Selection of Nickel-Base Alloys for

Metal Dusting Resistance

The causes of "metal dusting", a potentially severe corrosion problem resulting from the steam
reforming process, are examined. Long-term laboratory data is presented which shows how the
consideration of alloy content can be effective in resisting metal dusting attack. A newly-developed
alloy offers promising potential for metal dusting limited applications.

Brian A. Baker and Gaylord D. Smith

Special Metals Corporation

Stephen A. McCoy
Special Metals Wiggin, Limited

Introduction to examine the causes of, and potential solutions to,

this often-disastrous problem. Special Metals corpora-

T
he production of synthesized gas (syngas) tion has been actively involved in characterizing and
from methane or natural gas via steam reform- developing new materials to cope with the increasing
ing is a common step to begin production of demand on equipment in harsh metal dusting environ-
hydrogen and ammonia, as well as methanol and other ments. Research has been conducted on a number of
liquid hydrocarbons. The syngas stream consists of a alloys, apart from the traditionally used primary and
mixture of CO, H2 and H2O, with lower levels of CO2 secondary reformer material ENCOLOY alloy 800H, to
and some CH4. When such gas mixtures are present in better understand the role of alloying in resisting
attack. This article presents long-term laboratory data
the process stream in the critical temperature range of
and shows how the consideration of alloy content can
about 400° to 800°C, the phenomenon labeled "metal
be effective in resisting metal dusting attack. In addi-
dusting" can potentially be a severe corrosion problem.
tion, mechanical properties are shown for the newly-
Metal dusting can be described as a catastrophic car-
developed INCONEL alloy 693 in comparison with
burization phenomenon which occurs under conditions
other nickel-base alloys which exhibit enhanced resist-
where the carbon activity of the gaseous atmosphere is
ance to metal dusting.
greater than one. Metal dusting is manifested by the
disintegration of the affected metal into a powdery
Causes of Metal Dusting
mixture of graphite and metal particles. Oxides and
carbides are also often present in the powdery mixture,
The amount of steam used for the reforming process
or "dust".
has been driven lower by the need for greater efficien-
Past and present failures resulting from metal dusting
cy, resulting in lower steam-to-hydrogen ratios. Higher
have prompted end users and material producers alike
front-end pressures have also increased the CO content

AMMONIA TECHNICAL MANUAL 257 2002

of the syngas. Lower H2O/H2 ratios in combination
with higher CO/CO2 ratios result in lower oxygen par-
tial pressures and higher carbon activities, respectively,
and serve to increase the severity of metal dusting
attack. Metal dusting is often avoided in industry by
designing around the critical metal dusting temperature
range. Syngas is produced at temperatures above the
critical range (>800°C) and transferred to a boiler via a
short transfer line where it is rapidly quenched to tem-
peratures below the critical metal dusting range
(<400°C). Alloy ferrules that are used in the transfer
line do often experience metal dusting and are periodi-
cally replaced. The need to maximize the efficiency of
steam reforming technology has led to the development
of equipment which must be capable of operating with-
in the range of temperature and carbon activity which
can promote metal dusting. This necessitates the use of
materials which exhibit excellent resistance to metal
dusting attack.
The mechanism of metal dusting for iron-base alloys
begins with saturation of the alloy matrix with car-
bon/carbides, usually hi a localized manner, and subse-
quent formation of metastable Fe3C, or cementite.
Decomposition of the cementite as the carbon activity
approaches unity produces kon particles and powdery
carbon. The metal particles then strongly catalyze fur-
ther carbon deposition. A different mechanism is pro-
Figure 1. The equidistant diffusion of carbon from a localized defect in the protective oxide scale
that results in the saturation of a hemispherical region with carbon.
AMMONIA TECHNICAL MANUAL
posed for nickel-base alloys, which does not involve
the formation of an intermediate metastable carbide.
Such a mechanism begins in the same way as for iron-
base alloys, with saturation of the alloy matrix with
carbon/carbides. However, in the case of nickel-base
alloys, the saturated matrix directly decomposes into
metal particles and graphite (Grabke, 1998; Lai, 1990)
Figure 1 illustrates the equidistant diffusion (assuming
a material which exhibits isotropic diffusion behavior)
of carbon from a point defect hi the protective oxide
scale which results in saturation of a hemispherical
region with carbon. Subsequent decomposition of this
saturated area results hi disintegration of the alloy
matrix and not simply grain fallout, producing a pit
having the same hemispherical shape as the carbon-sat-
urated region.
The presence of an adherent, protective, oxide sur-
face layer that can heal is required for protection of an
alloy against metal dusting attack. While oxide forma-
tion may be stable, the oxide layer may still be suscep-
tible to disruption. Higher levels of the scale forming
element will then make the scale healing process more
rapid and complete. Ultimate resistance to metal dust-
ing may involve complex interactions of scale charac-
teristics, diffusivity of scale-forming elements and car-
bon through the alloy matrix and carbon saturation lim-
its.
Defect, fe Oxide
258 2002
 Figure 2. Cross section of metal dusting pit in alloy 600 reformer pig-
tail pipe which was in service for approximately 5 years.
Etchant: 5% HNO3 in methanol, electrolytic.

Figure 3. Cross section of metal dusting pit in alloy 800 reformer

pigtail connecting ring which was in service for approximately 5
years.
Etchant: 5% HNOß methanol, electrolytic.

AMMONIA TECHNICAL MANUAL 259 2002

Alloy Performance in Metal Dusting
Environments
The tendency of synthetic gas streams to promote
metal dusting has been heightened by changing pro-
cessing conditions and the use of advanced catalysts.
Techniques such as injection of additional steam and
sulfur-containing species into the gas stream, which are
known to minimize the effects of metal dusting, are no
longer viable in many processes due to the increased
use of catalysts which are sensitive to sulfur content.
Higher gas pressures and temperatures, in conjunction
with lower H2O/H2 ratios and higher CO/CO2 ratios,
are becoming the norm for modern reforming opera-
tions. This combination of conditions may necessitate
the use of materials which offer resistance to metal
dusting that is superior to that of previously commonly
used alloys such as the Cr-Mo stainless steels and Fe-
Ni-Cr alloys.
Two commonly used alloys that typically offer good
overall corrosion resistance as well as high temperature
strength under a wide range of conditions are
INCONEL alloy 600 and INCOLOY alloy 800H.
Metal dusting failures of alloy 800 are well document-
ed (Esaklul, 1992; Special Metals, 1997). and problems
with alloy 600 have been encountered as well. Figures
2 and 3 show metal dusting attack of both the alloy 600
piping and the alloy 800 connecting ring, respectively,
from a reformer pigtail (Strauss and Grabke, 1998).
The reformer had been in service for about 5 years.
Harsher conditions have necessitated the use of alloys
having high nickel contents in addition to higher levels
of scale-forming elements, such as chromium. Alloy
601 has been used for waste heat boiler shells and tub-
ing in ammonia plants, as well as for reformer compo-
nents which has resulted in greater production due to
decreases in downtime and repair costs. For even
greater protection against metal dusting under the
harshest conditions, alloys such as INCOTHERM alloy
TD, INCOLOY alloy MA956, INCOCLAD
671/800HT and INCONEL alloy 617, INCONEL alloy
690, and INCONEL alloy 693, offer promising poten-
tial as upgrades from the commonly used alloy 601.
AMMONIA TECHNICAL MANUAL
Laboratory Testing
Table 1 shows the chemical composition for each
alloy tested, as well as for other alloys mentioned in
this article. Test specimens were prepared from com-
mercially available material; sample dimensions were
approximately 2.5 cm X 2.5 cm X thickness. Samples
were ground to a 120-grit finish for standardization
purposes.
Exposures were performed in a horizontal tube fur-
nace at 621°C in an atmosphere generated from an inlet
gas mixture of CO-20% H2. Samples were cycled,
lightly brushed, and ultrasonically cleaned and mass
change was measured at approximately two-week
intervals. In addition to mass change, pitting depth was
also determined for each sample using an optical
microscope having a calibrated fine focus knob.
Figure 4 shows mass loss rates vs. time for the alloys
tested. The rate was calculated by dividing the mass
change (only shown if negative) per unit area by the
total test time. The maximum pit depth as a function of
time is plotted in Figure 5. Figure 6 shows the result of
multiple regression analysis of the log of the pit pro-
gression rate, calculated by averaging the pit depth
over the total testing duration, vs. a summation of the
wt. % of various alloying additions. Only the austentic
nickel-base alloys and Fe-Ni-Cr alloys were included
in this calculation; the ferritic alloy, MA956, and also
alloy 400 were excluded. The ferritic materials exhibit
a different mechanism due to the high diffusion rates of
the scale formers through the matrix. The best fit was
produced using the following summation:
(%Ni+%Co)+5*%Mo+20*%Ti+33*%Al+15*%Si+
2*%Cr-2*%Fe.
This same regression technique was used to charac-
terize the variation in mass loss rate with the percent-
age of certain alloying additions (Figure 7). Again,
only the austenitic nickel-base and Fe-Ni-Cr alloys
were included in the calculation. The best fit was
obtained using the following summation:
2*(%Ni+%Co)+6*%Mo+31 *%Ti+34*%Al+80*%
Si+9*%Cr-%Fe.
260 2002
 Table 1. Nominal Composition of Commercial Alloys

Alloy Ni Cr Fe Mn Si Al Ti C Other

INCONEL alloy MA754 78 20 - - - 0.3 0.5 0.05 0.5 Y2O3

INCOTHERM alloy TD 73 22 - - 1.4 - - 0.01 3.0 Mo
INCONEL alloy 600 72 15.5 8 0.3 0.3 0.3 0.3 0.08 -
INCONEL alloy MA758 67 30 - - - 0.4 0.5 0.05 0.5Y203
MONEL alloy 400 64 0.1 1.6 0.7 0.1 0.02 0.4 0.15 32.2 Cu
INCONEL alloy 693 62 30 4 - - 3 - - Nb,Zr
INCONEL alloy 625LCF 61 21.5 2.5 - 0.1 0.2 0.2 0.02 9 Mo, 3.6 Nb
INCONEL alloy 601 60.5 L 23 13 0.2 0.2 1.4 0.4 0.05 -
INCONEL alloy 690 59 29 9 0.2 0.1 0.3 0.3 0.02 -
INCONEL alloy 671 53 46 - - - 0.3 0.3 0.03 -
INCONEL alloy 6 17 55 22 1 - 0.1 1.2 0.4 0.08 12.5 Co, 9 Mo
NMONIC alloy 263 51 20 39 0.3 0.1 0.5 2.2 0.06 20 Co, 5.9 Mo
INCOLOY alloy 825 42 21.5 28 0.4 0.1 0.1 1 0.02 3 Mo, 2Cu
INCOLOY alloy DS 37 16 41 1.0 2.3 - - 0.08 -
INCOLOY alloy 330 35 19 44 1.0 1.3 - - 0.07 -
INCOLOY alloy 803 34 27 36 1.0 0.8 0.4 0.4 0.08 -
INCOLOY alloy 864 34 21 39 0.4 0.8 0.3 0.6 0.03 4.2 Mo
INCOLOY alloy 32 21 45 0.9 0.1 0.4 0.4 0.07 -
800/800HT
INCOLOY alloy MA956 - 20 75 - - 4.5 0.5 0.05 0.5Y203

T 1 1—<—'—>—i 1 1 r
O 2000 4000 6000 8000 10000 12000 14000 16000 18000
Exposure Time, hrs

Figure 4. Mass loss rate vs. exposure time for alloy samples exposed to CO-20% H2 at 621° C.

AMMONIA TECHNICAL MANUAL 261 2002

 1000 2000 3000 «X» 5000 6000 7000 1000 900Q 10000 11000 12000 13000 14000 15000 16000 17000 18000
Exposure Time, Hours

Figure 5. Maximum pit depth measurements for alloy samples exposed to CO-20% H2
at 621°C.

CO-20%H2at621°C

1 3
^«X)
4i -^ ** - *33Q
*DS 4803
c .4 . * * 825
_£ *
864 " .tȤP1
2
g - -*KQ
45- finn * 263MA754m
Coo '^* * ^625LCF
690 ^ 4671
1 -5-
1 MA758 ^61?
«
Q.
_R S .
* 693
a .6 .
3 0 50 100 150 200 250
0
(%NI+%Co)+5^oMo+20«%TI+33* /oAI+15*%Sl+2*(%Cf-%Fe)

Figure 6. Results of multiple linear regression for pit progression rate vs.
composition.

AMMONIA TECHNICAL MANUAL 262 2002

 CO-20%H2at621°C
I
1-1- 80
°*
• 330
825 «f600
8 c
-4- MA754 625LCF
« -5
100 200 300
Figure 7. Results of multiple linear regression for mass loss rate vs.
composition.
As a general trend, nickel-base alloys exhibited
lower mass loss rates and pitting progression rates than
iron-base alloys. Exceptions include alloys 600 and
MA754, which contain only 15% and 20% chromium,
respectively, have no other significant additions of
scale-forming or carbide-forming elements, and had a
fairly high mass loss rate resulting from numerous,
albeit fairly shallow, pits. Alloy 690 (Ni-29Cr-9Fe),
commonly used as heat-exchanger tubing in the
nuclear industry, exhibits much lower mass loss and pit
progression rates than alloy 601, which is increasingly
used as an upgrade from iron-base materials such as
Cr-Mo steels, austenitic stainless steels, and Fe-Ni-Cr
heat-resistant alloys such as alloy 800. The mechani-
cally alloyed corollary to alloy 690, alloy MA758, also
exhibits good performance. Alloy MA956, a ferritic
alumina former, also exhibited good performance.
Alloy 263 performed well despite its modest chromium
level of 20% and 39% iron content, and appears to pos-
sibly gain protection from its substantial titanium addi-
tion and possibly its molybdenum addition, which may
promote early carbide formation and provide diffusion-
aï blocking of the carbon flux (Fahrmann and Smith,
2000). Alloy 617, having 22% chromium and 1.2% alu-
AMMONIA TECHNICAL MANUAL 263
R71
671
693
400 500 600
minum, may also gain the same benefit from its molyb-
denum addition. Alloy 625LCF, a refined version of
alloy 625, has performed reasonably well and contains
21.5% chromium, 9% molybdenum, and 3.6% niobium
and probably benefits from its fine grain structure. The
high silicon content of alloy TD may have afforded
some enhancement in performance, in addition to its
3% molybdenum content.
The alloys 671 and 693 exhibit, overall, the most
promising performance trends in both resisting pitting
attack and resisting mass loss, and contain the highest
combinations of nickel, chromium, and aluminum. The
protection of the material is enhanced by the very high
level of scale-forming elements which form a dense,
adherent and self healing protective oxide surface
layer. Alloy 671, with its substantial chromium content,
could be utilized in the form of a bimetallic clad tube,
being mechanically very poor as a monolithic material
within the metal dusting temperature range. INCO-
CLAD tubes having an alloy 671 layer on the OD and
alloy 800HT at the ID have been successfully used in
coal-fired boilers with service times exceeding 20
years. The best performer overall in the laboratory test
was alloy 693, which possesses very high chromium
2002
 Table 2. Room-Temperature Tensile Properties for Annealed differences in properties between the
Material oxide scale and the substrate. The fol-
lowing data are intended to highlight
Alloy 0.2% Yield Ultimate Tensile Elongation, % the properties of the newly-intro-
Strength, MPa (ksi) Strength, MPa (ksi) duced alloy 693. Typical room-tem-
perature tensile properties for alloys
601 276 (40) 621 (90) 55 601, 617, 690, and 693 are shown in
617 379 (55) 827 (120) 53 Table 2. The properties for alloy 693
690 414 (60) 758 (110) 47 are similar to those for alloy 690.
693 414 (60) 827 (120) 43 Figure 7 compares the stress rupture
properties of alloys 601, 617, 690,
and 693 via a Larson-Miller plot.
and aluminum contents. ln-situ field exposures in syn-
Figure 8 shows stress vs. rupture life for alloys 601 and
gas environments have confirmed this alloy's superior
693. Low-cycle fatigue results for alloys 693 and
performance.
625LCF are shown in Figure 9. Alloy 693 possesses
fatigue properties at 593°C and 704°C which are supe-
Alloy Properties
rior to those of alloy 625LCF at 538°C. Alloy 625LCF
While corrosion resistance is of the utmost impor- is used as a benchmark and has been specifically
tance in applications which may cause metal dusting, developed and applied for the purpose of resisting
mechanical properties are also an important considera- fatigue in high temperature bellows applications.
tion when selecting a material. Materials with high Tables 3 and 4 show room-temperature and elevated
creep strength in the metal dusting temperature range temperature tensile results for alloy 693 after interme-
can be more resistant to scale disruption caused by the diate-temperature exposure. Respectable elongation

1000

100

20 30 40 50 60 70
LMP = (1.8T+490X20+log{Tr)yiOOO, T in °C, Tr in
Hours

Figure 8. Larson-Miller plot representing stress rupture data for alloys 617, 693, 601, and
690.

AMMONIA TECHNICAL MANUAL 264 2002

 1000

|693-704°C
601-704°CF
I•t
(A

ö>

10 100 1000 10000 100000

Rupture Life, Hours

Figure 9. Stress rupture data comparing alloys 601 and 693 at 649°C, 704°C, and 982°C.

Strain Control Low Cycle

Fatigue - Alloy 693
niA -,
O .Ul^
A U<U1
n 019 -
o>
£
Rj
\_,
^
n ni -
V/.UI • 693-593°C
dc_ n (V\Q .
C u.uvo • 693-704°C
1 0.006- X^
-. Ü0 UÜ4
^ ^^^^-
004. -
B ^
625LCF-538°C
— 625LCF-649°C
s - '" ••» • «

o n 002 -
0. T" " i ' 1 1

1000 10000 100000 1000000 1E+07

Cycles, Nf
Figure 10. Low-cycle fatigue data for alloys 625LCF and 693.

AMMONIA TECHNICAL MANUAL 265 2002

 Table 3. Room-Temperature Tensile Results for Alloy 693

Exposure 0.2% Yield Ultimate Tensile Elongation, % Exposure Time, h

Temperature, °C (°F) Strength, MPa (ksi) Strength, MPa (ksi)

As-Produced 410 (59.5) 834 (121.0) 43.0 As-Produced*

649 (1,200) 672 (97.5) 1,065 (154.5) 32.5 100
649 (1,200) 814(118.0) 1,179 (171.0) 28.8 500
704 (1,300) 645 (93.5) 1,045 (151.5) 48.9 100
704 (1,300) 816(118.3) 1,181 (171.3) 28.8 500
760 (1,400) 631 (91.5) 1,048 (152.0) 47.2 100
816(1,500) 634 (92.0) 1,048 (152.0) 55.2 100
* Hot-Rolled and Annealed Rod

Table 4. Elevated-Temperature Tensile Results for Alloy 693 (Tested at Exposure Temperature)

Exposure 0.2% Yield Ultimate Tensile Elongation, % Exposure Time, h

Temperature, °C (°F) Strength, MPa (ksi) Strength, MPa (ksi)

649 (1,200) 411 (59.6) 643 (93.2) 28.9 As-Produced*

649 (1,200) 534 (77.5) 776(112.5) 21.8 100
704 (1,300) 457 (66.3) 634 (91.9) 16.3 As-Produced
704 (1,300) 569 (82.5) 769(111.5) 14.6 100
704 (1,300) 561 (81.3) 776(112.5) 18.9 500
704 (1,300) 556 (80.7) 769(111.6) 20.3 1,000
760 (1,400) 461 (66.9) 607 (88.1) 9.9 As-Produced
760 (1,400) 524 (76.0) 658 (95.5) 17.2 100
760 (1,400) 338 (49.0) 549 (79.6) 29.1 500
760 (1,400) 322 (46.7) 542 (78.6) 32.0 1,000
816 (1,500) 431 (62.5) 558 (81.0) 15.3 As-Produced
816(1,500) 314 (45.5) 434 (63.0) 23.4 100

values are maintained after exposure times of up to
1,000 h hi the critical temperature range of 704°C to
760°C.
Conclusion
Due to development of advanced catalysts and efforts
to increase the efficiency of processes involving the
production of syngas, metal dusting corrosion has
become more prevalent. Failures of iron-base alloys, as
well as nickel-base alloys which contain insufficient
scale-forming elements, have prompted equipment
designers to seek materials that are more resistant to
metal dusting. Field and laboratory data confirm the
desirability of addition of certain scale-forming and
carbide-forming elements hi conjunction with a nickel-
base alloy matrix to limit pit progression rates. The
newly-developed alloy 693 offers promising potential
for metal-dusting limited applications.
Acknowledgments
The authors would like to thank Chad Clary for his
assistance with laboratory testing and sample evalua-
tion.

AMMONIA TECHNICAL MANUAL 266 2002

Literature Cited
Esakiul, K. A., ed., Handbook of Case Histories in
Failure Analysis, ASM, pp. 351-353 (1992).
Fährmann, M. G., and G. D. Smith, "Corrosion/2000,"
NACE International Paper No. 232 (2000).
Grabke, H. J., Materials and Corrosion, 49, 303
(1998).
QUESTIONS AND ANSWERS
Rob Gommans, Gommons Metallurgical Services:
How does alloy 602CA compare to the alloys tested?
Brian Baker, Special Metals Corporation: We have
actually tested the alloy 602CA material in our labora-
tory metal dusting test. The results from our test envi-
ronment show that the pitting rate of alloy 602CA is
intermediate between that of alloys 601 and 690, and
significantly higher than that of alloy 693.
Gommans: Alloy 693 contains 3% Al. How are the
welding properties? And loss of Al from the liquid
weld pool?
Baker: We have performed numerous welding trials,
using various filler metals and joining to various other
base metals. The weldability of alloy 693 has been
shown to be quite favorable. We are still working
toward commercialization of a matching filler materi-
al. If welding is performed with inert gas shrouding,
AMMONIA TECHNICAL MANUAL
Lai, G. Y., "High Temperature Corrosion of
Engineering Alloys," ASM, pp. 69-72 (1990).
Pippel, E., J. Woltersdorf, and R. Schneider, Materials
and Corrosion, 49, 309 (1998).
Special Metals, Private Communication (Dec. 1997).
Strauss, S. and H. J. Grabke, Materials and Corrosion,
49, 321 (1998).
one would not expect gross Al loss from the liquid
weld pool; the expected decrease in Al level would be
about 10%. Right now, we're recommending welding
with INCONEL filler metal 52 and capping the sur-
faces with INCONEL filler metal 72 for metal dusting
resistance.
Carl E. Jaske, CC Technologies: What is the relative
cost of alloy 693 vs. similar nickel-base alloys?
Baker: The cost of alloy 693 will be higher than that
of alloy 601. The specific cost factor will depend upon
the product form.
Jaske: What type and amount of field testing has been
done to evaluate real service performance vs. laborato-
ry testing?
Baker: Several field exposures in syngas environ-
ments have shown that the real service performance of
alloy 693 correlates quite well with laboratory results.
267 2002