The Relevance of Bicine in the Corrosion of Amine Gas Treating Plants

Sidney F. Bosen, Ph.D.

The Dow Chemical Company
820 Gessner, Suite 600
Houston, TX 77024
Stephen A. Bedell, Ph.D.
The Dow Chemical Company
2301 N. Brazosport Blvd., B-1605
Freeport, TX 77541
Abstract
In spite of the presence of bicine in a variety of gas treating amine solutions, both
formulated and generic, there is a lack of associated corrosion. An explanation for the
lack of corrosion and operating strategies for controlling corrosion in the few applications
that are affected will be presented.
Introduction
In the past bicine, N-N-bis(2-hydroxyethyl)glycine CAS# 150-25-4, has been reported
by a number of investigators as a degradation product of alkanolamine and
alkylalkanolamine solutions.
1,2
Several different mechanisms for producing bicine in the
amine solution were presented. These pathways include reaction of diethanolamine
(DEA) with glyoxal, a common H2S scavenger, and exposure of the heated gas treating
solution to an oxidizer, either dioxygen or sulfur dioxide. It should be noted that the two
oxidative pathways also result in the formation of secondary amines. When the amine
solution is composed of tertiary amines for selective H
2
S removal, even low levels of
secondary amines decrease CO
2
slip.
Bicine impacts gas treating amine solutions in two ways. First, it forms Heat Stable
Amine Salts (HSAS) since the carboxylic acid is a strong acid and the amino group is a
weaker base than commonly used gas treating amines. HSAS attributable to bicine can be
reduced like other HSAS acids by the addition of stronger bases to the gas treating
solution. Second, bicine has been thought to contribute to the corrosivity of gas treating
amine solutions. This hypothesis is supported by the findings of autoclave weight loss
corrosion studies
1
and experiments with a laboratory scale amine plant like test
apparatus equipped with electrochemical corrosion monitoring
3
. Both test methods show
increased carbon steel corrosion rates when bicine is present.
Based on the above, bicine was added to the list of routinely measured components of
1,739 amine samples submitted to the Dow Gas Treating Analytical Laboratory. The
survey of bicine content of these samples and the impact of bicine on amine plant
corrosion will be discussed.
Keywords: bicine, carbon steel corrosion, amine plant corrosion, oxalic acid, hydrogen
sulfide
Experimental Method
Bicine content of the amine solutions was determined by gas chromatography. Bicine
forms a volatile lactone (internal ester) in the GC injection port. Calibration standards
were prepared from authentic material purchased from the Aldrich Chemical Company.
The minimum level of detection for the GC method employed was 0.01 wt% (100
ppmw).
All of the GC sample results along with other usually measured parameters were stored in
NAUTILUS
1
, a laboratory information management system (LIMS). The LIMS data
was searched on bicine content and the search results were put into an Excel 97
2
(spreadsheet) workbook. Standard techniques, histogram and correlation, were used to
evaluate the spreadsheet data.
Results
The time period covered in the survey was 17 months. There were 1,739 samples
submitted by 273 different gas treating plants where bicine content was determined.
Since the routine testing protocol does not record bicine content of MEA solutions, a few
selected MEA solution samples were re-analyzed. In several of these MEA solutions low
levels of bicine were found and confirmed by gas chromatography/ mass spectrometry
(GC/MS). Gas chromatography and GC/MS were used to check a few DIPA solutions for
bicine. Since bicine was not found in these DIPA solutions and time did not permit
testing of all MEA solutions, both were excluded in this study.
Amine solution samples were from gas treating applications that covered the usual
spectrum from complete H
2
S and/ or CO
2
removal to selective H
2
S removal. Bicine was
found in samples from 163 or 59.7% of the plants. More than half of the plants using
generic amines, DEA and MDEA, had bicine in their samples. It was not unexpected that
for applications such as Claus Sulfur plant tail gas treating, all of the plants have bicine in
the samples. Operational upsets in the Claus Sulfur plant or quench water system can
result in SO
2
intrusion in the amine system.
Figure 1 gives the distribution of samples for ranges of bicine content. Note that 825
(47%) of the samples tested had 0.01 wt% (100 ppmw) or higher bicine. Figure 2 shows
the distribution of bicine content when it was found. It is interesting that 346 (42%) of the
samples that contained bicine had 0.25 wt% (2,500 ppmw) or higher.
Standard correlation data analysis were run on samples that had reported bicine along
with iron, H
2
S, CO
2
, oxalate, glycolate and thiosulfate. Correlation coefficients are given
in Table 1. There were no statistically significant correlations found between bicine and
the other analytes.
____________________________
1 Copyright Thermo LabSystems
2 Copyright Microsoft Corporation
While a number of laboratory studies have concluded that bicine content of gas treating
amine solutions should be kept at low levels, < 250 ppm, to minimize corrosion, 87% of
samples that contained bicine exceed the 250 ppm recommended maximum. While the
number of accounts with active corrosion among this sampling is not fully known, there
have not been an extraordinary number of reports from accounts of active corrosion. For
example a main amine system of a Southeastern United States refinery was the source of
75% of the samples that had 1.3 wt% (13,000 ppm) or more bicine. Past records indicate
bicine levels have been in excess of 250 ppm since 2000. Refinery personnel have not
reported any active corrosion of the refinery main amine system.
Another account that occasionally has bicine above 250 ppm is removing only CO2 from
a gaseous ethane stream. This plant is also monitoring corrosion with weight loss
corrosion coupons in the lean and rich streams. Typical coupon exposure time is 60 days.
Weight loss corrosion rates and bicine content are given in Figure 3. Lean solution
weight loss corrosion rates ranged from 2 to 12 mpy and rich solution weight loss
corrosion rates ranged from 0.4 to 1.0 mpy. The difference in corrosion rates is most
likely due to the difference in temperature of the corrosion probe locations, 250
o
F lean
and 180
o
F rich.
Discussion
Laboratory studies are used to model many process corrosion problems and provide some
guidance to plant operators. The current plant operating experience makes one question
what role bicine plays as a carbon steel corrodent in amine gas treating plants. Careful
examination of prior work does show some extrapolation of conditions is needed to
extend the results to operating amine units. Autoclave tests used by Rooney et al
1
did not
contain either CO
2
or H
2
S. While Cummings et al
3
have used testing apparatus that
appears to replicate the conditions in an amine plant, some of the behaviors observed do
not. For example Cummings et al show in Figure 4 of reference 3 an increase of
corrosion rate only when H
2
S is added to the system, not an increase followed by a
decrease. Since it was stated that the corrosion probes were the only carbon steel
components in contact with the solution and that bisulfide reacts almost instantaneously
with carbon steel, one would expect the corrosion rate to rapidly rise and then almost as
quickly fall to a very low observed rate. Instead the graph shows a constantly increasing
corrosion rate that is above 20 mpy after 24 hours. This reported behavior could be
rationalized by either electrochemical oxidation of H
2
S, which has been reported earlier
4
,
or it could be from constant mechanical removal of the iron sulfide film by the flow
dynamics of the corrosion cell. It is well known that iron sulfide films can be removed
from carbon steel surfaces when sufficient shear stress is applied to the film. This is the
basis for 6 ft/sec as the recommended maximum fluid velocity
5
in carbon steel piping for
lean amine solutions when H
2
S is present.
We believe a more detailed examination of what is occurring when bicine is present in
gas treating amine solutions is needed. In considering the effects of bicine on corrosion of
carbon steels, its high affinity
6
toward iron is often cited, especially the equilibria
between ferrous ion and bicine;
Fe(II) + bicine
-1
[Fe(bicine)]
+
K
ML
(1)
[Fe(bicine)]
+
+ bicine
-1
[Fe(bicine)
2
]
K
ML2
(2)
where K
ML
= 10
4.3

and K
ML2
= 10
3.0
.

bicine in these equations refers to the bicine
anion, which should be the dominant form at the pH encountered in the gas treating
solutions. The protonation equilibria clearly show this:
bicine
-1
+ H
+
[bicineH]
K
H1
(3)
[bicineH] + H
+
bicineH
2
+
K
H2
(4)
Where K
H1
= 10
8.1
and K
H2
= 10
1.7
. Thus at pH 10, 98.7% of the bicine is fully
deprotonated, with the remainder being in the neutral (zwitterionic) form.
The low corrosion rates that have been seen with relatively high levels of bicine caused
us to question the coordination chemistry of bicine. Though one can point to equilibrium
constants and show that bicine has a reasonable affinity toward iron, the chemistry of an
amine solution can be much more complicated. In theory, equilibrium modeling can
predict solution speciation for gas treating solutions if all component interactions are
known. Unfortunately this is not the case. When considering metal complexes, the effect
of alkali metal ions is often overlooked because the equilibrium contants for complex
formation are small. One gas treating solution had as much as 1.6wt% (0.097M) bicine
along with 0.54M alkali metal ions (a combination of Na
+
and K
+
which we will refer to
collectively as A
+
). The high levels of alkali metal ions were a result of heat stable salt
neutralization.
In order to consider the effects of alkali metal, a simple system was modeled with the
components of bicine, A
+
and Fe(II). Even this simple system requires some educated
estimates for interaction parameters. The complexation of bicine with an alkali metal ion
(A
+
) is represented by the following equilibrium and equilibrium constant.
K
A+
bicine
-
+ A
+
[A(bicine)]
(5)
Values for K
A+
have not been reported in the literature, so that a proper value must be
estimated by examining interactions between A
+
and similar compounds. Several of
these are shown in Table 2.
EDTA (ethylenediaminetetraacetic acid), with 6 donor groups, is not a good model for
bicine, though comparison with HEDTA (hydroxyethylethylene-diaminetriacetic acid -
also 6 donors, but one carboxylate of EDTA replaced with a hydroxyethyl group) shows
that the harder carboxylate group is much more effective at complexation of A
+
. This
same effect explains the low affinity of TEA relative to carboxylate containing ligands.
Monocarboxylic acids of amines such as glycine and alanine show quite different
affinities. In light of these model compounds, the value of log K
A+
for bicine has been
estimated for this study at between 0.1 and 0.5. It should be pointed at that the few
values reported for K
+
are slightly lower than those for Na
+
. The heats of complex
formation, also shown in the table, vary from negative in the case of EDTA to positive in
the case of alanine. Positive enthalpy values are also reported in the NIST database
6
for
Na
+
+ (acetate or oxalate). A positive H for complex formation means that the
equilibrium constant will increase as temperature is raised.
Figure 4 shows how the degree of complexation of bicine by alkali metal ions is affected
by the equilibrium constant for 0.097 molar bicine and 0.54 molar A
+
. In the chosen
range of log K
A+
(0.1-0.5), the percentage of complexed bicine ranges from 39-59%.
In this section, the effect of bicine-alkali metal complexation on FeS solubility will be
considered. The equilibrium constant equations and values shown above were used to
solve the following mass balance equations for a system containing bicine, ferrous ion
and sodium.
Total Fe = [Fe(II)] + [Fe(bicine)]
+
+ [Fe(bicine)] (6)
Total Bicine = bicine
-1
+ [bicineH] + [Fe(bicine)]
+
+ 2[Fe(bicine)] + [A(bicine)] (7)
Total Alkali Metal = A
+
+ [A(bicine)] (8)
The concentrations of bicine and A
+
are shown in the table 3 for each case. In
considering the amount of FeS which could dissolve in the solution, it was assumed that
the total Fe and total S were equal (in other words, the solution did not contain any iron
or sulfide other than what could dissolve). The amount of total Fe and sulfur was
changed until the product of free Fe(II) and S
-2
exceeded the solubility product constant
for FeS of 6.3x10
-18
.
Comparison of cases A-D as well as F-H show considerable reductions in the solubility
of FeS as the alkali metal concentration is increased. Both the low and high ranges of log
K
A+
were considered. Case I shows what would happen if the affinity of A
+
for bicine
was much larger. Cases J and K show that the same high level of A
+
would be more
effective in reducing FeS solubility at lower total bicine concentrations. It is clear from
comparison of cases J and A that increased bicine concentration results in increased FeS
solubility. The purpose of this table is to demonstrate that high levels of alkali metal salt
can greatly diminish the equilibrium solubility of FeS. A previous study by Comeaux
10
presented a similar analysis of metal complexants (other amines) in an MEA solution in
order to explain increased solubility of FeS. How the reduction of FeS solubility affects
corrosion is not entirely known, but in a transport controlled dissolution, the dissolution
rate can be increased by increasing the equilibrium solubility of the solute
11
.
The model used to generate the results above did not include many other possible
interaction parameters. As discussed earlier, the interaction of TEA with A
+
is probably
much less than that of bicine with A
+
, so interaction between MDEA and A
+
is also
probably low. More interesting is the possibility of interaction of bicine with protonated
amines (from acid gas loading or heat stable salts). De Robertis
12
et al studied the
complex formation between acetate and several amines and found interactions with
equilibrium constants in the same range as those used for bicine and A
+
. These other
binary interactions may also reduce the ability of bicine to complex iron. Of course it
should be noted that other anion interactions with A
+
could diminish the effects noted
between A
+
and bicine. Similar considerations can be given to other potential corrosive
anions such as oxalate. Values of equilibrium constants for oxalate complexes of Na
+
are
reported between 0.5 and 0.9 (log values) in the NIST database.
In summary, these calculations show that the presence of large amounts of cations may
account for reduced corrosion rates in gas treating solutions containing bicine. For the
lean solution cited above (1.6 wt% bicine, 381 ppm H2S, 0.54 molar alkali metal cation,
pH 10.0 at ambient temperature) the solubility of ferrous iron assuming log K for
(A+bicine) of 0.3 is 12 ppb (parts per billion); without the alkali metal 47 ppb.
Rooney et al
1
have shown that bicine and oxalic acid have similar molar effects on the
corrosion of carbon steel when acid gases are absent. This paper also shows that addition
of caustic to the test amine solution reduces the corrosivity of both bicine and oxalic acid.
Pourbaix diagrams [potential (volts versus standard hydrogen electrode) versus pH], can
be used to predict the thermodynamically stable species present in an amine solution in
contact with a carbon steel surface. Since bicine was not included in
CorrosionAnalyzer
TM 3
software used to generate Pourbaix diagrams, oxalic acid was
chosen as a surrogate chelant. The Pourbaix diagrams were further simplified by
substituting iron for carbon steel. Conditions and chemistry used to generate the Pourbaix
diagrams, Figures 5-8, are given in table 4. Please note the solutions also contained the
indicated amount of oxalic acid. The high temperature conditions were chosen to
replicate those given by Cummings et al
3
where possible with the exception of pressure at
the highest lean temperature. The software predicted solution boiling if the pressure was
not increased. The diagrams also show predicted amine solution pH at the specified
temperature. When H2S is present at the predicted solution pH for both 120
o
F and 250
o
F cases (Figures 5 and 7 respectively), protective iron sulfide films are
thermodynamically favored. When H
2
S is absent at the predicted solution pH for both
120
o
F and 250
o
F cases (Figures 6 and 8 respectively) protective films are not
thermodynamically favored. The reader is reminded that Pourbaix diagrams only show
what is thermodynamically stable, not the time to reach stability.
Figures 5 and 7 also predict loss of the protective iron sulfide films if the solution pH is
not between 7.2 and 12.3 for the 120
o
F case or 6.5 to 10.5 for the 250
o
F case.
Cummings et al
3
stated that corrosion rates increased when they attempted to neutralize
the small volume of circulated amine solution in the Mini Amine Plant. They may have
inadvertently raised the pH of the amine solution in the corrosion test cell high enough
and long enough to destabilize the protective iron sulfide film on the test electrodes.
Proper neutralization of amine solutions is more complicated than just adding caustic to
the amine solution. Mixing two liquids of significantly different densities and viscosities
requires some engineering especially when the system is in non-turbulent flow and has
___________________________
3 Copyright OLI Systems Inc.
very little back mixing.
Figures 6 and 8 also represent the environment occasionally found in Claus Sulfur plant
tail gas amine treating systems. The lean solution is completely stripped of H
2
S in order
to meet/ exceed the treated gas specification. If defects occur in the protective films on
the carbon steel surfaces and H
2
S is absent, the usual insoluble ferrous oxide/ hydroxide
film is formed. Bicine helps solubilize the ferrous oxide/ hydroxide corrosion product.
Bicine chelated iron is then transported to the absorber where H
2
S precipitates iron from
the complex. Bicine is then available to pick up more iron from the exposed freshly
corroding carbon steel surface. Sometimes it is easy to diagnose this corrosion
mechanism- the lean solution is clear and the rich solution is full of black particles.
Allowing more H
2
S to remain in the lean solution easily controls this type of corrosion.
The software predicts a protective iron sulfide film is thermodynamically favored with as
little as 10 ppm H
2
S (0.0001 m/m for 35 wt% MDEA) in the lean solution at both
temperatures. Many plant operators over strip amine solutions to give more reaction time
for controlling upsets. Since bisulfide reacts very rapidly with carbon steel, operating
with less reaction time cushion only has to be done for a relatively short time period.
Conclusions
Bicine is found in a significant percentage of amine gas treating solutions, both generic
and formulated amines.
The amount of iron that can be chelated by bicine is reduced when group 1 metal
cations (sodium/ potassium) are added to the amine solution when heat stable amine salts
are neutralized.
Corrosion of carbon steel attributable to bicine is minimized when 10 ppm or higher
H2S is present in the amine solution in contact with the metal surface.

References
1. Rooney, P.C., Bacon, T.R. and DuPart, M.S. Effect of Heat Stable Salts on Solution
Corrosivity of MDEA-Based Alkanolamine Plants. Part III, presented at the
Laurence Reid Gas Conditioning Conference, Norman, Oklahoma 1997
2. Critchfield, J. E. and Jenkins, J.L., Petroleum Technology Quarterly, spring 1999, pg.
87
3. Cummings, A.L. , Veatch, F.C. and Keller, A. E. ,1997, Corrosion and Corrosion
Control Methods in Amine Systems Containing H
2
S,paper No. 341 presented at
NACE CORROSION/97.
4. Asparger, R.G. , Liu, H.J. , and Dean, J.W. Accurate, Reproducible Measurement of
Reduced Corrosion in Gas Treating Amine Systems after Application of Judicious
Neutralization Techniques, presented at the AICHE 1995 Spring National Meeting,
Houston, Texas, March 1995.
5. Nielson, R.B. , Lewis, K. R., McCullough, J.G., and Hansen, D.A., 1995, Corrosion
in Refinery Amine Systems, paper No. 571 presented at NACE CORROSION/95.
6. Martell, A. E. and Smith, R.M., Critically Selected Stability Constants of Metal
Complexes, NIST Standard Reference Database No. 46, 2001, NIST, Gaithersburg,
MD.
7. James, T.L., and Noggle, J. H., J. Amer. Chem. Soc., 1969, 91, 3424.
8. Casale, A., De Robertis, A and De Steffano, C., Thermochimica Acta, 1995, 255,
109.
9. Casale, A., De Robertis, A and De Steffano, C., Thermochimica, Acta, 1989, 140, 59.
10. Comeaux, R. V., Preprint, Amer. Pet. Inst. 27
th
Meeting, May, 1962.
11. Stumm, W. and Morgan, J. J., "Aquatic Chemistry", 3rd Ed.John Wiley and Sons,
1996.
12. De Robertis, A. , De Stefano, C., Giuffre, O. and Sammartano, S., J. Chem. Soc.,
Faraday Trans., 1996, 92, 4219.
Table 1. Correlation Coefficients between Bicine and Other Analytes
Analyte Correlation Coefficient
with Bicine
Iron 0.0217
Sodium 0.0096
H2S -0.0158
CO2 -0.0268
Thiosulfate 0.1617
Oxalate 0.5627
Glycolate 0.3675
Thiocyanate -0.0365
Table 2. Values of KA+ for Model Chelants
Structure Log K
Na
(25
o
C,
K in L/mol).
H
(kJ/mol)
Reference
(CH
2
CH
2
OH)
2
N(CH
2
CO
2
H) (Bicine)
(HO
2
CCH
2
)
2
NCH
2
CH
2
N(CH
2
CO
2
H)
2
(EDTA)
1.8 -5.4 6
(HO
2
CCH
2
)
2
NCH
2
CH
2
N(CH
2
CO
2
H)(CH
2
CH
2
OH) (HEDTA)
1.0 7
HN(CH
2
CO
2
H)
2
(IDA) 0.36 6
H
3
CN(CH
2
CO
2
H)
2
(MIDA) 0.61 6
N(CH
2
CH
2
OH)
3
(TEA) -1.1 6
H
2
N(CH
2
CO
2
H) (Glycine) -0.5 8
H
2
NHCH(CH
3
)

CO
2
H (Alanine) 0.30 7 9
Table 3. Comparison of Ferrous Sulfide Solubility with Alkali Metal Ion Concentration
and Complexation Constants.
Case [Bicine]
total
[A
+
]
total
Log
K
A+
Total [FeS]
solubility
*
Ratio to
case A
A 0.097 0.00 9.8 x 10
-5
1.00
B 0.097 0.10 0.1 8.8 x 10
-5
0.90
C 0.097 0.30 0.1 7.3 x 10
-5
0.74
D 0.097 0.54 0.1 6.1 x 10
-5
0.62
F 0.097 0.10 0.5 7.8 x 10
-5
0.80
G 0.097 0.30 0.5 5.5 x 10
-5
0.56
H 0.097 0.54 0.5 4.0 x 10
-5
0.41
I 0.097 0.54 1.0 1.8 x 10
-5
0.18
J 0.030 0.00 3.1 x 10
-5
0.32
K 0.030 0.54 0.5 1.2 x 10
-5
0.12
*
[FeS] is the total amount of FeS which the solution could hold. Iron and sulfide will be
in several solution forms, such as Fe(bicine), Fe(bicine)
2
, H
2
S, HS
-
and S
=
.
All concentrations in moles/L.
Table 4. Pourbaix Diagram System Conditions and Chemistry
System Parameters Cold Lean Hot Lean
Temperature,
o
F (
o
C) 120 (48.9) 250 (121.1)
Pressure, psia (kPa) 29.7 (204.8) 35 (241.3)
Components
MDEA, wt% 38.50 38.50
H
2
O, wt% 58.49 58.49
H
2
S, wt% 0.038 0.038
Oxalic Acid present, wt% 0.68 0.68
Oxalic acid equivalent to
1.596 wt% bicine, wt% 0.86 0.86
Group 1 cation hydroxides,
wt%
1.43 1.43
Figure 1. Bicine Distribution of all Samples
Figure 2. Sample Bicine Distribution When Present
Figure 3. CO
2
Removal Plant Bicine and Corrosion Coupon Data
0
2
4
6
8
10
12
14
02/13/02 05/24/02 09/01/02 12/10/02 03/20/03 06/28/03 10/06/03
Date
C
o
r
r
o
s
i
o
n

R
a
t
e
,

m
p
y
0
500
1000
1500
2000
2500
3000
B
i
c
i
n
e

C
o
n
c
e
n
t
r
a
t
i
o
n
,

p
p
m
Lean Coupon
Rich Coupon
Bicine
0
10
20
30
40
50
60
70
80
90
100
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
log K
A+
%

o
f

b
i
c
i
n
e

c
o
m
p
l
e
x
e
d

b
y

A
+
Figure 4. Effect of log K
A+
on Bicine Complexation
Figure 5. Pourbaix Diagram Refinery Main System, 120
o
F Oxalate Substituted for
Bicine, H
2
S Present.
Figure 6. Pourbaix Diagram, Refinery Main System, 120
o
F Oxalate Substituted for
Bicine, H
2
S Absent
Figure 7. Pourbaix Diagram, Refinery Main System, 250
o
F Oxalate Substituted for
Bicine, H
2
S Present
Figure 8. Pourbaix Diagram, Refinery Main System, 250
o
F Oxalate Substituted for
Bicine, H
2
S Absent