In situ studies of carbon formation leading to
metal dusting in syngas processes
Olle Söderström
Department of Chemical Engineering, Lund University
February 2010
Abstract
Metal dusting corrosion begins with carbon deposition on the metal surface followed by diffusion of
the surface carbon into the metal which eventually causes supersaturation of carbon in the metal
and the metal dusting is initiated. Experiments have been carried out in a thermogravimetric analysis
equipment. A screening process was conducted of 4 different materials with various alloy
composition to choose for further investigation. It was found that the poorer the alloy composition,
the higher the carbon formation rate. The presence of ethylene increased the carbon formation rate
significantly but the ethylene content did not however seem to be decisive for the carbon formation
rate. Experiments were also carried out by the means of determining limits for carbon free
operation. It was found that for an atmosphere consisting of 50 % hydrogen and the rest inert
nitrogen, carbon starts to form over about 18 % ethylene. It was also found that, in an atmosphere
consisting of 10 % ethylene, when hydrogen was ramped from 39 -25 % and the rest was inert
nitrogen, the carbon formation starts at about 34 % hydrogen. For a syngas mixture of 21 % carbon
monoxide, 63 % hydrogen, 16-6 % steam and the rest inert nitrogen the amount of steam needed to
suppress carbon formation was found to be 6 %.
Introduction
Damaging carbon formation is a common and
well known problem in process units exposed
to carburising gases, especially in syngas
processes. Carbon formation in catalysts will
disintegrate the catalyst pellets, influence the
catalytic activity, and affect the pressure drop
across reactor tubes. Carbon formation on
metals can cause fouling on heat transfer
surfaces and may be a precursor to metal
dusting corrosion. Ultimately, the carbon
formation may result in equipment failures.
Thus, there is a strong motivation to
understand and determine the influence of
various parameters on carbon formation such
as
temperature,
pressure
and
gas
composition. This work has focused on carbon
deposition on materials to gain some general
understanding as well as to be able to set
limits for safe operation with respect to metal
dusting corrosion for some gas mixtures
relevant to syngas processes.
Theory
Syngas is primarily produced by steam
reforming of natural gas which mostly consists
of methane but also some impurities. The
general steps are that the feed gas first goes
through a purification step where the gas is
purified from sulphur and chlorine
compounds at about 380 °C before it enters
the
pre-reformer
in
which
higher
hydrocarbons are reformed into methane at
500-650 °C. After that the gas enters the
steam reformer in which methane reacts with
steam to form carbon monoxide and
hydrogen at 650-900 °C.
𝐶𝐻4 (𝑔) + 𝐻2 𝑂(𝑔) ⇌ 𝐶𝑂(𝑔) + 3𝐻2 (𝑔)
Depending of the end-use of the syngas
higher hydrogen content might be wanted.
This is done in the water-gas shift reactor
where excess steam reforms the carbon
monoxide to carbon dioxide and hydrogen at
about 250-300 °C.
𝐶𝑂(𝑔) + 𝐻2 𝑂(𝑔) ⇌ 𝐶𝑂2 (𝑔) + 𝐻2 𝑔
Carbon activity
The carbon activity is a way of quantifying the
thermodynamic driving force for the carbon
forming reactions and it is derived from the
equilibrium constants for the carbon forming
reactions. In syngas processes the reactions
relevant to carbon transfer from gas to metal
are as follows. [1], [2]
𝐶𝑂(𝑔) + 𝐻2 (𝑔) ⇌ 𝐻2 𝑂(𝑔) + 𝐶(𝑔𝑟𝑎𝑝ℎ𝑖𝑡𝑒)
2𝐶𝑂 𝑔 ⇌ 𝐶𝑂2 𝑔 + 𝐶(𝑔𝑟𝑎𝑝ℎ𝑖𝑡𝑒)
𝐶𝐻4 𝑔 ⇌ 2𝐻2 𝑔 + 𝐶(𝑔𝑟𝑎𝑝ℎ𝑖𝑡𝑒)
𝐶𝑛 𝐻𝑚 𝑔 ⇌
𝑚
2
Figure 1 Schematic illustration of the processes in
metal dusting of iron, low alloy steels and nickel. [2]
𝐻2 𝑔 + 𝑛𝐶(𝑔𝑟𝑎𝑝ℎ𝑖𝑡𝑒)
The last reaction generally describes carbon
deposition from higher hydrocarbons present
in the natural gas feed. An example of how to
calculate the carbon activity for the reduction
reaction (the topmost) is found below.
𝑎𝑐 = 𝐾
𝑝𝐶𝑂 ∙𝑝𝐻2
𝑝𝐻2 𝑂
log 𝐾 =
7100
𝑇
− 7.496
Mechanism
The reaction mechanism for carbon attack
that could lead to metal dusting corrosion on
iron, iron-based low alloys and nickel is
illustrated in figure 1 and is as follows: a)
carbon transfer from the gas phase and
oversaturation of the metal phase, b)
Nucleation and growth of cementite, Fe3C, at
the surface of Fe and low alloy steels, c)
Nucleation and growth of graphite into the
Fe3C, by d) C atoms attaching to the lattice
planes of graphite, growing more or less
perpendicular into Fe3C, resp. Ni and high-Ni
alloys (from TEM studies), e) carbon filaments
grown behind particles detached from the
metal phase by the graphite growth (SEM), f)
Steady state of metal dusting on Fe and low
alloy steels at temperatures < 600 °C, inward
growth of Fe3C which disintegrates outward
under coke formation, g) Steady state of
metal dusting on Fe and low alloy steels at
temperatures > 700 °C, formation of an iron
layer between Fe3C and coke, carbon diffusion
through this layer and final disappearance of
Fe3C. [2]
Prevention of metal dusting
Uses of coatings or changing the environment
are ways of preventing metal dusting. It is
mainly done through blocking the surface by
either a protective oxide scale or by adsorbing
sulphur. On surfaces protected by sulphur or
oxides, such as Cr2O3, Al2O3 and spinels,
adsorption and dissociation of CO, CH4 and
other hydrocarbons generally does not take
place. Thus, carbon is not transferred into the
metal and metal dusting is therefore not
initiated. [2]
Material and Method
Experiments have been conducted in a
thermogravimetric analysis, TGA, equipment
provided by Haldor Topsøe A/S. In the TGA,
weight changes of a material can be measured
as a function of time under controlled gas
composition
and
temperature.
The
temperature
is
controlled
with
a
thermocouple connected to a regulator and
the gas composition by well calibrated 5850
TR mass flow controllers from Brooks. The
weight changes are recorded with a Mark
2CT5 Microforce Balance from C.I. Electronics
Ltd. able to measure changes with a precision
of 0.01 mg in the sample weight.
The experiments are fully automated using a
number of softwares. All experimental steps
are set up in an Excel sheet. LabVIEW is used
to read the instructions from the Excel file and
pass the commands onto Automation Tool-kit
that continuously instructs the hardware how
to run the TGA equipment.
With the various gas lines connected to the rig
it was possible to run with a gas composed of
CH4, CO, C2H4/C4H10, H2, N2 and steam.
Sample preparation
Four materials were tested; two stainless
steels and two carbon steels. Their alloy
compositions are found in table 1.
Table 1 Alloy composition of the materials tested.
Fe
Cr
Ni
Mo
C
Mn
Si
P
S
N
SS308H
66,571
18-20
8-10.5
0.040.1
<2
<0.75
<0.045
<0.03
-
SS316
62-69
Fe-2.25Cr1.0Mo
95.4
Fe-1.25Cr0.5Mo
96.7
16-18
10-14
2.03.0
<0.08
2.25
0.20
1.0
1.25
0.20
0.5
0.15
0.15
<2
<0.75
<0.045
<0.03
<0.1
0.65
0.30
0.035
<0.035
-
0.65
0.5
0.035
<0.035
-
The Fe-1.25Cr-0.5Mo carbon steel samples
were cut from a larger plate of approximately
500x200x4 mm. Since this plate was rusty it
had to be ground and polished which was
done with an electrical grinder using first P50
grit sand paper to make a coarse grinding
getting the rust off and then P120 grit sand
paper to polish the scratches. Then samples of
dimensions 10x4x2 mm weighing about 1 g
were cut out with an angle grinder.
The Fe-2.25Cr-1.0Mo carbon steel was
delivered in a long rod with a rectangular
cross-section of 2x1.5 mm with a coarse grey
surface. The stainless steels 308 H and 316
were delivered in long cylindrical rods of 2
mm in diameter with smooth metallic-shiny
surface. Samples were cut to a weight of
about 1 g and bent into a U-shape so that
they would not fall through the bottom of the
sample basket.
Before inserting the samples in the reactor
they were washed with soap in distilled water,
wiped with paper and then also cleaned with
n-Hexane on a piece of paper. The sample was
placed in a glass basket using a clean pincer
and the glass basket was hung on a quartz
hook in the reactor connected to the micro
scale.
Results and Discussion
Screening of materials
A material susceptible to carbon formation
within 24 hour experiments is sought.
Therefore a screening process was conducted
of two stainless steels and two carbon steels
with various alloy compositions to choose for
further investigation. It was found that the
poorer the alloy composition, the higher the
carbon formation rate, as seen in table 2. The
carbon steel Fe-1.25Cr-0.5Mo was chosen for
further experiments.
Table 2 Shows the rate of carbon formation on tested
materials at pCH4=1.0 and T=750 °C.
Material
Rate(mg C/(h*gmat))
SS-316
SS-308H
Fe-2.25Cr-1.0Mo
Fe-1.25Cr-0.5Mo
0
0.00706
0.0824
0.129
Ethylene influencing the carbon
formation rate from methane
The influence of ethylene on the carbon
formation rate from methane was studied.
This is because the natural gas feed contains
higher hydrocarbons which affect the carbon
growth and ethylene was chosen to represent
the higher hydrocarbons since it is relatively
reactive. It was found that the presence of
ethylene increased the carbon formation rate
from methane but it was independent of the
concentration, see figure 2.
carbon starts to grow. Since hydrogen
suppresses the thermal cracking, the
hydrogen content has to be taken into
account as well. For a gas consisting of 50 %
H2, C2H4 ramping from 0-20 % in 10 hours and
the rest inert N2, carbon starts to grow at 18
% C2H4, see figure 3.
Figure 3 Shows that the carbon growth starts at 18 %
C2H4 in an atmosphere consisting of 50 % H2 and the
rest inert N2.
Another approach to determining the limit for
safe operation is to keep the ethylene
composition fixed and ramp the hydrogen
from a level where the hydrogen completely
suppresses carbon formation downwards until
the carbon starts growing on the surface. For
a gas consisting of 10 % C2H4, H2 ramping from
40-25 % in 8 hours and the rest inert N2,
carbon starts to grow at 34 % H2, see figure 4.
Figure 2 Shows how the carbon formation from
methane is influenced by the ethylene composition of
the gas. The gas flow through the reactor consists of 60
% CH4, 30 % C2H4 and 10 % H2 during initiation and then
steps through 15, 10, 5, 1 % C2H4 where the rest is CH4.
Limits for safe operation
In experiments conducted to set limits for safe
operation the materials were pre-treated by
heating in an atmosphere consisting of 50 %
H2 and 50 % N2 from 200-550 °C at 4 °C/min.
The heating in hydrogen partly reduces the
protective oxide layer providing a more
susceptible surface for carbon formation. It is
also providing more homogenous surface
conditions which give more consistent results.
An interesting point when studying higher
hydrocarbons is at what concentration the
Figure 4 Shows that carbon grows below 34 % H2 in an
atmosphere consisting of 10 % C2H4 and the rest inert
N2.
The product from the primary reformer is
syngas at about 900 °C that needs to be
cooled. Since steam suppresses the reduction
reaction of carbon monoxide a safe operation
limit for steam is sought. For a gas consisting
of 21 % CO, 63 % H2, steam ramping from 16-6
% in 10 hours and the rest inert N2, carbon
starts to grow at 6 % steam, see figure 5.
even though there is an enormous
thermodynamic potential for a carbon
forming reaction, the kinetics for the reaction
is extremely decisive to determine if carbon
will grow.
Sources of error
Figure 5 Shows that carbon grows below 6 % steam in
an atmosphere consisting of 21 % CO, 63 % H2 and the
rest inert N2.
Conclusions
A screening of various alloys showed that the
composition of the alloy has large influence
on the carbon formation rate. It was shown
that the poorer the alloy, the higher the
carbon formation rate.
It was also shown that the presence of a small
amount of ethylene in methane has a large
influence on the carbon formation rate. An
addition of x % ethylene in pure methane
increases the rate to the double. However,
the addition of more ethylene does not have
much effect on the carbon formation rate.
Limits for carbon free operation have been
investigated. For an atmosphere of 50 %
hydrogen one cannot go above 18 % ethylene,
where the rest is inert nitrogen, before carbon
starts to form (ac=8.4*1011). And for an
atmosphere of 10 % ethylene one cannot go
below 34 % hydrogen, where the rest is inert
nitrogen, before carbon starts to form
(ac=10.2*1011). For a syngas with an
atmosphere of 63 % hydrogen, 21 % carbon
monoxide one cannot go below 6 % steam,
where the rest is inert nitrogen, before the
carbon starts to form (ac=28.5). This shows
that it is irrelevant to compare carbon
activities from different carbon forming
reactions with one another. It also shows that
One of the largest sources of error that is also
the hardest to estimate is the incubation time
since it is no reading from the TGA indicating
how fast the carbon transfer from the gas to
the metal is before carbon protrusions starts
growing. Even though there is thermodynamic
potential for the carbon to form, it could take
years before the material is sufficiently
saturated with carbon for the cementite to
form, which is often the case in real process
equipment. Also, in the experiments
determining the carbon free operation limits,
the process gas compositions are constantly
varied. This means that the incubation time is
constantly varied. This may have an impact on
when carbon starts to grow, hence the carbon
free operation limits presented in this work.
The kinetics in the carbon forming reactions is
very slow which results in that the reactions
are far from equilibrium when carbon starts
to form. For example, in the experiments
determining the lower limits of steam for
carbon free operation, the slow kinetics of the
carbon formation reactions may give an
underestimate of the lower limit of steam at
which carbon starts to grow.
Future work
This project has focused on gaining general
knowledge of the influence of various gas
components not earlier treated in literature
and should therefore be considered a prestudy. Determining limits for carbon free
operation proved to be very time consuming
and since this project is set for a limited time,
further work is needed in order to develop a
severity function for carbon formation on
materials. Preferably, this could be done in a
pilot able to conduct experiments under
industrial conditions.
References
[1] M. L. Holland and H. J. de Bruyn, Metal
dusting failures in methane reforming plant,
Int. J. Pres. Vessels & Piping, 1996, 66, pp.
125-133.
[2] H. J. Grabke, Metal dusting, Material and
corrosion, 2003, 54, pp. 736-746.