Review of Studies on Corrosion of Steel by CO₂, Focussed on the Behaviour of API Steel in Geological CO₂ Storage Environment

When well stability is talked about, an important topic is corrosion (of either steel or cement). However, in this paper, the focus is on steel corrosion. It is important to mention that every completion in wells requires metallic components (wellhead, Christmas tree, tubing, casing, packer, etc.), which – in most of the cases (not always) – are made of some form of steel that have metallic alloy elements, i.e. chromium [5].

A CO₂ injector well can suffer corrosion in every component, from the wellhead to the down-hole completion components and, therefore, to achieve large success in the CO₂ storage project, the wells must bear the highest impact of corrosion [2]; thus, to know how to resist corrosion, it is necessary to know how it attacks.

CO₂is stable, inert and non-corrosive as a gas; however, when it is in the presence of water (either in the aqueous or vapour CO₂ phase), dissolution occurs, with subsequent hydration, forming H₂CO₃ (carbonic acid), which is the principal agent attacking steels [6]; the principal corrosion products are FeCO₃ (iron carbonate) and Fe₃C (iron carbide) [7] but, to understand this mechanism better, analysing how it proceeds step by step is required.

Following Table 1, first the CO₂ is hydrated by dissolving in water, forming carbonic acid (H₂CO₃), which is a weak acid because the CO₂ is partly dissociated in water. As carbonic acid is diprotic, it dissociates in two steps, which is considered the main source of acidity. The resulting pH is a function of the CO₂ partial pressure. In the first dissociation of carbonic acid, the bicarbonate is obtained (HCO₃), which, later on, dissociates into carbonate (CO₃) [6, 8].

Knowing how CO₂ reacts in the presence of water, we can see that Fe also reacts with its environment (Table 2). Initially, Fe reacts with carbonic acid, yielding ferrous bicarbonate (Fe(HCO₃)₂); after this, the precipitation of iron carbonate occurs, in two ways. First, when the concentrations of Fe²⁺ and CO²₃^– ions reach the maximum solubility limit, they combine forming iron carbonate (FeCO₃), with a consequent increase in pH [9]. Second, when the ferrous bicarbonate dissociates, it forms FeCO₃ + CO₂ + H₂O; this water could react with the resultant carbon dioxide, yielding, again, carbonic acid and, in this way, it could result in a cycle of corrosion.

The case when CO₂ and water are present has been explained, but, normally, water can have different kinds of ions, such as Cl^–, Na⁺, Ca⁺, SO²₁^– , etc., which can affect the equilibrium of CO₂ and the resulting pH [6]. On the other hand, the presence of FeCO₃ could lead to a reduction in the corrosion rate because it is formed on the steel surface [10]. Moreover, this protective film is dependent on low temperatures and high supersaturation of Fe²⁺ and CO²₃^– ; however, this protective film could grow without having any protective property, but, when it does have the property, it could reduce the corrosion rate between 5 and 100 times [8]. Adversely, even when the iron carbonate is precipitated, if it is not compacted on the surface, it cannot prevent the corrosion [10]. Therefore, some analyses have been performed, showing that at low temperatures, the FeCO₃ film gets dissolved continuously and the corrosion rate increases [11]. The precipitation kinetics of FeS is almost two orders of magnitude faster than that of FeCO₃ at the same conditions [12]; moreover, FeCO₃ was formed after 60 days, and even when there was a reduction in corrosion rate, there was still corrosion [13]. In addition, as shown in Figure 1, it is noted that the electrochemical dissociation of iron leads to surface corrosion.

Figure 1

Finally, although the Fe²⁺ ion could react later to form the protective film, this could lead to study the corrosion rate and rate of formation of protective film to know whether FeCO₃ is having a positive or negative effect, because as mentioned earlier [11, 12, 13], there is still corrosion on steel.

CO₂ can affect the different materials in CO₂ injector wells. Therefore, the metallurgical selection (as well as the type of cement) plays an important role [15] because the casing steel and other parts of the well can be subject to corrosion when they are exposed to CO₂ or formation fluids (or both) (Figure 2). This corrosion can be controlled by using corrosion-resistant alloys (CRAs) [16]. Furthermore, if a CRA is improperly selected, it can lead to mistakes in the future, affecting the performance; hence, taking care of the specific properties of the steels is required because a group of CRAs can be resistant at one temperature; however, this does not guarantee that the CRAs will work with the same performance in a different environment [17].

Figure 2

Casing and tubing play an important role in wells; so, ensuring their integrity is essential. Therefore, the American Petroleum Institute (API) has standardised some grades of steels, which are differentiated by groups, content of chemical compounds, manufacturing processes and mechanical properties [19], giving the API steel grade; different names are obtained, qualified by a number that represents the minimum yield strength and a letter chosen arbitrarily, without meaning. Therefore, they have been divided into four groups, categorised by resistance to sulphide stress cracking and working pressure [20]:

Furthermore, they can be divided by product specification level (PSL):

However, each carbon API steel has the following specific characteristics:

These products must comply with the required chemical composition (Table 3), tensile properties and thickness, as presented in Table 4.

Onoyama et al. [23] investigated the corrosion resistance of three different compositions of duplex stainless steel. They evaluated the first two in National Association of Corrosion Engineers (NACE) solution at 80 °C for 100 h to get the effect, in percentage, of different compositions of the alloying elements in the steel, finding that it is desirable to reduce the Si percentage (<0.5%), leading to the amount of Ni becoming 10% and increasing the resistance; moreover, Mo and N improve the steel and, Sn and Sb have to be correctly added, neither in excess or nor in lower amount, with the best values being 0.05% Sn and 0.15% Sb. On the other hand, duplex steel was tested in 2.8 × 10^–2 mol/L NaCl solution saturated with CO₂ and N₂ at 260 °C for 100 h at 0.10 and 0.05 MPa, obtaining the maximum value around 8.5 × 10^–3 mm/yr at 0.10 MPa of pressure and around 3 × 10^-3 mm/yr at 0.05 MPa of pressure.

Kimura et al. [24] studied the effect of CO₂ corrosion in 13Cr steel, modifying the chemical composition of Cu, Ni and Mo, at 3 MPa and 180 °C, inside an autoclave with 20% of NaCl solution saturated with CO₂ gas, for 7 days, evaluating the weight loss rate. When there was no addition of Cu, Ni and Mo, the corrosion rate was 1.71 mm/yr, but when these elements were part of the chemical composition, the corrosion rate was <0.3 mm/yr. Moreover, with the increase in Mo content at the same Ni percentage, the corrosion rate decreased. On the other hand, the addition of Cu did not affect the corrosion rate and- the effect of Ni-Cu was not clear.

Another study was carried out by Leth-Olsen [25], which compared carbon-steel, 13Cr and super 13Cr (S13Cr) in a potassium solution with CO_2(g) in the liquid and vapour phases, finding an excellent performance of S13Cr, with an average of 0.01 mm/yr, compared to 13Cr, which had a maximum value of 0.7 mm/yr and 0.3 mm/yr (in the liquid phase). Furthermore, the values of maximum corrosion were noted when the dummies were exposed in the liquid phase, the rates of corrosion in the vapour phase being 0.2 mm/yr for carbon-steel and 0.05 mm/yr for 13Cr; no corrosion was seen in S13Cr.

In 2008, the erosion-corrosion of 22Cr steel was analysed, compared with that of an X65 pipeline under the same conditions. This experiment was evaluated using sand grains under three special conditions. First, by varying the temperature from 20 to 70 °C at a sand concentration of 200 ppm and a CO₂ flow of 20 m/s, a constant corrosion rate of around 5 mm/yr was obtained for 22Cr, while X65 showed an increase from 12.25 to 43.5 mm/yr. Second, the solid particle concentration was varied from 30 to 200 ppm, with a temperature of 20 °C and flow velocity of 20 m/s; although having the same tendency, the corrosion rate for 22Cr increased from 0.1 to 2.4 mm/yr approximately, while for X65, it changed from 9 to 12 mm/yr. The last was the variation of velocity from 7 to 20 m/s, maintaining the temperature at 20 °C and concentration of sand particles at 100 ppm; the corrosion rate in 22Cr increased from 0.2 to 0.8 mm/yr, while for X65, it varied from 2.4 to 5.9 mm/yr approximately [26]. Then, it is important to note that an important parameter is considered in this study, namely, the solid particles.

Xu et al. [7] simulated the 3Cr steel used in the well casing in the Huizhou oilfield, with the same formation water composition to get the effect of CO₂ corrosion at different temperatures (45, 65, 85 and 105 °C), measuring every structure in a CO₂ environment, comparing the result in terms of temperature and CO₂ partial pressure (0.5 and 0.2 MPa). They concluded that by increasing the temperature and pressure, it is possible to gain higher corrosion rates. They observed a corrosion rate of 4.5 mm/yr at CO₂ partial pressure of 0.5 MPa and 100 °C; the highest value obtained at CO₂ partial pressure of 0.2 MPa was 1.14 mm/yr at 65 °C and approximately 1 mm/yr at 105 °C.

Cui et al. [27] evaluated the CO₂ corrosion at supercritical conditions (80 °C and 8.274 MPa) in J55, N80 and P110, at different water cuts (30, 50, 70, 90 and 100%), obtaining an increase in corrosion rate with higher water cut amount. Furthermore, N80 and J55 showed similar behaviour (7 and 4 mm/yr) and, at the same time, were better than P110 (9 mm/yr).

Furthermore, Lin et al. [28] compared the three steels mentioned herein at ion concentrations (g/l) of 19.0 Cl^–, 1.14 SO²₄^– , 0.6 HCO^–₃ , 1.05 Mg⁺₂ , 0.39 Ca⁺₂ , 11.99 Na⁺ and 0.12 CO²₃ . First, the pressures (6.89 and 10.34 MPa) were varied at 90 °C; the following values were obtained for the corrosion rate of N80, P110 and J55 steels: 1.752 mm/yr, 2.403 mm/yr, 1.854 mm/yr, 0.922 mm/yr, 1.054 mm/yr and 1.105 mm/yr, respectively. Furthermore, when the pressure was maintained constant (between 1.38 and 2.07 MPa CO₂ partial pressure), the corrosion peak was at 100 °C and, after that, it was lower, the best steels being N80 followed by J55.

Some studies have been carried out to compare J-55 and N-80. One on these studies was by Li et al. [29], who analysed the effect of temperature at 5 MPa of partial pressure and varied the partial pressure at 80 °C over a period of 72 h. Furthermore, these two kinds of steel were compared with carbon steel P110: when pressure is varied, the best result was obtained using J55; however, on varying the temperature, sometimes, the best result was obtained using N80.

When this group of steels is talked about, another important steel item referred to is K55. A study was carried out by Elramady [30], in which a piece was subjected in formation water with CO₂ at 40 psi (0.276 MPa) at ambient temperature, obtaining a rate varying from 0.4 to 0.74 mm/yr. On the other hand, Pehlke [31] used brine, CO₂, H₂S and N at 1000 kPa and 170 °C, obtaining a corrosion rate between 0.188 and 0.243 mm/yr for K55 steel.

Furthermore, when corrosion by CO₂ is studied, there are other acids that can affect the well apart from the carbonic acid. Jingen et al. [32] analysed the effect of the ratio of CO₂ partial pressure to H₂S partial pressure (10, 100, 200 and 400 MPa) in formation water, for N80 steel, at 90 °C and pCO₂ equal to 0.4 MPa, having a flow rate of 1.7 m/s; they found that when H₂S is not present, the corrosion rate is 1.689 mm/yr, while the ratio is 10; the rate is 0.172 mm/yr, changing to 0.789, 0.621 and 0.511 mm/yr for the subsequent ratios. Additionally, another work studied N80 steel at the same temperature, but at 4 MPa in the present of acetic acid with CO₂ in brine. A value of 0.55 mm/yr was obtained in the absence of the acid, but when solid particle concentration was increased to 1000, 3000 and 5000 ppm, the corrosion rate was 1.25, 3.55 and 4.95 mm/yr approximately [33].

In this group, one of the most commonly used steels is L80. One of the studies on L80 was by Choi et al. [34], where the temperature was set at 65 and 90 °C and the CO₂ pressures used were 4 MPa (gaseous phase) and 8 and 12 MPa (supercritical phase), in a saline solution of 25% HCl; at 65 °C, the corrosion rate was 8.7, 9.9 and 11 mm/yr, respectively, while at 90 °C, the corrosion rate was 6.1, 3.4 and 4.8 mm/yr, respectively. A similar study was done by Lopes et al. [35], which analysed the steel at 15 and 30 MPa in brine at 50 °C, obtaining 3.534 mm/yr at 15 MPa after 7 days, but 0.375 mm/yr after 30 days and 2.774 mm/yr at 30 MPa after 7 days.

The corrosion process could be affected by other factors, such as sulphur and water. Qiu et al. [36] studied corrosion at three temperatures – 60, 90 and 150 °C – and 0.1 MPa of H₂S partial pressure and 0.5 MPa of CO₂ partial pressure, varying the sulphur content (2 and 4%) and the water content (30 and 70%); the results are shown in Table 5. When the chromium composition of this steel is varied, the corrosion rate changes too, and an example of this is the work by da Silva [37], which used three different CO₂ partial pressures (0.1, 0.3 and 0.65 MPa) at 20 °C with 2, 3 and 5% of NaCl for 72 h, having 0.82% of Cr in twocoupons of L-80 steel (Table 6).

Two other steels in this group are T95 and C110, which were investigated by Elgaddafi [38], where they varied the H₂S concentration (0, 10, 50 and 150 ppm) at 38 °C and 41.37 MPa; for T95, the CO₂ corrosion rate was 15, 20, 17 and 17.5 mm/yr, while for C110, the values were 13, 14.5 and 9 mm/yr (there was no test for 150 ppm of H₂S).

The last two groups contain only two steels, which are used for deep well casing. Regarding the studies on P110, the first corresponds to analysis at 100 and 160 °C with a pressure of 4 MPa in a brine solution for 120 h, obtaining a marked contrast in corrosion rate, with values of 6.1843 and 0.8754 mm/yr [39]. Similarly, Guan [40] – at 90 °C and 4 MPa – obtained a corrosion rate around 0.8261 mm/yr; at 110 °C with the same pressure, the rate was 0.2489 mm/yr. On the other hand, if the pressure was varied (0.1, 2, 4 and 6 MPa), maintaining the same temperature (100 °C) for 168 h, the results were 0.64, 0.76, 0.91 and 0.83 mm/yr, respectively [41].

The other steel is Q125. This steel was studied in a CO₂ environment, in a brine solution simulated from the Jinlin Oil Field at 1 MPa and 30 °C and a flow of 1 m/s having a corrosion rate of 1.012 mm/yr [42]. Furthermore, varying the temperature (30, 60, 90 and 120 °C) and keeping the pressure at 2 MPa, using formation water from an oilfield, the values obtained were 0.9, 2.8, 3.6 and 2.4 mm/yr [43]. Giving continuity to the investigation from Elgaddafi [38], increasing the H₂S concentration, the corrosion rates were approximately 3.2, 9.5, 10 and 10 mm/yr.

The different values of corrosion rates indicate that corrosion could be dependent on the test time. Hence, these values were plotted against the days on linear (Figure 3) and semi-log (Figure 4) scales to get better information. There is a clear tendency versus the time in the graphs; in Figure 4, four regions were identified:

Figure 3

Figure 4

With the already-identified four zones of the corrosion mechanism, the tendency was modelled, yielding two curves; the first equation shows a slightly better fit in the initial region, and the second equation is the best simplified tendency approximation along time (blue dotted curve).

To analyse the long-term corrosion rates, these have to be in their plateau; it means, they have to be stabilised. Therefore, every rate with a test time below 45 days was stabilised by the mean of the tendency in Eq. (2) (because it is simpler and well fitted). Furthermore, they were separated by the CO₂ phase (vapour and supercritical). Then, to conclude the results observed in both graphs (Figures 5 and 6), it is necessary to refer to the NACE corrosion category, which defines the corrosion rate as low if the value is <0.025 mm/yr, moderate if it ranges between 0.025 and 0.15 mm/yr and high if the value is from 0.16 to 0.25 mm/yr. Thus, the steels could be classified as in Table 7, and in general, the corrosion rate is lower in the supercritical condition, which is the phase of CO₂ when it is injected, and higher in the vapour when it goes through the cement paste.

Figure 5

Figure 6

Addition of chemicals to form steel alloys permit better response against corrosion, especially when chromium is added (it forms stainless steel) and, furthermore, the quantity of nickel and molybdenum could help to prevent CO₂ corrosion. The effect of these elements when they are added to steels was analysed (Figures 7–9). There is a clear effect of chromium as a corrosion-protective element. Only its presence in steel reduces corrosion significantly, and at a percentage >20%, it presents a plateau but has a very low corrosion rate. When nickel is added, the rate reduces faster, helping the chromium as a protective element; however, the molybdenum has no significant effect, showing the same tendency as when it is not added. This is in the supercritical condition, because when the vapour phase is observed, the effect is not as high as in the supercritical condition.

Figure 7

Figure 8

Figure 9

Studying the effect of these two factors is difficult, because the idea is to analyse by varying one of them and maintaining the other one constant; to observe the effect, however, it was decided to graphically represent them as separate parameters (Figures 10 and 11). Then, varying the pressure (>7.38 MPa is supercritical condition), the lowest corrosion rates are found in the supercritical conditions, and as the pressure becomes higher, lower rates are obtained. However, when the temperature is evaluated, it is difficult to get a tendency probably because different values of pressure could be studied at one temperature.

Figure 10

Figure 11