報 文Boshoku Gijutsu, 38, 9-18 (1989)

UDC 620.194:539.42:620.193.4:669.14:547.23

Stress Corrosion Cracking of Carbon Steel in Aqueous

Diethanolamine and Monoethanolamine Solutions

Z. A. Foroulis*

* Exxon Research Engineering Co.

Electrochemical voltametric scans and slow-strain rate tests have been used to characterizethe stress corrosion cracking behavior of carbon steel in diethanolamine (DEA) and mono-ethanolamine (MEA) solutions. Pure DEA and MEA solutions show a weak electrochemicalreactivity toward carbon steel and do not promote cracking. However, solutions of theseaminoalcohols saturated with CO2 show considerable electrochemical reactivity and a strongtendency to cause stress corrosion cracking in the potential range -0.8 to -0.6 V (SCE). InCO2 saturated MEA solutions cracking was intergranular with a minor tendency for trans-

granular cracking only at lower potentials. In CO2 saturated DEA solutions only trans-granular cracking was observed in the whole potential range of cracking. The mechanisticsignificance is discussed.

Key words: stress corrosion cracking, monoethanolamine, diethanolamine, carbon steel

1. Introduction

The occurrence of stresscorrosion cracking ofmild steel in hot potassium carbonate/bicarbon-

ate solutions in practice has been described in

the literature1)-3). Laboratory studies have con-firmed this phenomenon and provided reason-

able rationalization of the cause and probablemechanism4)-7). More recently the occurrence

of stress corrosion cracking was also reported innon-stress relieved welds in aqueous monoetha-

nolamine3)9) (MEA) solutions used in acid gas

(C02 and/or H2S) treating facilities. Laboratorystudies were carried out by several investigatorsin an effort to gain an understanding of the

cause and probable mechanism of stress corro-sion cracking in MEA solutions and more gen-

erally aminoalcohols. Gutzeit and Johnson9)using slow-strain-rate (SSR) tests at various po-

tentials were able to induce cracking in leanMEA solutions containing an unspecified cor-

rosion inhibitor, but they report that cracking

occurred to a lesser degree in the absence ofthe inhibitor. Parkins and Foroulis, were able

to simulate the stress corrosion cracking byMEA solutions in the laboratory and concluded

that this phenomenon is probably another form

of the C032-/HC03- cracking10). However, ques-tions are raised by the results of Lenhart etal.11) regarding the above interpretation whichassumes that the amine plays no role other thanto provide C032-/HC03- ions in the presence ofdissolved C02. These investigators were unableto produce stress corrosion cracking of carbonsteel in 20% diethanolamine (DEA) solutionsusing slow-strain-rate (SSR) tests. To get abetter understanding of the role of DEA andMEA on the stress corrosion cracking behaviorof mild steel, a laboratory program was there-fore undertaken to study the SCC of carbonsteel in DEA and MEA solutions with andwithout the absorption of C02. The investiga-tion involved a combination of electrochemicalvoltametric studies and slow-strain-rate (SSR)tests at the free corrosion potential and at vari-ous controlled potentials. This report summa-rizes the results of this study and providesfurther insight on the mechanism of SCC ofmild steel in aminoalcohols solutions.

2. Experimental approachThe experimental technique used was the

slow-strain-rate test in which smooth tensilespecimens are strained to failure at a fixedrate in the environment of interest. The test* P.O. Box 101, Florham Park, NJ 07932, U.S.A.

10 Boshoku Gijutsu

solution was maintained in a cylindrical tefloncell of one liter capacity which fitted on andaround the test rod via a water tight seal.The cell has a closed teflon lid equipped withsuitable openings for introduction and outflowof purge gas (N2 or C02), a platinum counterelectrode, a bridge connected to an externalsaturated calomel electrode, a reflux condenserand a thermocouple for temperature control.Heating was achieved with an external heatingmantle, and the test temperature was controlledto +2C.

Electrochemical voltametric studies using car-bon steel electrodes in the various test solutionswere carried out in the active and passive po-tential range in order to define the electro-chemical behavior of mild steel in the amino-alcohol solutions.

Slow-strain-rate (SSR) tests were carried outboth at the free corrosion potential and at vari-ous controlled potentials, which electrochemicalvoltametric studies have defined as areas ofmaximum anodic electrochemical reactivity.The design of electrodes and cell design usedfor the voltametric polarization studies was de-scribed previously12)

In slow-strain-rate tests carried out at the freecorrosion potential, the potential was monitoredcontinuously. In tests carried out at constant

potential, the potential was controlled by meansof a potentiostat. The stress corrosion testspecimens were prepared from cold drawn car-bon steel (AISI 1018) rods 0.63 cm diameter.The rods were out into 33 cm long pieces andmachined into tensile specimens having a gagediameter of 0.25 cm and a gage length of 1.27cm. After machining, the specimens were usedwithout any further heat treatment. Prior torunning the tests, the specimens were surfac

polished with 600 grit SiC paper, rinsed withwater and finally degreased with acetone.

SSR tests were carried out in solutions of25 wto and 43 wto diethanolamine (DEA) inwater at 99C. In addition, several tests were

also carried out with a 25 wto monoethanol-amine (MEA) solution. Most of the tests werecarried out with 43o DEA solution which re-

pesents an equal molar concentration of the25 wto MEA. Tests were run both with solu-tions saturated with nitrogen and solutions sat-urated with C02 at test temperature and atmos-

pheric pressure. Several tests were also run inan inert parafinic base turbine oil (Tereso 32,

product of Exxon Corporation) to simulate theductile fracture expected in a non-aggressiveenvironment. Tests were carried out at strainrates in the range 2 x 10.7 to 4 x 10.6 sec-1. Thesestrain rates were selected because experience inthis and other laboratories indicated that stresscorrosion cracking of carbon steel generally oc-curs at strain rates between about 10.5 to 10.7sec-1.

The fracture and side surfaces of all speci-mens tested were examined with an optical anda scanning electron microscope (SEM) followingcathodic cleaning of the specimens using a pro-

prietary alkaline solution (Endox 214). In addi-tion, metallographic examination of longitudinalsections through the fractured test pieces wasalso carried out for evidence of side cracks.

The assessment of the results was carried outin terms of the nature of fracture (micro-voidcoalescence vs. multiple cracking with inter-

granular or transgranular facets), morphology ofside cracks, percent reduction in area, percentelongation, time to failure and maximum nomi-nal stress calculated on the basis of the originalcross section area. Average crack propagationrates were determined from the deepest crackmeasured on the final fracture surface.

3. Results and discussionAs shown previously by several investiga-

tions4)-7),10) the SCC behavior of mild steel inalkaline media such as carbonate/bicarbonatesolutions and aqueous MEA is likely to occuronly in a restricted range of electrochemical

potential. To define the potential range forpossible SCC tendency in aqueous DEA andMEA solutions, both N2 and C02 sat'd, anodicvoltametric scans were run. Typical anodiccurves depicting the anodic scans in (N2 sat'd)solution and (C02 sat'd) solution are shown

schematically in Figure 1 for 43o DEA solution.A speed of about 50 my/min was used. Figure1 shows that in C02 sat'd solutions the potentialzone of maximum anodic electrochemical re-

activity lies between -0.8 to -0.6 V (SCE).In N2 sat'd solution the potential range formaximum anodic reactivity is somewhat shiftedto lower potential. The maximum anodic cur-rent density within the potential zone of electro-chemical reactivity is about 8 x 10 A/cm2 forthe C02 sat'd solution which is considerablyhigher than the 8x10-4 A/cm2 for the N2 sat'dsolution.

Vol. 38, No. 1 11

Several tests were also run in this study using25 wto MEA solution so that a comparisoncould be made with a DEA solution of equiva-lent molar concentrations. The results of sev-eral voltametric scans of carbon steel electrodesin 25 wto MEA solutions both N2 sat'd andC02 sat'd at 99C at atmospheric pressure areshown in Figure 2 in which the results of volta-metric seams with equivalent molar concentra-tions of 43o DEA are also included. Curve

(C) shows voltametric scan in 25o MEA solu-tion saturated in C02 and curve (D) shows thescan in N2 sat'd MEA solution. Comparisonof the MEA and DEA voltametric scans showsthat the MEA solution exhibits higher electro-chemical reactivity within the active potentialloop than an equivalent molar concentrationDEA solution. Furthermore, in both MEA andDEA solutions the reactivity whthin the active

potential loop is much more pronounced whenthe solution is saturated with C02 than whensaturated with nitrogen.

For the same amine solution, MEA or DEA,the much weaker electrochemical reactivity inthe active potential loop when the solution isN2 sat'd is attributed to much higher solution pH(about 11.4 for 43% DEA and 12.2 for 25%MEA) which tends to promote spontaneous pas-

sivation and decreases the maximum critical

current density for passivity corresponding to

the peak current on the active potential loop.

In the C02 sat'd solutions the pH values of

these solutions are considerably lower, pH 9.6

for C02 sat'd MEA and pH 9.5 for C02 sat'd

DEA. The lower pH alone could not explain

the much higher critical current density for

passivity observed within the active potential

loop in these solutions. As discussed subse-

quently, the presence of HC03- ions which are

produced by saturating the solutions with C02are likely contributors to accelerated anodic re-

activity of these solutions.

It is worth noting that in the C02 sat'd solu-

tions there is not significant difference in solu-

tion pH between the respective MEA and DEA

solutions to account for the higher electrochemi-

cal reactivity within the active loop of the MEA

solution in comparison with a solution of equi-

valent molar DEA concentration. The differ-

ence in reactivity must therefore be attributed

to some other effect of the amine molecule.

The potentials corresponding to the peak

anodic current density in the polarization curves

shown in Figures 1 and 2 lie in the range of

Fig. 1 Voltametric potential-current scans ofcarbon Steel in 43% DEA solution C02sat'd (Curve A) and N2 sat'd (Curve B)at 99C.

Fig. 2 Voltametric potential-current scans ofcarbon steel in amine solutions at 99C.Curves A and C for C02 sat'd solutionsof 43% DEA and 25% MEA and CurvesB and D for N2 sat'd solutions of 43%DEA and 25% MEA respectively.

12 Boshoku Gijutsu

Table 1 Slow-strain-rate tests of carbon steel (AISI 1018) in 25 and 45% DEA solutions at99C.

NC-No Cricks; I-Irinspranulu Crick;-date not iullable.

Fig. 3 SEM microphotograph showing the trans-

granular crack at the fracture surfaceof test 28D.

Fig. 4 Cross section microphotograph showing

transgranular side cracks in the neck-

down area of carbon steel in 25% DEA

CO sat'd at -0.7 V (SCE). Test 31D,

2% nital.

Vol. 38, No. 1 13

-0.80 to -0.85 V (SCE) for N 2 sat'd solutionsand about -0.65 to -0.70 V (SCE) for C02sat'd solutions. These potentials corrected forthe effect of temperature, are near the thermo-dynamic equilibrium potentials corresponding toformation of Fe304 surface films13).

Several slow-strain-rate tests were carried out

primarily with DEA solutions saturated withC02 (1 Atm) at 99C at the free corrosion

potential and at various strain rates. In addi-tion, a test was also carried out with N2 sat'dsolution at the free corrosion potential.

A further series of tests was carried out withthe potential controlled in the potential zone of-0.8 to -0.6 V (SCE) which the electrochemical

scans indicate as a potential zone of maximumanodic reactivity. Several tests were also car-ried out at higher potentials which lie within thezone of lower electrochemical reactivity and

passivation. In addition, tests were also run inan inert oil, Tereso 32, for comparison. Mostof the tests were run with a 43o DEA solutionwhich is an equal molar concentration to 250MEA solution. However, one test was also runwith 25% DEA solution which exhibited similarbehavior.

The results of these tests are summarized inTable 1. In test 6D which was run at the freecorrosion potential the solution was purged withN2 prior to and during the test. This testexhibited a purely ductile failure with no evi-dence of secondary cracking. This is not sur-

prising since the solution pH was about 11.4which tends to cause the spontaneous passiva-tion of the steel surface at the free corrosion

potential -0.40 V (SCE), which lies within thepassive potential region.

Tests SD and 10D to 14D were carried out atthe free corrosion potential in C02 sat'd solu-

tions at different strain rates. With the excep-tion of test 13D, all tests exhibited a ductilefracture with no secondary cracking. In thesetests the free corrosion potential was about-0.83 to -0.84 V which is slightly outside the

potential zone for maximum electrochemical re-activity and thus the lack of clear evidence ofcracking is understandable. In test 13D thefree corrosion potential during the test drifted toa value of about -0.64 V which lies within thezone of maximum anodic reactivity. It is notknown why in the case of test 13D the Eerrdrifted within the SCC potential range whereasin the case of tests SD, 11D, 12D and 14D thecorrosion potentil remained outside the SCCrange. However, it is possible to speculate thatthis may be related to a combination of thedegree of solution saturation with C02 and thestrain rate used in this experiment. SEM exami-nation of the fracture and side surfaces in-dicated the presence of transgranular cracks.

To explore the tendency for SCC within the

potential zone of maximum electrochemical re-activity, several tests were carried out at acontrolled potential and two different strainrates. At the strain rate of 2.8x10-6 only test2D, which was run at -0.6 V, showed evidenceof small transgranular cracks. Tests run a -0.9, -0.8, -0.5, -0.4, -0.2 and +0.4V

exhibited no evidence of cracking with the frac-ture being purely ductile. To better establishthe cracking tendency within the potential zoneof maximum reactivity (-0.8 to -0.6 V (SCE)),tests 28D and 29D were run at a lower strainrate of 3 x 10-7. Both tests exhibited strongevidence of transgranular cracking; secondarycracks were also evident in the neckdown area.Figure 3 shows an SEM microphotograph of thetransgranular crack at the fracture surface of a

Table 2 Slow strain rate tests for carbon steel (AISI 1018) in 25% mear solution N2

sat'd and in an inert oil (Tereso 32).

NC=No cracking.

14 Boshoku Gijutsu

specimen tested at -0.7 V (Test 28D). Figure4 shows in cross section the appearance oftransgranular cracks in the neckdown area inspecimen of test 31D.

From the foregoing discussion it is clear thatwithin the potential zone (>-0.8 to -0.6 V) ofmaximum electrochemical reactivity in C02sat'd DEA solution, carbon steel is susceptibleto stress corrosion cracking. At potentials of-0.5 V and higher the steel surface remains

passive and stress corrosion cracking tendencywas not observed. Likewise at potentials lowerthan -0.8 V the anodic electrochemical reactiv-ity is less pronounced and the tendency to SCCis negligible.

Table 1 also includes data on several para-meters obtained from the SSR tests such as timeto failure, percent reduction in area at fracture

(RA), percent elongation, and RA/RA inert oil,which is the ratio of the percent reduction inarea in test solution to that in inert oil.

These parameters can also be used as a co-operative assessment of the ability of an en-vironment to induce cracking. However, theyare generally more useful in hydrogen inducedcracking which is accompanied by considerablereduction in ductility. These parameters are

generally less important by themselves as adiagnostic tool of stress corrosion cracking dueto anodic dissolution as can be seen by compar-ing the data in Table 1 which show that the

small reduction in ductility observed is not con-sistently related to SCC. This is not surprisingsince these parameters are probably markedlyaffected by the juxtaposition of cracks wheremultiple cracking occurs.

Table 2 shows a summary of SSR test dataobtained in 25% MEA N2 sat'd at 99C inwhich some data obtained in an inert oil (Tereso32) are also included. The data show that mildsteel in 25% MEA solution N2 sat'd at the freecorrosion potential as well as at controlled po-tentials within the potential range of maximumelectrochemical reactivity at E=-0.6V (SCE)and -0.8 V (SCE) did not show evidence ofstress corrosion cracking. Table 3 shows asummary of SSR test data in C02 sat'd 25%MEA solution, at both the free corrosion po-tential and at electrochemical potential control-led within the active and passive potential re-

gions. The data show that within the potentialrange -0.6 to -0.8 V (SCE) intergranularcracking was observed. At even lower potentialof -0.9 V both transgranular and intergranularcracking was noted. At potentials in the range-0.5 to +0.6 V (SCE) no cracking was ob-

served. At the free corrosion potential, crackingwas found only when considerable concentra-tion of C02 was absorbed by room temperature

presaturation (test 1G RR).The data in Table 3 also shows that two of

the three specimens tested in the potential region

Table 3 Slow strain rate tests for carbon steel (AISI 1018) in 25% MEA solution sat'dwith C02 at 99C, at 2.8 X 10-6 sec-1 strain rate.

X Solution presaturated with C02 at RT before test.I=Intergranular cracking, T=transgranular cracking.

Vol. 38, No. 1 15

of -0.6 to -0.8 V (SCE) have also exhibited

about a ten percent reduction in ductility asindicated by the percent reduction in cross sec-tion area and elongation. The third specimen,

test 7G, exhibited negligible change in ductility.At the slightly lower potential of -0.9 V (SCE)the specimen exhibited even slightly higher re-

duction in ductility as measured by the percentreduction in area. The intergranular morphol-

ogy of cracks observed in test 6G, is shown inFigures 5 and 6.

A comparison of the MEA with the DEA

cracking data presented thus far, shows that CO,

sat'd solutions of equal molar concentrations of

both MEA and DEA exhibit stress corrosioncracking tendency within the potential zone-0.8 to -0.6 V (SCE) of maximum electro -chemical anodic reactivity.

However, where as in the case of MEA astrain rate of 2.8 x 10-1 was adequate to causecracking in this potential range, in the case ofDEA solutions a much lower strain rate of about 3 x 10-7 was required to consistently pro-duce cracks. In addition the cracks observed inDEA solutions were transgranular whereas inthe case of MEA solution in this potential range,the cracks were intergranular. A tendency fortransgranular cracking in MEA solutions was only observed at lower potentials (<-0.8 V).

These differences in cracking behavior of MEAand DEA solutions suggest that the nature ofthe aminoalcohol molecule maybe of some signif-icance in the SCC processes. The evidence

presented thus far indicates that in the po-tential region of about -0.8 to -0.6V (SCE)carbon steel is susceptible to stress corrosioncracking in CO, saturated MEA and DEA solu-tion. In addition, the polarization data alsoindicate that carbon steel is subjected to activedissolution in the potential range of about -0.8to -0.70 V (SCE) and to an active-passive tran-sition (probably leading to formation of Fe3O4surface film) in the approximate potential rangeof -0.7 to -0.6V (SCE). Assuming that the

polarization curves correspond to near quasi-equilibrium conditions given the relatively low

polarization speed used it can be therefore con-cluded that SCC was observed both within theactive-passive potential zone (about -0.7 to-0.6 V) as well as within the active zone (about-0.8 to -0.7V). A plausible explanation of

these observations is discussed in subsequent

paragraphs.To gain a more quantitative measure of the

propensity for cracking of these solutions, anestimate of the average crack propagation rateat the potential of maximum electrochemicalreactivity was made from the deepest cracksmeasured on the final fracture surface in thecase of 43% DEA and 25% MEA sat'd with CO2

at 99C. The crack velocity was estimated tobe about 2.4 x 10-7 mm/sec in the DEA solutionand about 1 x 10-6 mm/sec in the equivalent mo-lar concentration MEA solution. The abovecrack propagation rates are lower than the crack

growth rate in 28% K2CO3 solutions sat'd withCO, which was found to be about 3x10-6mm/

Fig. 5 SEM microphotograph of the fracturesurface showing the depth of intergra-nular crack penetration of the fractureof a specimen (Test 6G) tested in 25%MEA solution C02 sat'd at 99C at E=-0.600 V (SCE).

Fig. 6 SEM microphotograph showing at higher

magnification the Intergranular nature

of the crack shown in Fig. 5.

16 Boshoku Gijutsu

sec in the same potential range6),7).The ratio of measured crack growth rates of

about 4.2 in equal molar MEA and DEA solu-tions compares reasonably well with the ratio ofthe maximum critical current densities for pas-sivity of about 5 estimated from the data shownin Figure 2 for MEA and DEA solutions re-spectively (curves A and C). This agreementcould be considered as a reasonable confirma-tion that the mechanism of cracking within theactive potential region involves an anodic dis-solution mechanism.

4. General discussion

The kinetics of anodic dissolution of iron inthe active potential loop in carbonate/bicarbon-ate solutions which have been reported previ-ously14,15) are of some significance in inter-

preting the SCC behavior of carbon steel inMEA and DEA solutions. Of particular signi-ficance in this regard are the findings by Daviesand Burnstein14) that in the potential regioncorresponding up to the anodic current peak,bicarbonate ions participate in the active dis-solution process which increase the iron dissolu-tion rate via formation of soluble Fe(CO3)22-species. The probable reaction sequence is giveby Eq- [1]-[3].

Fe-1-H20-Fe(OH)2(S)+2H++2e [1]

Fe(OH)2(S)+HC03-→FeCO3(s)+H2O+OH- [2]

FeCO3( S)+HCO3=Fe(CO3)22+H+ [3]

In addition to the effect of HCO3- on anodic

iron reactivity, other factors such as formationof reactive intermediates by the reaction of CO2

and aminoalcohols or even the influence of theaminoalcohol molecules on the iron dissolution

process may be important and their possibleeffect should be examined.

A comparison of the stress corrosion cracking

and electrochemical behavior of carbon steel in

MEA & DEA solutions summarized in Tables1 to 3 and Figure 2 in the present report with

that in CO2 sat'd K2CO3 solutions reported else-where6),7) indicates several pronounced similar-ities and also some differences.

The similarities relate to the fact that in allinstances carbon steel is susceptible to crackingin these solutions only when the solutions are

saturated with CO2, which undoubtedly pro-

duces HCO3-, and remains immune to stress

cracking in the absence of CO2. Another im-

portant similarity involves the fact that the threesolutions considered, namely CO2 saturated so-lutions of MEA, DEA and K2CO3, exhibit pro-nounced anodic electrochemical reactivity andstress corrosion cracking tendency within thesame potential range of about -0.8 to -0.6 V

(SCE).There are some significant differences how-

ever in regards to the electrochemical reactivityand stress corrosion cracking behavior of car-bon steel in these solutions that need to beconsidered. The differences include, the lowerelectrochemical reactivity of CO2 sat'd solutionsof DEA in comparison with MEA solutions ofessentially the same solution pH (Figure 2) andthe lower crack propagation rates estimated inthe case of DEA solutions as compared to MEAand K2CO3 solutions within the potential zone

(-0.8 to -0.6 V) of maximum cracking pro-pensity. Another important difference is incrack morphology. In the case of K2CO36),7)and MEA solutions (Figure 5, 6) cracking is

predominantly intergranular, whereas in DEAsolutions only transgranular cracks were ob-served (Figure 3, 4) and in this instance slowerstrain rates were required during testing to con-sistently produce cracking.

In nitrogen sat'd MEA and DEA solutionsthe low electrochemical reactivity and negligiblecracking tendency in the potential range -0.8to -0.6 V (SCE) is attributed to the high pH(11.5-12.2) of these solutions and the absenceof HCO3- or other surface active anions whichcould accelerate the dissolution process. Theslightly higher anodic current density observedwithin the active potential loop in the case ofMEA solution as compared to DEA (Curves Band D) in Figure 2 indicates a slightly lowerreactivity in the case of DEA solution.

The exact cause of this effect is not known.However, since in this pH range aminoalcoholsare likely to exist predominantly as neutral mol-ecules it is reasonable to assume that the ob-served difference in reactivity is caused by dif-ferences in chemisorption tendencies of thesemolecules on the metal surface which are likelyto slow down the rate of metal dissolution.Diethanolamine as a secondary aminoalcohol islikely to have a stronger tendency, than MEAwhich is a primary aminoalcohol, for chemisorp-tion, via the N atom on the iron surface.

In the case of CO2 sat'd MEA and DEA

Vol. 38, No. 1 17

solutions the cause of the higher electrochemicalreactivity and strong tendency for cracking inthe potential range of -0.8 to -0.6 V (SEC) isnot completely understood. There is little doubtthat the presence of HCO3- which is producedby saturating these solutions with CO2 con-tributes to the higher electrochemical reactivityand cracking tendency observed in these solu-tions as compared to the nitrogen saturatedsolutions. In this regard, the SCC in these

solutions can be regarded as a case of bicar-bonate cracking as proposed earlier by Parkinsand Foroulis10). The overlaying of potentialzones for maximum reactivity and SCC tend-ency of CO2 sat'd solutions of MEA and DEAwith that of CO2 sat'd K2CO3 solution, providessupport for this interpretation. Added supportfor this mechanism is also emanating from ob-servations which indicate the presence of thefamiliar10) black-gray surface films of Fe3O4/FeCO3 on test specimens which were subject tostress corrosion cracking.

To explain the difference in electrochemicaland SCC reactivity between CO2 sat'd MEAand DEA solutions, it is postulated that thereactivity differences is due at least in part todifferences in chemical behavior of MEA andDEA solutions when saturated with CO2 whichare likely to result in different bicarborate anionconcentrations.

Diethanolamine solutions are known to be lessefficient for the absorption of CO2 than mono-ethanolamine solutions. Data by Shneerson andLeibush16) clearly shows that the absorption co-efficients KG(a) for CO2 obtained under thesame conditions and in the same apparatus inMEA solutions are about 2 to 21/2 times greaterthan these for DEA solutions. From CO2 vapor-liquid equilibrium data over MEA and DEAsolutions, the equilibrium concentration of CO2 atatmospheric pressure and 100C is estimated at0.5 mole of CO2 per mole of MEA vs. about

0.32 mole of CO2 per mole of equal molar con-centration of DEA. Assuming that all of theabsorbed CO2 is converted to bicarbonate ionsat equilibrium, this then would indicate an over-all higher concentration of HCO3- anions inthe MEA vs, a DEA solution of equivalentmolar concentration. Higher concentration ofHCO3- would then explain the higher SCC tend-ency and electrochemical reactivity of CO2 sat'dMEA vs. DEA solutions.

Other factors, such as absorption of MEA,

DEA molecules or carbonate anions (which areformed as intermediates in the reaction of CO2with aminoalcohols) on the iron surface whichmay substantially influence the electrochemicalbehavior and dissolution kinetics of iron are

possible but probably not as pronounced.The above rationalization of the apparent dif-

ferences between CO2 sat'd MEA and DEAsolutions as far as electrochemical reactivity andSCC tendency is considered as tentative andwill need to be further re-examined as newexperimental data become available.

Environments which promote intergranularstress corrosion cracking of mild steel such as

C032/COSH, NO3-, OH-, etc., have also beenreported14),17) to produce structurally dependentattack frequently in the form of intergranularcorrosion of varying intensity in unstressedspecimens and generally only within the po-tential range in which they promote stress cor-rosion cracking. A likely explanation why inter-

granular stress corrosion cracking, by a dissolu-tion mechanism such as the one observed in thecase of CO2 sat'd MEA solutions, progresses

along grain boundaries is probably related tochemical impurity segregation at grain bound-aries. This explanation has been advanced

previously18) and is supported by several experi-mental observations19),20). However, other ex-

planations are possible for the higher reactivityof grain boundary regions. Such explanationsare based on the higher degree of lattice imper-fections such as dislocations, vacancies, etc.,occuring in grain boundaries, which are likelyto be energetically and kinetically preferredsides of anodic metal dissolution.

The observed change of the cracking mor-

phology from intergranular to transgranular inCO2 sat'd MEA solutions at lower potentials

(Table 3) is consistent with previous reports inboth MEA solution10) and in bicarbonate solu-tions4) in which a transition from intergranularto transgranular cracking tendency wasobserved by lowering the potential to a lessreactive potential region.

One important difference between MEA andDEA SCC behavior remains the apparent vari-ance in crack morphology between MEA solu-tions (predominantly intergranular) and DEAsolutions in which only transgranular crackswere detected throughout the SCC potential re-

gion (-0.8 to -0.6 V (SCE). The reason forthis change in crack morphology with the nature

18 Boshoku Gijutsu

of the aminoalcohol is not known. It is pos-sible to speculate that chemisorption of DEAmolecules on the steel surface may have ren-dered inoperative preexisting active paths, e.g.

grain boundaries, on the steel surface whichtend to promote intergranular cracking. Undersuch conditions stress corrosion cracking is like-ly to initiate in strain generated active pathswhich are likely to be produced by film ruptureand slip step emergence on the metal surfacethus providing the circumstances for transgran-War cracking.

Intergranular cracking resulting from the in-

gress of hydrogen into the steel (HIC) as apossible mechanism to explain the transgranularmorphology of the cracks in DEA solutions isconsidered not likely because the potential range-0.8 to -0.6 V (SCE) where cracking was ob-

served is, outside the potential zone of hydrogenevolution. It seems therefore more likely thatthe mechanism of transgranular cracking ob-served in DEA solutions is one related to local-ized dissolution with the crack tip region beingactive while the crack surfaces and other metalsurfaces remain relatively passive.

Acknowledgement

The author would like to acknowledge thecontributions of Mr. David Rewick who carriedout the electrochemical and stress corrosioncracking tests and of Mr. David Lattig who

performed the SEA and Metallographic work.(Received July 2, 1988)

References

1) K.T.G. Atkins, D. Fyfe & J. D. Rankin: AICHESymposium on Safety in Ammonia Plants andRelated Facilities, CEP Technical Manual

16, 39 (1974).2) K. Naito, T. Hashimoto & T. Kihara: Japan

Petr. Inst., 14, 672 (1971).3) R. N. Parkins & R. R. Fessler: Materials in

Engineering Applications, 1, 80 (1978).4) J. M. Sutcliffe, R. R. Fessler, W. K. Boyd &

R. N. Parkins: Corrosion, 28, 313 (1972).5) R. N. Parkins, Corr. Sci. 20, p. 147 (1980).6) Z. A. Foroulis, Metallurgical Transaction, to be

submitted.7) Z. A. Foroulis, Boshoku Gijutsu, 36, 689 (1987).8) D. G. Huges: "Stress Corrosion Cracking in

MEA Units" Proceedings 1982 UK CorrosionConference, p. 85, November, 1982.

9) J. Gutzeit, J. M. Johnson: Corrosion/86, PaperNo. 206, NACE, Houston, TX, 1986.

10) R. N. Parkins & Z. A. Foroulis: Mater. Perform.27, 19 (1988).

11) S. J. Lenhart, H. L. Craig & J. D. Howell: Cor-rosion/86, Paper No. 212, NACE, Houston, TX,1986.

12) Z. A. Foroulis: J. Electrochem. Soc., 113, 532

(1966).13) M. Pourbaix : Atlas of Electrochemical Equili-

bria in Aqueous Solutions, Pergamon Press, Ox-ford (1966).

14) D. H. Davies & G. T. Burnstein: Corrosion, 36,416 (1980).

15) Z. A. Foroulis: Proc. Intern. Symposium "Sur-faces, Inhibitors and Passivation" The Electroch.Soc., Inc. Pennington N.J. 1986, p. 539.

16) A. L. Shneerson & A. G. Leibush: J. Appl.Chem. (USSR) 19, 869 (1946).

17) R. N. Parkins: The Theory of Stress CorrosionCracking in Alloys, NATO, 1971, p.167.

18) R. N. Parkins: in Stress Corrosion Cracking andHydrogen Embrittlement of Iron Base Alloys,Ed. R. W. Staehle, et al., NACE, 1977, p. 601.

19) K. Bohnenkamp: Fundamental Aspects of StressCorrosion Cracking, NACE, 1969, p. 374.

20) L. M. Long & N. A. Lockington: Corr. Sci. 11,853 (1971).