NACE_2011_Paper_11072

2011

1

Effect of FeCO3 Supersaturation and Carbide Exposure on the CO2 Corrosion Rate of Carbon Steel

Tonje Berntsen Statoil ASA

P.O. Box 308 N-5501 Haugesund

Norway

Tor Hemmingsen University of Stavanger

N-4036 Stavanger Norway

Marion Seiersten Institute for Energy Technology

P.O. Box 40 N-2007 Kjeller

Norway

ABSTRACT The pH stabilization technique is a widely used corrosion protection method for multiphase gas pipelines with glycol as hydrate inhibitor. It implies to increase the pH by addition of 3HCO in order to enhance the formation of protective iron carbonate films. The protection mechanism at ~20C is of concern because the conditions for precipitating protective corrosion film are less favorable compared to higher temperatures due to the increasing solubility of FeCO3 with decreasing temperature. The scope of the ongoing work is to investigate whether corrosion mitigation of pipelines at ~20C relies on the formation of protective corrosion films or if the corrosion rate is sufficiently lowered by the elevated pH. This paper discusses the corrosion rate and corrosion potential observed on carbon steel exposed to varying concentrations of 3HCO and Fe2+ at 20C in a 1wt% NaCl and 50wt% glycol solution purged with CO2 at 1 atm partial pressure. The objective was to promote protective FeCO3 films by high iron and bicarbonate concentrations and study the effect of supersaturation and variations in iron and bicarbonate concentration. Protective films did not form despite high supersaturation and long exposure times. The reason for this is discussed in light of exposed iron carbide (Fe3C) and prerequisites for iron carbonate growth. Key words: CO2 corrosion, pH stabilization, FeCO3 supersaturation, film formation, precipitation.

2011 by NACE International. Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to NACEInternational, Publications Division, 1440 South Creek Drive, Houston, Texas 77084. The material presented and the views expressed in this paper aresolely those of the author(s) and are not necessarily endorsed by the Association.

1

Paper No.

11072

2

INTRODUCTION Pipelines for oil and gas are often designed with a maximum allowable corrosion tolerance of 0.1 mm/y to allow for process upsets or unpredictable incidents despite corrosion mitigation actions. Previous experiments [1] have shown that the unmitigated corrosion rate at ~20C, 1 atm CO2 and 1 wt% NaCl is about 1 mm/y. It decreased to values between 0.1 mm/y (this work) and 0.2 mm/y [1] under pH stabilized conditions without a protective FeCO3 film. Once a truly protective FeCO3 film had precipitated on the steel surface, the corrosion rate decreased by one order of magnitude to less than 0.01 mm/y. As the corrosion rate reached this low level, the corrosion potential increased rapidly to -0.45 V, and then stabilized around -0.5 V. When a protective film covers the surface, it is film properties like porosity, thickness and composition which control the transport of reactants and corrosion products through the film that governs the corrosion rate. These properties depend on the FeCO3 precipitation process [2]. The precipitation is facilitated by increased CO2 partial pressure, pH, bicarbonate concentration, temperature and Fe2+ concentration; consequently increased supersaturation, and all other measures which can reduce the transport of reactants and corrosion products to and from the steel surface [3]. Dugstad and Drnen [2] studied precipitation of FeCO3 film on one steel with carbon content of 0.057% and another with carbon content of 0.080% at low temperature; i.e. 20C (pH 6.5, 0.6 MPa CO2 partial pressure). A protective corrosion film formed on the 0.08% C steel giving a corrosion rate well below 0.1 mm/y. Only small FeCO3 crystallites had formed on the 0.057% C steel after ~4 months exposure, and the corrosion rate was still well above 0.1 mm/y; i.e. a non-protective film. It is a prerequisite for initiating growth of FeCO3 film that the solution must be supersaturated with regards to iron carbonate, implying that the saturation ratio (SR) of FeCO3 must be >1. The saturation ratio is defined as

sp

COFe

K

aaSR

23

2 where 2Fea is the activity of Fe

2+, 23CO

a is the activity of 23CO and Ksp is the solubility product of FeCO3 [3]. The concentration-temperature curve for the solubility of FeCO3 is inverse compared to most salts, meaning the solubility increases with decreasing temperature. This means that the driving force for FeCO3 precipitation, consequently SR, decreases with falling temperature. The precipitation process involves both nucleation and particle growth. According to Sun [4], the nucleation rate is primarily important in homogeneous crystallization processes. In the case of crystallization onto a metal surface, the crystallization process is classified as heterogeneous and the overall process kinetics is dominated by crystal growth. According to Johnson and Tomson [5], FeCO3 has extremely slow precipitation kinetics at temperatures below 75C. They claim that increased SR, i.e. high Fe2+ and 23CO concentrations and high pH, might improve the adherence of such a film [5]. This is in agreement with studies of the induction time for precipitation of FeCO3 that have been performed at T> 60C. At constant MEG and Fe2+ concentrations, the induction time for measurable precipitation decreased as a function of increasing 3HCO concentrations; consequently increasing SR [6]. The growth of an FeCO3 layer on steel is strongly affected by the corrosion rate at low supersaturation; consequently the rate at which Fe2+ is released from the surface. At high supersaturation, the corrosion rate has less of an effect on the corrosion layer accumulation rate [4]. The main difference between protective and non-protective corrosion layer morphologies is the absence or presence of empty Fe3C (i.e. not filled with FeCO3) in contact with the steel surface,

2

3

respectively. In the absence of Fe3C (freshly polished specimen), the precipitation of FeCO3 can only occur at the steel surface where the concentrations of Fe2+ from the corrosion and 3HCO from the cathodic reduction of CO2 are at its maximum. This would lead to protective FeCO3 film when the saturation limit is exceeded. Due to galvanic coupling between the Fe3C and the steel in a direction perpendicular to the surface, preferential corrosion of the steel matrix will result in an empty Fe3C layer in contact with the steel surface. The electron conductive Fe3C layer provides additional surface area to the steel surface for the cathodic reduction of CO2 to 3HCO . This leads to local alkalinization at a certain distance from the steel surface. If the saturation limit is exceeded, precipitation of FeCO3 can take place inside, or more likely, on the surface of the Fe3C. This gives a non-protective layer that even enormous iron supersaturations cannot subsequently render protective. Only layers of FeCO3 that are directly in contact with the steel can be protective [7]. Formation of protective FeCO3 film at 20C Experiments performed at similar conditions to the experiments presented in this paper are discussed in a previous publication [1]. The conclusion from the previous work was that a protective FeCO3 film (corrosion rate lower than 0.01 mm/y) formed on X-65 steel under pH stabilized conditions. Figure 1 shows the corrosion rate and SR variations for a long term experiment performed at room temperature. A summary of the experimental conditions is provided in Table 3 for comparison with the experiments discussed here. A SR above 300 was required for the FeCO3 film to precipitate, refer to Figure 1.

0.001

0.01

0.1

1

0 30 60 90 120 150Time / days

CR

/ (m

m/y

)

0

50

100

150

200

250

300

350

400SR

CR p.a grade MEGBicarb added 22->67 mmol/lSR

670 hrs

1000 hrs

1650 hrs

a)

-0.70

-0.65

-0.60

-0.55

-0.50

-0.45

0 30 60 90 120 150Time / days

Ecor

r / V

Bicarb added 22->67 mmol/l

Ecorr

b)

Figure 1: a) Corrosion rate and SR and b) corrosion potential as a function of time in ~50 wt% MEG at room temperature, 1wt% NaCl and 1 atm CO2. No pre-corrosion. Concentration of 3HCO

was 22-88 mmol/l. The precipitation and film growth process was monitored by measuring the Fe2+ and 3HCO concentrations and calculating the corresponding SR. From the onset of precipitation, the decrease in the corrosion rate from ~0.1 mm/y down to a steady level below 0.01 mm/y was slow and took ~30 days. The film growth process on the metal surface was followed by retrieving samples at ~monthly intervals (refer to the original paper for pictures covering the whole time span of the experiment [1]). Figure 2 shows Scanning Electron Microscopy (SEM) pictures from a specimen that was immersed for 84 days, consequently after the corrosion rate had decreased to below 0.01 mm/y. The SEM images in Figure 2 a) and c), in combination with Energy Dispersive Spectroscopy (EDS) analysis of the phases, clearly show that the surface was covered with a 10-15 m thick FeCO3 (d) film of very small (1-3 m) cubic crystals (b). The Fe3C structure is visible as a light grey network integrated in the iron carbonate in Figure 2 d). Fe3C was exposed on the surface at an early stage when the SR was still low and the corrosion rate quite high [1].

3

4

a) b)

c) d) Figure 2: The surface of a specimen from a previous experiment [1]. The specimen was exposed

to a solution with 1 wt% NaCl, 22-88 mmol/l 3HCO , ~50wt% MEG and 1 atm CO2 for 84 days at room temperature. Magnification 100x, 1000x, 500x and 2000x (top left-lower right).

This long term experiment showed that it is possible to form a protective FeCO3 film under pH stabilized conditions at temperatures close to 20C since the corrosion rate decreased to less than 0.01 mm/y and a dense film covered the surface. The other experiments discussed in [1] indicated that precipitation of FeCO3 film occurred more easily on a pre-corroded surface with some exposed Fe3C compared to a freshly ground surface. In addition to an elevated pH, a high iron concentration was needed to promote precipitation. These results formed the background for the conditions chosen for the new experiments presented in this paper; pre-corrosion, 20C, 1wt% NaCl, ~50wt% MEG, target SR 300 and SR 500 with 70 and 100 mmol/l 3HCO , and corresponding iron concentrations. The objective of the experiments presented here was to reproduce the film formation process observed in the long term experiment under controlled temperature and bicarbonate concentration, searching to map the effects of SR and iron concentration on the precipitation and growth processes of FeCO3.

FeCO3

Fe3C

4

5

EXPERIMENTAL PROCEDURE Apparatus and corrosion measurement method The tests were performed at atmospheric pressure of CO2 in 3 liter jacketed glass cells as shown in Figure 3. The cells were thermostatically controlled at 20C and equipped with magnetic stirring. A standard three-electrode setup was used for all the electrochemical measurements with the cylindrical steel specimen as the working electrode. The counter electrode was a titanium ring. A saturated Ag/AgCl reference electrode was connected to the cell via a Luggin capillary. A potentiostat with an eight channel multiplexer was used for the electrochemical corrosion rate measurements.

Figure 3: Glass cell with lid equipped for corrosion testing.

The corrosion rate was monitored by linear polarization resistance (LPR) measurements and electrochemical impedance (EIS) measurements were performed to assess the IR drop in the solution, the experimental settings are listed in Table 1. The weight loss was measured and the corresponding corrosion rate calculated. The reported LPR corrosion rates in this work are corrected for the measured IR drop and weight loss.

Table 1: Experimental settings for the electrochemical measurements. LPR Potential ramp 5 mV vs. Ecorr Scan rate 0.1 mV/s EIS Initial frequency 10 000 Hz Final frequency 0.001 / 0.01 Hz AC voltage 10 mV vs. Ecorr

5

6

Specimen material and preparation The chemical composition of the X65 subsea pipeline steel used in the specimens is given in Table 2.

Table 2: Alloying elements in the X65 steel used for the corrosion tests. C Si Mn S P Cr Ni V Mo Cu Al Sn Nb 0.08 0.25 1.54 0.001 0.019 0.04 0.03 0.045 0.01 0.02 0.038 0.001 0.043 The working electrode was a cylindrical-shaped specimen with dimensions 1 cm x 1 cm , and the exposed surface area was 3.14 cm2. The specimens were washed in soap, ground with 500 and 1000 grit silicon carbide abrasive paper, and rinsed in an ultrasonic bath with acetone for 10 minutes. It was weighed and mounted on a steel bar covered with PEEK and sealed in both ends by Teflon rings. A PEEK piece covered the end of the steel bar and isolated it electrically from the solution. The specimen was rinsed with ethanol just prior to immersion. Chemicals and monitoring of test solution The test electrolyte consisted of distilled water, 1 wt% NaCl, ~50 wt% mono ethylene glycol purged with CO2 gas at 1 atm for at least 24 hours. The mono ethylene glycol (MEG) used was routinely checked to make sure it was not corrosion inhibiting. NaHCO3 was used to increase the pH. In order to increase the concentration of dissolved Fe2+, small amounts of a 3 mol/l FeCl24H2O purged with N2 were added to the cells by the use of a titrator, or iron powder (>99 wt%) was dissolved in the cells. The water/MEG content was monitored by regular Karl Fischer titration analysis of solution samples. The evaporated water was not replaced. The solution pH was measured at regular intervals with a pH electrode. The pH meter was calibrated in aqueous buffer solutions, and the pH value was corrected for the MEG content in solution by the equation given by Sandengen [8]. The Fe2+ concentration in the test solution was measured at regular intervals. According to an in-house procedure, a small volume was withdrawn from the cell and the Fe2+ concentration in the sample was determined spectrophotometrically. Surface examination of specimens The specimen was rinsed immediately in ethanol upon removal from the cell. It was dried, weighed and stored in a dry atmosphere before the surface was examined using a Scanning Electron Microscope (SEM). Elemental analysis was performed with Energy Dispersive Spectroscopy (EDS). Detailed description of the experiments Experiments 1-2: Target SR 300; 70 and 100 mmol/l bicarbonate concentration The specimens were pre-corroded in a 1 wt% NaCl and distilled water solution at 20C according to the exposure time given in Table 3. Then the NaHCO3 salt, MEG and NaCl solutions were added to the cells, maintaining the NaCl concentration at 1 wt% and reaching a MEG concentration of 50 wt% and bicarbonate concentrations of 70 and 100 mmol/l in the two cells. The MEG/NaCl solution was transferred using a N2 gas lift arrangement. The Fe2+ concentration was increased during the experiment by adding known amounts of 3 mol/l FeCl2 solution to reach SR 300. The bicarbonate concentration (total alkalinity) was measured by titration shortly after the NaHCO3 salt was dissolved, and at regular intervals throughout the experiment. Experiments 3-4: Target SR 500; 70 and 100 mmol/l bicarbonate concentration The procedure was changed slightly compared to experiment 1-2. Iron powder was considered to be a better choice as a source for Fe2+ since some of the FeCl24H2O salt might have oxidized to trivalent iron during storage. Prior to immersing the steel specimen, iron powder was dissolved in the 1 wt% NaCl solution purged with CO2 in order to obtain the desired SR by adding bicarbonate only after the pre-corrosion period. The concentration of Fe2+ was monitored until all the iron powder was completely dissolved. The specimens were then pre-corroded in the 1 wt% NaCl solution with dissolved iron

6

7

according to the exposure times given in Table 3. The amount of iron powder was calculated so as to give the Fe2+ concentration required to reach SR 500. The rest of the procedure was the same as for experiments 1-2. In addition to the experiments primarily presented here, Table 3 lists the experimental conditions for the previous long term experiment mentioned in the introduction for comparison [1].

Table 3: Experimental conditions for experiments 1- 4 a previous experiment [1]. All the experiments had X-65 steel, ~50 wt% MEG and 1 atm CO2.

Experiment Previous exp.

1 2 3 4

SRtarget - 300 300 500 500 [ 3HCO ]target [mmol/l] 70 70 100 70 100 [Fe2+]target [mg/l] - 64 21 105 34 Iron source Corroding

specimens FeCl2 FeCl2 Iron powder Iron powder

Temperature [C] room 20 20 20 20 Pre-corrosion time [h] - 24 24 44* 41** * 1 wt% NaCl with ~200 mg/l Fe2+ ** 1 wt% NaCl with ~80 mg/l Fe2+

RESULTS AND DISCUSSION Effect of saturation ratio on the corrosion rate at 20C Four corrosion experiments were conducted under the experimental conditions described in the previous section. A survey of corrosion rates, corrosion potentials and other key data is provided in Table 4. A summary of the results from the previous long term experiment mentioned in the introduction is included for comparison.

Table 4: Overview of corrosion rates based on weight loss, weight loss calibrated LPR corrosion rates, corrosion potentials, exposure times, measured concentrations, max SR,

temperature and average pH for experiments 1-4 and a previous experiment [1]. Experiment Previous

exp. 1 2 3 4

CRcal,pre-corr [mm/y] - 0.81 0.98 0.74 0.60 CRcal,final 100 h [mm/y] 0.006 0.06 0.13 0.15 0.07 CRcal,average [mm/y] 0.04 0.09 0.12 0.14 0.07 CRWL [mm/y] 0.04 0.13 0.16 0.16 0.11 Ecorr, last 100 h [mVAg/AgCl] -486 -676 -685 -671 -686 Total duration [days] 162 48 48 84 85 [Fe2+]max measured [mg/l] ~100 93 32 97 41 SRmax ~343 385 434 498 530 [ 3HCO ] [mmol/l] 79 7* 68.1 1.3 98.6 0.3 71.7 0.6 98.8 1.5 MEG conc. [wt%] 53 11 50.8 0.3 50.5 0.5 51.9 1.2 51.3 0.6 Temperature [C] 22 1 20.4 0.2 20.4 0.1 20.8 0.7 20.8 0.7 pH 7.0 0.2* 6.89 0.01 7.05 0.01 6.92 0.02 7.05 0.01 *After increasing the bicarbonate concentration to the desired level, see Figure 1. In the following, the various data from each experiment will be discussed.

7

8

Pre-corrosion period The corrosion rates for the pre-corrosion period for experiments 1-4 are listed in Table 4 and shown in Figure 4, 6, 8 and 10. In experiments 3-4, the corrosion rate was slightly lower than in experiments 1-2. This is probably due to the presence of the already-dissolved iron, which gives a lower driving force for corrosion. Corrosion period at high SR Experiment 1: FeCl2 was added successively in small portions, but because the true Fe2+ concentration of the FeCl2 solution was lower than intended, it took ~ one week before enough Fe2+ was added to reach SR 300; refer to Figure 4 a). From that point on, the Fe2+ concentration stabilized around 83 mg/l and the SR value remained at ~330. The corrosion rate dropped to ~0.1 mm/y when the MEG and bicarbonate solution was added, and then decreased slowly and reached 0.06 mm/y at the end of experiment 1, as shown in Table 4 and Figure 4 a). The corrosion potential shifted in the positive direction when Fe2+ was added to the solution; see Figure 4 b). It then decreased slightly as the corrosion rate decreased, and was 676 mV at the end of the experiment.

0.01

0.1

1

10

0 30 60Time / days

CR

/ (m

m/y

)

0

100

200

300

400

500

600

SR

CR Exp 1

SR Exp 1

a)

-0.70

-0.69

-0.68

-0.67

-0.66

-0.65

-0.64

-0.63

0 30 60Time / days

Ecor

r / V

Ag/

AgC

l

Ecorr Exp 1

b) Figure 4: a) Corrosion rate, calculated SR and b) corrosion potential of experiment 1 as a

function of time. Test conditions are given in Table 3. The SEM pictures in Figure 5 show the corroded surface of the specimen from experiment 1. The surface consists of mostly bare steel, but has small areas covered by clusters of small cubic (1-2 m) and larger barrel-shaped FeCO3 crystals (5-10 m). There was practically no Fe3C film on the surface, but the FeCO3 present had precipitated on carbide structures, see Figure 5 c) and d). The composition of the two phases on the steel surface was analyzed by EDS, confirming that the crystals were FeCO3 and the network in between was iron carbide; see Figure 5 d). The decreasing corrosion rate and the fact that FeCO3 had precipitated on small areas on the surface, shows that the crystallization process had started, but FeCO3 did not grow to a continuous layer despite of the high SR. The corrosion rate was low, but not as low as expected if a protective film had formed. The change in corrosion potential was also marginal compared to when a protective film forms, refer to Figure 1.

8

9

a) b)

c)

d)

Figure 5: Surface, a) and b), and cross-section, c) and d) of the specimen from experiment 1. Test conditions are given in Table 3.

There was less carbide on the surface than expected from the amount of removed iron. The corrosion rate in the pre-corrosion period corresponds to about 2 m removed steel, and by the end of the experiment about 12 m. Since the Fe3C phase is cathodic with respect to the steel, this should theoretically leave a Fe3C layer of 12 m on the surface. The SEM images illustrate the lack of such a layer, and this indicates that the carbide structure was fragile and had fallen off during the experiment, or when it was taken out of the solution and rinsed in ethanol. The lack of Fe3C and the decreasing corrosion rate suggests that the corrosion rate was not influenced by galvanic coupling between Fe3C and the bare steel, which is expected to increase the corrosion rate. Experiment 2: As in experiment 1, the FeCl2 solution was added in several steps to reach and exceed SR 300, refer to Figure 4 a). After about 10 days, the Fe2+ concentration more or less stabilized at about 28 mg/l and the SR value remained at ~370.

FeCO3 Fe3C

9

10

0.01

0.1

1

10

0 30 60Time / days

CR

/ (m

m/y

)

0

100

200

300

400

500

600

SR

CR Exp 2

SR Exp 2

a)

-0.70

-0.69

-0.68

-0.67

-0.66

-0.65

-0.64

-0.63

0 30 60Time / days

Ecor

r / V

Ag/

AgC

l

Ecorr Exp 2

b) Figure 6: a) Corrosion rate, calculated SR and b) corrosion potential of experiment 2 as a

function of time. Test conditions are given in Table 3. The corrosion rate dropped to ~0.1 mm/y when the MEG and bicarbonate solution was added and increased slightly toward the end of the exposure, becoming 0.13 mm/y in the final 100 hours of exposure. This is significantly higher than in experiment 1. The corrosion potential was shifted in the positive direction when Fe2+ was added to the solution; see Figure 4 b) at ~5-8 days. It then increased slightly as the corrosion rate increased, and was 685 mV at the end of the experiment. The SEM images in Figure 7 show the corroded surface of the specimen from experiment 2. Figure 7 c) indicate that more carbide was present on the surface in experiment 2 compared to experiment 1. This is expected since the corrosion rate was higher. The iron carbide and carbonate phases were identified using EDS. The surface has a rough appearance (a-b) with a thin (~6-8 m), but continuous film of Fe3C (c) and scattered, barrel-shaped particles of FeCO3 (5-10 m) (b) precipitated on top of it. A few clusters (~20 m) of cubic crystals (~5 m) are also observed; see Figure 7 b). The corrosion potential responded in the same way to the added FeCl2 in this experiment as in experiment 1. As the corrosion rate increased toward the end, the corrosion potential also increased as expected from the Nernst equation. The pre-corrosion treatment should have removed about 3 m of metal, and 16 m should have been removed by the end of exposure. The SEM images show that the Fe3C film was much thinner than 16 m, implying that some of the Fe3C had fallen off at the time of SEM and EDS analysis. Still, the existence of a thin, continuous layer of Fe3C might have contributed to the increasing corrosion rate toward the end of exposure, either by the increase in area for the cathodic reduction or by an accelerated corrosion rate caused by galvanic coupling of Fe3C and the steel. The SR levelled out in the same range as in experiment 1, but as intended, the bulk Fe2+ concentration was considerably lower in experiment 2. The corrosion rate was higher in experiment 2; the surface concentration of Fe2+ and local SR should be higher. The existence of a Fe3C layer would in experiment 2 represent a physical separation between the steel surface, where the Fe2+ concentration is the highest, and the cathodic sites on the carbide. Only a few small crystallites of FeCO3 had precipitated in the very outer parts of the Fe3C layer, which is in accordance with the theory of local alkalinization in the Fe3C layer leading to precipitation if the local SR is high enough. Evidently, the local SR was not high enough to promote FeCO3 film formation in this case. This may indicate that a higher Fe2+ concentration in the bulk is needed in order to form a FeCO3 film.

10

11

a) b)

c) d) Figure 7: Surface, a) and b), and cross-section, c) and d) of the specimen from experiment 2.

Test conditions are given in Table 3. Experiment 3: The SR was raised to ~500 by adding bicarbonate together with MEG immediately after the pre-corrosion period; refer to Figure 8 a). The iron concentration was already 90 mg/l when the specimen was immersed due to the addition of iron powder as described above. The SR decreased steadily between 10 and 40 days of exposure, to below 100. The solution was observed to be cloudy around 28 days. This implies that precipitation of FeCO3 occurred, which was confirmed by the SEM images in Figure 9, and EDS analysis. The corrosion rate dropped to ~0.1 mm/y when the MEG and bicarbonate solution was added and remained stable for ~30 days, refer to Figure 8 a). It then increased to ~0.14 mm/y in the course of about 10 days, coincident with the observed decrease in SR. The corrosion rate stabilized, and the final corrosion rate was 0.15 mm/y. The corrosion potential was ~30 mV higher after the MEG and bicarbonate was added compared to experiments 1-2, and decreased to a minimum at around -675 mV after ~35 days of exposure. This coincided with the rise in corrosion rate and the decrease in SR. The corrosion potential continued to

Fe3C

FeCO3

11

12

increase toward the end of the experiment, and reached a final value of -671 mV for the last 100 hours of exposure.

0.01

0.1

1

10

0 30 60 90Time / days

CR

/ (m

m/y

)

0

100

200

300

400

500

600

SR

CR Exp 3

SR Exp 3

a)

-0.70

-0.69

-0.68

-0.67

-0.66

-0.65

-0.64

-0.63

0 30 60 90Time / days

Ecor

r / V

Ag/

AgC

l

Ecorr Exp 3

b) Figure 8: a) Corrosion rate, calculated SR and b) corrosion potential of experiment 3 as a

function of time. Test conditions are given in Table 3. The SEM images in Figure 9 show the surface of the corroded specimen from experiment 3. The corrosion film was loosely adhered and came off easily when the specimen was rinsed with ethanol. The film consisted of a discontinuous Fe3C layer 4-20 m thick with regions of 5-10 m thick FeCO3 film sitting on top, i.e. the FeCO3 was not integrated into the carbide structure. Figure 9 b) shows that the carbonate crystals were barrel-shaped (5-10 m) with small cubic crystals (1-2 m) on top of the barrels. The iron carbide and carbonate phases were identified using EDS. The somewhat higher initial corrosion potential in this experiment compared to the two previous ones is attributed to the high initial Fe2+ concentration (~90 mg/l) in the same way as before. The corrosion potential followed the decrease in the SR; i.e. the falling Fe2+ concentration in the bulk, until day 35. The pre-corrosion period should have produced a 4 m Fe3C layer, and after 35 days, it should have been about 13 m thick, provided it was intact on the steel surface. By the end of exposure, 32 m of material was removed. The theoretical Fe3C thickness at 35 days corresponds well to the thickness of the Fe3C layer observed underneath the FeCO3 film in the SEM images, and together with the SR decrease, this indicates that the FeCO3 film precipitated quite early in the process. The total amount of removed material corresponds well to the distance between the steel surface and the outer edge of the FeCO3 film. The average corrosion rate in experiment 3 was 0.14 mm/y. This is the highest corrosion rate seen among the experiments presented here, see Table 4. This should give a higher surface concentration of Fe2+ compared to the other experiments, and better conditions for precipitation of FeCO3. The fact that the FeCO3 had precipitated on top of the carbide, suggest that a local alkalinization occurred there as a result of the cathodic reduction reaction occurring on the Fe3C, and that a critical SR for precipitation was reached locally. The sharp increase in the corrosion rate to ~0.14 mm/y at ~30 days, is coincident with the decrease in SR and a decrease in the corrosion potential. The decrease in SR is due to precipitation which removes a considerable amount of the Fe2+ from the bulk solution while the 3HCO concentration and pH change is minor. Some of the FeCO3 precipitated in the outer part of the Fe3C layer.

12

13

a) b)

c) d) Figure 9: Surface, a) and b), and cross-section, c) and d) of the specimen from experiment 3.

Test conditions are given in Table 3. Beyond day 35, the Fe2+ concentration and SR continued to decrease as FeCO3 precipitated, gradually covering the Fe3C and possibly trapping the Fe2+ released by corrosion inside the solution volume close to the surface. Simultaneously, the corrosion potential started to increase as the corrosion rate stabilized. The observed increase in corrosion potential could be ascribed to the possible accumulation of Fe2+ on the steel surface, but this is not fully understood. Experiment 4: As in experiment 3, iron powder was dissolved prior to immersing the specimen. The first SR reading was ~300 and the corresponding Fe2+ concentration ~25 mg/l; refer to Figure 10 a). The SR was ~500 after 8-9 days of exposure and remained high for ~45 days before it decreased at approximately the same rate as in experiment 3. It was below 150 when the experiment was terminated. This implies that precipitation of FeCO3 occurred toward the end of the experiment. The presence of FeCO3 crystals on the surface was confirmed by the SEM images shown in Figure 11, and EDS analysis. The corrosion rate dropped to well below 0.1 mm/y when the MEG and bicarbonate solution was added; refer to Figure 10 a). It remained stable for about 10 days, and then it decreased slightly down to a minimum of 0.05 mm/y after 45 days.

Fe3C

FeCO3

Fe3C

FeCO3

13

14

0.01

0.1

1

10

0 30 60 90Time / days

CR

/ (m

m/y

)

0

100

200

300

400

500

600

SR

CR Exp 4

SR Exp 4

a)

-0.70

-0.69

-0.68

-0.67

-0.66

-0.65

-0.64

-0.63

0 30 60 90Time / days

Ecor

r / V

Ag/

AgC

l

Ecorr Exp 4

b) Figure 10: a) Corrosion rate, calculated SR and b) corrosion potential of experiment 4 as a

function of time. Test conditions are given in Table 3. The corrosion potential was significantly lower and more stable after the MEG and bicarbonate was added compared to experiment 3. The corrosion rate increased slowly after 45 days and the final corrosion rate was 0.07 mm/y, see Table 4. This period of increasing corrosion rate coincided with the decrease in SR from ~500 down to ~150 and the corrosion potential decreased, analogous to experiment 3, only it occurred later. The SEM images in Figure 11 show the surface of the corroded specimen from experiment 4. The corrosion film was physically more adherent compared to the specimen from experiment 3. Figure 11 a-b) show that there are scattered, 10-20 m large barrel-shaped FeCO3 crystals on the practically bare steel surface. The steel has a fuzzy appearance where the fine Fe3C structure protrudes, with some shallow grooves where preferential corrosion has taken place. Figure 11 c-d) illustrates how the FeCO3 crystals are clearly integrated into the thin Fe3C structure. This film is 10-20 m thick. The iron carbide and carbonate phases were identified using EDS. The corrosion potential initially in experiment 4 was lower compared to experiment 3, as expected from the much lower initial iron concentration (~25 mg/l). At a later stage, the corrosion potential decreased coincident with the decreasing SR, caused by decreasing Fe2+ concentration on the steel due to precipitation of FeCO3. A slight increase in the corrosion rate at the same time, similar to the period between 15-45 days in experiment 3, but the corrosion rate increase was less. The pre-corrosion period should have produced a 3 m Fe3C layer, and at the onset of precipitation (at ~45 days) it should have been about 10 m. By the end of exposure, 16 m of material was removed. The SEM images show that the Fe3C film was discontinuous and thinner than 16 m, implying that some of the Fe3C had fallen off at the time of SEM analysis. The average corrosion rate was 0.07 mm/y, so it was considerably lower than in experiment 3 with similar SR. The lower corrosion rate gives a lower surface concentration of Fe2+, and thus a lower driving force for precipitation compared to experiment 3. Despite this, the SEM pictures reveal, some precipitation occurred in the outer part of the apparently Fe3C structure. The crystallites were evenly distributed over the surface and quite adherent.

14

15

a) b)

c) d) Figure 11: Surface, a) and b), and cross-section, c) and d) of the specimen from experiment 4.

Test conditions are given in Table 3. Overall discussion It was not possible to obtain a protective FeCO3 film under the conditions and time span of the four experiments discussed here. This was despite having a bulk solution with the same Fe2+ and 3HCO concentration as in the previous experiment with the same steel, where the corrosion rate was below 0.01 mm/y [1]. A high supersaturation (SR~500) induces bulk precipitation and precipitation of FeCO3 in the outer part of the Fe3C layer. That may lead to high local Fe

2+ concentration close to the metal, but insufficient carbonate concentration to achieve the growth rate required for the formation of a protective film. Whether this was due to local acidification as proposed by some [7] or was due to physical hindrance remains an open question. The lower supersaturation (SR~300) leads to carbonate precipitation within the carbide layer, but the time seemed to be insufficient in order to grow a protective carbonate film. These experiments show that the kinetics of FeCO3 precipitation is very slow at the 20C, in accordance with Johnson and Tomson [5]. It seems that a high SR must be maintained for a certain time before precipitation occurs and SR decreases, suggesting that there might exist some sort of

Fe3C

FeCO3

15

16

induction time also for the heterogeneous crystallization process of FeCO3 as well as for the homogeneous one [6]. A striking difference between the previous long term experiment [1] and all the four experiments presented here is the size of the crystallites. In the case of the protective film, the crystals are solely cubic and 1-3 m, whereas they are both cubic and barrel-shaped and generally larger in these experiments. This point needs further studies. Dugstad and Drnen did the same type of experiment with two different types of steel [2]. The 0.08% C steel they studied had the same level of carbon as the X-65 steel used in the experiments presented in this work, but it had somewhat higher levels of carbide-forming alloying elements, and hence it might have contained more carbide. They obtained protective film on the 0.08% C steel, but not on the 0.057% C steel. This indicates that a key parameter for growth of FeCO3 film, apart from the supersaturation, is the presence of carbide.

CONCLUSIONS It is clear that it was not possible to obtain a protective FeCO3 film under the conditions and time span of the four experiments discussed here. Neither of the experiments where the SR for FeCO3 was increased by adding Fe2+ to the solution instead of letting it build up only due to corrosion, resulted in a protective FeCO3 film and corrosion rates less than 0.01 mm/y at 20C. Apart from the supersaturation, presence of carbide is believed to be a key parameter for growth of FeCO3 film. The effect of carbide content and the morphology of carbide film will be studied further by corroding the matrix in a controlled manner, giving a reproducible amount of Fe3C on the steel surface.

ACKNOWLEDGEMENTS This work has been carried out at the department of Materials and Corrosion Technology at Institute for Energy Technology in Norway as part of my ongoing PhD project. The financial support from Statoil ASA is highly appreciated.

16

17

REFERENCES 1. T. Berntsen, M. Seiersten and T. Hemmingsen, CO2 Corrosion of Carbon Steel at Lower Temperatures, The European Corrosion Congress (Edinburgh, UK, European Federation of Corrosion, Event. No. 299, 7-11 September 2008), pp. 20.

2. A. Dugstad and P. E. Drnen, Efficient Corrosion Control of Gas Condensate Pipelines by pH-stabilisation, CORROSION/99, paper no. 20 (Houston, TX: NACE, 1999), pp. 10.

3. A. Dugstad, Mechanism of Protective Film Formation during CO2 Corrosion of Carbon Steel, CORROSION/98, paper no. 31 (San Diego, CA: NACE, 1998), pp. 11.

4. W. Sun, and S. Nesic, Kinetics of Corrosion Layer Formation: Part 1 - Iron Carbonate Layers in Carbon Dioxide Corrosion,Corrosion, Vol. 64, No. 4 (2008): p. 334-346.

5. M. L. Johnson, and M. B. Tomson, Ferrous Carbonate Precipitation Kinetics and Its Impact CO2 Corrosion, CORROSION/ 91, paper no. 268 (Houston, TX: NACE, 1991), pp.15.

6. G. Watterud, Precipitation of FeCO3 from MEG solutions, Institute for Energy Technology, Kjeller, Norway: IFE/KR/F2009/087.

7. J.L. Crolet, N. Thevenot and S. Nesic, Role of Conductive Corrosion Products on the Protectiveness of Corrosion Layers, CORROSION/96, paper no. 4 (Houston, TX: NACE, 1996), pp. 14.

8. K. Sandengen, B. Kaasa, T. stvold, pH Measurements in Monoethylene Glycol (MEG) plus Water Solutions, Industrial & Engineering Chemistry Research, Vol. 46, No. 14 (2007): p. 4734-4739.

17