Effect of major degradation products of ethylene glycol aqueous solutions on steel corrosion

Effect of major degradation products of ethylene glycol aqueous solutions on steel corrosion

G Model EA 26974 No. of Pages 12 Electrochimica Acta xxx (2015) xxx–xxx Contents lists available at ScienceDirect Electrochimica Acta journal homep...

2MB Sizes 136 Downloads 217 Views

G Model EA 26974 No. of Pages 12

Electrochimica Acta xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta

Effect of major degradation products of ethylene glycol aqueous solutions on steel corrosion M. Santambrogioa , G. Perruccia , M. Truebaa,* , S.P. Trasattia , M.P. Casalettob a

Dipartimento di Chimica, Università degli Studi di Milano, Via Golgi 19, 20133 Milan, Italy Consiglio Nazionale delle Ricerche (CNR), Istituto per lo Studio dei Materiali Nanostrutturati (ISMN), Area della Ricerca di Palermo, Via Ugo La Malfa 153, 90146 Palermo, Italy b

A R T I C L E I N F O

A B S T R A C T

Article history: Received 1 October 2015 Received in revised form 23 March 2016 Accepted 24 March 2016 Available online xxx

The effect of major degradation acid products of ethylene glycol (EG) aqueous solutions, namely glycolic, formic, acetic, and oxalic acids, on the corrosion behavior of low carbon steel was investigated under stirring conditions at 80  C by means of well-established techniques for electrochemical, physicochemical, and surface analyses. The electrochemical behavior of steel under polarization conditions is dominated by active dissolution of iron with Fe2+ production leading to oxide products and H+ reduction as the cathodic counterpart. No correlation was found between the corrosion current density estimated by Tafel extrapolation method and that determined by the rate of production of Fe2+ under free corrosion conditions. The latter experiments revealed that the nature and the relative proportion of carboxylic acids influence the corrosion behavior of steel. The rate of production of soluble corrosion products increases with the stability and complexation ability of the organic anion towards Fe3+, being more significant in the case of glycolic acid in excess with high chelation propensity. Conversely, formation of Fe2+ and Fe3+ oxalates on iron surface is promoted in the presence of oxalic acid due to a catalytic action on magnetite dissolution. The extent of above processes is compromised if hydrogen bonding interactions between different carboxylic acids are privileged. ã 2016 Elsevier Ltd. All rights reserved.

Keywords: steel corrosion ethylene glycol organic acids

1. Introduction Glycols are among the most versatile compounds [1,2]. Their uniqueness is due to properties such as low volatility, viscosity, and electrical conductivity, in addition to excellent hygroscopicity, complete miscibility with water, good heat transfer capacity, and low cost. Glycols are a class of organic compounds belonging to the family of alcohols called diols of general formula OH-Rn-OH, where the aliphatic carbon chain connects two  OH groups. They are colourless liquids with practically no odor, and also excellent solvents for many organic compounds. Glycols are widely used as anti-freeze and cooler, chemical intermediate, gas dehydration agent, heat transfer fluid, solvent, etc. The most important member of glycols family is the mono ethylene glycol HO-CH2-CH2-OH (EG) [2]. EG has been proposed as the best choice for heat transfer applications because of the lower viscosity and superior heat transfer efficiency with respect to other glycols such as propylene

* Corresponding author. E-mail address: [email protected] (M. Trueba).

glycol, which is considered when toxicity is a concern [2,3]. A minimum concentration of 25–30 wt% in water has been recommended for obtaining thermal performances comparable to that of water, bactericide and fungicide actions, and corrosion protection. The quality of water is an important issue, in particular the contents of Cl (corrosive), Ca2+ and Mg2+ (scale-forming) should be below 25 and 50 ppm, respectively, if tap water is used. The heat transfer performance of water-based EG fluids decreases with time because of absorption of moisture from the environment and, more importantly, due to EG thermal oxidation in the presence of oxygen leading to carboxylic acids as main degradation products [4,5]. The most common acids found in degraded real fluids are acetic (AA), formic (FA), glycolic (GA), and oxalic (OA). Table 1 reports the corresponding schematic structures, dissociation reactions and standard equilibrium dissociation constants (pKa) [6]. The former three acids are monoprotic carboxylic acids, being GA an hydroxy acid (a-hydroxyacetic acid) and AA the most basic according to the pKa values (Table 1). The acid strength of dicarboxylic OA, if completely dissociated, decreases and becomes comparable to that of FA and GA. Another important issue is the metallic material of the heat transfer system that may catalyze the hydrothermal oxidation of EG and in turn corrode faster as a result

http://dx.doi.org/10.1016/j.electacta.2016.03.144 0013-4686/ ã 2016 Elsevier Ltd. All rights reserved.

Please cite this article in press as: M. Santambrogio, et al., Effect of major degradation products of ethylene glycol aqueous solutions on steel corrosion, Electrochim. Acta (2016), http://dx.doi.org/10.1016/j.electacta.2016.03.144

G Model EA 26974 No. of Pages 12

2

M. Santambrogio et al. / Electrochimica Acta xxx (2015) xxx–xxx Table 1 Carboxylic acids structures, dissociation reactions, and standard equilibrium pKa values. Carboxylic acid

Dissociation reaction

pKa [5]

Acetic acid (AA)

Schematic structures

CH3COOH $ CH3COO + H+

4.76

Glycolic acid (GA)

OHCH2COOH $ OHCH2COO + H+

3.83

Formic acid (FA)

HCOOH $ HCOO + H+

3.75

Oxalic acid (OA)

H2C2O4 $ HC2O4 + H+ HC2O4 $ C2O42 + H+

1.25 3.81

of pH decrease, among other factors [[4,5,7–10], and ref. therein]. Major limitations of mild steel and cast iron are the low heat capacity and the high susceptibility to corrosion, in comparison to Cu and Al alloys. Nonetheless, iron-based materials are still the most used due to practical considerations. Buffering and inhibition are common practices for corrosion prevention [11]. Besides specific operative conditions (EG concentration, temperature, pressure, O2, additives), the nature of the metal in contact with EG aqueous solution contributes to the variety of possible reactions that lead to different organic sub-products. The mechanistic pathways of EG hydrothermal oxidative decomposition involve aldehydes such as glycoaldehyde, acetaldehyde and formaldehyde, in addition to FA, as intermediate compounds. Aldehydes being involved in low and high barrier paths of EG degradation have been theoretically predicted also [12]. These sub-products may undergo self- and cross-disproportionation reactions in hot water in absence of catalysts to form FA and methanol [13]. In addition, acid-catalyzed C C bond formation from formaldehyde and FA may leads to GA, being the reaction more selective with FA in excess. From studies of thermal degradation of EG in the presence of iron based materials [4 and ref. therein], the following general sequence has been proposed: formation of OA and AA, then conversion of OA into FA with prolonged heating, and finally dissociation of AA and FA to form carbon dioxide, which converts into carbonic acid upon reaction with water. Salts sub-products of ferrous acetate, formate and carbonate were considered also. Ion chromatography analysis of uninhibited EG solutions (75 wt%) after 90 days of operation at 95  C in crude oil preheating tubes indicated AA and FA as major sub-products [4]. Different results have been obtained from systematic studies using real fluids of solar collector systems [5] and laboratory-made solutions [7–10]. Ion chromatography and mass spectrometry analyses of degraded aqueous solutions of EG (50–70 wt%) after operation within common temperature range of service (75–130  C) and in the presence of dissolved O2 indicated GA as main constituent while OA and FA were detected in smaller amounts. The hydrothermal degradation of EG was importantly accelerated after placing metallic Cu in the reaction medium, differently from Al for which higher amounts of FA rather than of GA were detected. In addition, production of CO2 on the expense of

O2 consumption by reactions with free radical intermediates of EG decomposition was more important as temperature increased and in the presence of Cu [10]. Investigations on electrochemical corrosion of Al alloys have shown that O2 reduction controls metal dissolution in EG solutions at ambient temperature [14,15]. Higher susceptibility to corrosion has been reported in hot alkaline EG solutions to which GA was intentionally added, being associated to Al3+ complexation by glycolate [16]. A very recent investigation of EG electrooxidation on iron-group nanoalloy catalysts have demonstrated, both experimentally and theoretically, that GA is generated at the first stage of the EG oxidation to form OA [17]. Participation of organic intermediates during corrosion of steel is not limited to EG solutions but extends to aqueous CO2 solutions [18] and alcohols [19]. The electrochemical corrosion of steel in acid solutions is subject of intense research, in part due to its significance in oilfield applications [11]. Studies in aqueous solutions of EG have considered the effect of galvanic corrosion [20], acidity (HCl) [21], EG concentration [20,22], solution rotation speed [23,24] and presence of inhibitors [22–24]. Concerning the carboxylic acids degradation products, OA [25–27], AA [28–31] and FA [29,31–33] have attracted major interest, differently from GA [34]. In addition, mineral acids and/or supporting electrolytes, as well as Cl, have been intentionally added to the test solutions. Electrochemical methods such as potentiodynamic polarization and electrochemical impedance spectroscopy (EIS), in combination with weight loss measurements, have been mostly used. The polarization behavior usually manifests active corrosion. The possible relation between electrochemical decomposition of EG and electrochemical corrosion of steel, in particular under hydrodynamic conditions at high temperatures, has ever been suggested. In this paper, the corrosion of steel is investigated at 80  C in stirred solutions of EG at 30 wt% in tap water containing mixtures of AA, GA, FA and OA (Table 1) in trace amounts to simulate degraded EG solutions of real fluids. In addition to EIS and potentiodynamic polarization techniques, experiments under free corrosion conditions were carried out. These consisted in the measurement as a function of time of the free corrosion potential, solution pH and conductivity, and concentrations of Fe2+ and Fe3+. Possible alteration of the acid-based equilibrium with corrosion

Please cite this article in press as: M. Santambrogio, et al., Effect of major degradation products of ethylene glycol aqueous solutions on steel corrosion, Electrochim. Acta (2016), http://dx.doi.org/10.1016/j.electacta.2016.03.144

G Model EA 26974 No. of Pages 12

M. Santambrogio et al. / Electrochimica Acta xxx (2015) xxx–xxx

was evaluated by potentiometric titration. The chemical composition of selected corroded surfaces was estimated by X-ray photoemission spectroscopy (XPS). 2. Experimental The material used was 0.5 mm thick sheet of commercial low carbon steel (Marcora Spa, Italy), whose chemical composition was determined by optical emission spectrometry (Quantron Magellan Q8, Bruker) and is reported in Table 2. Unless otherwise indicated, all reagents were of high purity grade (> 99%) and supplied by Sigma-Aldrich, Fluka and Merk. Ethylene glycol (EG) was provided by Cillichemie Spa (Italy). The purity level of EG (> 99.9%) was verified by HPLC analysis and Karl Fischer titration. All solutions were prepared using water of quality Milli-Q, with the exception of EG aqueous solutions. These were made at 30 wt % using synthetic tap water (EG 30-TW) following the standard recipe reported in Table 3 [35]. The carboxylic acids were added in different proportions for a total concentration of 0.058 mol L1, using 1 mol L1 aqueous stock solutions of a given carboxylic acid (Table 1), namely GA, AA, OA (97%, Fluka) and FA (96%, Sigma-Aldrich). Table 4 reports the designation and composition of EG 30-TW solutions containing the acid mixtures, in addition to solutions pH and conductivity (s) at 80  C. The organic acids proportions and acid mixture total concentration were chosen by considering the major constituents detected in uninhibited EG real fluids after service [4,5,7], as well as preliminary results (unpublished) of HPLC analysis of EG at 30 wt% in water after accelerated degradation experiments (autoclave at 200  C, 5 bar of compressed air, 24 hours) that indicated no OA but GA in excess with respect to AA and FA. Electrochemical measurements were performed at 80  1  C under continuous stirring (300 rpm) using a double-walled cylindrical cell containing 600 mL of the test solution that was placed over a magnetic stirrer (VELP Scientific) and connected to a JULABO thermostatic water bath. Fig. 1 shows the schematic representation of the electrochemical cell setup. A rubber cap was used as cover and for fixing the electrodes. EIS and potentiodynamic polarization experiments were carried out using a three-electrode configuration, namely Lugging probe with a saturated calomel reference electrode (SCE), Pt spiral as counter electrode, and the steel sheet with 1 cm2 of active surface area as working electrode (Fig. 1). The opening of the Luggin probe that accommodates the SCE was well above the rubber cap suchas to keep the reference electrode nearly at room temperature(25–30  C) in order to guarantee (electro)chemical stability between the experiments [36]. The working electrode was prepared before use as follows. The steel specimen (30  30 mm) was wet ground with abrasive silicon carbide paper up to 800 grit using a lubricant as wetting agent and then cleaned in an ultrasonic bath for 15 min with acetone. Thereafter, a rectangular piece of the same steel sheet was arc-joined to the upper side of the specimen for external electrical connection. Finally, the active surface area of 1 cm2 was delimited with the aid of Kapton adhesive tape. The electrodes were connected to a computer driven Solartron 1286/1250 potentiostat/frequency generator. For a given test solution (Table 4), the open circuit potential (Eoc) was monitored during 30 min, thereafter the impedance spectrum was registered between 65 kHz and 102 Hz with an AC perturbation amplitude of  15 mV, and finally anodic (or cathodic) polarization curve was Table 2 Chemical composition (wt %) of the steel substrate. C

Mn

Si

P

S

Cr

Ni

Mo

Cu

<0.01

0.45

0.80

0.10

0.01

0.03

0.02

<0.01

0.01

3

Table 3 Recipe of synthetic tap water (TW). Chemical

Stock (mg L1) Stock Added (ml) Final concentration (mg L1)

NaHCO3 MgSO47H2O K2HPO4 KH2PO4 (NH4)2SO4 NaCl FeSO47H2O NaNO3 CaSO4 Humic acid Fulvic acid

10 1 1 1 100 100 10 1 1 1 1

10.0 13.4 0.7 0.3 0.1 0.1 0.1 1.0 27.0 1.0 1.0

100 13.4 0.70 0.30 0.01 0.01 0.001 1.00 27.0 1.00 1.00

recorded at a scan rate of 0.5 mVs1. Each set of experiments was repeated twice using freshly prepared specimens. The same experimental setup described above was used for monitoring as a function of time (2 hours) the free corrosion potential (Ecorr), solution pH and s, and iron ions concentrations ([Fe2+], [Fe3+]). A pH glass electrode and a conductimetric cell were inserted through the rubber cap into the test solution, while Pt counter electrode was removed. A potentiometer/pHmeter and a conductimeter (AMEL instruments) were used for measuring Ecorr, pH and s every 20 min after 5 min of conditioning in the test solution (t = 0). After each measurement, volumes of 5 mL were accurately collected in two 25 mL volumetric flasks and the total test volume (600 mL) was replenished by adding 10 mL of fresh test solution pre-heated at 80  C. The collected samples were used for the analysis of [Fe2+] and [Fe3+], following common spectrophotometric methods for the determination of trace amounts of iron ions in water [37]. In particular, 1 mL of 1,10-phenanthroline/sodium acetate reagent and thereafter water up to 25 mL were added to one group of samples for the analysis of Fe2+. The reagent was prepared in a 200 mL volumetric flask by dissolving 1 g of 1,10 phenanthroline (o-phen), 2.3 g of sodium acetate and 6 mL of acetic acid (100%) in 100 mL of methanol. Thereafter, water was added up to 200 mL. The reagent solution was saturated with argon before use to avoid oxidation of Fe2+ by dissolved oxygen. Complexation of Fe2+ with o-phen produces a red-orange complex, as represented by the following equation: Fe2+ + 3 o-phen ! Fe(o-phen)32+

(1)

The second group of collected samples was used for the determination of total Fe3+ ([Fe3+]tot). To each volumetric flask the following reagents were added in the order: 6 mL of 1.25 mol L1 H2SO4 (95-97%, Fluka), 3 mL of 0.05 mol L1 K2S2O8, 6 mL of 200 g L1 NH4SCN (> 97%, Sigma-Aldrich), and finally water up to 25 mL. Persulfate ions oxidize Fe2+ to Fe3+, which reacts with thiocyanate ions to form a red-brown complex: S2O82 + 2 Fe2+ + 2 H+ ! 2HSO4 + 2 Fe3+

(2)

Fe3+ + SCN fi [FeSCN]2+

(3) 2+

The absorbance of sample solutions containing Fe and Fe3+ complexes was measured at 510 nm and at 474 nm, respectively, using a Beckman Coulter DU800 spectrophotometer and glass cuvettes. For construction of calibration curves (Supporting Information, Fig. S1), solutions with known concentrations of Fe2+ and Fe3+ (between 0.5–6 mg L1 and 5–250 mg L1, respectively) were prepared as indicated above, using stock solutions of 50 mg L1 of (NH4)2Fe(SO4)2 and 5 g L1 Fe(NO3)3. The later stock solution was standardized by titration with HCl/KI. The effective [Fe3+] was calculated by subtracting [Fe2+] from [Fe3+]tot.

Please cite this article in press as: M. Santambrogio, et al., Effect of major degradation products of ethylene glycol aqueous solutions on steel corrosion, Electrochim. Acta (2016), http://dx.doi.org/10.1016/j.electacta.2016.03.144

G Model EA 26974 No. of Pages 12

4

M. Santambrogio et al. / Electrochimica Acta xxx (2015) xxx–xxx

Table 4 Designation, composition, pH, and conductivity, of as-prepared solutions of GE 30 wt% in synthetic tap water containing mixtures of different carboxylic acids (total concentration 0.058 mol L1). Designation of test solution

EG 30-TW G10A3F1 G1A3O10 G2A10F2 G1A3F10 G6A2F6 G4A4F6 G1F8O5 *

Carboxylic acid concentration/mol L1 Glycolic (G)

Acetic (A)

Formic (F)

Oxalic (O)

– 0.041 0.004 0.008 0.004 0.025 0.016 0.004

– 0.013 0.013 0.041 0.013 0.008 0.016 –

– 0.004 – 0.008 0.041 0.025 0.025 0.021

– – 0.041 – – – – 0.033

Acid(s) in excess

*

– G O A F GffiF GffiAffiF FffiO

6.5 3.1 1.7 3.5 3.2 3.0 3.1 2.1

pH

*

s/mS cm1

0.3 0.3 1.1 0.2 0.3 0.3 0.3 0.6

at 80  C.

attenuation length [39] and a standard set of VG Escalab sensitivity factors. The uncertainty on the atomic quantitative analysis is about  10%. All potentials in the text are referred to SCE (+ 0.244 V vs SHE at room temperature) with no correction due to thermal diffusion effects. The potential of SCE at 80  C should be smaller by about 40 mV (ESCE  + 0.205 V at 80  C) [36]). However, such effects should not influence potential measurements between experiments carried out under identically controlled conditions and using test solutions containing weak acids. Unless otherwise indicated, graphical and quantitative analyses of the experimental data were carried out using OriginPro 9.1 program (OriginLab, Northampton, MA). 3. Results and discussion 3.1. Electrochemical behavior under AC perturbation and DC polarization

Fig. 1. Schematic representation of the double-walled one-compartment cylindrical three-electrode cell configuration. WE  working steel electrode with 1 cm2 of active surface area, CE  Pt spiral counterelectrode; RE  Luggin probe with saturated calomel reference electrode. Magnetic stirring and water inlet and outlet for heating the test solution at 80  C are indicated.

The buffer capacity of the test solutions was determined at room temperature (23  C) before and after the exposure test (2 hours) by potentiometric titration, using 0.01 mol L1 KOH as titrant and the glass electrode connected to the Amel pH meter. The surface chemical composition of selected samples was investigated by XPS in an ultrahigh vacuum (UHV) chamber with a base pressure lower than 1 106 Pa during data collection. Photoemission spectra were collected by a VG Microtech ESCA 3000 Multilab spectrometer, equipped with standard Al Ka excitation source (hn = 1486.6 eV) and multi-channeltron detection system. The hemispherical analyser operated in the CAE mode at constant pass energy of 20 eV. The binding energy (BE) scale was calibrated by measuring C 1s peak (BE = 285.1 eV) from the adventitious carbon and the accuracy of the measure was  0.1 eV. Photoemission data were collected and processed by using the VGX900 software. Data analysis was performed by a nonlinear least square curve-fitting procedure using a properly weighted sum of Lorentzian and Gaussian component curves, after background subtraction according to Sherwood [38]. Surface relative atomic concentrations were calculated by a standard quantification routine including Wagner’s energy dependence of

Fig. 2 shows the average curves of steel electrode potential variation with time that were calculated from experimental open circuit potential (Eoc)  time curves registered during 30 min in the given test solution at 80  C under continuous stirring. In the case of EG 30-TW solution, the potential decreases with time and tends to stabilize at  718  20 mV, indicating some inhibition of iron dissolution due to adsorption of glycol molecules on the metal surface. With the addition of carboxylic acid mixtures to EG 30-TW, all systems but those containing OA reach a stationary state at less negative potentials almost instantaneously. The electrode potential increases and tends to stabilize at less negative values in the

Fig. 2. Average open circuit potential (Eoc)  time curves of steel in stirred EG 30 wt % in TW solutions with and with no addition of carboxylic acids mixtures at 80  C.

Please cite this article in press as: M. Santambrogio, et al., Effect of major degradation products of ethylene glycol aqueous solutions on steel corrosion, Electrochim. Acta (2016), http://dx.doi.org/10.1016/j.electacta.2016.03.144

G Model EA 26974 No. of Pages 12

M. Santambrogio et al. / Electrochimica Acta xxx (2015) xxx–xxx

case of OA in excess (G1A3O10). Similar trend is indicated for the test solution containing smaller concentration of OA (G1O5F8). The predominant reactions at the metal/solution interface should correspond to iron dissolution and hydrogen evolution [40]: Fe ! Fe2+ + 2 e

(4)

2 H+ + 2 e ! H2

(5)

with spontaneous oxidation of Fe2+ to Fe3+ by dissolved oxygen according to the reactions: FeOH+ ! FeOH2+ + e

(6)

Fe(OH)2 ! Fe(OH)2+ + e

(7)

Although the concentration of dissolved oxygen decreases with temperature, continuous supply of this additional cathodic reactant is promoted with stirring. At pH 5, hydrolyzed ferrous iron species like Fe(OH)+ and Fe(OH)2 are more readily oxidized by dissolved oxygen, though their contribution is expected to be much smaller than that of Fe2+ [41]. During measurements, a decrease of free metal area with time due to formation of black corrosion products was noticed in all cases, being the extent of this process less important in the case of EG-30TW and more evident for OA in excess. Black corrosion products indicate prevalent formation of iron oxides, more likely magnetite Fe3O4. This mixed (Fe2+, Fe3+) oxide is a typical component of sludge, deposits and thick oxide build up on carbon steel components of heat exchangers [42]. Thus, under present conditions, iron oxidation proceeds via oxolation rather than hydroxylation reactions [43], that is condensation leads prevalently to oxo bridges, favored also with stirring and possible buffer effect of carboxylic acids that limit local pH increase. Accordingly, the prevalent reactions that lead to formation of magnetite can be represented as follows: FeOH+ + 1/2 O2 + H2O ! 2 FeOH2+ + 2 OH

(8)

FeOH+ + 2 FeOH2+ ! Fe3O4 + 4 H2O + 3 H+

(9)

where the second step corresponds to dehydrationn and deprotonation of hydroxy species. These reactions reasonable explain the formation of oxides products in the case of EG-30TW also. Concerning the effect of the acid strength of the investigated carboxylic acids (Table 1), some correlation is indicated between this property and the shift of Eoc  t traces to less negative potentials (Fig. 2), being in accordance with the decrease of the initial pH of the corresponding test solutions (Table 4). In particular, for the solutions containing a pertinent carboxylic acid in excess, mean Eoc values (at about 30 min) increase as follows:  723  5 mV (G2A10F2) < - 677  2 mV (G10A3F1) < - 667  2 mV (G1A3F10) < - 620  6 mV (G1A3O10). In the case of acid mixtures with similar proportion of GA and FA (G4A4F6 and G6A2F6), the average Eoc  t traces lie between those of the corresponding carboxylic acid in excess (G10A3F1, G1A3F10). However, Eoc decreases towards the E  t trace of G10A3F1 at the beginning of the exposure test (Fig. 2), being more evident for G6A2F6. Similarly, a less steep increase of Eoc with time is obtained for the acid mixture with OA  FA (G1O5F8) in comparison to G1A3O10, despite the similar initial pH (Table 4). The high propensity of FA to form intermolecular hydrogen bonds [44] could explain the differences above. Such interactions are expected to be more favored in the case of OA as this acid presents two COOH groups (Table 1), while less privileged with GA since disruption of

5

intramolecular hydrogen bonds of type O H  O = is more difficult. In addition, the more important shift of Eoc  t trace as the concentration of OA is raised suggests participation of oxalate ions in the corrosion film growth, probably due to chemisorption [45]. The results point out that the nature and the relative proportion of the carboxylic acids in the acid mixtures influence the free corrosion behavior of steel. The impedance spectra collected after 30 min of exposure were featureless for most conditions but OA in excess. Qualitative examination of Bode plots (Fig. S2) indicated a decrease of the impedance modulus |Z| in the limit of high frequencies ( 65 kHz) as the initial pH of the test solution decreases (Table 4), thus revealing smaller electrolyte resistance for OA containing solutions. However, the impedance response was distinctly different for OA in excess (G1A3O10) as compared to the other acid solutions. That is, |Z| increased more importantly in the limit of low frequencies (up to about 1800 Vcm2 for f < 101 Hz). Similarly, a well-defined peak in the phase F between 103–101 Hz was recorded. The other test conditions showed almost flat and closely overlapped |Z|  f traces that shifted up to 1000 Vcm2 at f  104 Hz, depending on the test solution composition. Similarly, almost overlaying f - f plots at f  0 were obtained. The differences above indicate that mixed charge transfer and mass transfer control of steel corrosion becomes significant as the concentration of OA increases in EG solutions containing acid mixtures. Thus, in the presence of OA in excess, surface confined reactions promoting corrosion film growth support possible formation of iron oxalate products, as indicated the variation of Eoc with time (Fig. 2). The electrochemical response of steel to subsequent polarization was very similar for all the investigated solutions, as shown in Fig. 3. As obtained by others [22,23,28,30], the current density rapidly increase towards limiting values, regardless the direction of the potential scan. In addition, the quite symmetric anodic and cathodic curves nearly overlap, regardless the composition of the test solution. Some differences are indicated for excesses of OA (G1A3O10) and AA (G2A10F2), as well as for tests solution containing GA  FA (G6A2F6). In the case of the former, the potential at the beginning of the polarization scan (anodic and cathodic) is shifted to higher values, in accordance with the E  t responses (Fig. 2). For G2A10F2 and G6A2F6, the current density limits of anodic and cathodic branches are shifted to higher and smaller values, respectively. Nonetheless, anodic and cathodic Tafel slopes (ba, bc), estimated with the help of the Tafel analysis tool of CView software (version 3.4a, Scribner Associates Inc.), were similar for all the investigated systems (ba  |bc|  110  15

Fig. 3. Representative anodic and cathodic polarization curves of steel in stirred EG 30 wt% in TW solutions with and with no addition of carboxylic acids mixtures at 80  C.

Please cite this article in press as: M. Santambrogio, et al., Effect of major degradation products of ethylene glycol aqueous solutions on steel corrosion, Electrochim. Acta (2016), http://dx.doi.org/10.1016/j.electacta.2016.03.144

G Model EA 26974 No. of Pages 12

6

M. Santambrogio et al. / Electrochimica Acta xxx (2015) xxx–xxx

mV dec1), in agreement with reported data for iron in stirred acid solutions [21]. Although the limiting current behavior may result from convective diffusion forced with stirring, visual inspection of the corroded surfaces after the test confirmed the presence of black corrosion products, being more uniform in the case of OA in excess. Accordingly, with anodic polarization, active dissolution of steel on oxide-free metal surface leads prevalently to magnetite (9) and possibly to higher valence oxides such as Fe2O3 at less negative potentials [40]. Cathodic polarization promotes reduction of H+ on the free metal surface, though Fe2+ species may form due to magnetite reduction (oxide covered surface): Fe3O4 + 2e + 8H+ ! Fe2+ + 4H2O

(9)

Fig. 4 reports the plots of the corrosion potential Ecorr, corrosion current density jcorr, and of limiting anodic and cathodic current densities (jLa and |jLc|, respectively), as a function of the initial pH of the corresponding test solution at 80  C (Table 4). The former electrochemical properties were estimated by Tafel extrapolation method as indicated above, whereas the limiting current densities correspond to the potential values of + 0.4 and  1.6 V, respectively, for the pertinent polarization curve (Fig. 3). From the analysis of Ecorr (Fig. 4a), only for solutions containing OA, namely G1O5F8 and G1A3O10, the potential deviates importantly from the corresponding Eoc by about + 50 mV (Fig. 2), which implies that electrons removal due to metal oxidation proceeded with time during the EIS experiment, though highlights some catalytic effect of OA on the corrosion film growth. Nonetheless, the increase of Ecorr with the acid strength of test solutions is reproduced (Fig. 2). Conversely, the current densities show no correlation with the initial pH of the test solution (Fig. 5b,c). For a given test condition but OA in excess, jcorr, jLa and |jLc| either increase or decrease, indicating fast charge transfer reactions leading to a depletion of reacting species on the surface. The opposite variation between jcorr and jL in the case of G1A3O10 suggests more limited mass transport due to the thickening of the corrosion film. Considering the initial conductivity of the test solutions at 80  C (Table 4), some correlation between s and jcorr is indicated for G2A10F2 and EG 30TW, being both properties smaller in the case of the former (AA in excess) and higher in the case of EG-30TW. These differences could be associated to the less favored dissociation of AA at low concentrations and to the possible ionic nature of EG due to disruption of intramolecular hydrogen bonds by water [46]. Note that tap water was used for preparing all test solutions (Table 3). Additional observation regards the much higher jLa value in the case of EG 30-TW (Fig. 5b,c), being comparable to those obtained

for most acid mixtures. To be noticed also is the increase of jcorr, jLa and |jLc| as both GA and FA concentrations are raised. These results bear to suspect that electrooxidation of EG is concomitantly promoted with anodic polarization with possible formation of aldehydes, FA and GA [12,13,17]. Considering that chelation of FeOH2+ by GA is highly favored [47], complexation would acts as an activating reaction of iron dissolution as both FA and GA concentrations are raised. The smaller aggressiveness of the EG solution containing AA in excess may result from the lower acid strength and the higher stability of the corresponding carboxylate anion in comparison to formate [44]. Overall, the electrochemical kinetics of steel corrosion in EG solutions containing carboxylic acid mixtures does not solely depend on the acid strength. Specific interactions between the given organic anion and the metal surface seems to play an important role. Possible EG decomposition by electrooxidation deserves a more systematic investigation for a better understanding of metal corrosion in heat transfer fluids. 3.2. Effect of carboxylic acid nature on freely corroding steel Measurements of Ecorr, solution pH and s, as well as of Fe2+ and Fe concentrations, were carried out during two hours after 5 min of conditioning in the test solution (t = 0), using the same experimental setup. Major attention was given to solutions containing the carboxylic acids in excess, and those with OA  FA and GA  FA (Table 4). Fig. 5 reports the variation as a function of time of Ecorr, pH and s. Although Ecorr values (Fig. 5a) are higher than Eoc by about 10–20 mV (within 30 min) (Fig. 2), indicating that a systematic error has occurred during potential measurement due to instrumental precision (see section Experimental), the trend of Ecorr  t traces with the initial pH of test solution is reproduced (Fig. 2), also for longer times. In addition, the shift of Ecorr to higher values with time in the presence of OA confirms that corrosion film growth is more important in relation to the other test conditions. The pH and s of solutions tend to increase linearly with time (Fig. 5b,c), though not very significantly due to the large volume of test solution employed and possible buffer action of a given carboxylic acid as either H+ or organic anion are consumed. In summary, the shift of Ecorr  t and s  t traces to higher values, and that of pH  t to smaller values, correlate with the increase of the acid strength (Table 4). A different result is obtained from the analysis of Fe2+ and Fe3+ concentrations. For all test conditions, [Fe3+] is much higher than [Fe2+] during all the experiment time (Fig. 6), indicating a rapid 3+

Fig. 4. Relationships between electrochemical properties (a) Ecorr, (b) jcorr, and (c) jLa, |jLc|, as a function of the initial pH (at 80  C) of the test solutions. Labels indicate the corresponding EG solution (see Table 4 for clarity).

Please cite this article in press as: M. Santambrogio, et al., Effect of major degradation products of ethylene glycol aqueous solutions on steel corrosion, Electrochim. Acta (2016), http://dx.doi.org/10.1016/j.electacta.2016.03.144

G Model EA 26974 No. of Pages 12

M. Santambrogio et al. / Electrochimica Acta xxx (2015) xxx–xxx

7

Fig. 5. Variation as a function of time of (a) Ecorr, (b) pH, and (c) s, during exposure of steel substrates in stirred EG 30% TW solutions at 80  C containing different carboxylic acids mixtures: (&) G2A10F2, (*) G1A3O10, (~) G1A3F10, (!) G10A3F1, ( ) G6A2F6, (^) G1O5F8 (see Table 4 for clarity).

Fig. 6. Variation as a function of time of (a) [Fe2+] and (b) [Fe3+] during exposure of steel substrates in stirred EG 30% TW solutions at 80  C containing different carboxylic acids mixtures: (&) G2A10F2, (*) G1A3O10, (~) G1A3F10, (!) G10A3F1, ( ) G6A2F6, (^) G1O5F8.

conversion of Fe2+ into Fe3+ due to the action of dissolved oxygen. In addition, both [Fe2+] and [Fe3+] prevalently increase linearly with time, with the exception of OA in excess (G1A3O10). In this case, no Fe2+ were detected by the analytical procedure employed, which reveals that the concentration of this specie is below the instrumental detection limit (0.05 mg L1). Similarly, [Fe3+] is smallest among all the investigated systems. Accordingly, transport controlled reactions are determined by the limited release of iron ions as the corrosion film grow. More importantly, no correlation between production of iron soluble species and acid

strength is found, which points out that iron dissolution is influenced by the nature of the organic anion. From the slope of [Fe2+]  t plots (Fig. 7a) calculated by linear regression (R2 > 0.99), the estimated dissolution rates d [Fe2+]/dt (in g cm2 s1) are 4.6  107 for G10A3F1, 3.5  107 for G2A10F2 ( G1O5F8), 2.9  107 for G1A3F10, and 4.5  108 for G6A2F6. This trend is in contradiction with that derived from the calculated jcorr values by Tafel extrapolation method (Fig. 6b). A strong disagreement was indicated also from jcorr values calculated from the dissolution rate d[Fe2+]/dt using Faraday’s law [48], as these were much higher (mA cm2 range)

Fig. 7. Variation as a function of time of instantaneous rates of (a) iron dissolution ([Fe2+]/dt) and (b) Fe2+ oxidation (d[Fe3+]/dt), during exposure of steel substrates in stirred EG 30% TW solutions at 80  C containing different carboxylic acids mixtures: (&) G2A10F2, (*) G1A3O10, (~) G1A3F10, (!) G10A3F1, ( ) G6A2F6, (^) G1O5F8.

Please cite this article in press as: M. Santambrogio, et al., Effect of major degradation products of ethylene glycol aqueous solutions on steel corrosion, Electrochim. Acta (2016), http://dx.doi.org/10.1016/j.electacta.2016.03.144

G Model EA 26974 No. of Pages 12

8

M. Santambrogio et al. / Electrochimica Acta xxx (2015) xxx–xxx

Fig. 8. Potentiometric titration curves of EG solutions containing excesses of AA, OA, FA and GA, as well as with OA  FA and GA  FA, before (open symbols) and after (closed symbols) the exposure experiment.

though smaller than iLa. Besides the Tafel method for estimation of jcorr, as well as of participation of dissolved oxygen as an additional cathodic reactant, fast electrochemical reactions promoted by changes in the electrode potential in relation to steady state freely corroding conditions more likely determine the above mentioned disagreements. That is, the systems evolve differently if electrochemical corrosion is accelerated with polarization. Nonetheless, present experiments show no correlation between the variation of Ecorr, pH and s, and that of the dissolution rate also. Ecorr values are quite similar for GA, AA and FA in excess, as well as for the acid mixture containing GA  FA, but d[Fe2+]/dt consistently differ (Figs. 5 and 6a). Similarly, the pH is higher and s is smaller for EG solution containing AA in excess in comparison to that with FA in excess (Fig. 5b,c), but d[Fe2+]/dt is higher in the case of the former. It is to be noticed also that, differently from [Fe2+], the variation of [Fe3+] with time resembles closely for most conditions (Fig. 6b). In order to explore further the effect of the organic anion on iron dissolution and, in particular, on the conversion of Fe2+ into Fe3+, the instantaneous dissolution rates (d[Fe2+]/dt and d[Fe3+]/dt) were calculated by derivation of the corresponding concentration  time curves (Fig. 5) and plotted as a function of time. As shown in Fig. 7a, the instantaneous production of Fe2+ changes little with the exposure time in the case of solutions containing excesses of GA and AA, as well as for the acid mixture with GA  FA. Conversely, d [Fe3+]/dt initially decreases and subsequently increases (Fig. 7b), being this trend more evident for GA in excess with higher d[Fe2 + ]/dt values. An opposite variation is obtained for FA in excess (Fig. 7). Both d[Fe2+]/dt and d[Fe3+]/dt initially increase and then tend to decrease with time. In the case of OA in excess, fast consumption of Fe2+ to form Fe3+ during the first minutes of

exposure is likely. However, the instantaneous release of Fe3+ progressively decreases towards stationary values. Similarly, for the acid mixture with OA  FA, both d[Fe2+]/dt and d[Fe3+]/dt decrease towards values similar to that of OA in excess by the end of the exposure time. The above results can be reasonably explained by considering the nature of the organic anions (Table 1). Acetate is more stable than formate (CH3 group push more negative charge towards the already negative COO- end). Similarly, the former anion is less prone to formation of dimeric pairs due to solvation. Thus, Fe3+  acetate complexes are more favored, though to a lesser extent than in the case of GA because of the higher propensity of glycolic acid (a-hydroxy acid) to chelate surface FeOH2+ [47]. This consideration is supported by the analysis of the buffer capacity of the test solutions before and after the exposure test. The titration curves closely overlap for all test conditions but GA in excess (Fig. 8). In this case, the curve shifts to higher volumes of titrant (0.01 mol L1 KOH) since chelation reaction produces free H+ [47] and limits reestablishment of GA dissociation equilibrium. In the presence of OA, formation of oxalate-rich corrosion film involving both Fe2+ and Fe3+ assisted by complexation and chemisorption of hydrogen oxalate species (pH < 2) (Table 1) is more likely. For mixed solutions containing GA  FA and OA  FA, chelation and chemisorption reactions are less favored because intermolecular hydrogen bonding with FA provides higher energetic stability than that involving water molecules. Based on the considerations above, present findings show that the stability and complexation affinity of the organic anion, importantly influence the dissolution kinetics of iron under free corrosion conditions in the presence of small amounts of carboxylic acids in EG aqueous solutions.

Table 5 XPS analysis of steel surfaces exposed in EG solutions containing excesses of GA and OA. Sample

GA

Signal

Binding energy (eV)

Intensity (%)

Binding energy (eV)

Intensity (%)

C1s

285.1 286.7 288.5

71.9 13.2 14.9

285.1 286.7

65.3 11.0

289.0

23.7

CC, CH, CH2 CO C¼O O¼CO

OA

Chemical assignment

O1s

529.7 531.4 532.3

36.4 37.0 26.6

529.7 531.4 532.3

18.5 50.6 30.9

O2 (FeO, Fe2O3) OH (FeOOH) O¼CO

Fe 2p3/2

706.6 709.8 711.1 713.4

7.0 49.9 33.8 9.3

709.8 711.1 713.4

63.4 19.5 17.1

Fe metal FeO, Fe2O3 Fe3O4 (FeOOH) FeO, Fe2O3

Please cite this article in press as: M. Santambrogio, et al., Effect of major degradation products of ethylene glycol aqueous solutions on steel corrosion, Electrochim. Acta (2016), http://dx.doi.org/10.1016/j.electacta.2016.03.144

G Model EA 26974 No. of Pages 12

M. Santambrogio et al. / Electrochimica Acta xxx (2015) xxx–xxx

9

Fig. 9. Curve-fitting of C 1s, O 1s and Fe 2p3/2 photoelectron peaks for steel surfaces after exposure in EG solutions containing excesses of (a) GA and (b) OA.

3.3. Surface chemical composition The chemical composition of the surfaces exposed in solutions with GA and OA in excess was analyzed by XPS. Survey spectra showed C 1s, O 1s and Fe 2p as main peaks, with comparable elemental composition (on average 45, 40 and 15 atomic %, respectively) for both surfaces. Results of the curve fitting of each photoelectron signal for a given carboxylic acid are reported in Table 5 and Fig. 9 [49]. Deconvolution of C1s line leads to three major components that can be attributed to carbon skeleton (CC, CH, CH2), C O and C¼O/O¼C O bonds [49]. The O 1s photoelectron peak can be deconvoluted by three different components: the first located at a binding energy (BE) of 529.7 eV is assigned to O2 ions of iron oxides on the surface

(FeO, Fe2O3), that at BE = 531.4 eV is attributed to hydroxyl groups (OH) of FeOOH species, and the component located at the highest binding energy BE = 532.3 eV is assigned to surface oxygen species of carboxyl groups C¼O/OC¼O [49]. The analysis of Fe 2p3/2 line revealed the presence of Fe2+ and Fe3+ ions. The component located at BE = 709.8 eV is assigned to FeO and Fe2O3, and the component at BE = 711.1 eV can be attributed to FeO(OH) and to magnetite Fe3O4 (Fe2+,Fe3+). Metallic iron (BE = 706.6 eV) is detected only for the surface exposed to GA in excess (Fig. 9a). Although the contribution of this component to Fe 2p3/2 line is quite low (Table 5), this result agree with the less uniform corrosion film formed on the surface and with the higher release of soluble corrosion products (Fig. 6), showing correlation with the E  t trace positioned at more negative potentials with respect to G1A3O10 (Fig. 5a). The highest

Please cite this article in press as: M. Santambrogio, et al., Effect of major degradation products of ethylene glycol aqueous solutions on steel corrosion, Electrochim. Acta (2016), http://dx.doi.org/10.1016/j.electacta.2016.03.144

G Model EA 26974 No. of Pages 12

10

M. Santambrogio et al. / Electrochimica Acta xxx (2015) xxx–xxx

Fig. 10. RAIR spectrum of iron surface after 2 hours of exposure in EG solution containing OA in excess.

binding energy peak at BE = 713.4 eV (Fig. 9) originates from the peak asymmetry observed in the Fe oxide band envelope (FeO/ Fe2O3) [50]. Comparison of the relative concentration of surface species (Table 5) indicates higher amount of FeOOH in the case of OA. Similarly, the intensity ratio of Fe 2p3/2 peaks at 709.8 and 711.1 eV is higher, pointing out smaller surface concentration of Fe3O4. In addition, the corresponding C 1s peak associated to O¼CO group is shifted to higher binding energies, suggesting stronger adsorption of oxalate species on the surface due to chemisorbed oxalate corrosion products. To verify this indication, RAIR analysis at near grazing incidence angle (75 ) was carried out to evaluate the bulk composition of the corrosion film, considering also that FeC2O4 and Fe2(C2O4)3 present well-different IR bands between 1750–1200 cm1. From XPS analysis with a sampling depth of few nm it is difficult to distinguish Fe2+ from Fe3+ in the de-convoluted Fe 2p3/2 peak. According to reported IR studies of iron oxalates [51–53], the most straightforward difference between FeC2O4 and Fe2(C2O4)3 vibrations is given by peaks at 1730 cm1 (terminal C¼O groups) and 1265 cm1 (C-O and C-C stretching) that are absent in the case of FeC2O4. As shown in Fig. 10, these characteristics are well discerned. A characteristic peak of FeC2O4 is detected at 1625 cm1. Additional vibrations at 1230/1200 cm1 and 1380/1335 cm1 are also ascribable to Fe2(C2O4)3 and FeC2O4, though are detected at smaller and higher cm1, respectively, due to possible effects of water motions (1830, 1100 and below 700 cm1) [52]. The band at about 1230 cm1 suggests outer-sphere oxalate  Fe3+ complexes. The presence of both FeC2O4 and Fe2(C2O4)3 in the corrosion film is indicated further by the OH/H2O stretching vibrations in the high frequency region (Fig. 10, inset), where the peak at 3330 cm1 is assigned to FeC2O4 and the shoulder at 3450 cm1 to Fe2(C2O4)3 [53]. The shift of the latter vibration to higher wave numbers by as much as 200 cm1 has been reported (see also [51]). Nonetheless, the overlapping peaks between 1750 and 1650 cm1 indicates several complex structures involving adsorption of oxalate on FeOOH. This is supported by the very intense bands between 1100  900 cm1 associated to OH out-of-plane bending of iron hydroxycompounds, in particular the peak at 990 cm1 correspond to OH deformation of FeOOH. The importance of this characteristic suggests contribution of stabilizing hydrogen bonding interactions with participation of mononuclear complex forms between a

single oxygen moiety and the metal atom [Fe  OOC-COO]. To notice also is the wide relatively weak band below 650 cm1 typical of Fe-O stretching vibration of Fe3O4 [55], whereas the absence of peaks between 1600–1500 cm1 exclude absorption on Fe2O3 (pH < 3) [54]. Based on surface analysis and dissolution kinetics results, different reaction pathways are involved in the corrosion mechanism of iron in the presence of excesses of OA and GA. In the case of the former, the autoreductive dissolution of magnetite represented by the following general equation [45]: Fe3O4 + 8HC2O4 ! 2Fe(C2O4)33 + Fe(C2O4)2 + 4H2O

(10)

justify the negligible concentration of Fe2+ and Fe3+ in solution, as compared to other test conditions, and the presence of mixed oxalate complexes in the corrosion film. As soon as H+ are discharged on the metal, the resulting atomic and molecular hydrogen blocks the iron surface protecting it against re-oxidation and promotes the reductive mechanism with concomitant chemisorption of oxalate on the oxide covered surface. The reductive mechanism involves interfacial electron transfer between ferrous oxalate complexes and surface ferric ions, the onset of this process being determined by the reduction of Fe3+ by oxalate. Thus, the extent of the reaction (10) depends on both Fe3+ and oxalic acid concentrations. In the case of GA, such blocking is limited due to the lower acid strength (Table 1), but more importantly due to the high chelation affinity of glycolate for FeOH2+ surface sites with formation of soluble Fe3+ complexes [46]: FeOH2+ + HOCH2COOH $ [(OCH2COO)FeOH] + 2H+

(11)

2+

Thus, desorption of FeOH by GA activates iron dissolution while limits hydrolysis reactions. 4. Conclusions The corrosion behavior of low carbon steel in stirred ethylene glycol (EG) aqueous solutions containing mixtures of different carboxylic acids was investigated under accelerated and free corrosion conditions at 80  C. The electrochemical behavior under polarization conditions manifests active dissolution leading to formation of oxide sub-products and H+ reduction as the cathodic

Please cite this article in press as: M. Santambrogio, et al., Effect of major degradation products of ethylene glycol aqueous solutions on steel corrosion, Electrochim. Acta (2016), http://dx.doi.org/10.1016/j.electacta.2016.03.144

G Model EA 26974 No. of Pages 12

M. Santambrogio et al. / Electrochimica Acta xxx (2015) xxx–xxx

counterpart, regardless the nature of the test solution. Differently from the corrosion potential, corrosion and limiting current densities showed no correlation with the acidity of the test solutions. In addition, the corrosion current densities calculated from iron dissolution rates were in disagreement with those estimated by Tafel extrapolation method. The relative stability and complexation ability of the investigated carboxylic acids determine the kinetics of iron dissolution. The analysis of freely produced Fe2+ and Fe3+ with time complemented by surface analysis techniques indicated that formation of soluble Fe3+ complexes prevails in the presence of FA, AA and GA. Higher soluble species are promoted in the presence of the latter a-hydroxy acid with high chelation propensity. The opposite is obtained in the case of FA due to the lower stability of formate as compared to acetate. Conversely, in the case of OA, formation of surface-confined Fe2+ and Fe3+ oxalate complexes prevails due to magnetite autoreductive catalytic decomposition. The importance of the mechanistic path depends on the extent of hydrogen bonding interactions between different carboxylic acids, thus highlighting further the complexity of corrosion processes driven by EG decomposition in real fluids. Present results provide some evidence on the need of fundamental studies in the field of organic electrochemistry in relation to the corrosion phenomena, not limited to water based glycol fluids. Acknowledgments The authors wish to thank Prof. Marina Cabrini for helpful discussions. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j. electacta.2016.03.144. References [1] Glycols—Advances in research and application, Q. Ashton Acton (Ed), ScholarlyEditionsTM, Atlanta, GA,US, 2013. [2] H. Yue, Y. Zhao, X. Ma, J. Gong, Ethylene glycol: properties, synthesis, and applications, Chem. Soc. Rev. 41 (2012) 4218–4244. [3] The Dow Chemical Company, “How to choose the right heat transfer fluid”, Process Heating, vol. 16, January 2008, Troy, MI, p. 52. [4] K. Ranjbar, A. Abasi, Failure assessment of crude oil preheating tubes in mono ethylene glycol-water mixture solution, Eng. Failure Anal. 31 (2013) 161–167. [5] M. Madera, W. Hoflinger, R. Kadnar, Ion chromatographic identification and quantification of glycol degradation products, J. Chromatography A 997 (2003) 279–284. [6] Handbook of Chemistry and Physics, 84 th ed., CRC Press, LLC, Boca Raton, FL, 2004. [7] W.J. Rositter Jr., P.W. Brown, M. Godette, The determination of acidic degradation products in aqueous ethylene glycol and propylene glycol solutions using ion chromatography, Solar Energy Mater. 9 (1983) 267–279. [8] W.J. Rossiter Jr., M. Godette, P.W. Brown, K.G. Galuk, An investigation of the degradation of aqueous ethylene glycol and propylene glycol solutions using ion chromatography, Solar Energy Mater. 11 (1985) 455–467. [9] J.R. Clifton, W.J. Rossiter Jr., P.W. Brown, Degraded aqueous glycol solutions: pH values and the effect of common ions on suppressing pH decreases, Solar Energy Mater. 12 (1985) 77–86. [10] P.W. Brown, W.J. Rossiter Jr., K.G. Galuk, A mass spectrometric investigation of the thermal oxidative reactivity of ethylene glycol, Solar Energy Mater. 13 (1986) 197–202. [11] M. Finšgar, J. Jackson, Application of corrosion inhibitors for steels in acidic media for the oil and gas industry: A review, Corros. Sci. 86 (2014) 17–41. [12] L. Ye, L. Zhao, L. Zhang, F. Qi, Theoretical studies on the unimolecular decomposition of ethylene glycol, J. Phys. Chem. A 116 (2012) 55–63. [13] S. Morooka, C. Wakai, N. Matubayasi, M. Nakahara, Hydrothermal carboncarbon bond formation and disproportionations of C1 aldehydes: Formaldehyde and formic acid, J. Phys. Chem. A 109 (2005) 6610–6619. [14] G.A. Zhang, L.Y. Xu, Y.F. Cheng, Mechanistic aspects of electrochemical corrosion of aluminium alloy in ethylene glycol–water solution, Electrochim. Acta 53 (2008) 8245–8252.

11

[15] L.Y. Xu, Y.F. Cheng, Effect of fluid hydrodynamics on flow-assisted corrosion of aluminum alloy in ethylene glycol–water solution studied by a microelectrode technique, Corros. Sci. 51 (2009) 2330–2335. [16] D.Q. Zhang, L.X. Zhou, C.Y. Rao, L.X. Gao, Effect of pH and glycolic acid on corrosion of AA6061 alloy in ethylene glycol–water solution, Mater. Corros. 65 (2014) 31–37. [17] T. Matsumoto, M. Sadakiyo, M.L. Ooi, T. Yamamoto, S. Matsumura, K. Kato, T. Takeguchi, N. Ozawa, M. Kubof, M. Yamauchi, Atomically mixed Fe-group nanoalloys: catalyst design for the selective electrooxidation of ethylene glycol to oxalic acid, Phys. Chem. Chem. Phys. 17 (2015) 11359–11366. [18] S. Neši c, Key issues related to modelling of internal corrosion of oil and gas pipelines – A review, Corros. Sci. 49 (2007) 4308–4338. [19] R.C. Newman, Review and hypothesis for the stress corrosion mechanism of carbon steel in alcohols, Corrosion 64 (2008) 819–823. [20] V.N. Esenin, L.I. Denisovich, Contact corrosion of metals in aqueous and aqueous-glycolic solutions, Russian J. Appl. Chem. 81 (2008) 1796–1802. [21] A. Kriaa, N. Hamdi, K. Jbali, M. Tzinmann, Corrosion of iron in highly acidic hydro-organic solutions, Corros. Sci. 51 (2009) 668–676. [22] I. Danaee, M.N. Khomami, A.A. Attar, Corrosion behavior of AISI 4130 steel alloy in ethylene glycole water mixture in presence of molybdate, Mater. Chem. Phys. 135 (2012) 658–667. [23] M.N. Khomami, I. Danaee, A.A. Attar, M. Peykari, Corrosion of alloy steel in 30% ethylene glycol solution and CrO42 under hydrodynamic condition, J. Iron & Steel Res. Int. 20 (2013) 82–87. [24] I. Danaee, M.N. Khomami, A.A. Attar, Corrosion of AISI 4130 steel alloy under hydrodynamic condition in ethylene glycol + water + NO2 solution, J. Mater. Sci. Technol. 29 (2013) 89–96. [25] I. Betova, M. Bojinov, P. Kinnunen, P. Pohjanne, T. Saario, Influence of the electrolyte composition and temperature on the transpassive dissolution of austenitic stainless steels in simulated bleaching solutions, Electrochim. Acta 47 (2002) 3335–3349. [26] C. Giacomelli, F.C. Giacomelli, J.A.A. Baptista, A. Spinelli, The effect of oxalic acid on the corrosion of carbon steel, Anti-Corrosion Methods & Materials 51 (2004) 105–111. [27] B.J. Wiersma, J.I. Mickalonis, K.H. Subramanian, A.G. Hansen, Corrosion testing of carbon steel in oxalic acid that contains dissolved iron, Conference Paper, NACE  International Corrosion Conference Series (2013) 17. [28] A. Turnbull, M. Ryan, A. Willetts, S. Zhou, Corrosion and electrochemical behaviour of 316L stainless steel in acetic acid solutions, Corros. Sci. 45 (2003) 1051–1072. [29] A.J. Invernizzi, E. Sivieri, S.P. Trasatti, Corrosion behaviour of Duplex stainless steels in organic acid aqueous solutions, Mater. Sci. Eng. A 485 (2008) 234– 242. [30] S.K. Singh, A.K. Mukherjee, Kinetics of mild steel corrosion in aqueous acetic acid solutions, J. Mater. Sci. Technol. 26 (3) (2010) 264–269. [31] M.N. Katariya, A.K. Jana, P.A. Parikh, Corrosion inhibition effectiveness of zeolite ZSM-5 coating on mild steel against various organic acids and its antimicrobial activity, J. Ind. Eng. Chem. 19 (2013) 286–291. [32] G.E. Badea, D. Ionita, P. Cret, Corrosion and passivation of the 304 stainless steel in formic acid solutions, Mater. Corr. 65 (2014) 1103–1110. [33] M.M. Singh, A. Gupta, Corrosion behaviour of mild steel in formic acid solutions, Mater. Chem. Phys. 46 (1996) 15–22. [34] I. Sekine, T. Hayakawa, T. Negishi, M. Yuasa, Analysis for corrosion behavior of mild steels in various hydroxy acid solutions by new methods of surface analyses and electrochemical measurements, J. Electrochem. Soc. 137 (1990) 3029–3033. [35] Standard specification for reagent water, ASTM D1193-99 e1, ASTM International, Philadelphia, 2001. [36] T.R. Yu, G.L. Ji, Electrochemical methods in soil and water research, Pergamon Press, U.K, 1993, pp. 73 Ch. 3. [37] D. Harvey, Modern Analytical Chemistry, The McGraw-Hill Companies, Inc., US, 2000. [38] P.M.A. Sherwood P.M.A, Data Analysis in X-ray Photoelectron Spectroscopy in: Practical Surface Analysis by Auger and X-ray Photoelectron Spectroscopy, D. Briggs & M.P. Seah (Eds.), Wiley New York, 1990. [39] C.D. Wagner, L.E. Davis, W.M. Riggs, The energy dependence of the electron mean free path, Surf. Interface Anal. 2 (1980) 53–55. [40] I.H. Plonski, Effect of bare metal surface on the dissolution in aqueous citrate solutions of magnetite films on carbon steel, J. Appl. Electrochem. 27 (1997) 1184–1192. [41] B. Morgan, O. Lahav, The effect of pH on the kinetics of spontaneous Fe(II) oxidation by O2 in aqueous solution – basic principles and a simple heuristic description, Chemosphere 68 (2007) 2080–2084. [42] W.W. Frenier, F.B. Growcock, Mechanism of iron oxide dissolution – A review of recent literature, Corrosion 40 (1984) 663–668. [43] J.-P. Jolivet, C. Chanéac, E. Tronc, Iron oxide chemistry. From molecular clusters to extended solid networks, Chem. Commun. (2004) 481–483. [44] H.W. Gibson, Chemistry of formic acid and its simple derivatives, Chem. Rev. 69 (1969) 673–692. [45] M.A. Blesa, H.A. Marinovich, E.C. Baumgartner, A.J.G. Maroto, Mechanism of dissolution of magnetite by oxalic acid – ferrous ion solutions, Inorg. Chem. 26 (1987) 3713–3717. [46] S. Rondinini, P. Longhi, P.R. Mussini, T. Mussini, Autoprotolisis constants in nonaqueous solvents and aqueous organic solvent mixtures, Pure & Appl. Chem. 59 (1987) 1693–1702.

Please cite this article in press as: M. Santambrogio, et al., Effect of major degradation products of ethylene glycol aqueous solutions on steel corrosion, Electrochim. Acta (2016), http://dx.doi.org/10.1016/j.electacta.2016.03.144

G Model EA 26974 No. of Pages 12

12

M. Santambrogio et al. / Electrochimica Acta xxx (2015) xxx–xxx

[47] E. Mentasti, Equilibria and kinetics of the complex formation between iron(III) and a-hydroxycarboxylic acids, Inorg. Chem. 18 (1979) 1512–1515. [48] E. McCafferty, Validation of corrosion rates measured by the Tafel extrapolation method, Corros. Sci. 47 (2005) 3202–3215. [49] J.F. Moulder, W.F. Stickle, P.E. Sobol, K.D. Bomben, Handbook of X-Ray Photoelectron Spectroscopy, in: J. Chastain, R.C. King Jr. (Eds.), Physical Electronics Inc., Eden Prairie, MN, USA, 1995. [50] A. Bozzi, T. Yuranova, J. Mielczarski, J. Kiwi, Evidence for immobilized photoFenton degradation of organic compounds on structured silica surfaces involving Fe recycling, New J. Chem. 28 (2004) 519–526. [51] W. Su, J.O. Iroh, IR and XPS studies on the interphase and poly(Nmethylpyrrole) coatings electrodeposited on steel substrates, Electrochim. Acta 44 (1999) 3321–3332.

[52] M.J. Borda, D.R. Strongin, M.A. Schoonen, A novel vertical attenuated total reflectance photochemical flow-through reaction cell for Fourier transform infrared spectroscopy, Spectrochim. Acta A 59 (2003) 1103–1106. [53] M.C. D’Antonio, A. Wladimirsky, D. Palacios, L. Coggiola, A.C. González-Baró, E. J. Baran, R.C. Mercader, Spectroscopic investigations of iron(II) and iron(III) oxalates, J. Braz. Chem. Soc. 20 (2009) 445–450. [54] P. Persson, K. Axe, Adsorption of oxalate and malonate at the water-goethite interface: molecular surface speciation from IR spectroscopy, Geochim. Cosmochim. Acta 69 (2005) 541–552. [55] M. Goti c, S. Musi c, Mössbauer, FT-IR and FE SEM investigation of iron oxides precipitated from FeSO4 solutions, J. Mol. Struct. 834–836 (834) (2007) 445– 453.

Please cite this article in press as: M. Santambrogio, et al., Effect of major degradation products of ethylene glycol aqueous solutions on steel corrosion, Electrochim. Acta (2016), http://dx.doi.org/10.1016/j.electacta.2016.03.144