See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/316860841

Corrosion of metals exposed to 25% magnesium chloride solution and tensile

stress: Field and laboratory studies

Article  in  Case Studies in Construction Materials · May 2017

DOI: 10.1016/j.cscm.2017.04.001

CITATIONS

0

READS

50

3 authors:

Some of the authors of this publication are also working on these related projects:

Synthesis of Material Application Methodologies for Winter Operations. View project

WSU Excellence Fund. Gift fund provided by the Simpson-Strong Tie, LLC. View project

Xianming Shi

Washington State University

185 PUBLICATIONS   2,853 CITATIONS   

SEE PROFILE

Guoxiang Zhou

Harbin Institute of Technology

6 PUBLICATIONS   11 CITATIONS   

SEE PROFILE

Anburaj Muthumani

Montana State University

31 PUBLICATIONS   74 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Xianming Shi on 11 May 2017.

The user has requested enhancement of the downloaded file.

Contents lists available at ScienceDirect

Case Studies in Construction Materials

journal homepage: www.elsevier.com/locate/cscm

Case study

Corrosion of metals exposed to 25% magnesium chloride solutionand tensile stress: Field and laboratory studies

Xianming Shia,⁎, Guoxiang Zhoub, Anburaj Muthumanic

a Laboratory of Corrosion Science & Electrochemical Engineering, Department of Civil and Environmental Engineering, P. O. Box 642910, WashingtonState University, Pullman, WA 99164-2910, USAb School of Civil Engineering & Architecture, Wuhan Polytechnic University, 430023, Wuhan, Chinac Western Transportation Institute, Montana State University, P. O. Box 174250, Bozeman, MT 59717, USA

A R T I C L E I N F O

Keywords:Corrosion test-bedMagnesium chlorideField investigationTensile stressSteelAluminum alloy

A B S T R A C T

The use of chemicals for snow and ice control operations is a common practice for improving thesafety and mobility of roadways in cold climate, but brings significant concerns over their risksincluding the corrosive effects on transportation infrastructure and motor vehicles. The vastmajority of existing studies and methods to test the deicer corrosivity have been restricted tolaboratory environments and unstressed metals, which may not reliably simulate actual serviceconditions. As such, we report a case study in which stainless steel SS 304 (unstressed andexternally tensile stressed), aluminum (Al 1100) and low carbon steel (C1010) coupons wereexposed to 25% MgCl2 under field conditions for six weeks. A new corrosion test-bed wasdeveloped in Montana to accelerate the field exposure to this deicer. To further investigate theobserved effect of tensile stress on the corrosion of stainless steel, SS 304 (unstressed andexternally stressed) coupons were exposed to 25% MgCl2 solution under the laboratoryconditions. The C 1010 exhibited the highest percentage of rust area and suffered the mostweight loss as a result of field exposure and MgCl2 sprays. In terms of ultimate tensile strength,the Al 1100 coupons saw the greatest reduction and the unstressed and externally stressed SS 304coupons saw the least. The ability of MgCl2 to penetrate deep into the matrix of aluminum alloyposes great risk to such structural material. Tensile stressed SS 304 suffered more corrosion thanunstressed SS 304 in both the field and laboratory conditions. Results from this case study mayshed new light on the deicer corrosion issue and help develop improved field testing methods toevaluate the deicer corrosivity to metals in service.

1. Introduction

Clearing snow and ice from roadways is critical for the public safety [1,2] and improved mobility [3] in cold climate. Largeamount of solid and liquid chemicals (also known as deicers) are commonly used for such operations. Deicers applied on theroadways often contain chlorides as freezing point depressant due to their affordable cost. A recent survey of highway maintenanceagencies indicated that sodium chloride (NaCl) was the commonly used deicer, followed by magnesium chloride (MgCl2), agro-basedproducts, Calcium Chloride (CaCl2) and others [4]. However, NaCl is rarely used and minimally effective below pavementtemperatures of 10 °F [5]. MgCl2 is the next best deicer which exhibits better ice melting performance at cold temperatures [6,7] andit is also less expensive than CaCl2 and other agro-based products. Recently, many transportation agencies have been using MgCl2

http://dx.doi.org/10.1016/j.cscm.2017.04.001Received 9 February 2017; Received in revised form 27 March 2017; Accepted 10 April 2017

⁎ Corresponding author.E-mail address: xianming.shi@wsu.edu (X. Shi).

Case Studies in Construction Materials 7 (2017) 1–14

Available online 27 April 20172214-5095/ © 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/BY/4.0/).

MARK

directly or combining MgCl2 with other chloride based salts for winter road operations.Despite the benefits, the use of chemical deicers brings significant concerns over their risks including the corrosive effects on

transportation infrastructure and other assets [8–10]. Corrosion of bare, coated, or embedded metals in the presence of deicers posesa significant risk to the value, performance, serviceability and reliability of motor vehicles, steel bridges, and reinforced concretestructures [11–15]. While this issue has been extensively studied as early as in 1970s [16], recent decades have seen increased use ofchloride deicers in the roadway environment and the increased use of MgCl2 in place of traditional salt (NaCl) or salt brine.

Considering the substantial hidden costs associated with deicer corrosion, it is important to test the corrosivity of chemical deicersbefore their use on the road surface. To this end, there are four common test methods used, i.e., the Pacific Northwest Snowfighters(PNS)/NACE TM0169-95 corrosion test method, the Society of Automotive Engineers (SAE) J2334 test method, Strategic HighwayResearch Program (SHRP) H-205.7 test method and the American Society of Testing and Materials (ASTM) B117 test method [17,18].

The vast majority of existing studies and methods to test the deicer corrosivity have been restricted to laboratory environmentsand unstressed metals, which may not reliably simulate actual service conditions. Most methods measure the weight loss ofunstressed metals as a result of deicer exposure and mainly differ in the exposure regime of corrosion coupons. They are conductedunder well-controlled laboratory conditions and thus fail to mimic the corrosion performance of metals in the field. There are a widearray of factors affecting the corrosion behavior of metals in the field deicer environment, which may include changes in temperatureand relative humidity, time of wetness, wind speed, solar radiation, and presence of other contaminants. Additionally, the metallicstructures or components in the service environment are often subject to mechanical stresses, instead of being unstressed. It has beenreported that an externally applied tensile stress can aggravate the corrosion of certain metallic alloys in neutral 3.5% NaCl solution[19]. In the extreme case, the joint action of corrosion and tensile stress can result in a form of corrosion known as stress corrosioncracking.

In this context, the main objectives of this case study are to:

• present a test-bed developed for controlled field deicer exposure of metals;• compare the corrosion behavior of bare metals in the field and in the laboratory; and• develop a test method of bare metals with both deicer exposure and tensile stress incorporated.

To this end, we present the development of an atmospheric exposure site with custom-made test racks in Montana, a cold-climatestate in the United States. With this test-bed, the corrosion of externally tensile stressed Type 304 stainless steel (SS 304), andunstressed SS 304, C 1010 carbon steel, and 1100 aluminum alloy periodically exposed to neutral 25% MgCl2 solution wasinvestigated. In addition, we present an accelerated laboratory test protocol that integrates external tensile stress withelectrochemical corrosion of metallic coupons.

2. Methodology

2.1. Capabilities of the test-bed and laboratory facility

The main purpose of the outdoor facility was to evaluate the corrosion behavior of bare metals in a field environment withcontrolled deicer exposure. The corrosion test-bed was deployed in Lewistown, MT and featured the capacity to simultaneously testthe corrosivity of a given chemical to eighteen unstressed metallic coupons and four externally stressed coupons over time. In thepreliminary setup, a constant tensile stress was applied by hanging a concrete block on one side of the corrosion coupon which wasfixed to a rigid pulley on the rack using a plastic coated steel rope. Such external loads providing external stress for the corrosioncoupons can be readily replaced, depending on the testing requirements.

To simulate the periodical exposure of metallic structures or components to deicers, an automated spray system was built on topof the test tack with polyvinyl chloride (PVC) pipes, spray nozzles, timers and a 689-kPa (100-psi) water pump. The spray nozzles arewell spaced so as to provide a uniform distribution of deicer spray over the corrosion coupons. There are ten spray nozzles on top ofthe rack which could spray the deicer solution at a pre-determined rate. The replaceable weight providing stress to metallic couponsand the automated spray system are illustrated in Fig. 1(A) and (B). Fig. 1(C) and (D) show the unstressed coupons placed on thewooden holder and the schematic top view of the test rack and coupon layout, respectively. The test racks have been designed toensure that the corrosion coupons do not have direct contact with any other metal and to minimize any potential cross-contaminationbetween corrosion coupons. The corrosion coupons have direct contact either with wood (unstressed coupons) or with plasticshielded ropes (externally stressed coupons). The sprayer system can automatically spray the chemical deicer periodically orrandomly based on the input provided to the timer.

Furthermore, the test-bed facility has a weather station that measures parameters such as air temperature, solar radiation, windspeed, humidity, and precipitation. The outdoor facility can be monitored all day by using a remotely operated camera.

To compare the results of externally stressed coupons from the field test, a customized laboratory test was developed. Specifically,an electrochemical cell was designed as shown in Fig. 2, featuring a hollow cone-shaped plastic tube with the corrosion coupon insert.Narrow end of the plastic tube is sealed with rubber. Both ends of the coupon extending beyond the cone are gripped by the MaterialsTesting Systems (MTS) machine which can apply various mechanical loads to simulate the external tensile stress as deployed in thefield test. A deicer solution can fill inside the cone during the test duration.

X. Shi et al. Case Studies in Construction Materials 7 (2017) 1–14

2

Fig. 1. The field test-bed in Lewistown, Montana: A) Side view of replaceable load and automated spray system; B) Side view of externally stressed coupons; C) Topview of unstressed coupons; D) Schematic top view of test rack illustrating the layout of metal coupons.

X. Shi et al. Case Studies in Construction Materials 7 (2017) 1–14

3

2.2. Field exposure and associated investigation

The field test includes the exposure of metal coupons to the atmospheric conditions at Lewistown, MT using custom made testracks as shown in Fig. 1. The steel frame and timber used in the test tracks were treated for enhanced corrosion resistance consideringtheir exposure to the deicer spray. Table 1 provides the weekly atmospheric conditions of the test-bed during the six-week testingperiod. During the exposure duration, the air temperature ranged from −9.7 °C (14.6 °F) to 30.4 °C (86.8 °F), the wind speed rangedfrom 0 to 59.2 km/h (36.8 miles/h), and the solar radiation was in the range of 14.41–44.08 KW/m2/day.

Three types of metal coupons were chosen for the corrosion investigations in this study. The metal coupons were made of stainlesssteel Type 304 (SS 304), a basic aluminum soft sheet (Al 1100) and a hot rolled low carbon steel (C1010), with their chemical

Fig. 2. Electrochemical cell to deploy tensile stress on SS304 coupons immersed in deicer solution.

Table 1Climatic conditions in Lewistown, MT during the field test.

Weather Information (Lewistown, MT)

Duration of field test (09/12/2012–10/23/2012)

Average Temperature(°F)

Average RelativeHumidity(%)

Average Solar Radiationper day(KW/m2)

Total Precipitation(cm)

Average wind speed(MPH)

Week 1 60.39 ± 12.19 33.72 ± 14.35 22.74 ± 4.34 0 9.61 ± 5.36Week 2 62.41 ± 10.96 41.58 ± 15.20 20.93 ± 2.58 0 8.92 ± 4.00Week 3 60.42 ± 9.75 51.88 ± 21.63 18.71 ± 3.10 0 10.58 ± 6.19Week 4 39.10 ± 7.98 70.60 ± 17.92 12.85 ± 6.43 1.68 10.86 ± 5.42Week 5 51.80 ± 10.40 55.71 ± 18.06 12.77 ± 5.39 0.76 16.05 ± 8.66Week 6 42.68 ± 10.87 60.40 ± 19.41 10.31 ± 4.00 0.25 15.01 ± 9.28

X. Shi et al. Case Studies in Construction Materials 7 (2017) 1–14

4

composition shown in Table 2. Fig. 3 illustrates the two-dimensional configuration of the metal coupons, with the approximate areaexamined being the center portion of 12.9 cm2 (2 inch2). The thickness of SS 304, Al 1100 and C1010 coupons were 2.29 cm (0.90inch), 0.2.29 cm (0.90 inch) and 2.67 cm (0.105 inch), respectively. Six coupons were selected for each metal type and additionalfour SS 304 coupons were selected to test the corrosion rate under external tension (10% of the Ultimate Tensile Strength or UTS).External stress was applied by hanging a concrete block (approximately 90.7 kg or 200 lbs) on one side of the stainless steel which isfixed to a rigid pulley on the test rack as shown in Fig. 1(B).

The liquid deicer used in this study was 25 wt.% MgCl2 solution prepared with 5338 g magnesium chloride hexahydrate (fromIntegra Chemical, WA, USA) mixed with every 4662 g tap water. To accelerate the simulation of deicer exposure of one winter seasonwithin the six week testing period, the MgCl2 solution was sprayed every 8 h on the coupons for 1 min, at the rate of 0.19 L/min (0.05gallon per minute). The parameters measured from the field test include the open circuit potential (OCP) vs. saturated calomelelectrode (SCE) and digital photographs of metal coupons approximately every week. Other parameters include the percent rust area(by digital photo analysis), weight loss (after cleaning), and tensile strength.

2.3. Laboratory investigation

To better replicate and compare the results of externally stressed SS 304 coupons from the field test, a customized laboratory testwas developed, using the electrochemical cell as shown in Fig. 2. Laboratory tests were conducted for SS 304 coupons in theexternally stressed (shown in Fig. 2) and unstressed environment, respectively. For externally stressed coupons, four different tensileloads i.e., 5% of UTS (45.4 kg or 100 lbs), 10% of UTS (90.7 kg or 200 lbs), 15% of UTS (136.1 kg or 300 lbs) and 20% of UTS(181.4 kg or 400 lbs) were applied at 48 h, 72 h, 96 h, and 120 h, respectively. For each periodical Linear Polarization Resistance(LPR) measurement, the stainless steel was polarized around its corrosion potential (−15 mV to 15 mV/SCE vs. OCP by a directcurrent (DC) signal at a scan rate of 0.2 mV/s. The polarization resistance (Rp) is defined by the slope of the potential vs. currentdensity plot at the corrosion potential. Corrosion current density (icorr) is subsequently calculated from icorr = B/Rp, assuming aStern–Geary constant (B) of 26 mV for the steel corrosion [20]. These corrosion parameters were measured at the end of each externalstress application and digital photographs at the end of the experiment.

Another laboratory test was designed to measure the corrosion rate of unstressed SS 304 placed inside the deicer solution. In thistest, SS 304 coupon and reference electrode (SCE) was placed in the beaker containing deicer solution for a period of 5 days. The

Table 2Composition of (weight%) of stainless steel, aluminum and low carbon steel coupons.

Alloy Si Cu Mg Cr Ni C Mo Zn S P Fe Al Other

Al1100 0.05–0.2 0–0.05 0–0.1 Remainder 0.15SS 304 0–1.0 0–1.0 0–2.0 18–20 8–12 0-0.08 0–1.0 None 0–0.03 0–0.0045 Remainder None Co: 0–0.2; N: 0–0.1C 1018 0.15–0.3 0.3–0.9 0.1 0.5 0.04 Remainder

Fig. 3. Two-dimensional configuration of the corrosion coupons.

X. Shi et al. Case Studies in Construction Materials 7 (2017) 1–14

5

Fig. 4. Photos of SS304, Al 1100 and C1010 during 0th, 3rd and 6th week after exposure to MgCl2 solution and atmospheric conditions.

Fig. 5. Percent of weight loss and rust area for metallic coupons after the field exposure.

X. Shi et al. Case Studies in Construction Materials 7 (2017) 1–14

6

parameters measured are corrosion parameters from LPR measurements every 24 h and digital photographs at the end of theexperiment.

2.4. SEM/EDX measurements

Following the field test, the corrosion coupons were cleaned with acetone to remove the rust from the entire metal surface. For thesurface analysis, a small portion (0.38 cm by 0.76 cm) was cut from the center of each coupon and its exposed top surface was usedfor analysis. For cross-sectional analysis, the small portion of coupon was cut in to two-halves using saw blade and then polished withdry sandpapers, cleaned with pressurized air to remove debris. Note that no epoxy impregnation was used as that would compromisethe C and O signals from the coupon surface.

The newly cut top surface and cross-sectional surface were subjected to scanning electron microscopy (SEM) and energy-dispersive X-ray spectroscopy (EDX), in order to examine their localized morphology and elemental distributions at the microscopicscale. A Zeiss Supra 55VP Field Emission SEM system was employed with 1 kV accelerating voltage and vacuum pressures between10−5–10−6 torr. For surface analysis, SEM was used to investigate the effects of deicer exposure on the metal’s morphology, bycollecting data from sites randomly selected from the top surface exposed to atmosphere and MgCl2 sprays. The seven randomlyselected sites consisted of three pit surface, three non-pit surface and one with combination of pit and non-pit surface atmagnifications beginning at 500 times and up to 10,000 times. For cross-sectional analysis, SEM was performed on three sites evenlydistributed along the depth, starting from MgCl2 exposed top surface to bottom surface at magnifications beginning at 500 times and

Fig. 6. Change in the ultimate tensile strength of metallic coupons and associated percent loss of tensile strength after the field exposure.

Fig. 7. OCP values of SS304 (Unstressed and tensile stressed coupons), Al 1100 and C1010 after field exposure and MgCl2 sprays.

X. Shi et al. Case Studies in Construction Materials 7 (2017) 1–14

7

up to 10,000 times.The EDX data were obtained using a PGT Si(Li) detector that featured the ability to detect small variations in trace element

content. The detection limit was generally 1–2 atomic%. For the EDX analysis, an accelerating voltage of 20 kV was used with a scantime of 60 s per sampling area. Areas used for EDX analysis corresponded directly to the morphological examination by SEM.

3. Results and discussion

3.1. Corrosion behavior of different metals in the field environment

3.1.1. Visual analysis and mechanical evaluationFig. 4 presents the photos of exposed metal coupons in the field during the 0th, 3rd and 6th week, from which the ImageJ™

software and its color deconvolution function were employed to quantify the percent rust area on the coupon surface. During the 0thweek (i.e., prior to MgCl2 exposure), all the corrosion coupons appeared shiny and free of any apparent corrosion distress. Once themetals were exposed to atmospheric conditions and MgCl2 sprays, there was a certain degree of corrosion showing up in all the metalcoupons. By the end of 3rd week, the Al 1100 and C1010 coupons exhibited high percent rust area of 40% and 60%, respectively,whereas SS304 (both unstressed and externally tensile stressed) exhibited much less signs of rusting (< 10% by surface area). By theend of 6th week, the Al 1100, C1010, SS 304 (unstressed) and SS304 (externally stressed) coupons exhibited significant difference inpercent rust area, i.e., 60–70%, 90–100%, 30–40% and 10–20%, respectively.

Fig. 5 presents the results of average percentage of weight loss and rust area after the six-week field exposure. The Al 1100 and SS304 (unstressed and externally stressed) coupons suffered minor weight loss ( < 0.2%), whereas the C 1010 suffered a moresignificant weight loss of 2.43%. Yet, all the metallic coupons had a significant amount of rust developed on their surface. The Al1000, C 1010, SS 304 (unstressed), and SS 304 (externally stressed) exhibited an average of 63.5%, 89.5%, 47.7%, and 8.5% area ofcorrosion, respectively. Among them, the carbon steel (C 1010) exhibited the highest percentage of rust area and suffered the mostweight loss as a result of field exposure and MgCl2 sprays. In addition, the seemingly more severe corrosion of aluminum alloy byMgCl2, relative to stainless steel, is consistent with a previous study [18]. In the case of SS304 coupons, it is interesting to note thatthe presence of tensile stress induced slightly more weight loss but substantially less area of rust. In other words, visual observation

Fig. 8. SEM images of unstressed SS304 coupon after field exposure. a) top surface with pit and non-pit (1000×). b) top surface with pit (7000×). c) top surface withnon-pit (5000×). d) Cross-sectional surface 1 (5000×).

X. Shi et al. Case Studies in Construction Materials 7 (2017) 1–14

8

failed to serve as a reliable indicator for the extent of accumulative corrosion damage in this case. Note that the average corrosionrate of the C1010, Al 1100, externally stressed SS 304, and unstressed SS 304 over the six-week accelerated field exposure durationwas estimated to be 5.6 mm/year, 2.9 mm/year, 0.28 mm/year, and 0.12 mm/year, respectively, based on the weight lossmeasurements. These corrosion rates are significantly higher than typical atmospheric corrosion rates of these metals, revealingthe detrimental role of MgCl2.

Fig. 6 shows the change in the ultimate tensile strength (UTS) of metallic coupons and percent UTS loss after the six-week fieldexposure. The Al 1100 coupons saw the greatest reduction in UTS (34.7%) and the unstressed and externally stressed SS 304 couponssaw the least loss in UTS (12.9% and 12.5%, respectively). Despite their relatively high weight loss and percent rust area, the C 1010coupons suffered only an average 18. 5% loss in UTS. The high UTS reduction of Al 1100 can be attributed to the intergranularcorrosion along the grain boundaries [21,22] and the attack of MgCl2 to the matrix underneath, as confirmed by the SEM datareported in a later section.

3.1.2. Open circuit potential (OCP) analysisFig. 7 shows the temporal evolution of OCP of all the metallic coupons during the field exposure and MgCl2 prays. The Al 1100

coupons exhibited significantly more negative potentials than other metallic coupons and their average OCP value did not changesignificantly over the exposure duration. Visual inspection revealed the presence of corrosion products (e.g., Fig. 4) as well as theformation of pits on the surface of aluminum alloy. Previous research reported the presence of OCP transients, i.e., sudden potentialdrops during corrosion initiation followed by stabilizing potential during corrosion growth [23]. The periodical OCP measurementswere not frequently enough to capture the phase characteristic of pit initiation.

As shown in Fig. 7, the unstressed SS 304 coupons exhibited significantly more noble potentials than other metallic coupons andtheir average OCP value did not show any significant drop over the exposure duration. Yet, visual inspection revealed the presence ofpits on the surface of unstressed SS 304. The periodical OCP measurements failed to capture the pit initiation on the stainless steelsurface exposed to MgCl2, which is known to induce significant drop in the OCP [24]. The observed increase in the average OCPduring the last 10 days may be attributed to the formation of cathode scale or some other film during the atmospheric exposure.

In comparison, both the C1010 and stressed SS 304 coupons exhibited a significant drop in their average OCP during the first10 days, which was followed by a significant increase and then stabilization (Fig. 7). These are likely associated with the formation ofmetastable pits on these metallic surfaces during the early stage of field exposure and MgCl2 sprays, followed by the stable growth of

Fig. 9. SEM images of externally stressed SS304 coupon after field exposure. a) top surface with “pit” and non-pit (5000×). b) top surface with “pit” (5000×). c) topsurface with non-pit (500×). d) Cross-sectional surface 1 (5000×).

X. Shi et al. Case Studies in Construction Materials 7 (2017) 1–14

9

corrosion and corrosion products. It is noteworthy that the externally stressed SS 304 coupons consistently exhibited much lower OCPvalues than their unstressed counterparts, confirming the role of tensile stress in aggravating the corrosion attack by MgCl2.

3.1.3. SEM/EDX analysesFigs. 8–11 present the representative SEM images of unstressed SS 304, externally stressed SS 304, Al 1100, and C 1010 coupons

after their six-week field exposure and MgCl2 sprays. Overall, the SS304 seemed to be most corrosion-resistant in the givenenvironment. The aluminum alloy (Al 1100) and carbon steel (C 1010) coupons exhibited the most apparent pitting on their surfaces(Figs. 10(a) and 11(a)) and deep into the material’s matrix (Figs. 10(d) and 11(d). Figs. 10(b) and 11(b) reveal the presence of largeand deep pits on the surface of Al 1100 and C 1010, with both diameter and depth of up to approximately 160 μm and 25 μm,respectively. These help explain the observed high percentage of tensile strength reduction, 34.7% and 18.5%, respectively (Fig. 6).In contrast, Fig. 8(b) and (d) reveal the presence of large yet shallow pits on the surface of unstressed SS 304, whereas Fig. 9(b) and(d) reveal significant localized corrosion attack which was not in the form of apparent pits. The application of external tensile stressinduced the formation of orderly yet strained microstructure on the SS 304 coupons along the direction of tension and significantlyaltered the material’s matrix, as evidenced in Fig. 9(c) and (d), compared with Fig. 8(c) and (d). This is consistent with previousstudies indicating the change of SS 304 microstructure by the tensile stress [1]. These help explain the observed lower percentage oftensile strength reduction on unstressed and externally stressed SS 304 coupons, 12.9% and 12.5%, respectively (Fig. 6). Themorphological examinations at the microscopic level (Figs. 11(b), 10(b), 8(b) and 9(b)) also help explain the relative order ofobserved percent rust area on the C 1010, Al 1100, unstressed SS 304, and externally stressed SS 304, being 89.5%, 63.5%, 47.7%,and 8.5%, respectively (Fig. 5). It is noteworthy that the corrosion attack of externally tensile stressed SS 304 resulted in the gradualdisintegration of the material’s matrix (Fig. 9), instead of apparent pitting (as shown in Fig. 8). For the externally stressed SS 304, theattack by MgCl2 was not apparent from surface examination (e.g., by percent rust area) but led to as significant a strength reductionas the unstressed SS 304 (see Fig. 6). As such, visual inspection may give misleading indication of the corrosion condition of SS 304components in a service environment with both tensile stress and MgCl2 exposures.

Table 3 summarizes the chemical traces present in all the metal coupons after the six-week exposure, as quantified by EDXanalysis. For both the unstressed and stressed SS 304 coupons, the surface (both pit and non-pit) featured the significant presence ofMg and Cl residuals as well as significantly higher O, C, and Si signals and lower Fe and Cr signals, relative to the cross-sectional

Fig. 10. SEM images of Al 1100 coupon after field exposure. a) top surface with pit and non-pit (500×). b) top surface with pit (1500×). c) top surface with non-pit(1500×). d) Cross-sectional surface 1 (5000×).

X. Shi et al. Case Studies in Construction Materials 7 (2017) 1–14

10

surfaces indicative of the bulk matrix. For both types of SS 304 coupons, the pit surface featured significantly higher O, and C signalsand lower Fe, Cr, and Ni signals, relative to its non-pit counterpart. These results suggest that the Mg2+ and Cl− ions onlycompromised the surface layer of SS 304, and the localized corrosion attack of Cr- and Ni- rich passive film resulted in the formationof oxides in the surface layer, especially within the pits [25]. The high O, C and Si signals may be attributed to the formation ofcarbonates and airborne dust on the surface, as a result of atmospheric exposure.

In contrast, for the Al 1100 coupons, both the surface and the cross-sectional surfaces featured the significant presence of C, O, Mgand Cl signals, suggesting that the Mg2+ and Cl− ions penetrated deep into the matrix and compromised much more than the surfacelayer to form oxides as corrosion products. This helps to explain why the use of a salt remover could significantly enhance thecorrosion resistance of carbon steel and stainless steel in 30% MgCl2, but not that of the aluminum alloy [26]. The ability of MgCl2 topenetrate deep into the matrix of aluminum alloy poses great risk to such structural material, as confirmed by the considerable loss intensile strength (34.7%, shown in Fig. 6) and misleadingly lower weight loss (Fig. 5). For Al 1100, the pit surface featuredsignificantly higher O and lower Al signals, relative to its non-pit counterpart (Table 3).

For the C 1010 coupons, both the surface and the cross-sectional surfaces featured the significant presence of C and O signals,suggesting that the pitting corrosion had compromised much more than the surface layer. Different from the case of Al 1100,however, Mg signals were not detected in any of the examined surfaces and Cl signals were only detected in the surface layer, whichimplies a less devastating mechanism at work and may help explain the relatively moderate loss in tensile strength (18.5%, shown inFig. 6). For C 1010, the pit surface featured significantly higher O and C signals and lower Fe signals, relative to its non-pitcounterpart (Table 3).

3.2. Corrosion behavior of SS 304 in the laboratory environment

From the laboratory test, visual inspection of externally tensile stressed and unstressed SS304 coupons did not reveal anysignificant corrosion by continuous immersion in 25% MgCl2 for 120 h. However, the main objective of this experiment is to identifythe difference in corrosion behavior of externally stressed and unstressed SS304 coupons. Table 4 provides the data that illustrates thetemporal evolution of Rp, icorr and corrosion rate (CR) of tensile stressed SS 304 and unstressed SS 304 coupons, based on periodicalLPR measurements. It is noteworthy that the instantaneous corrosion rates of both stressed and unstressed SS 304 coupons tested inthe continuous immersion laboratory test were about three orders of magnitude lower than their average corrosion rates over the six-

Fig. 11. SEM images of C 1010 coupon after field exposure. a) top surface with pit and non-pit (500×). b) top surface with pit (7000×). c) top surface with non-pit(20000×). d) Cross-sectional surface 1 (5000×).

X. Shi et al. Case Studies in Construction Materials 7 (2017) 1–14

11

week field exposure duration. This highlights the limitations of the laboratory test and its tendency to greatly underestimate thecorrosion risk of MgCl2 and thus greatly overestimate the durability and reliability of metals in such service environments. The greatdiscrepancy between the laboratory and field obtained corrosion rates also highlight the role of uncontrolled field factors (e.g., wet/dry cycling, temperature cycling, atmospheric pollutants such as sulfur and nitrogen oxides, and deposition of aerosol particles andhydroscopic magnesium chloride) in exacerbating the atmospheric corrosion of metals.

Based on the data in Table 4, the corrosion current density (icorr, in 10−2 μA/cm2) of unstressed SS 304 in 25% MgCl2 followed alogarithmic increase with the increase in immersion time (t, in seconds), i.e., icorr = 0.7574 ln(t) − 6.0459, R2 = 0.985. This impliesthat the formation of corrosion product on the metallic surface worked to slow down the corrosion attack. In contrast, the corrosioncurrent density of increasingly tensile stressed SS 304 in 25% MgCl2 followed an exponential increase with the increase in immersiontime (t, in seconds), i.e., icorr = 1.8599 e^(4 × 10−6 × t), R2 = 0.953. The presence of externally applied tensile stress substantiallyincreased the corrosion rate of SS 304 and facilitated the propagation of corrosion attack, which is consistent with the commoncorrosion knowledge [28] as well as the observed weight loss measurements from field exposure (Fig. 5). The detrimental role ofexternal stress in aggravating the corrosion of 304 austenitic stainless steel is possibly due to “susceptible paths for stress corrosioncracking” [27]. After 120 h of immersion in 25% MgCl2, the corrosion resistance of SS 304 decreased from 6.20 × 10−5 Ω cm2

(unstressed) to 2.22 × 10−5 Ω cm2 (stressed) and the corrosion rate increased from 3.91 × 10−4 mm/yr (unstressed) to1.21 × 10−3 mm/yr (stressed). In contrast, the six-week exposure to atmosphere and MgCl2 sprays did not result in any significantdifference in the ultimate tensile strength between the unstressed and stressed SS 304 (Fig. 6), likely attributable to the fact that thecorrosion attack of SS 304 remained mainly on the surface layer (Fig. 8, Fig. 9, and Table 3).

Table 3Chemical traces of various elements from EDX analysis.

Group Elements

C O Cr Ni Si Fe K Al Mg ClSS 304 (unstressed)

Pit surface 11.72 27.5 12.05 4.2 0.91 40.38 0.4 0.51 0.88 0.83Non-pit surface 4.57 8.92 15.88 6.85 0.43 60.55 0.07 0.21 0.33 0.21Cross-section surface 1 0.01 2.29 18.69 8.7 – 70.31 – – – –Cross-section surface 2 0 4.13 18.12 8.51 – 69.17 – – – –Cross-section surface 3 0 5.4 18.24 7.83 – 69.73 – – – –

SS 304 (tensile stressed)Pit surface 6.3 23.51 14.13 5.16 0.68 47.71 0.3 0.41 0.79 0.73Non-pit surface 3.06 2.29 15.69 8.7 0.65 68.95 0.06 0.05 0.42 0.12Cross-section surface 1 0 2.55 19.1 8.3 – 65.1 – – – –Cross-section surface 2 0 3.27 17.67 8.16 – 70.83 – – – –Cross-section surface 3 1.38 8.96 16.82 7.53 – 65.23 – – – –

Al 1100Pit surface 7.23 58.58 – – – 0.495 – 26.91 2.65 0.76Non-pit surface 9.86 39.31 – – – 0.24 – 47.48 1.24 0.76Cross-section surface 1 7.38 30.65 – – – – – 59.71 1.69 0.48Cross-section surface 2 4.98 27.03 – – – – – 66.32 1.32 0.33Cross-section surface 3 6.58 26.98 – – – – – 64.87 1.18 0.39

Carbon Steel 1010Pit surface 7.91 21.5 – – 0.07 66.63 0.06 – – 0.46Non-pit surface 5.37 4.78 – – 0.1 87.69 – – – 1.85Cross-section surface 1 11.66 19.52 – – 0.23 68.53 – – – –Cross-section surface 2 7.63 21.21 – – 0.2 70.78 – – – –Cross-section surface 3 7.75 14.28 – – 0.14 77.62 – – – –

Table 4Corrosion parameters of the externally stressed and unstressed SS 304 in the laboratory environment, based on LPR measurements.

Measurement Time Externally Stressed SS304 Unstressed SS304

Tensile stress Rp

(Ω cm2)icorr (μA/cm2) CR (mm/yr) Rp

(Ω cm2)icorr (μA/cm2) CR (mm/yr)

48 h 5% of UTS 6.20 × 105 4.19 × 10−2 4.33 × 10−4 8.26 × 105 3.15 × 10−2 3.26 × 10−4

72 h 10% of UTS 5.51 × 105 4.73 × 10−2 4.89 × 10−4 7.53 × 105 3.45 × 10−2 3.57 × 10−4

96 h 15% of UTS 3.29 × 105 7.90 × 10−2 8.22 × 10−4 7.35 × 105 3.54 × 10−2 3.67 × 10−4

120 h 20% of UTS 2.22 × 105 1.17 × 10−1 1.21 × 10−3 6.89 × 105 3.78 × 10−2 3.91 × 10−4

Note: UTS = Ultimate Tensile Strength.

X. Shi et al. Case Studies in Construction Materials 7 (2017) 1–14

12

4. Conclusions

• A prototype field test-bed for corrosion studies was showcased in the State of Montana, which may be further upgraded to includeother desirable features (e.g., remote control of spray frequency).

• During the six-week field exposure to periodical MgCl2 sprays, the carbon steel (C 1010) suffered the most corrosion in terms ofpercent rust area and weight loss, whereas the aluminum alloy (Al 1100) suffered the most percent loss in tensile strength. Fromvisual inspection, the C1010 and Al 1100 suffered significantly more corrosion than the stainless steel (SS304, both unstressed andtensile stressed).

• In the field environment, the OCP values of tensile stressed SS 304 were significantly lower than those of their unstressedcounterparts, confirming the more active corrosion. Yet, the two groups exhibited comparable reductions in tensile strength as aresult of atmospheric exposure and periodical sprays of 25% MgCl2.

• From SEM/EDX analyses of field coupons, the Mg2+ and Cl− ions only compromised the surface layer of SS 304, but penetrateddeep into the matrix of Al 1100 and compromised much more than the surface layer to form oxides as corrosion products. For theC 1010 coupons, the pitting corrosion had compromised much more than the surface layer. Different from the case of Al 1100,however, Mg signals were not detected in any of the examined surfaces and Cl signals were only detected in the surface layer,which implies a less devastating mechanism at work and may help explain the relatively moderate loss in tensile strength.

• From the five-day laboratory test, the LPR test indicates the presence of externally applied tensile stress significantly decreased thecorrosion resistance and increased the corrosion rate of SS 304 and this detrimental effect increased with the level of stress. Theinstantaneous corrosion rates of both stressed and unstressed SS 304 coupons tested in the continuous immersion laboratory testwere about three orders of magnitude lower than their average corrosion rates over the six-week field exposure duration.

Acknowledgements

The authors acknowledge the financial support by the Western Transportation Institute and the China National Natural ScienceFoundation (Grant No. 51278390). They would like to thank Dr. Yongxin Li for his valuable inputs. Also, the authors would like tothank Michelle Akin and Zachary Zupan for their assistance in field and laboratory testing.

References

[1] L. Qiu, W. Nixon, Performance measurement for highway winter maintenance operations, Iowa Highway Research Board Technical Report, (2009), p. 474.[2] T. Usman, L. Fu, L. Miranda-Moreno, Quantifying safety benefit of winter road maintenance: accident frequency modeling, Accid. Anal. Prev. 42 (6) (2010)

1878–1887.[3] L. Qiu, W. Nixon, Modeling the causal relationships between winter highway maintenance, adverse weather and mobility, Proceedings: 87th Annual Meeting of

the Transportation Research Board, Washington D.C., 2008.[4] X. Shi, L. Fay, C. Gallaway, K. Volkening, M.M. Peterson, T. Pan, A. Creighton, C. Lawlor, S. Mumma, Y. Liu, T.A. Nguyen, Evaluation of alternate anti-icing and

deicing compounds using sodium chloride and magnesium chloride as baseline deicers, Final Report for the Colorado Department of Transportation, ColoradoDepartment of Transportation, Denver, CO, 2009 Report No. CDOT-2009–01. February 2009.

[5] NCHRP, Guidelines for the selection of snow and ice control materials to mitigate environmental impacts, National Cooperative Highway Research Program(NCHRP) Report 577, Transportation Research Board of the National Academies, Washington, D.C, 2007.

[6] E.V. Baroga, 2002–2004 salt pilot project, A Final Report Prepared for the Washington State Department of Transportation, Washington State Department ofTransportation, Olympia, WA, 2005.

[7] X. Shi, K. Fortune, R. Smithlin, M. Akin, L. Fay, Exploring the performance and corrosivity of chloride deicer solutions: laboratory investigation and quantitativemodeling, Cold Reg. Sci. Technol. 86 (1) (2013) 36–44.

[8] L. Sutter, T. Dam, K. Peterson, D. Johnston, Long-term effects of magnesium chloride and other concentrated salt solutions on pavement and structural Portlandcement concrete: phase I results, Transp. Res. Rec.: J. Transp. Res. Board 1979 (2006) 60–68.

[9] X. Shi, D. Veneziano, N. Xie, J. Gong, Use of chloride-based ice control products for sustainable winter maintenance: a balanced perspective, Cold Reg. Sci.Technol. 86 (2013) 104–112.

[10] X. Shi, S. Jungwirth, M. Akin, R. Wright, L. Fay, D.A. Veneziano, Y. Zhang, J. Gong, Z. Ye, Evaluating snow and ice control chemicals for environmentallysustainable highway maintenance operations, J. Transp. Eng. 140 (11) (2014) 05014005.

[11] Y. Li, Y. Fang, N. Seeley, S. Jungwirth, E. Jackson, X. Shi, Corrosion of chloride deicers on highway maintenance equipment: renewed perspective and laboratoryinvestigation, Transp. Res. Rec. 2361 (2013) 106–113.

[12] J.T. Johnson, Corrosion Costs of Motor Vehicles, (2002) http://www.corrosioncost.com/pdf/trans.portation.pdf . (Accessed July 2007).[13] G.H. Koch, M.H. Brongers, M.G. Thompson, Y.P. Virmani, J.H. Payer, Corrosion Cost and Preventative Strategies in the United States. Federal Highway

Administration. Report # FHWA-RD-01-156, 2002.[14] X. Shi, L. Fay, Z. Yang, T.A. Nguyen, Y. Liu, Corrosion of deicers to metals in transportation infrastructure: introduction and recent developments, Corros. Rev. 27

(1–2) (2009) 23–52.[15] X. Shi, Y. Liu, M. Mooney, M. Berry, B. Hubbard, T.A. Nguyen, Laboratory investigation and neural networks modeling of deicer ingress into Portland cement

concrete and its corrosion implications, Corros. Rev. 28 (3–4) (2010) 105–154.[16] T.R. Menzies, National cost of motor vehicle corrosion from deicing salt, in: R. Baboian (Ed.), Automotive Corrosion and Protection (NACE International,

Proceedings of the CORROSION/91 Symposium, Houston, TX, 1992.[17] S.S. Pathak, M.D. Blanton, S.K. Mendon, J.W. Rawlins, Investigation on dual corrosion performance of magnesium-rich primer for aluminum alloys under salt

spray test (ASTM B117) and natural exposure, Corros. Sci. 52 (4) (2010) 1453–1463.[18] Y. Xi, Z. Xie, Corrosion effects of magnesium chloride and sodium chloride on automobile components, Colorado Department of Transportation, (2002) Report #

CDOT-DTD-R-2002-4.[19] J.F. Li, W.J. Chen, X.S. Zhao, W.D. Ren, Z.Q. Zheng, Corrosion behavior of 2195 and 1420 Al-Li alloys in neutral 3.5% NaCl solution under tensile stress, Trans.

Nonferr. Met. Soc. China 16 (5) (2006) 1171–1177.[20] C. Andrade, J.A. Gonzalez, Quantitative measurements of corrosion rate of reinforcing steels embedded in concrete using polarization resistance measurements,

Werkst. Korros. 29 (1978) 515–519.[21] T.D. Burleigh, E. Ludwiczak, R.A. Petri, Intergranular corrosion of an aluminum-magnesium-silicon-copper alloy, Corrosion 51 (1) (1995) 50–55.[22] V. Guillaumin, G. Mankowski, Localized corrosion of 2024 T351 aluminum alloy in chloride media, Corros. Sci. 41 (3) (1998) 421–438.

X. Shi et al. Case Studies in Construction Materials 7 (2017) 1–14

13

[23] J.F. Li, B. Maier, G.S. Frankel, Corrosion of an Al–Mg–Si alloy under MgCl2 solution droplets, Corros. Sci. 53 (6) (2011) 2142–2151.[24] B. Maier, G.S. Frankel, Pitting corrosion of bare stainless steel 304 under chloride solution droplets, J. Electrochem. Soc. 157 (2010) C302–C312.[25] H.E. Evans, D.A. Hilton, R.A. Holm, S.J. Webster, The development of localized pits during stainless steel oxidation, Oxid. Met. 14 (3) (1980) 235–247.[26] M.H. Nazari, X. Shi, E. Jackson, Y. Zhang, Y. Li, Laboratory investigation of washing practices and bio-based additive for mitigating metallic corrosion by

magnesium chloride deicer, J. Mater. Civ. Eng. 29 (1) (2016) 04016187.[27] D.A. Vaughan, D.I. Phalen, C.L. Peterson, W.K. Boyd, Relationship between hydrogen pickup and susceptible paths in stress corrosion cracking of type 304

stainless steel, Corrosion 19 (9) (1963) 315t–326t.[28] D.T. Spencer, M.R. Edwards, M.R. Wenman, C. Tsitsios, G.G. Scatigno, P.R. Chard-Tuckey, The initiation and propagation of chloride-induced transgranular

stress-corrosion cracking (TGSCC) of 304L austenitic stainless steel under atmospheric conditions, Corros. Sci. 88 (2014) 76–88.

X. Shi et al. Case Studies in Construction Materials 7 (2017) 1–14

14

View publication statsView publication stats