calculation of neces- manganese recovery in the flux coating.
tool to help design flux coatings with sary additions in the flux coating for Since there were no calculated values of
enhanced slag detachability. final weld composition. recovery percentages for niobium, the
Another series of recovery calcula- ASM Metals Handbook Vol. 6 value of
Weld Metal Recovery tions was performed on the button 70% was taken as the recovery rate for
melts to understand the percentage of alloy additions in the flux coating.
Following the series of calculations recovery directly from the core rods Using the recovery rates determined
discussed in the experimental proce- without other influences. These re- previously, mass balance calculations
dures section, Table 7 gives the weld sults are shown in Table 9. Chromium were performed to determine the
recovery results of importance to the and iron contents were almost 100% amounts for replenishing the alloying
integrity of the weld metal. Chromium in recovery in an ideal welding atmos- elements. Two empirical values were
gives an experimentally determined phere. The recovery of manganese used to predict the percentage of the al-
average of 97% weld recovery with re- from the commercial core rods (Alloy loying elements additions to achieve the
spect to the three sources of contribu- 1) averaged about 95% whereas the re- target weld composition. Each electrode
tion, the core rod, the flux coating, covery of manganese from the in- weighed an average of 20.5 g and the av-
and the base metal. Manganese dis- house manufactured Heat 2 (Alloy 2) erage weight of the flux coating was 10
played an average weld recovery of averaged about 89%. It is clear with g. These data originated from average
about 54%. This value was slightly this information that the replenishing calculations of weighed core rods and
lower than that typically used (around of manganese must be made via flux. electrodes manufactured in-house at
70 to 80%) in the industry. However, The ASM Metals Handbook reported CSM. For chromium, a calculated 5%
being the major deoxidizer in the weld recovery from ferromanganese flux as addition into the flux coating is re-
system, a low recovery rate is not en- about 75%. Alloying through the flux quired. For manganese and niobium,
tirely surprising. Niobium was only must be done with the appropriate in- the calculated values were 3 and 4.1%,
calculated for the in-house electrodes gredients to minimize iron increase. respectively. Using these calculated val-
at an average recovery value of about Niobium recovery from the experi- ues, the target end weld composition
94%. The commercial electrodes were mental core rods was relatively high at was reached. These three main alloying
not taken into consideration because an average of 94%. Iron recovery from additions were reached in each of the
the niobium content in those fluxes the core rods was reported as an aver- large batch extrusions demonstrating
was not determined. The ASM Metals age of 76%. The different recovery the robustness of the composition
Handbook Vol. 6 provided niobium re- rates from the core rod and from the range. Note also that the last three ex-
covery from the flux ingredient in the flux coating reported in this work are truded flux formulations possessed the
form of ferroniobium as 70% along invaluable for future efforts in flux de- same cryolite, calcium carbonate, rutile,
with additional known alloy recovery sign and formulation. ferroniobium, chromium, and man-
from SMAW flux coating (Table 8) The findings showed that it was pos- ganese additions with small adjust-
(Ref. 5). Another element of impor- sible to obtain reliable recovery values ments to other minor flux ingredients
tance is the iron content, which must from all four of the most important al- to obtain the correct amount of silicon
be carefully monitored because it too, loying elements. These values were used and iron in the target weld composition.
like chromium, forms the undesirable to incorporate alloying additions in the The final base flux coating also
M23C6-type carbide. Average meas- flux coating design and optimization served as the starting point for alloy-

AUGUST 2014 / WELDING JOURNAL 279-s

Sham 8-14_Layout 1 7/14/14 4:39 PM Page 280

WELDING RESEARCH

Table 11 — Welding Observations at Optimal Welding Parameters Table 12 — Quantitative Values Related to the
Slag
Electrode E­1 E­2 Heat 2 Heat 3 Heat 4 CSM Final
Electrode Electrode Average
Slag Self Good Not Good Not Good Good Self Melt Rate Thickness
Detachability Release to Poor to Poor Release (mm*s­1) (mm)
Slag Thinner, Thicker Thicker Thicker Thicker Thicker
E­1 4.9 1.52
Thickness Porous
E­2 4.9 1.91
Slag Flow Less Viscous Viscous Viscous Viscous Viscous CSM Final 5.1 2.03
Viscous
Arc Soft, Harsh, Harsh, Unstable Unstable Soft,
Stable Stable Less Arc Arc Stable thickness of 1.91 mm. The highest aver-
Stable age slag thickness was the CSM Final at
2.03 mm.
ing. Table 10 gives the chemical com- Comparison of Qualitative and
position of the experimental high- Arc Quality
Quantitative Welding
nickel core rod (Alloy HD 52) used for
manufacturing the SMAW consumable
Characteristics The arc characteristics were de-
electrodes. Even though most of the scribed previously in a qualitative man-
elements are very close to the target General qualitative and quantitative ner (Table 11). Arc voltage data were
weld composition (also defined in welding observations were documented collected to characterize the arc stability
Table 10), alloying elements were lost for the characterization of the overall and relate with the metal transfer
through the SMAW process, thus re- welding characteristics that each elec- modes during welding. During welding,
quiring replenishment of chromium, trode displayed. Table 11 gives the qual- E-1 and CSM Final both exhibited soft
niobium, and manganese as done in itative characteristics for the and stable arcs. E-2, however, displayed
Heat 6. Using 95% recovery rate for commercial and in-house manufactured a harsh, bright, but stable arc. As to the
chromium, 70% recovery rate for nio- electrodes according to input from a intermediate flux systems, Heat 2 gave
bium, and 60% recovery rate for man- group of professional welders. Compar- a less stable arc whereas Heats 3 (Fig. 9)
ganese, the compositions of these ing four experimental electrodes and and 4 gave unstable arcs due to elec-
three elements were restored. Table 10 two commercial electrodes, the best trode eccentricity issues. When an elec-
shows that the chromium, niobium, performing electrode was the CSM Final trode is eccentric, the filler metal core
and manganese values in the no dilu- with a soft and stable arc, self-release rod is not centered in the flux coating
tion weld composition paralleled those slag, and thick, homogenous slag cover- causing the thinner side of the flux
of the target weld composition. ing. Professional welders use the terms coating to burn away faster and the arc
The overall weight-percentage of “harsh” and “soft” to describe their ob- to blow out on that side. This causes the
rutile was increased approximately 5 servation of the welding arc. A harsh arc arc to widen and wander producing an
wt-% to increase the oxygen content, generally implies a stronger arc force, unacceptable bead shape. Voltage signal
which subsequently lowered the iron but is more erratic in nature, and more analysis was not performed on Heat 3
recovery without compromising nio- difficult for a welder to control. A soft and 4 because of the difficulties in-
bium recovery. Silicon dioxide was re- arc, on the other hand, is more uniform volved in welding with eccentric elec-
moved to decrease the amount of and easier for a welder to control. Feed- trodes. The voltage data should assist in
silicon in the weld, but increasing the back from several industrial welders quantitatively substantiating the de-
rutile content compensated for the re- confirmed these observations. Figure scriptive welding lingo of a harsh and
moval of silicon dioxide to minimize 12 shows the slags collected from E-1 soft arc.
silicon recovery since rutile is also a and CSM Final to demonstrate good
network former. Talc was also re- slag detachability. In contrast, slag de- Comparison of Welding Arc
moved and one percentage of alginate tachability for Heat 2 was very poor as Signal Processing
was added instead as a better slipping shown in Fig. 13, which shows the slag
agent without introducing more sili- adhered tenaciously to the weld metal. The commercial electrodes, E-1 and
con into the flux system. The amount All slag-related measurements were E-2, exhibited lower average voltages of
of bentonite clay was decreased a few taken at a constant electrode melt rate. 24 and 23 V with a standard deviation
pct. to lower the aluminum and silicon Slag thickness measurements were of ±1.5 and ±2 V, respectively, and ex-
in the flux coating. For CSM Final for- taken at various locations along the slag hibited consistent and regular short-cir-
mulation, the iron decreased from 4.9 and consisted of the center and the side cuiting events — Figs. 14, 15. The
to 3.6 wt-% and the silicon content walls along the slag. The average thick- in-house manufactured electrodes using
was decreased from 0.6 to 0.1 wt-%. nesses of E-1, E-2, and the CSM Final Heat 2 displayed a higher average volt-
To achieve the predefined target com- slag systems are given in Table 12. The age of 29 V with a standard deviation of
position of the no dilution weld metal, commercial consumable designated E-1 ±1.9 V. Heat 2 displayed mainly short-
talc was completely removed and ben- had the thinnest average slag thickness circuiting metal transfer mode with
tonite clay was decreased significantly. at 1.52 mm. E-2 had an average slag some globular events occurring in-be-

280-s WELDING JOURNAL / AUGUST 2014, VOL. 93

Sham 8-14_Layout 1 7/14/14 4:39 PM Page 281

WELDING RESEARCH

tween. Figure 16 shows a portion of the The amount must be optimized for weld orado School of Mines, Golden, Colo.
voltage data from Heat 2 displaying the cleanliness without affecting the viscos- 4. Liu, S., Frederickson, G. L., Johnson,
change from short-circuiting to globular ity of the slag flow since fluorides act as M. Q., and Edwards, G. R. 1994. Shielded
metal transfer mode. Changes in metal network breakers and decrease the vis- Metal Arc Welding Consumables for Advanced
High Strength Steels (HSLA-130 Steel Welding
transfer modes throughout the welding cosity of the slag.
Consumables). Center for Welding and Join-
of an electrode show problems with arc • The CTE values for the slag and the ing Research, Colorado School of Mines.
stability. Considering an electrode feed weld metal is important for good slag 5. ASM Handbook. Welding, Brazing, and
rate of 5.5 mm/s for Heat 2, these detachability. Perovskite, with a CTE Soldering. 1993, Vol. 6, ASM International,
droplets were quite sizeable. Large glob- value of 15.8 × 10–6/°C, had a larger CTE Materials Park, Ohio.
ular transfer is generally an indication value than the weld metal (CTE value of 6. Vornovitskii, I. N., Medvedev, A. Z.,
of transfer instability and spatter. CSM 10.4 × 10–6/°C) which exhibited the and Cherkasskii, A. L. 1973. Influence of co-
Final average voltage was 22.5 V with a most desirable slag detachability. With a efficient of thermal expansion of slag on its
standard deviation of ±1.6 V — Fig. 17. CTE value of 6 × 10–6/°C, Sphene, a detachability from weld metal. Svarochn.
The dominant transfer mode for E-1, E- dominant slag phase, displayed poor Proizvodstvo 3: 35–37.
7. Vornovitskii, I. N., Malashonok, V. A.,
2, and CSM Final is short-circuiting oc- slag detachability.
and Cherkasskii, A. I. 1975. Procedure for
curring at similar frequencies. The • A series of mass balance equations quantitative evaluation of slag detachability.
frequencies of the short-circuiting is presented to calculate accurate recov- Svarochn. Proizvodstvo 2: 47–48.
metal transfer mode for E-1, E-2, and ery rates for alloying in the flux coating 8. Olson, D. L., Liu, S., and Fleming, D. A.
CSM Final are 3, 4, and 6 Hz, respec- developed in this research. 1993. Welding Flux: Nature and Behavior. Cen-
tively. Calculating the actual amounts of • Flux alloying recovery rate for ter for Welding and Joining Research, Col-
transfer per short-circuiting event, E-1 chromium in this high-nickel alloy weld orado School of Mines.
on average transferred 0.19 grams (0.57 system was determined as 95% same as 9. Donchenko, E. A., Panasenko, L. K.,
g/s, Table 3, divided by three short-cir- for low-carbon steel SMAW consumable and Larochkin, V. K. 1978. Some properties
cuiting events per second) while the electrodes. of welding slag. Svar. Proiz. 6: 38–40.
10. Claussen, G. E. 1949. The metallurgy
CSM Final transferred 0.075 g/s (0.45 • Manganese was calculated as 60%
of covered electrode weld metal. Welding
g/s, Table 3, divided by six short-circuit- flux alloying recovery for this high- Journal 28: 12–24.
ing events per second). The smaller nickel alloy weld system which is much 11. Scott, J., and Moline, K. 2008.
mass transferred is expected to result in lower than the 75% reported in the Devasco International, Inc., private
smoother transfer and less erratic arc ASM Handbook. communication.
behavior. In summary, the CSM Final • Electrode formulation “CSM Final” 12. Liu, S., Dallam, C. B., and Olson, D. L.
displayed the lowest overall average exhibited the best arc stability with the 1982. Performance of the CaF2-CaO-SiO2
voltage and also the highest and most lowest average voltage at 22.5 V and the system as a submerged arc welding flux for a
consistent short-circuiting metal trans- lowest mass per transfer event at 0.075 niobium based HSLA steel. Proc. Intl. Conf.
fer mode throughout the entire weld. g per droplet. Flux formulation is a com- on Welding Technology for Energy Applications,
Gatlinburg, Tenn., AWS-ORNL, ASM,
plex, interwoven, balancing process to
p. 445.
Conclusions ensure good extrudability, welding char- 13. Indacochea, J. E., and Olson, D. L.
acteristics, and weld integrity. 1983. Relationship of weld metal mi-
The major conclusions of this re- crostructure and penetration to weld metal
search follow: Acknowledgments oxygen content. Journal of Materials of En-
• A design methodology for develop- ergy System 5: 139.
14. Perez Guerrero, F. 2003. Effect of
ing SMAW consumable electrode flux
The authors acknowledge the help nickel additions of rutile electrodes for un-
coatings for high-nickel alloys is pre- derwater welding. MS thesis, Colorado
and discussions of Joe Scott and Keith
sented focusing on enhancing extrud- School of Mines.
Moline of Devasco International, Inc.,
ability, weld cleanliness, slag 15. Devries, R. C., Roy, R., and Osborn, E.
and Bob Hamilton of Johns Manville
detachability, alloying recovery, welding F. 1955. Phase Equilibria in the System CaO-
during the course of this research. The
characteristics, and arc quality. TiO2-SiO2. American Ceramic Society 38(5):
authors also acknowledge the support 161.
• A GTAW wire can be converted into
of Dr. George Young for the useful 16. Wu, C., et al. 2009. Plasma-sprayed
a SMAW consumable electrode, thereby
discussions. CaTiSiO5 ceramic coating on Ti-6Al-4V with
proving that a unified welding wire is vi-
able for both flux coated and bare wire excellent bonding strength, stability, and cel-
welding processes. References lular bioactivity. J. R. Soc. Interface 6(31):
159–168.
• In enhancing extrudability, an ap- 17. Redfern, S. A. 1996. High-tempera-
propriate amount of binder with respect ture structural phase transitions in per-
to relative humidity is important, ade- 1. Jackson, C. E. 1973. Fluxes and Slags in ovskite (CaTiO3). J. Phys.: Condensed Matter
quate slipping agent, and a dry mix with Welding. Weld Res. Bull. 1973, No. 190, 8: 8267–8275.
a normal particle size distribution at Welding Research Council, New York, N.Y. 18. Special Metals. 2008. Inconel® 600
200 mesh is desirable. 2. Linnert, G. E. 1965. Welding Metal- Alloy: Technical Bulletin.
• Adequate amount of cryolite is nec- lurgy. Vol. 1, Ch. 8, American Welding Soci- 19. Natalie, C. A., and Olson, D. L. 1985.
ety, Miami, Fla., pp. 367–396. Physical and Chemical Behavior of Welding
essary for enhanced weld cleanliness for 3. Natalie, C. A., and Olson, D. L. 1985.
high-nickel SMAW consumable elec- Fluxes. Center for Welding Research, Col-
Physical and Chemical Behavior of Welding orado School of Mines, Golden, Colo.
trodes and maintaining slag viscosity. Fluxes. Center for Welding Research, Col-

AUGUST 2014 / WELDING JOURNAL 281-s