Carbon Steel Corrosion in Piperazine-Promoted Blends under CO2 Capture Conditions

Louis C.Y. Yu, Kyra L. Sedransk Campbell*, Daryl R. Williams

Department of Chemical Engineering, Imperial College, South Kensington Campus, London,

SW7 2AZ, United Kingdom

*Corresponding author: k.sedransk@imperial.ac.uk

Abstract

Aqueous amine promoter blends have improved CO2 absorption capacity and uptake. Tertiary (3o) and sterically-hindered (SH) amines are favoured for their molar absorption ratio, (i.e. CO2 absorption capacity). With a promoter, namely piperazine (PZ), the ordinarily slow reaction kinetics of a 3o/SH amine is accelerated. Amine blends of 30 and 50 % by weight, MDEA+PZ and AMP+PZ, were tested using immersion corrosion techniques at 120 oC. In all cases, a siderite (FeCO3) product layer was formed on the surface of the carbon steel coupons. Aqueous PZ solutions produced thin layers with comparatively lower Fe ion concentrations than blended solutions. The fast CO2 capture kinetics of PZ, and therefore carbonate formation, makes the rapid reaction possible due to readily available Fe ions oxidised on the surface. The replacement of PZ content in a blend, by MDEA or AMP, resulted in slower formation of siderite and variably poorer corrosion protection. Critically, the use of AMP in the blend offers better protection against corrosion, shown by lower concentration of Fe ions in the bulk solution than parallel MDEA solutions. This can be attributed to the faster formation of carbonate species by AMP, as a SH amine, which also results in more imperfect crystals.

Keywords: post-combustion carbon capture, amine, piperazine, corrosion, carbon steel, stripper, reboiler, siderite

Introduction

There is increasing consumption of fossil fuels in industrial manufacturing, power generation and transport worldwide, all of which are contributing to the release of CO2 into the environment [1].The emission of greenhouse gases, CO2 in particular, is at the root of global climate change. This devastating problem requires an immediate solution: a mature approach to minimize the release of CO2 into the environment. Post-combustion CO2 capture (PCCC) with amine solvents is the most developed technology for CO2 capture, originally developed for the natural gas sweetening industry.

It has widespread applicability and requires relatively straightforward retrofitting onto existing infrastructure. While amine technology is well-established, widespread large-scale deployment remains a challenge due both the capital and operational costs. One approach to reducing operating costs has focused on the development of improved amine solutions. In particular blends, created with a tertiary (3o) or sterically-hindered (SH) amine in combination with a promoter amine, have been shown to improve CO2 carrying capacity while retaining an effective uptake rate. Additional optimisation of these mixtures, for high CO2 capacity and uptake, has received considerable attention [2], [3], [4], [5]. In working to reduce capital costs, the potential of less expensive construction materials has been recently reported [6]. However, the integrity of the infrastructure must be maintained with no impact on the solvent beyond what is observed with stainless steel.

The selection of primary (1o) and secondary (2o) amines has dominated solvent choice for PCCC due to the fast reaction kinetics which are far superior to 3o or SH amines. While 1o (and 2o) amines directly react with CO2 to form carbamate ions (Equations 1 and 2), two amine molecules are required for each CO2 molecule. Solutions containing methyldiethanolamine (MDEA) or 2-amino-2- methyl-1-propanol (AMP), which are 3o and SH amines respectively, are valuable for their high CO2 capacity [7].This high absorption capacity is achieved because of a one to one mole ratio of amine to CO2 (Equations 3 and 4). Additionally beneficial is the considerably lower heat of CO2 absorption via 3o pathways (carbamate formation), which lowers the regeneration energy [8].The main drawback of these amines is the slower CO2 hydrolysis reaction step (Equation 3). However, it has been demonstrated that with the addition of rate promoters this limitation can be overcome to make capture industrially feasible [9], [10]. A popular rate promoter for use with 3o and SH amines has been piperazine (PZ), which contains two 2o nitrogen groups [9], [10]. Its use is generally at low concentrations ranging from 5 to 20 % by weight, principally because higher concentrations have encountered insolubility [4]. The benefits of PZ as a promoter are due to its thermal stability up to 150oC [11] and the observed reduced levels of oxidative degradation [4]. In particular, PZ added to MDEA solutions were identified to reduce the concentration of degradation products, as compared to pure MDEA by both Closmann et al. 2009 [12] and Nguyen 2013 [13]. Recently, the use of 3 mixtures of PZ with MDEA or AMP has proven popular due to their potential commercial viability and improved CO2 absorption capacity (Table 1). It should be noted that there are variations amongst the reported capacity of CO2 measured in the literature across apparently identical solutions. It is suggested that additional variables unaccounted for the loading process or detection method may lead to these discrepancies. For example, in the case of the three 30 % by weight experiments the CO2 pressures are notably different. Alternatively, the measurement method in Adeosun et al. 2013 was with phosphoric acid titration. By contrast, Chen et al. 2011 [14] continuously measured CO2 absorption at 100 oC using an infrared CO2 analyser. Finally, in most studies reported error analyses for the measurement of the CO2 loading are minimal, further contributing to uncertainty in the data reported.

(1)

(2)

(3)

(4)

1

Table 1. Literature values for maximum CO2 loading of blends of MDEA/PZ and AMP/PZ

†all tests had a system pressure of 101.3 kPa

‡reference experiments for both MEA and PZ were included in some studies and the results noted here

As previously introduced, reduction of both operating and capital costs are highly desirable. The heavy reliance placed on the use of stainless steel contributes to a high capital cost [6]. This choice is based on the knowledge that even an aqueous system in the presence of CO2 can contribute to corrosion through the formation of carbonic acid and subsequent attack on carbon steel (Equations 5 to 9) [18]. This problem can be worsened by the presence of CO2 rich amine solvents [19] and in turn, can accelerate amine degradation [20], [21], [22]. Whilst the solution blends previously discussed were developed to optimize operation (and reduce operating costs), they have great potential for use with carbon steel, and consequently reducing capital costs.

While amines have been traditionally viewed to be corrosive, this is a gross generalization that applies primarily to 1o and 2o amines. For both 3o and SH amines, carbonate formation occurs, and subsequently the reaction between CO32- and iron(II) (Fe2+) ions produces siderite (FeCO3) (Equations 5 to 9), a known corrosion inhibitor [21], [22], [23], [24]. It has been suggested that the growth of a siderite layer is temperature dependent; formation at low temperatures is favourable due to ion saturation requirements, whereas at elevated (>80 oC) temperatures higher ion saturation levels are required [25]. The research presented in this paper investigates the corrosion products formed on carbon steel surfaces after exposure to the aforementioned blend solutions and determines the effectiveness of these surface products in corrosion inhibition for high temperature sections of the PCCC process.

(5)

(6)

(7)

(8)

(9)

The promise of corrosion reduction for popular promoter blends in literature using PZ as the rate promoter (Table 1) is based on PZ’s 2o polyamine structure, where siderite has been reported to form in pure PZ solutions loaded with CO2 [26]. This anachronistic behaviour has been attributed to multiple capture mechanisms of CO2, suggested by the presence of both carbonate and carbamate species [27], [28], [29]. Capture of one CO2 molecule by two PZ molecules produces a protonated PZ (PZH+) and PZ carbamate (PZCOO-) (Equation 10) [28]. The hydrolysis of one CO2 molecule in the presence of one PZ molecule produces free protons, which in turn produce one PZH+ and one bicarbonate species (HCO3-) (Equation 11). Due to the polyamine structure of PZ, an existing additional PZH+ molecule can react with CO2 on the second nitrogen to form protonated PZ carbamate (H+PZCOO-) (Equation 12 and 13) [28]. In addition to the reaction with PZ, the aqueous balance of the solution also can produce carbonate species through hydrolysis of CO2 (Equation 14 and 15) [28]. The presence of an excess of carbonate species from all of these pathways provide the necessary reactants for the formation of siderite.

(10)

(11)

(12)

(13)

(14)

(15)

However, the kinetics of these competitive pathways remains only minimally understood. Some recent work has indicated that the concentration of PZ carbamates and the hydrolysis of carbonates are a function of CO2 loadings. Such variations could impact the production of siderite at varying levels of CO2 [29], [30].

Methods

Amine solvents monoethanolamine (MEA, ≥99% Sigma), piperazine (PZ, 99% Sigma), methyldiethanolamine (MDEA, ≥99% Sigma) and 2-amino-2-methyl-1-propanol (AMP, ≥90% Sigma) were used as supplied (Figure 1 A to D).

Figure 1. Chemical structures of (A) MEA, (B) PZ, (C) MDEA and (D) AMP

Table 2. Amine solution component weights.

Aqueous amine ‘promoter’ style blends were prepared where either MDEA (3o) or AMP (SH) amine was added to PZ. Solutions were diluted in DI water to amine mass fractions of 0.3 and 0.5 [2], [31] (Table 2). The specific compositions selected were based on literature (Table 1) [2]. Reference solutions of PZ and MEA at both 30% by weight and 50% by weight were also studied.

Solutions (200 mL) were tested in 250 mL round-bottom flasks fitted with reflux condenser at atmospheric pressure (Figure 2). Nitrogen gas (N2) was bubbled into the solution at 20 mL min-1 (1 barg) for four hours at room temperature to remove any dissolved oxygen in the solution. Subsequently, solutions were loaded by bubbling CO2 at 100 mL min-1 (1 barg) for two hours and upon saturation, heated to 120 oC. Temperature control of the solutions was maintained using Dry-sin blocks, fitted with a temperature probe and seated on a hot plate.

Figure 2. Illustration of test rig (not to scale).

Carbon steel C1018 corrosion coupons (76.25x12.62x1.63 mm, 98.83-99.13% Fe, 0.18% C, 0.6-0.9% Mn, ≤0.04 P and ≤0.05 S, Alabama Specialty Metals) were washed with acetone, dried and weighed. The coupons were submerged into the solution and suspended using a thin unobtrusive PTFE thread from a stopper in the centre neck of the round bottom flask. Throughout the experiment CO2 was bubbled continuously through the liquid at 20mL min-1 to maintain saturation.

Experiments were run for seven days (168 hours) with liquid samples drawn on days two, four and seven. The coupons were removed at the end of the trial then weighed, washed and re-weighed. The weight loss was converted to a metric of material size change rate (MSCR) which is based on the corrosion rate equation (Equation 16) [32]. This provides a practical insight into the net material gain or loss of carbon steel while subjected to CO2 loaded amine solutions. The MSCR is calculated by the difference in final and initial masses (mFINAL, mINITIAL) divided by the steel density (ρ), exposed surface area (A) and exposure time (t) in years. In the case of poor adhesion of product layers, the discrepancy is reported using both crude and clean weights.

(16)

The metal ion content of the liquid samples was measured using Inductively Coupled Plasma Optical Emissions Spectroscopy (ICP-OES). The solution was prepared in diluted nitric acid matrix (7.433M) where calibration standards (Fe, Mn, Cr and Al in the same matrix solution) were used for quantification. Triplicate measurements were taken to indicate homogeneity of the solution and instrument precision.

The loading of CO2 was determined using a Gas Chromatograph fitted with a Thermal Conductivity Detection (GC-TCD) (HP6890) for the liquid samples. Additionally, solution pH was monitored to confirm that loading remained constant throughout the duration of the experiment.

The surface of the corrosion coupons were prepared with a 10 nm gold coating for imaging using Scanning Electron Microscopy (SEM). Additionally, cross-sections of the coupons were embedded in Cold-Fix resin perpendicular to the surface area to measure siderite layer thickness using SEM. A cobalt standard was used before each Energy Dispersive X-ray (EDX) (JEOL6400) measurement to ensure the most accurate quantification as is possible. For X-ray diffraction (XRD), corrosion coupons were mounted on a solid sample holder using plasticine and the 2θ angles from 10 to 100 analysed at a step size of 0.334 and scan time of 40 seconds (PAnalytical X-Ray Diffractometer). The baseline for the corrosion products show fluorescence due to the primarily Fe substrate [33].

Results and Discussion

The CO2 loading for all solutions was measured using GC-TCD. As this study was designed based on literature studies, a comparison of the loading values is given (Table 3). Literature values for CO2 capacity were loaded and measured at 40 oC. In this work, solution loading took place at ambient temperature and subsequently heated to 120 oC which may reduce CO2 solubility and limit the solution to lean loading (the conditions of interest for testing in this work). It has been previously noted that inconsistencies exist widely in the literature (Table 1), with this study no different. This demonstrates a need for understanding of this phenomenon, but will not be addressed further herein.

Material size change rates (MSCR) offer a crude, but traditional, method of establishing the changes occurring to the metal coupon. The calculation uses the weight change of steel coupons after seven days of exposure to the aqueous amine blends. The MSCR of 30 % by weight solutions all reported negative values, indicating weight gain. This gain appears to be due to the growth of a product layer on the coupon surface which is visible macroscopically. The extent to which product layers formed varied amongst the different blends. The rate measured for 30 % by weight PZ is the same order of magnitude as the blend 20MDEA/10PZ, after the coupon prescribed cleaning (Figure 3). Comparatively, blends of 20AMP/10PZ, 10MDEA/20PZ and 10AMP/20PZ were smaller, thereby suggesting significantly higher levels of product layer growth. Blended amines comprising a total of 50 % by weight in solution show comparatively lower MSCR (i.e. more weight gain) to those of 30 % by weight. Critically, solutions containing a lower concentration of PZ (10 % by weight) appear to have less adhesive product layers whereas at 20 % by weight PZ the product layer is less likely to be removed during the routine coupon cleaning process. Whilst observed qualitatively, this trend also can be seen quantitatively from the reduced size of the error bars (Figure 3) where the discrepancy between cleaned and crude coupon weights is denoted using the error bars. The surprisingly distinct result for 50 % by weight PZ is important to consider. The MSCR indicates a significant net loss of 0.05 mm y-1 of material to solution. This loss suggests that at such a high concentration of PZ (at a high temperature) a strong driving force results in dramatically increased oxidation at the surface of the metal coupon. Additionally, other forms of corrosion (e.g. pitting) may also be occurring.

Figure 3. MSCR of C1018 steel in aqueous amine blends.

Table 3. GC-TCD measured CO2 loading capacities at 120oC, saturated

†This work where loading shown is the average and standard deviation over the seven day test.

The Fe ion concentration, in the liquid samples collected, was measured using ICP-OES for all 30 % and 50 % by weight solutions (Figure 4). In aqueous solutions with only PZ, the Fe ion concentration reaches a peak at day four in 30 % by weight and at day two in 50 % by weight (Figure 3). The concentration of available Fe ions in the bulk represents those which are not reacted with carbonate species in the formation of siderite. At a lower concentration of PZ (30 % by weight), carbonate and bicarbonate species formed are readily consumed by the available Fe ions oxidised on the surface of the steel coupon. However, an insufficient solution concentration of carbonate species results in an increasing bulk concentration of unconsumed Fe ions. As the formation of siderite covers the surface, reduced Fe oxidation and subsequent flow to the bulk is observed. As such, the concentration of Fe ions in the bulk begins to be reduced from available CO32- ions still being produced by the PZ present. By contrast, a higher concentration of PZ (50 % by weight) clearly serves as a driving force for a more significant oxidation at the surface of the coupon, resulting in a higher early bulk concentration of Fe ions (which helps to explain the apparent significant mass loss to the coupon). However, this concentration does rapidly decrease due to the consumption by a large concentration of available CO32- ions. After seven days (168 hours), the remaining bulk concentration for 50 % by weight PZ is only double that of 30% by weight.

Aqueous amine solutions of PZ blended with MDEA or AMP categorically demonstrate increased Fe ion concentration in the solution bulk. From observation, the inclusion of MDEA or AMP has an adverse effect on carbon steel oxidation, possibly due to competing capture mechanisms between PZ and the additives resulting in slowed precipitation and increased exposure of carbon steel to the solution. In turn, this behaviour has the potential to decrease efficacy in protection against corrosion. Solutions comprised of 30 % by weight amines, generally, demonstrate an order of magnitude lower Fe ion concentration in solution than those at 50 % by weight (Figure 4); though, the trends observed based on composition are parallel. For solutions with 30 % by weight amine, the Fe ion concentration was statistically significantly higher for solutions containing MDEA, rather than AMP (at the same % by weight), blended with PZ. The solution 10MDEA/20PZ reached 2.208 ± 0.098 mg L-1 and 20MDEA/10PZ 2.190 ± 0.123 mg L-1 after seven days (168 hours) (Figure 4 A); by contrast, 20AMP/10PZ only reached 1.468 ± 0.075 mg L-1 at seven days and 10AMP/20PZ reached low levels commensurate with 30 % by weight PZ. (Figure 4 B).

Figure 4. Fe ion concentration in MDEA/PZ, AMP/PZ, and PZ for (A) 30% and (B) 50 % dilution by weight.

In these blended solutions of PZ with either MDEA or AMP, there are competing CO2 reaction pathways. The formation of PZ-carbamate (Equations 10 - 15) is a preferred reaction when competing against bicarbonate formation via MDEA (Equations 5 - 7) [12]. Therefore, in a blend of MDEA/PZ, CO2 reacts with PZ molecules preferentially, only undergoing hydrolysis through MDEA capture with excess CO2 after PZ molecules have been exhausted [12]. As a SH amine, AMP demonstrates a comparatively faster uptake of CO2 than the 3o amine MDEA. This increases the rate of carbonate species production and therefore siderite, resulting in a lower net Fe ion concentration in solution. This difference is likely caused by the higher activation energy requirement for MDEA between 298K and 313K in CO2 capture than for AMP at identical conditions [10]. In turn, the capture reaction kinetics of AMP exceed those of MDEA [14]. Therefore, the capture of CO2 in MDEA is initially slower than AMP, consequently producing a lower concentration of CO32- . This limits the overall rate of siderite formation which allows Fe ions to build up to higher concentrations in the bulk solution.

Similarly in the case of AMP blends, replacement of PZ content adversely affects the solution’s capacity for producing siderite. Specifically in blends 20AMP/10PZ and 10AMP/20PZ, through an increase in AMP concentration and replacing PZ content, the dominant capture occurs through the SH mechanism. As the SH reaction is kinetically slower than that used by PZ, the rate of formation of CO32- is insufficient to form a siderite layer as quickly and hence the Fe ions in solution maintain a higher concentration.

For solutions of pure PZ, both 30 % and 50 % by weight produced pure siderite layers on the coupon surfaces, confirmed through XRD (Figure 5 C and F). This indicates the presence of CO32- in solution, which is a product of the PZ capture mechanism (Equations 10 – 15). Additional CO32- formed through CO2 hydrolysis, traditionally attributed to 3o and SH amines, is likely occurring according to kinetic experiments reported in literature [9], [24], [25]. SEM images of the coupon surfaces taken from these two solutions (Figure 5 A and D) show near identical siderite layers with no outstanding features (crystal size, shape and inter-crystal spacing). Additionally, similar layer thicknesses in 30 % by weight PZ (5 ± 1 μm) and 50 % by weight PZ (4 ± 1 μm) are also observed (Figure 5 B and E) which suggest that layer thicknesses have reached a maximum. At this point, the layer has reached its full potential effectiveness as Fe ions are no longer oxidised into solution, evidenced by the decreasing net Fe ion concentration (Figure 3) as discussed previously. Given similarities in product layers, the drastic MSCR for 50 % by weight PZ, in contrast to 30 % by weight, remains surprising. However, it is impossible in such a study to close the mass balance where the formation of siderite could also occur in solution or on other available solid surfaces, as well as being shed from the coupon surface while in the reaction vessel. To ascertain more about this, an indepth study focusing on PZ is essential and should also assess beyond the general corrosion mechanism to include such behaviours as localised pitting and grain boundary corrosion.

Figure 5. 30 % by weight PZ solution (A) Surface at 1000x, (B) cross-section thickness at 2000x, (C) XRD and 50 % by weight PZ solution (D) surface at 1000x, (E) cross-section thickness at 2000x and (F) XRD. R=resin, P=product, S=steel substrate.

A

B

D

E

F

C

S

R

P

S

R

P

9 ± 1 µm

8 ± 1 µm

40 μm

20 μm

40 μm

20 μm

There is no apparent direct correlation between the thicknesses of siderite layers formed in 30 % by weight blends (Figure 6) and the Fe ion concentrations measured from samples of the bulk solution. The concentration of PZ has the only apparent, but minimal, influence where an increased PZ concentration shows an increase in the final siderite layer thickness (Figure 6 B, D, F and H). However, the thickness does not account for surface variations (crystal structure, crystal size, surface trenches, defects and layering) which are apparent and more closely linked to Fe ion concentration. Among the siderite crystals formed, solution type is a critical factor in the quality of the layer produced as determined by both surface characteristics and layer thickness. In aqueous solutions containing only PZ, equivalent product layer thicknesses were produced. However, the discrepancies in crystal habit are the likely source of different Fe ion concentrations produced for 30 % and 50 % by weight PZ.

Figure 6. SEM imaging of coupon surface and cross section in 30 % by weight solutions by order of decreasing solution Fe ion concentration. (A/B) 10MDEA/20PZ at 2000x and 4000x, (C/D) 20MDEA/10PZ at 2000x and 4000x, (E/F) 20AMP/10PZ at 2000x and 2000x and (G/H) 10AMP/20PZ at 2000x and 5000x. R=resin, P=product, S=steel substrate.

7 ± 1 µm

5 ± 1 µm

4 ± 1 µm

4 ± 1 µm

R

P

S

R

P

S

R

P

S

R

P

S

A

B

C

D

E

F

G

H

20 μm

20 μm

20 μm

20 μm

10 μm

10 μm

10 μm

9 μm

The siderite layers on the coupon surfaces are formed through both crystal nucleation near the coupon surface and deposition of particles from bulk solution nucleation. In siderite deposition, Fe2+ and CO32- in the bulk solution precipitate and settle onto the coupon surface. These two mechanisms were observed and discussed in Zheng et al. 2015, and suggested siderite deposits from solution formed superficial layers. Ideally, precipitation of siderite nuclei near the carbon steel surface, where ion saturation levels were higher, can promote the formation of densely packed layers [20]. Microscopic defects, such as trenches, are commonplace for samples containing MDEA: e.g. a large trench measuring 25 ± 1 μm long and 7 ± 1 μm wide on the 10MDEA/20PZ coupon is shown in Figure 6 A. The presence of such significant gaps in the siderite layer allows the solution to continuously oxidise the base metal, reducing the protection afforded by the crystalline product. Previous work on corrosion of carbon steel in MDEA and AMP solutions suggested crystal growth mechanisms had an impact on the effectiveness of siderite layers [34]. The cuboidal habits observed in 10MDEA/20PZ and 20MDEA/10PZ suggests that there is either reduced crystal collision and/or no significant additional deposition for nucleation in the gaps. Leaving a less densely packed product layer allows the corrosive solution to access the steel surface and continue oxidation.

The Fe ion concentrations in 50 % by weight solutions are closely correlated to the thickness of siderite layers measured. Solutions with a lower concentration of Fe ions in solution also had thinner product layers (Figure 7 and Figure 5 E). A further relationship between Fe ion concentration and crystal shape of siderite is observed. In blends of MDEA the crystals are more uniform in shape and size (Figure 7 A, B, C and D); whereas, those formed in pure PZ and AMP blends have a larger size distribution and are poorly formed (Figure 5 D and E; Figure 7 E, F, G and H). This distinction is the result of different crystallisation mechanisms between MDEA, which favours crystal growth, and PZ or AMP, which promote nucleation, also observed by Yu et al. 2016 [34]. Notably, the addition of a kinetic retardant has apparent correlation with layer thickness, which increases from the minimum in pure PZ (kinetically fastest) to a maximum in MDEA blends (kinetically slowest).

Figure 7. SEM imaging of coupon surface and cross section in 50 % by weight solutions by order of decreasing solution Fe concentration. (A/B) 40MDEA/10PZ at 2000x and 2000x, (C/D) 30MDEA/20PZ at 2000x and 2000x, (E/F) 30AMP/20PZ at 2000x and 2000x and (G/H) 40AMP/10PZ at 2000x and 2000x. R=resin, P=product, S=steel substrate.

12 ± 1 µm

15 ± 1 µm

9 ± 1 µm

7 ± 1 µm

A

B

C

D

E

F

G

H

R

R

P

S

P

S

R

P

S

R

P

S

20 μm

20 μm

20 μm

20 μm

20 μm

20 μm

20 μm

20 μm

As previously described, slower capture rates of CO2 by MDEA result in lower rates of carbonate formation. At reduced concentrations of CO32-, the growth kinetics of siderite crystals are slowed, providing favourable conditions for growth of existing nuclei [35]. These relatively fast reactions between AMP and CO2 increase carbonate formation, thereby leading to higher ion saturation and promoting formation of new siderite nuclei [35]. The Fe ion concentrations in 50 % by weight solutions are higher in 40MDEA/10PZ and 30MDEA/20PZ whilst lower in 30AMP/20PZ and 40AMP/10PZ (Figure 4 B). Similarities also exist across 30 and 50 % by weight blends and can be seen in 20AMP/10PZ, 10AMP/20PZ (Figure 6 E and G), 30AMP/20PZ and 40AMP/10PZ (Figure 7 E and G) where siderite crystals have similar average sizes of 3 ± 1 μm. This observation suggests that siderite products in AMP based solutions are more likely to nucleate and deposit upon on existing layers.

Overall, the trends in solutions of 30 % by weight are less well defined than solutions of 50 % by weight. However, parallels between the two cases are clear. From cross-sectional SEM imaging of the siderite, the thinnest layers of 4 ± 1 μm were observed on coupons from 20MDEA/10PZ, 20AMP/10PZ and 30PZ (Figure 6 C, D, E and F; Figure 5 A, B). The crystal structures appear well packed with a higher density of nuclei per unit area, which suggests continued nucleation and growth of siderite in the surface sub-layer. By contrast, the siderite layer was 7 ± 1 μm in 10MDEA/20PZ solution (Figure 6 A and B). This case also shows much larger inter-crystal spacing and trenching, distinguishing the growth behaviour from those with lower PZ concentrations. As such, it is likely that precipitation on the surface of pre-existing siderite layers causes the marginal increase in thickness, but reduced efficiency in protection from oxidation of the metal. An important case can be seen in 10AMP/20PZ (Figure 6 G and H) where a well packed layer also appears covered in new nuclei. The Fe ion concentration and resulting siderite layer in this blend closely match with that of 30 % by weight PZ on which the corresponding siderite layer was referred to as having maximum effectiveness. The combination of crystal growth and surface precipitation could be a key to eliminating the net flux of Fe ions into solution. The correlation between nuclei density and layer thickness suggests that the location of crystal nucleation and growth is a factor in determining the operational effectiveness of the siderite layer formed.

Conclusion

The highly favourable CO2 capacity of SH (AMP) and 3o (MDEA) amines blended with a 2o kinetic rate promoter (PZ) have improved efficiencies of the PCCC process. Specific blends, while deemed promising in literature, have limited data on their corrosiveness. Future industrial deployment of PCCC may be more widespread upon minimising the corrosion of inexpensive carbon steels. This work draws upon a selection of blends in literature; 30 % and 50 % by weight solutions of MDEA and PZ and AMP and PZ investigated at 120 oC and atmospheric pressure.

The SH (AMP) and 3o (MDEA) structures and the capture mechanism of 2o PZ provide the basis for formation of surface corrosion inhibiting layers of FeCO3. In all of the blends examined here, only siderite could be detected on the surface of the carbon steel coupons. Both 30 % and 50 % by weight PZ solutions produced siderite layers sharing similar crystal habit, size, and layer thickness, and additionally, decreasing Fe ion concentrations in solution. The rapid reaction kinetics between PZ and CO2 enable a high rate of carbonate speciation and guarantee carbonate ion saturation in the bulk solution. This in turn readily reacts with Fe ions at the immediate surface of the coupon, forming an effective product layer sooner and thus reducing Fe ions in solution. Within blends, the thickness of siderite layers and corrosion inhibition were uncorrelated and resulted in solution Fe ion concentrations higher than PZ solutions. The reaction of CO2 with MDEA and AMP are kinetically slower than PZ due to the 3o and SH capture mechanisms. Blending at 30 % and 50 % by weight amine replaces PZ with the slower reacting amine and reducing overall CO2 capture rate which in turn lowers carbonate concentrations and thus slows siderite growth.

Blend type appears to influence corrosion protection with AMP based blends outperforming those with MDEA. Faster reaction kinetics between AMP and CO2 increase the rate at which the bulk is saturated with carbonate species. This in turn increases the rate of siderite formation and lowers net Fe ions in solution. Crystal habit differences formed between AMP and MDEA blends are noticeable, where AMP blends promote irregular shaped crystals covering a larger size distribution. Additional deposits of smaller siderite crystals were observed to fill inter-crystal gaps which contributed to further corrosion protection. Considering the different blends and siderite formations, AMP and PZ systems show the most promise for use with carbon steel infrastructure, characterised by rapid siderite formation and having high CO2 capacity.

Acknowledgements

The authors would like to acknowledge the Imperial College Department of Materials X-ray Diffraction and Electron Microscopy facilities, as well as the Department of Chemical Engineering’s Analytical Laboratory and Mechanical Workshop. In addition, at the University of Surrey’s Mechanical Engineering Department, the authors would like to thank Mr. D. Jones for his assistance in preparation of the epoxy-mounted samples. Dr. Sedransk Campbell would like to thank the Royal Society and EPSRC for support of her Dorothy Hodgkin Research Fellowship. Finally, the authors would like to thank Prof. Gary Rochelle for his discussions about piperazine.

References

[1]. Sreenivasulu, B., Gayatri, D.V., Sreedhar, I. & Raghavan, K.V. (2014) A journey into the process and engineering aspects of carbon capture technologies. Renewable and Sustainable Energy Reviews. 41, 1324-1350.

[2]. Adeosun, A., El Hadri, N., Goetheer, E. and Abu-Zahra, M.R.M. (2013). Absorption of CO2 by Amine Blends Solution: An Experimental Evaluation. International Journal of Engineering and Science. 3(9), 12-23.

[3]. Adeosun, A. and Abu-Zahara, M. (2013). Absorption in Alkanolamine Blends for CO2 Capture. 2nd Post Combustion Capture Conference.

[4]. Li, L., Voice, A.K., Li, H., Namjoshi, O., Nguyen, T., Du, Y. and Rochelle, G.T. (2013). Amine Blends Using Concentrated Piperazine. Energy Procedia. 37, 353-369.

[5]. Bruder, P and Hallvard, F.S. (2012) Capacity and kinetics of solvents for post combustion CO2 capture. Energy Procedia. 23. 45-54.

[6]. Billingham , M.A., Lee, C.H., Smith, L., Haines, M., James, S.R., Goh, B.K.W., Dvorak,K., Robinson, L., Davis, C.J. and Peralta-Solorio, D., (2011). Corrosion and materials selection issues in carbon capture plants. Energy Procedia. 4, 2020-2027.

[7]. Kohl, A. & Nielsen, R. (1997) Gas Purification. Gulf Publishing Company Houston. Texas, USA

[8]. Dubois, L and Thomas, D. (2009). CO2 Absorption into Aqueous Solutions of Monoethanolamine, Methyldiethanolamine, Piperazine and their Blends. Chemical Engineering & Technology. 32(5). 710-718.

[9]. Jacques-Derks, P. W. (2006). Carbon Dioxide Absorption in Piperazine Activated N-Methyldiethanolamine. PhD Thesis. The University of Twente.

[10]. Sodiq, A., Rayer, A.V. and Abu-Zahra, M.R.M. (2014). The Kinetic Effect of Adding Piperazine Activator to Aqueous Tertiary and Sterically-Hindered Amines Using Stopped-Flow Technique. Energy Procedia. 63, 1256-1267.

[11]. Rochelle, G., Chen, E., Freeman, S., Van Wagener, D., Xu, Q. and Voice, A. (2011). Aqueous Piperazine as the New Standard for CO2 Capture Technology. Chemical Engineering Journal. 171, 725-733.

[12]. Closmann, F., Nguyen, T. and Rochelle, G.T. (2009) MDEA/Piperazine as a solvent for CO2 capture. Energy Procedia. 1. 1351-1357.

[13]. Nguyen, B. T. N. (2013) Amine Volatility in CO2 Capture. PhD thesis. The University of Texas at Austin.

[14]. Chen, X., Closmann, F. Rochelle, G.T. (2011) Accurate Screening of Amines by the Wetted Wall Column. Energy Procedia. 4. 101-108.

[15]. Abdulkadir, A, Al-Ali, K., Hajaj, A.A., Nashef, E. and Zahra, M.A. (2015) CO2 enthalpy and heat capacity measurements in aqueous Piperzine blends with different Alkanolamines. In: International Energy Agency Greenhouse Gas R&D Programme, 3rd Post Combustion Carbon Capture Conference, Saskatchewan, Canada.

[16]. Dash, S.K.,Samanta, A.,Samanta, A.N. and Bandyopadhyay, S.S. (2011) Absorption of carbon dioxide in piperazine activated concentrated aqueous 2-amino-2-methyl-1-propanol solvent. Chemical Engineering Science. 66. 3223-3233.

[17]. Yang, Z.Y., Soriano, A.N.,Caparanga, A.R. and Li, M.H. (2010) Equilibrium solubility of carbon dioxide in (2-amino-2-methyl-1-propanol + piperazine + water). The Journal of Chemical Thermodynamics. 42. 659-655.

[18]. Gouedard, C., Rey, A., Ledirac, N., Cohen, A., Dugay, J., Vial, J., Pichon, V., Bertomeu, L., Picq, D. & Carrette, P.L. (2013) Amine degradation in CO2 capture. 2. New degradation products of MEA. Pyrazine and alkylpyrazines analysis, mechanism of formation and toxicity. International Journal of Greenhouse Gas Control. 19. 576-583.

[19]. Hatcher, N.A., Jones, C.E., Weiland, G.S. & Weiland, R.H. (2014) Predicting corrosion rates in amine and sour water systems, Optimised Gas Treating.

[20]. Zheng, L., Matin, N.S., Landon, J., Thomas, G.A. and Liu, K. (2015) CO2 loading-dependent corrosion of carbon steel and formation of corrosion products in anoxic 30 wt.% monoethanolamine-based solutions. Corrosion Science. 102, 44-54.

[21]. Choi, Y.S., Nešić, S., Duan, D. & Jiang, S. (2012) Mechanisitic Modelling of Carbon Steel Corrosion in MDEA-based CO2 Capture Process. Corrosion NACE International Conference and Expo. C2012-0001321.

[22]. Xiang,Y., Choi, Y.S., Yang, Y. & Nešić, S. (2015) Corrosion of Carbon Steel in MDEA-Based CO2 Capture Plants Under Regenerator Conditions: Effects of O2 and Heat-Stable Salts. Corrosion. 70 (1). 30-37

[23]. Tanupabrungsun, T., Brown, B. & Nesic, S. (2013) Effect of pH on CO2 Corrosion of Mild Steel at Elevated Temperatures. Institute of Corrosion and Multiphase Technology. 2348.

[24]. Ruzic, V., Veidt, M. and Nešić, S. (2006) Protective Iron Carbonate Films—Part 1: Mechanical Removal in Single-Phase Aqueous Flow. Corrosion 62(5). 419-432.

[25]. Nešić, S., Nordsveen, M., Nyborg, R. and Stangeland, A. (2003) A Mechanistic Model for Carbon Dioxide Corrosion of Mild Steel in the Presence of Protective Iron Carbonate Films—Part 2: A Numerical Experiment. Corrosion. 59(6) 489-497.

[26]. Zheng, L., Landon, J., Matin, N., Li, Z., Qi, G. and Liu, K. (2014). Corrosion Behaviour of Carbon Steel in Piperazine Solutions for Post-Combustion CO2 Capture. The Electrochemical Society. 61(20), 81-95.

[27]. Bougie, F and Iliuta, M.C. (2011). CO2 Absorption in Aqueous Piperazine Solutions: Experimental Study and Modelling, Journal of Chemical and Engineering Data. 56. 1547-1554.

[28]. Zheng, L., Landon, J., Zou, W. & Liu, K. (2014) Corrosion benefits of Piperazine as an Alternative CO2 Capture Solvent. Industrial & Engineering Chemistry Research. 53. 11740-11746.

[29]. Freeman, S.A. (2011). Thermal Degradation and Oxidation of Aqueous Piperazine for Carbon Dioxide Capture. PhD thesis. The University of Texas at Austin.

[30]. Zheng, L., Landon, J., Matin, N., Li, Z., Qi, G. and Liu, K. (2014). Corrosion behaviour of carbon steel in piperazine solutions for post-combustion CO2 capture. ECS Transactions. 61(20). 81-95.

[31]. Meuleman, E., Artanto, Y., Jansen, J., Osborn, M., Pearson, P., Cottrell, A. & Feron, P. (2010). CO2 capture performance of MEA and blended amine solvents in CSIRO’s pilot plant with flue gas from a brown coal-fire power station. 6th International Technical Conference on Clean Coal & Fuel Systems. 3.

[32]. Herrag, L., Hammouti, B., Elkadiri, S., Aouniti, A., Jama, C., Vezin, H. and Bentiss, F. (2010). Adsorption properties and inhibition of mild steel corrosion in hydrochloric solution by some newly synthesized diamine derivatives: Experimental and theoretical investigations. Corrosion Science. 52(9). 3042-3051.

[33]. Fransen, M.J. (2004) 1 and 2 Dimensional Detection Systems and the Problem of Sample Fluorescence in X-Ray Diffractometry. International Centre for Diffraction Data. 47, 224-231.

[34]. Yu, L.C.Y., Campbell, K.L.S. and Williams, D.R. (2016). Using carbon steel in the stripper and reboiler for post-combustion CO2 capture with aqueous amine blends. International Journal of Greenhouse Gas Control. 51, 380-393.

[35]. Mullin, J.W. (2001) Crystallization. 4th edition. Butterman Heinemann publishing.

-6.351146028964679E-3-2.1911453799920527E-2-1.0479390947786361E-23.1755730144738777E-4-1.587786507236939E-3-3.4931303159325476E-3-1.0479390947786361E-2-1.1432062852134167E-2-1.9053438086899677E-3-3.4931303159325476E-3-9.5267190434216342E-4-1.5877865072425799E-3-6.351146028964679E-3-2.1911453799920527E-2-1.0479390947786361E-23.1755730144738777E-4-1.587786507236939E-3-3.4931303159325476E-3-1.0479390947786361E-2-1.1432062852134167E-2-1.9053438086899677E-3-3.4931303159325476E-3-9.5267190434216342E-4-1.5877865072425799E-330PZ20MDEA/10PZ20AMP/10PZ10MDEA/20PZ10AMP/20PZ50PZ40MDEA/10PZ40AMP/10PZ30MDEA/20PZ30AMP/20PZ-2.2229011101317147E-3-1.2702292057951922E-3-1.2702292057923718E-2-1.3019849359376745E-2-8.5740471390963937E-34.8903824423016176E-2-1.4290078565166297E-2-2.3181683005710076E-2-2.3499240307163105E-2-1.0479390947786361E-2Dummy0.25

MSCR (mm y-1)

-6.351146028964679E-3-2.1911453799920527E-2-1.0479390947786361E-23.1755730144738777E-4-1.587786507236939E-3-3.4931303159325476E-3-1.0479390947786361E-2-1.1432062852134167E-2-1.9053438086899677E-3-3.4931303159325476E-3-9.5267190434216342E-4-1.5877865072425799E-3-6.351146028964679E-3-2.1911453799920527E-2-1.0479390947786361E-23.1755730144738777E-4-1.587786507236939E-3-3.4931303159325476E-3-1.0479390947786361E-2-1.1432062852134167E-2-1.9053438086899677E-3-3.4931303159325476E-3-9.5267190434216342E-4-1.5877865072425799E-330PZ20MDEA/10PZ20AMP/10PZ10MDEA/20PZ10AMP/20PZ50PZ40MDEA/10PZ40AMP/10PZ30MDEA/20PZ30AMP/20PZ-2.2229011101317147E-3-1.2702292057951922E-3-1.2702292057923718E-2-1.3019849359376745E-2-8.5740471390963937E-34.8903824423016176E-2-1.4290078565166297E-2-2.3181683005710076E-2-2.3499240307163105E-2-1.0479390947786361E-2Dummy0.25

MSCR (mm y-1)

A

10MDEA/20PZ07.0216332383095381E-26.8549252366455413E-29.8475039138521417E-207.0216332383095381E-26.8549252366455413E-29.8475039138521417E-2024701.36166666666666661.83899999999999972.208333333333333520MDEA/10PZ03.6555893277737513E-20.104160133128435130.1230541886053998803.6555893277737513E-20.104160133128435130.12305418860539988024701.20033333333333331.79233333333333362.189666666666666210AMP/20PZ02.8112867753634408E-23.8279672586548415E-29.0184995056457971E-302.8112867753634408E-23.8279672586548415E-29.0184995056457971E-3024700.163666666666666650.289333333333333390.2573333333333333620AMP/10PZ04.8294237061302987E-26.4267669425095278E-27.5498344352707566E-204.8294237061302987E-26.4267669425095278E-27.5498344352707566E-2024700.833333333333333371.21333333333333321.46830PZ01.0969655114602885E-24.7444704657105818E-26.557438524302006E-301.0969655114602885E-24.7444704657105818E-26.557438524302006E-3024700.137333333333333340.4367.8E-2DUMMY01

Duration (days)

Iron concentration (mg L-1)

B

30MDEA/20PZ00.195167449471814080.374477413649119350.1702967997350508500.195167449471814080.374477413649119350.17029679973505085024708.266333333333333810.4533333333333339.066999999999998440MDEA/10PZ00.237506491139365530.740810367098085340.7600219295081773700.237506491139365530.740810367098085340.76002192950817737024709.645666666666665311.7511.40333333333333430AMP/20PZ00.446788913619543540.28661704996970.272472628594751300.446788913619543540.28661704996970.2724726285947513024703.61533333333333353.87233333333333363.591666666666666840AMP/10PZ00.736882849124159130.405832888432336560.1148956047897394900.736882849124159130.405832888432336560.11489560478973949024703.7476666666666663.95633333333333333.813000000000000250PZ06.1784571968520795E-20.141877881762216757.5718777944003713E-306.1784571968520795E-20.141877881762216757.5718777944003713E-3024701.44066666666666680.767666666666666610.19466666666666668DUMMY01

Duration (days)

Iron concentration (mg L-1)

A

10MDEA/20PZ07.0216332383095381E-26.8549252366455413E-29.8475039138521417E-207.0216332383095381E-26.8549252366455413E-29.8475039138521417E-2024701.36166666666666661.83899999999999972.208333333333333520MDEA/10PZ03.6555893277737513E-20.104160133128435130.1230541886053998803.6555893277737513E-20.104160133128435130.12305418860539988024701.20033333333333331.79233333333333362.189666666666666210AMP/20PZ02.8112867753634408E-23.8279672586548415E-29.0184995056457971E-302.8112867753634408E-23.8279672586548415E-29.0184995056457971E-3024700.163666666666666650.289333333333333390.2573333333333333620AMP/10PZ04.8294237061302987E-26.4267669425095278E-27.5498344352707566E-204.8294237061302987E-26.4267669425095278E-27.5498344352707566E-2024700.833333333333333371.21333333333333321.46830PZ01.0969655114602885E-24.7444704657105818E-26.557438524302006E-301.0969655114602885E-24.7444704657105818E-26.557438524302006E-3024700.137333333333333340.4367.8E-2DUMMY01

Duration (days)

Iron concentration (mg L-1)

B

30MDEA/20PZ00.195167449471814080.374477413649119350.1702967997350508500.195167449471814080.374477413649119350.17029679973505085024708.266333333333333810.4533333333333339.066999999999998440MDEA/10PZ00.237506491139365530.740810367098085340.7600219295081773700.237506491139365530.740810367098085340.76002192950817737024709.645666666666665311.7511.40333333333333430AMP/20PZ00.446788913619543540.28661704996970.272472628594751300.446788913619543540.28661704996970.2724726285947513024703.61533333333333353.87233333333333363.591666666666666840AMP/10PZ00.736882849124159130.405832888432336560.1148956047897394900.736882849124159130.405832888432336560.11489560478973949024703.7476666666666663.95633333333333333.813000000000000250PZ06.1784571968520795E-20.141877881762216757.5718777944003713E-306.1784571968520795E-20.141877881762216757.5718777944003713E-3024701.44066666666666680.767666666666666610.19466666666666668DUMMY01

Duration (days)

Iron concentration (mg L-1)

H

2

NOH

HNNH

HONOH

HO

H

2

N