Chemosphere 134 (2015) 598–605

Contents lists available at ScienceDirect

Chemosphere journal homepage: www.elsevier.com/locate/chemosphere

Field methods for rapidly characterizing paint waste during bridge rehabilitation Zhan Shu a,⇑, Lisa Axe a, Kauser Jahan b, Kandalam V. Ramanujachary c a

Department of Civil and Environmental Engineering, Newark College of Engineering, New Jersey Institute of Technology, Newark, NJ 07102, USA Department of Civil and Environmental Engineering, Rowan University, Glassboro, NJ 08028, USA c Department of Chemistry and Biochemistry, Rowan University, Glassboro, NJ 08028, USA b

h i g h l i g h t s  Metal leaching mechanisms from paint waste generated during bridge rehabilitation were investigated.  The diffuse layer model demonstrated the importance of complexation with the iron oxide surface.  Statistically-based models developed in this research predicted metal leaching from paint waste.  The study may assist DOT agencies with applying a predictive tool for waste classification.

a r t i c l e

i n f o

Article history: Received 19 May 2014 Received in revised form 13 October 2014 Accepted 27 October 2014 Available online 27 December 2014 Handling Editor: I. Cousins Keywords: Waste classification Leaching Bridge paint Modeling FP-XRF

a b s t r a c t For Department of Transportation (DOT) agencies, bridge rehabilitation involving paint removal results in waste that is often managed as hazardous. Hence, an approach that provides field characterization of the waste classification would be beneficial. In this study, an analysis of variables critical to the leaching process was conducted to develop a predictive tool for waste classification. This approach first involved identifying mechanistic processes that control leaching. Because steel grit is used to remove paint, elevated iron concentrations remain in the paint waste. As such, iron oxide coatings provide an important surface for metal adsorption. The diffuse layer model was invoked (log KMe = 4.65 for Pb and log KMe = 2.11 for Cr), where 90% of the data were captured within the 95% confidence level. Based on an understanding of mechanistic processes along with principal component analysis (PCA) of data obtained from fieldportable X-ray fluorescence (FP-XRF), statistically-based models for leaching from paint waste were developed. Modeling resulted in 96% of the data falling within the 95% confidence level for Pb (R2 0.6–0.9, p 6 0.04), Ba (R2 0.5–0.7, p 6 0.1), and Zn (R2 0.6–0.7, p 6 0.08). However, the regression model obtained for Cr leaching was not significant (R2 0.3–0.5, p 6 0.75). The results of this work may assist DOT agencies with applying a predictive tool in the field that addresses the mobility of trace metals as well as disposal and management of paint waste during bridge rehabilitation. Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction The majority of the steel bridges in the interstate system were constructed between 1950 and 1980 (Appleman, 1997), all were protected from corrosion with paint coatings containing lead and chromate (Strivens and Lambourne, 1999). Currently, a number of transportation agencies (i.e., New York State Department of Transportation (NYSDOT, 2008); Minnesota Department of Transportation (MnDOT, 2004); and Missouri Department of Natural Resources, 2006) apply a conservative approach by assuming all ⇑ Corresponding author. Tel.: +1 (973) 597 6077. E-mail address: [email protected] (Z. Shu). http://dx.doi.org/10.1016/j.chemosphere.2014.10.081 0045-6535/Ó 2014 Elsevier Ltd. All rights reserved.

waste generated from bridges rehabilitated before 1988 or with lead concentrations greater than 5000 ppm as hazardous. This practice stems from the fact that there is no approved reliable, fast, and efficient method for classifying paint waste in-situ as nonhazardous. In earlier work (Shu, 2014), elevated iron was observed (as great as 80 wt.%) in the paint waste due to the use of steel grit as blasting abrasive during bridge rehabilitation. Although magnetic separation is applied to remove the steel grit, a fraction remains in the paint waste. Iron oxide formed on the steel grit surface plays an important role in the system (Shu et al., 2015). In our companion paper (Shu et al., 2015) despite the elevated concentrations of trace metals, leached concentrations evaluated using the U.S. Environmental

599

Z. Shu et al. / Chemosphere 134 (2015) 598–605

Protection Agency (U.S. EPA) toxicity characteristic leaching procedure (TCLP) (U.S. EPA, 1992) were less than the toxicity characteristic (TC) levels. Furthermore, the multiple extraction procedure (MEP) (U.S. EPA, 1986, 2004) studies simulating long-term leaching behavior in a landfill environment demonstrated that 19 out of 24 samples resulted in concentrations less than the TC level. To better understand the phases metals and metalloids were associated with, sequential extraction (Tessier et al., 1979) revealed the importance of the iron oxide surfaces in the leaching process (Shu et al., 2015). The objective of this study was to develop a model that can predict the leachability of metals in paint waste and to apply this model in the field on waste generated during bridge rehabilitation. This research builds on previous work (Shu, 2014; Shu et al., 2015). Through mechanistic modeling, the iron oxide coating formed on the steel grit surface was assessed for its role in metal mobility. Subsequently, statistically-based models that address leaching from paint waste were developed for field application. Results of this work assist in applying a predictive tool in the field that addresses the mobility of trace metals as well as disposal and management of paint waste during bridge rehabilitation.

2. Materials and methods 2.1. Data collected A total of 117 paint waste samples were obtained from 24 bridges under rehabilitation in seven regions (Regions 1, 2, 3, 5, 7, 10, and 11) in NYS between October 2010 and November 2011 (Shu, 2014). To investigate metal distribution in the paint waste, eight Resource Conservation and Recovery Act (RCRA) metals (i.e., As, Ba, Cr, Cd, Pb, Hg, Se, and Ag) along with Fe, Ca, Ti, and Zn were analyzed with the NITON XL3t-600 series FP-XRF following Method 6200 (U.S. EPA, 1998) using either Soil Mode (metal concentrations < 2 wt.%) or Mining Mode (metal concentrations P 2 wt.%) (Shu, 2014). Additional details including detection limits for the FP-XRF in these two modes are provided in Supporting Information (Tables S1 and S2). Results from using FP-XRF on eight paint samples with Pb concentrations ranging from 210 to 168,093 mg/kg were compared to applying digestion with hydrogen fluoride (HF) (Method 3052) (U.S. EPA, 2004) followed by inductively coupled plasma mass spectroscopy (ICP-MS) (Method 6020A) (U.S. EPA, 2007). FP-XRF correlated (R = 0.85–0.98) (Details are provided in Supporting Information) with the ICP-MS analysis (Figs. S1 and S2 in Supporting Information). This work demonstrates the effectiveness of using FP-XRF as a field method to analyze the RCRA metals as well as iron and zinc concentrations in bridge paint waste. In managing paint waste disposal from bridge structures, state DOT agencies are required to investigate waste leaching behavior and determine waste classification. In this study, trace metal concentrations in the leachate solution were detected using ICP-MS (Method 6020A; U.S. EPA, 2007). This instrument was calibrated using National Institute of Standards and Technology (NIST)

Standard Reference Material (SRM). The calibration curve was developed with ten standards covering the range of 2–1700 lg L1 using Multi-Element Solution 2A (Spex Certiprep). Detection limits included As = 0.7 lg L1, Ba = 0.4 lg L1, Cd = 0.5 lg L1, Cr = 0.7 lg L1, Pb = 0.5 lg L1, Se = 0.2 lg L1, Ag = 0.3 lg L1, Fe = 2.5 lg L1, Zn = 1.0 lg L1, and Hg = 2.0  103 lg L1. Data collected (Shu, 2014; Shu et al., 2015) from studying the leaching behavior were applied in mechanistic modeling, in the analysis of key variables, and then in the resulting statistically basedmodel. 2.2. Mechanistic processes Based on the results from sequential extraction along with leaching studies (Shu et al., 2015), adsorption/desorption and dissolution/precipitation are considered plausible processes. Because ferrihydrite is a dominant surface (Shu et al., 2015) and because of its significant adsorbent characteristics including a large surface area and high affinity for metal ions, this oxide surface was used in modeling sorption (Kendall, 2003; Apul et al., 2005). The hydrous ferric oxide (HFO) surface has low-affinity and high-affinity sites (Dzombak and Morel, 1990), represented as FeWOH and FeSOH, respectively. The weak-affinity site density of 0.2 mol/mol Fe and the high-affinity site density of 0.005 mol/mol Fe were used in this study (Dzombak and Morel, 1990; Meima and Comans, 1998; Apul et al., 2005). Dominant surface complexes with ferrihydrite have been shown to include FeOPb+ and FeOCrOH+ (Supporting Information, Tables S3 and S4) (Dzombak and Morel, 1990; Kendall, 2003; Jing et al., 2006). In this study, two models were trialed for Pb and Cr leaching over the pH of 4.5–7. Sorption onto the ferrihydrite surface was considered using the surface complexation diffuse layer model (DLM) coupled with 2-pK formalism. Surface acid–base reactions and equilibrium constants used in the DLM are from the compilation of studies reviewed in Dzombak and Morel (1990). Inputs such as background electrolyte, adsorbate, and sorbent concentrations were based on the TCLP leaching experiments, XRF analyses, and sequential extraction results (Table 1) (Shu et al., 2015). An ionic strength of 0.1 was applied in the calculations to simulate the leaching conditions considered. The second modeling approach involved precipitation/dissolution. Barnes and Davis (1996) demonstrated PbCO3(s) (cerrusite) (pH < 8) and Pb3(OH)2(CO3)2(s) (hydrocerussite) (pH P 8) as the dominant Pb minerals in leadbased paint. In fact, the primary lead compound used in paints was white lead (2PbCO3Pb(OH)2(s)) with concentrations as great as 40 wt.% of dry paint (Gooch, 1993). Lead chromate (chrome yellow PbCrO4(s)) was used in (colored) paint at 5–7% (Clark, 1976), while lead tetraoxide (red lead Pb3O4, Pb2O4, PbO22PbO) was also a component of paints (Clark, 1976; Gooch, 1993). Boy et al. (1995) found Cr(OH)3(s) (chrome oxide green) as the dominant phase when they investigated chromium stabilization in paint waste with Portland cement and blast furnace slag. Jing et al. (2006) further demonstrated that Cr(OH)3(s) and Ca2Cr2O5(s) were the dominant phases when they evaluated Cr leaching behavior in

Table 1 Total and leached concentrationsa used in surface complexation modeling. Pb

Cr

Mean 1 b

Total Concentrations in the paint (mg kg ) Leached concentrations over TCLP procedure (M or mol/L) Desorbed metal concentrations over TCLP procedure (% of total metal) a b c

Shu et al. (2014, 2015). The values are based on results from FP-XRF. BDL refers to below detection limit.

Min 4

4.6  10 1.4  106 0.011

5 BDLc BDL

Max 5

1.7  10 1.0  105 0.048

Mean

Min

Max

3018 1.5  105 0.62

21 BDL BDL

1  104 1.8  104 3.7

600

Z. Shu et al. / Chemosphere 134 (2015) 598–605

the solidified soil. In this research, the solubility of the following minerals was considered: PbCrO4(s) (lead chromate), Pb3O4(s) (red lead), PbCO3(s) (cerrusite), Pb(OH)2(s), Pb3(OH)2(CO3)2(s) (hydrocerussite), PbO(s) (litharge), and Cr(OH)3(s) (or Cr2O3(s) chrome oxide green). Based on paint formulations and mineral thermodynamic stability (Baes and Mesmer, 1976; Ball and Nordstrom, 1991; Schecher and McAvoy, 1992; Marani et al., 1995; Stumm and Morgan, 1996) the following minerals were applied in this study: PbCO3(s) (pKso = 1013.13) (Benjamin, 2002), Pb3(OH)2(CO3)2(s) (pKso = 1045.46) (Benjamin, 2002), and Cr(OH)3(s) (or Cr2O3(s)) (pKso = 1033.13) (Benjamin, 2002) (Supporting Information, Tables S3 and S4). 2.3. Principal component analysis (PCA) and statistical modeling Mechanistic modeling was first invoked to address dominant processes. Data obtained from macroscopic experiments were applied (Shu et al., 2015). Because the main objective of this study was to develop a model that can predict the leachability of metals from paint waste in the field and generated during bridge rehabilitation, mechanistic processes help in resolving critical variables. Our approach involves developing a statistically-based model based on field data obtained from FP-XRF. Specifically, PCA (Montgomery et al., 2009) is used to identify and reduce the dimensionality of data, from which a model is developed to predict metal leaching and therefore classification of the paint waste. The analysis involves a mathematical procedure that transforms a number of potentially correlated variables into a smaller number of uncorrelated variables (principal components (PCs)) (Torrecilla et al., 2009), which are linear combinations of the original variables. In this study, correlation-based PCA (Torrecilla et al., 2009) was applied to understand variables (i.e., As, Ba, Cr, Cd, Fe, Pb, Hg, Ag, Se, Zn, Ti, and Ca) that play a significant role in addressing total variances in the statistical model. Based on an understanding of mechanistic processes along with PCA analysis, leaching data were subjected to multivariate statistical analyses to evaluate the effect of statistically significant variables on metal leaching (additional details are discussed in Sections 3.3 and 3.4). Multivariate statistical approaches such as multiple linear regression analysis (MLRA) are used to determine the significance of specific parameters among the datasets. The total metal concentrations from FP-XRF analysis are applied as inputs in the leaching models as follows: (i) Multivariate regression is tested in the first step of modeling. Leached metal = a + b(total Ca) + c(total Fe) + d(metali,total) + e(metali+1,total) + . . . + n(metaln,total). (ii) Box–Cox transformation (Kutner et al., 2005) is applied according to the residual analysis if necessary, where leached metal y is transformed to yk: (Leached metal)k = a + b(total Ca) + c(total Fe) + d(metali,total) + e(metali+1,total) + . . . + n(metaln,total). (iii) Transformed model: Log(leached metal) = a + blog(total Ca) + clog(total Fe) + dlog(metali,total) + elog(metali+1,total) + . . . + nlog(metaln,total). where the leached metal concentration is in mg L1 and the total metal concentration is in mg kg1 based on FP-XRF. Coefficients a–n are determined by a series of statistical analyses. Specifically, the F-test (ANOVA), goodness of fit, and t-test were applied to evaluate the significance of regression, individual coefficients, and subset of coefficients (F-test) (details are provided in Supporting Information). Based on the PCA analysis, highly correlated metals were removed from the model to reduce the number of variables. The significance of a restricted model was assessed using the partial F-test. Furthermore, residual plots were applied to investigate

the reasonableness of the restricted model. The model in (iii) corresponds to the transformed model in (i). Total Fe reflects the presence of iron oxides in the model. A fraction of metal was also observed to be associated with carbonates in the paint waste, which may be due to the application of calcite (CaCO3(s)) (12 wt.% in the paint waste (Shu, 2014) as an extender (supplementary pigments) in the paint). The dissolution of calcite (CaCO3(s)) affects the pH of the system during the leaching procedure. Consequently, the Ca concentration is expected to be an important variable in the model. Other groups of metals (metali,total) such as Zn and Ti present at elevated concentrations in the paint waste may be important variables in the model accounting for their potential influence on leaching as a result of adsorption competition and in the case of Ti (TiO2) surface interactions (details are provided in Sections 3.3 and 3.4). 3. Results and discussion 3.1. Metal association with iron oxides form on the steel grit surface The TCLP results revealed that metal concentrations ranged from less than 0.5 lg L1 (detection limit) to 1.46 mg L1 for Pb, less than 0.7 lg L1 (detection limit) to 9.52 mg L1 for Cr (Shu et al., 2015), less than 0.4 lg L1 (detection limit) to 9.60 mg L1 for Ba, and from 20.4 to 1307 mg L1 for Zn (Figs. S5 and S6). Shu et al. (2015) found the iron oxide coatings on the steel grit surface were an important phase in the leaching process. As a result, a maximum of 22.6 mg L1 of Pb and 9.52 mg L1 of Cr leached from the paint waste (Shu et al., 2015) in the long-term leaching experiments (i.e., MEP). It is important to note that Cr(III) is expected to be the dominant form in the leaching procedure based on the previous studies conducted (Shu et al., 2015). Briefly, the trend observed during the leaching studies revealed that the Cr sorption capacity was consistent with Cr(OH)2+ sorption to the iron oxide formed on the steel grit surface (Luther et al., 2013; Shahriari et al., 2014). The formation of the iron oxide coatings on steel substrates has been observed by many researchers (Stipp et al., 2002; Kelly et al., 2007; De La Fuente et al., 2011; Lu et al., 2011); ferrihydrite and lepidocrocite are typical morphological structures observed under atmospheric conditions (Monnier et al., 2010; De La Fuente et al., 2011; El Hajj et al., 2013). Lu et al. (2011) hypothesized that Pb2+ was first adsorbed onto the nanometer-sized, metastable, iron oxyhydroxide polymers of 2-line ferrihydrite. As these nano-particles assembled into larger particles, Pb2+ was trapped in the iron oxyhydroxide structure and re-arranged to form a solid solution. On the other hand, the metastable ferrihydrite transforms into the more stable and crystalline Fe(III) oxides goethite [a-FeO(OH); KSO = 1041] (Das et al., 2011), hematite (a-Fe2O3; KSO = 1043) (Das et al., 2011), and lepidocrocite (c-FeOOH; KSO = 1039) (Lu et al., 2011) at neutral pH (Jambor and Dutrizac, 1998; Shaw et al., 2004). The crystallization process may be inhibited through the presence of impurities such as adsorbates resulting in ferrihydrite being stable for long periods of time (Schwertmann and Murad, 1983; Axe and Anderson, 1995). 3.2. Modeling of metal leaching Monodentate and bidentate surface complexes have been applied in modeling adsorption in several studies (Villalobos et al., 2001; Trivedi et al., 2003; Xu et al., 2006a,b). To resolve mechanistic processes in this study, modeling revealed monodentate surface complexes provided the best fit. FeOH0 represents a surface hydroxyl group and the surface reaction for either Pb2+ or Cr(OH)2+ (Me2+) on HFO is the following:

601

Z. Shu et al. / Chemosphere 134 (2015) 598–605

B Fe  OH0 + Me2+ = B Fe  OMe+ + H+ Morel, 1990).

log KMe (Dzombak and

In our study, up to 0.048 wt.% of the total Pb leached from the paint waste for the pH range of 4.5–7 (Fig. 1). Although XRD analysis reveals the presence of PbCrO4 and Pb3O4 in the paint waste (Shu, 2014), these forms of Pb are associated with residual phase. Based on lead solubility (Fig. S3 Supporting Information), PbCO3 (cerussite) is considered as the dominant solid for pH less than 8, while Pb3(OH)2(CO3)2 (hydrocerussite) controls solubility for pH greater than or equal to 8 (Barnes and Davis, 1996). However, precipitation did not capture leaching results from the TCLP procedure and overestimated the dissolved lead in the solution. The diffuse layer model (log KMe = 4.65 for Pb) (Dzombak and Morel, 1990) on the other hand captured 90% of the data within the 95% confidence level, suggesting that sorption to the iron oxides plays an important role for Pb. Interestingly, leaching from pH 4.5 to 5.0 resulted in less desorption than predicted with the DLM. This result is likely attributed to the continuum of sorption reactions between specific adsorption and surface precipitation/coprecipitation of the Pb mineral on the oxide coated steel grit surface. Apul et al. (2005) and Meima and Comans (1998) found both adsorption and coprecipitation are important processes that affect Pb leaching from the incinerator bottom ash and steel slag. Similar with the Pb work, Cr precipitation overestimated aqueous concentrations in the system. Although chromium hydroxide Cr(OH)3(s) (or Cr2O3(s)) was observed in paint waste using XRD (Shu, 2014), based on its solubility (Supporting Information, Fig. S4) Cr(OH)3(s) may influence leaching for pH greater than 6.8 (Fig. 1). The desorbed Cr concentration decreased as pH increased from 4.5 to 7 and adsorption captured 90% of the desorption data within the 95% confidence interval. The DLM model adequately predicted the observed leaching (Fig. 1) in the paint waste suggesting adsorption/desorption from the iron oxide surface is the dominant process. This result is similar to that found by Jing et al. (2006) where the leaching behavior of Cr was dominated by the adsorption on iron oxides in the solidified soil. In addition, the trend is consistent with the trivalent Cr leaching (i.e., cation

6

0.1

A

Pb adsorption with ferrihydrite

0.08 95%

prediction 0.06 interval

4

0.04

Pb adsorption to ferrihydrite

0.02 0 4.5

5

5.5

6

6.5

Desorbed %

2 Pb(CO )

Pb (OH) (CO ) 3

3 2(s)

2

3 2(s)

7

90% of data were captured within 95% PI

0 5

5 4

3

3

2

Cr(III) adsorption to ferrihydrite

Cr(OH)

3(s)

1 0 4.5

0

90% of data were captured within 95% PI

95% prediction interval

2

1

5

0

5.5

6

6.5

2

7

4

6

8

10

12

14

pH Region 1

Region 2

3.3. Principal component analysis Applying PCA to the raw data showed essentially three main constituent axes with eigenvalues greater than 1 (Table 2), together explaining 85% of the data variance. The first PC represents 55.3% of the total variability and strong relationships with the associated total concentration present in the samples. The positive weights (i.e., As, Cr, Cd, Pb, and Ag) are associated with the bridge blasted with surface preparation (SP) standard SSPC (The Society for Protective Coatings) SP-6, while the negative weights (i.e., Hg, Se, and Zn) were related to SP-10 during bridge rehabilitation. SP-6 (Commercial Blast Cleaning) (NYSDOT, 2008) has been applied to bridges in New York State before 2006, where paint and rust from steel were removed to a remaining residual of 33% of the total removal area. After 2006, SP-10 (Near White Blast Cleaning) was required in the blasting procedure (NYSDOT, 2008) for all regions in NY. SP-10 restricts the visible residues remaining on the bridge surface to 5% of the total removal area. The second PC in PCA revealed the influence of Fe in the paint waste demonstrating that it is an important factor impacting model variability in paint waste. Steel grit (96% iron) is applied as blasting material during the paint removal procedure. Although magnetic separation is used to remove steel grit particles, the

Table 2 Principal component analysis of total metals in paint waste.

adsorption B Cr(III) to ferrihydrite

4

adsorption/desorption). Compared to Pb, Cr (as great as 9.52 mg L1) leached to a greater degree than lead (as great as 1.46 mg L1). Because of the difference between Pb and Cr surface complexation constants KMePb = 104.65 > KMeCr(III) = 102.11 (intrinsic constant) (Dzombak and Morel, 1990) these results are expected. Adsorption and precipitation mechanisms were also investigated for Ba and Zn (details are provided in the Supporting Information Figs. S5 and S6). Because of the heterogeneity of solid phase in paint sample, adsorption, coprecipitation as well as precipitation are important processes that affect Ba and Zn leaching from the waste. These results are in agreement with other studies on the leaching behavior in soils (Dijkstra et al., 2009) and coal ash (Mudd et al., 2004). While mechanistic modeling (i.e., adsorption and precipitation) describes metal leaching, this approach requires macroscopic studies to predict behavior. The purpose of this study is to develop a predictive tool for field application. Therefore, total metal concentrations obtained from FP-XRF are applied to predict waste behavior and waste classification. PCA was used to isolate variables (i.e., As, Ba, Cr, Cd, Fe, Pb, Hg, Ag, Se, Zn, Ti, and Ca) that play a significant role in addressing total variances in the statistical model.

Region 3

Region 5

Region 7

Region 10

Region 11

Fig. 1. Desorbed Pb (A) and Cr (B) in the presence of steel grit associated with paint waste as a function of pH after 18 h using the TCLP. CrT = 1.8  104 M, PbT = 1.0  105 M, FeT = 0.07 M, ionic strength = 0.1 M, surface area = 600 m2/g, KMePb = 104.65 (Dzombak and Morel, 1990), KMeCr = 102.11 (Dzombak and Morel, 1990), K soPbCO3 = 1013.13 (Benjamin, 2002), K soPb3 ðOHÞ2 ðCO3 Þ2 = 1045.46 (Benjamin, 2002), and K soCr2 O3 = 1033.13 (Benjamin, 2002).

Variable

PC 1

PC 2

PC 3

As Ba Ca Cd Cr Fe Pb Hg Ag Se Ti Zn Eigenvalue Proportion Cumulative

0.829 0.658 0.655 0.877 0.854 0.375 0.863 0.698 0.886 0.787 0.297 0.851 6.63 0.553 0.553

0.433 0.429 0.165 0.342 0.432 0.705 0.448 0.477 0.358 0.518 0.665 0.418 2.64 0.22 0.773

0.22 0.19 0.672 0.243 0.004 0.437 0.027 0.191 0.228 0.127 0.156 0.034 1.00 0.076 0.849

Absolute value of correlation greater than 0.66 is highlighted in bold.

602

Z. Shu et al. / Chemosphere 134 (2015) 598–605

smallest fraction remains with paint waste (Shu, 2014). Iron oxides form on the steel grit surface providing a highly reactive surface for metal sorption that further controls the degree of metal leaching from the paint waste. In addition to Fe, Ti was also observed to be an important factor in the paint waste based on PC2 (Table 2). This result is attributed to the application of TiO2 as extenders in paint, which may provide a sorption surface for trace metals as well. The third PC exhibited the effect of Ca in the paint waste. The observed behavior of Ca is consistent with Andra et al. (2011), where it was found to affect mobilization of Pb from alkaline soils in San Antonio, TX. Ca is attributed to the application of calcite (CaCO3(s)) (12 wt.% in paint waste (Shu, 2014)) as an extender (supplementary pigments) in the paint (Strivens and Lambourne, 1999). The release of the CaCO3(s) from paint waste does not directly affect the metal leaching. However, the dissolution of the CaCO3(s) results in an increase in pH and alkalinity during the leaching procedure. Because metal leaching is a function of pH, dissolution of the CaCO3(s) reflects this pH change and hence metal leaching. Therefore, CaCO3 is expected to be an important factor in the leaching model. Using PCA analysis, the most important factors accounting for total variability are the surface preparation standard (reflected in Table 2 (PC1)) and steel grit (iron) remaining in the paint waste (reflected in Table 2 (PC2)). Other factors such as Ti and Ca (reflected in Table 2 (PC2 and PC3)) also impact the variability observed. Given the mechanistic processes (Section 3.2), metal leaching depends on (a) total metal concentrations (group of metals based on surface preparation); (b) Fe oxides providing a highly reactive surface for metal sorption; (c) CaCO3(s) affecting the pH and alkalinity; (d) other metals, Zn and Ti, in the paint waste.

100

In this study, a MLRA modeling approach (discussed in Section 2.3) was applied to establish a single correlation between the dependent (TCLP and MEP results) and several independent variables (i.e., As, Ba, Cr, Cd, Fe, Pb, Hg, Ag, Se, Zn, Ti, and Ca). Because the surface preparation standard was observed to be an important factor in PCA analysis, leaching results were sorted into groups with respect to the two methods: SP-6 and SP-10 (Supporting Information, Table S5). (The surface preparation methods are the ones applied in the previous rehabilitation efforts in NYS, which determine the residual waste remaining on the bridge.) In sorting the data, visually, the threshold Fe concentration is estimated to range from 17 to 24 wt.% (Supporting Information, Fig. S7). Therefore, a range of threshold values at 1% intervals were tested to examine the significance of regression. Comparing the mean square errors (MSE) (regime 1 and regime 2) obtained from each threshold considered, the breakpoint (regime 1 and regime 2) with the smallest MSE was determined as the select threshold for this study. Consequently, two regime regressions were obtained with Fe concentrations (620% or >20 wt.%) in the paint waste. A series of residual plots revealed the reasonableness of the two regime models (details are provided in Supporting Information). In addition, the Chow test (F test on extra sum of squares) (Chow, 1960) indicated that two regime models were more adequate compared to a simpler model (single regime) (details are provided in Supporting Information). Based on the two regime models developed, one formulation with an indicator variable is presented as follows:

100

A

Pb Total Fe 20% SP-6

10

-1

Predicted leached concentration (mg L )

TC level

TC level

1

1

0.1

0.1

0.01

0.01

0.001

0.001

0.0001 0.0001

0.001

B

log y = 0.99 log x + 0.0054 R² = 0.68

log y = 1.0 log x - 0.0023 R² = 0.57

0.01

0.1

1

10

10

0.0001 0.0001

100

0.001

Pb SP-10 TC level log y = 0.98 log x + 0.0012 R² = 0.84

0.01

0.1

1

10

100

C

1

0.1

0.01

0.001 0.001

0.01

0.1

1

10 -1

Observed Leached Pb (mg L )

TCLP samples: MEP sampbles:

Reigon 1

Region 2

Region 3

Region 5

Region 7

Region 10

Region 11

Region 1

Region 2

Region 3

Region 5

Region 7

Region 10

Region 11

Fig. 2. Comparison of the results from predicted and observed leached Pb concentrations. The samples represent the TCLP and first day of the MEP extraction conducted on the paint waste samples. Bridges were blasted to (A) surface preparation SP-6 with total Fe concentration less than or equal to 20 wt.%, number of observations N = 20; (B) SP6 with total Fe concentration greater than 20 wt.%. N = 28; and (C) SP-10 N = 11. TC level for Pb is 5 mg L1. The dash line represents the 95% prediction interval.

603

Z. Shu et al. / Chemosphere 134 (2015) 598–605

(iv) Y ¼ y1  ð1  kÞ þ y2  k

when Fe  20%; k ¼ 0 when Fe > 20%; k ¼ 1

k is an indicator variable in the two regimes; y1 and y2 represent the two regime models derived from MLRA in Section 2.3. The statistical models developed (Supporting Information, Tables S6 and S7) for metal leaching demonstrated 96% of the data fall within the 95% confidence level for Pb (R2 0.6–0.9, p 6 0.04), Ba (R2 0.5–0.7, p 6 0.1), and Zn (R2 0.6–0.7, p 6 0.08). However, the regression model obtained for Cr leaching was not significant (R2 0.3–0.5, p 6 0.75) (Table S6), suggesting the regression is not significant for Cr. Significant correlations were observed between predicted and observed metal leaching for bridge samples blasted with SP-10 preparation (R2 = 0.84 for Pb, R2 = 0.70 for Ba, and R2 = 0.71 for Zn) (Fig. 2) (Figs. S8 and S9 in Supporting Information). This result may be attributed to the Cr reactions in the leaching process. Specifically, surface precipitation, 2Cr(OH)+2 + Fe2+ = FeCr2O4(s) + 4H+ is expected in the system (Jing et al., 2006) along with Cr(III) sorption. These processes cannot be quantitatively captured in the statistical modeling effort. In addition, although Cr(III) is the dominant form leached, a small percent may be present as Cr(VI) (Peterson et al., 1996, 1997; Du et al., 2012). Again as a result, these precipitation and complexation reactions were poorly captured in the statistical model. Compared to the bridges cleaned with the SP-6 approach, the bridges blasted using SP-10 revealed lower leached metal concentrations. For example, Pb concentrations ranged from less than 0.5 lg L1 (detection limit) to 0.83 mg L1, which are less than TC level of 5 mg L1. Similarly, these samples revealed less than 0.7 lg L1 (detection limit) to 0.98 mg L1 for Cr, and less than 0.4 lg L1 (detection limit) to 1.06 mg L1 for Ba; these fall below

-1

4. Conclusions and implications Overall, this study provides a predictive tool to estimate the metal leaching in the field without additional laboratory studies

A

Cr Total Fe 20 wt.%). These results are consistent with the importance of iron oxide in surface complexation with trace metals.

12

12

8

8

TC level

4

TC level

4

0

B

Cr Total Fe > 20% SP-6 y = 1.0 x - 0.099 R² = 0.64

0 0

4

8 1.5

12

0

4

Cr SP-10 y = 1.0 x - 0.0016 R² = 0.44

8

12

C

1

0.5

0

-0.5 0

0.5

1

1.5 -1

Observed Leached Cr (mg L )

TCLP samples: MEP samples:

Reigon 1 Region 1

Region 2 Region 2

Region 3 Region 3

Region 5

Region 7

Region 10

Region 11

Region 5

Region 7

Region 10

Region 11

Fig. 3. Comparison of the results from predicted and observed leached Cr concentrations. The samples represent the TCLP and first day of the MEP extraction conducted on the paint waste samples. Bridges were blasted to (A) surface preparation SP-6 with total Fe concentration less than or equal to 20 wt.%, number of observations N = 21; (B) SP6 with total Fe concentration greater than 20 wt.%. N = 20; and (C) SP-10 N = 8. TC level for Cr is 5 mg L1. The dash line represents the 95% prediction interval.

604

Z. Shu et al. / Chemosphere 134 (2015) 598–605

(e. g., TCLP). Modeling resulted in 96% of the data falling within the 95% confidence level for Pb (R2 0.6–0.9, p 6 0.04), Ba (R2 0.5–0.7, p 6 0.1), and Zn (R2 0.6–0.7, p 6 0.08). However, the regression model obtained for Cr leaching was not significant (R2 0.3–0.5, p 6 0.75). The statistical models are based exclusively on data collected from bridges undergoing rehabilitation where steel grit was used as the blasting material. Therefore, for other state DOTs working with similar structures and rehabilitation procedures, this research may be beneficial in supporting a field analysis for waste classification. Results of this work may assist DOT agencies with applying an approach and predictive tool in the field that addresses the mobility of trace metals as well as disposal and management of paint waste during bridge rehabilitation. Acknowledgements The authors would like to thank NYSDOT – United States for providing funding for this research. The contents of this report reflect the views of the authors who are responsible for the facts and the accuracy of the data presented herein. The contents do not necessarily reflect the official views or policies of the NYSDOT or the Federal Highway Administration.

Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.chemosphere. 2014.10.081. References Andra, S.S., Datta, R., Reddy, R., Saminathan, S.K., Sarkar, D., 2011. Antioxidant enzymes response in vetiver grass: A greenhouse study for chelant-assisted phytoremediation of lead-contaminated residential soils. CLEAN–Soil, Air, Water 39, 428–436. Appleman, B.R., 1997. Lead-based paint removal for steel highway bridges. Transportation Research Board. Apul, D.S., Gardner, K.H., Eighmy, T.T., Fällman, A.-M., Comans, R.N., 2005. Simultaneous application of dissolution/precipitation and surface complexation/surface precipitation modeling to contaminant leaching. Environ. Sci. Technol. 39, 5736–5741. Axe, L., Anderson, P.R., 1995. Sr diffusion and reaction within Fe oxides: evaluation of the rate-limiting mechanism for sorption. J. Colloid Interface Sci. 175, 157– 165. Baes, C.F., Mesmer, J.R.E., 1976. The Hydrolysis of Cations. Wiley, New York. Ball, J.W., Nordstrom, D.K., 1991. User’s manual for WATEQ4F, with revised thermodynamic data base and test cases for calculating speciation of major, trace, and redox elements in natural waters. Barnes, G.L., Davis, A.P., 1996. Dissolution of lead paint in aqueous solutions. J. Environ. Eng. 122, 663–666. Benjamin, M.M., 2002. Water Chemistry. McGraw-Hill, New York. Boy, J.H., Race, T.D., Reinbold, K.A., Bukowski, J., Zhu, X., 1995. Chromium stabilization chemistry of paint removal wastes in Portland cement and blast furnace slag. Hazard. Waste Hazard. Mater. 12, 83–95. Chow, G.C., 1960. Tests of equality between sets of coefficients in two linear regressions. Econometrica 28, 591–605. Clark, H.A., 1976. Paint compositions. Google Patents. Das, S., Hendry, M.J., Essilfie-Dughan, J., 2011. Transformation of two-line ferrihydrite to goethite and hematite as a function of pH and temperature. Environ. Sci. Technol. 45, 268–275. De La Fuente, D., Díaz, I., Simancas, J., Chico, B., Morcillo, M., 2011. Long-term atmospheric corrosion of mild steel. Corros. Sci. 53, 604–617. Dijkstra, J.J., Meeussen, J.C., Comans, R.N., 2009. Evaluation of a generic multisurface sorption model for inorganic soil contaminants. Environ. Sci. Technol. 43, 6196– 6201. Du, J., Lu, J., Wu, Q., Jing, C., 2012. Reduction and immobilization of chromate in chromite ore processing residue with nanoscale zero-valent iron. J. Hazard. Mater. 215, 152–158. Dzombak, D.A., Morel, F.M., 1990. Surface Complexation Modeling: Hydrous Ferric Oxide. Wiley, New York. El Hajj, H., Abdelouas, A., El Mendili, Y., Karakurt, G., Grambow, B., Martin, C., 2013. Corrosion of carbon steel under sequential aerobic–anaerobic environmental conditions. Corros. Sci. 76, 432–440. Gooch, J.W., 1993. Lead-Based Paint Handbook. Plenum Press, New York.

Jambor, J.L., Dutrizac, J.E., 1998. Occurrence and constitution of natural and synthetic ferrihydrite, a widespread iron oxyhydroxide. Chem. Rev. 98, 2549– 2586. Jing, C., Liu, S., Korfiatis, G.P., Meng, X., 2006. Leaching behavior of Cr(III) in stabilized/solidified soil. Chemosphere 64, 379–385. Kelly, S.D., Lu, P., Bolin, T., Chattopadhyay, S., Newville, M.G., Shibata, T., Zhu, C., 2007. Molecular structure of lead(II) coprecipitated with iron(III) oxyhydroxide. Dev. Earth Environ. Sci. 7, 67–93. Kendall, D.S., 2003. Toxicity characteristic leaching procedure and iron treatment of brass foundry waste. Environ. Sci. Technol. 37, 367–371. Kutner, M.H., Nachtsheim, C.J., Neter, J., Li, W., 2005. Applied linear statistical models. Lu, P., Nuhfer, N.T., Kelly, S., Li, Q., Konishi, H., Elswick, E., Zhu, C., 2011. Lead coprecipitation with iron oxyhydroxide nano-particles. Geochim. Cosmochim. Acta 75, 4547–4561. Luther, S., Brogfeld, N., Kim, J., Parsons, J.G., 2013. Study of the thermodynamics of chromium(III) and chromium(VI) binding to iron(II/III) oxide or magnetite or ferrite and magnanese(II) iron(III) oxide or jacobsite or manganese ferrite nanoparticles. J. Colloid Interface Sci. 400, 97–103. Marani, D., Macchi, G., Pagano, M., 1995. Lead precipitation in the presence of sulphate and carbonate: testing of thermodynamic predictions. Water Res. 29, 1085–1092. Meima, J.A., Comans, R.N., 1998. Application of surface complexation/precipitation modeling to contaminant leaching from weathered municipal solid waste incinerator bottom ash. Environ. Sci. Technol. 32, 688–693. Missouri Department of Natural Resources, 2006. Environmental Regulations for Paint Removal on Outdoor Structures, Hazardous waste program fact sheet, 6/ 2006. Minnesota Department of Transportation (MnDOT), 2004. Air quality and waste management for the removal of paint (using dry abrasive blasting) on steel bridge structures. Monnier, J., Neff, D., Réguer, S., Dillmann, P., Bellot-Gurlet, L., Leroy, E., Foy, E., Legrand, L., Guillot, I., 2010. A corrosion study of the ferrous medieval reinforcement of the Amiens cathedral. Phase characterisation and localisation by various microprobes techniques. Corros. Sci. 52, 695–710. Montgomery, D.C., Runger, G.C., Hubele, N.F., 2009. Engineering Statistics. John Wiley & Sons. Mudd, G.M., Weaver, T.R., Kodikara, J., 2004. Environmental geochemistry of leachate from leached brown coal ash. J. Environ. Eng. 130, 1514–1526. New York State Department of Transportation (NYSDOT), 2008. Standard Specification. Peterson, M.L., Brown, G.E., Parks, G.A., 1996. Direct XAFS evidence for heterogeneous redox reaction at the aqueous chromium/magnetite interface. Colloids Surf., A 107, 77–88. Peterson, M.L., Brown, G.E., Parks, G.A., Stein, C.L., 1997. Differential redox and sorption of Cr(III/VI) on natural silicate and oxide minerals: EXAFS and XANES results. Geochim. Cosmochim. Acta 61, 3399–3412. Schecher, W.D., McAvoy, D.C., 1992. MINEQL+: a software environment for chemical equilibrium modeling. Comput. Environ. Urban Syst. 16, 65–76. Schwertmann, U., Murad, E., 1983. Effect of pH on the formation of goethite and hematite from ferrihydrite. Clays Clay Miner. 31, 277–284. Shahriari, T., Bidhendi, G.N., Mehrdadi, N., Torabian, A., 2014. Effective parameters for the adsorption of chromium(III) onto iron oxide magnetic nanoparticle. Int. J. Environ. Sci. Technol. 11 (2), 349–356. Shaw, S., Benning, L., Terrill, N., Davidson, L., 2004. Nucleation and growth of iron oxyhydroxide nanoparticles from solution: an in situ time-resolved small angle X-ray scattering (SAXS) study. Geochem. Cosmochim. Acta 68, A158. Shu, Z., 2014. Field methods for rapidly characterizing contaminant mobility inpaint waste during bridge rehabilitation. Ph.D. dissertation, New Jersey Institute of Technology, Newyork, NJ. Shu, Z., Axe, L., Jahan, K., Ramanujachary, K.V., 2015. Metal Leaching from the Bridge Paint Waste in the Presence of Steel Grit. Chemosphere 119, 1105–1112. Stipp, S., Hansen, M., Kristensen, R., Hochella, M., Bennedsen, L., Dideriksen, K., Balic-Zunic, T., Leonard, D., Mathieu, H.-J., 2002. Behaviour of Fe-oxides relevant to contaminant uptake in the environment. Chem. Geol. 190, 321–337. Strivens, T., Lambourne, R., 1999. Paint and Surface Coatings: Theory and Practice. Woodhead. Stumm, W., Morgan, J., 1996. Aquatic chemistry, chemical equilibra and rates in natural waters. Environ. Sci. Technol.. Tessier, A., Campbell, P.G., Bisson, M., 1979. Sequential extraction procedure for the speciation of particulate trace metals. Anal. Chem. 51, 844–851. Torrecilla, J.S., García, J., Rojo, E., Rodríguez, F., 2009. Estimation of toxicity of ionic liquids in Leukemia Rat Cell Line and Acetylcholinesterase enzyme by principal component analysis, neural networks and multiple lineal regressions. J. Hazard. Mater. 164, 182–194. Trivedi, P., Dyer, J.A., Sparks, D.L., 2003. Lead sorption onto ferrihydrite. 1. A macroscopic and spectroscopic assessment. Environ. Sci. Technol. 37, 908–914. U.S. EPA, 1986. Method 1320 Multiple Extraction Procedure, Revision 0, Washington, DC. U.S. EPA, 1992. Method 1311 Toxicity Characteristic Leaching Procedure (TCLP), SW-846. Federal Register, 55 (March 29), Washington, DC. U.S. EPA, 1998. Method 6200, field portable X-ray fluorescence spectrometry for the determination of elemental concentrations in soil and sediment (EPA/600/R-97/ 150, Revision 0), Retrieved 01.11.07. U.S. EPA, 2004. Method 1310B Extraction Procedure (EP) Toxicity Test Method and Structural Integrity Test, Revision 2, Washington, DC.

Z. Shu et al. / Chemosphere 134 (2015) 598–605 U.S. EPA, 2007. Method 6020A inductively coupled plasma-mass spectrometry. SW846 (Chapter 3). Villalobos, M., Trotz, M.A., Leckie, J.O., 2001. Surface complexation modeling of carbonate effects on the adsorption of Cr(VI), Pb(II), and U(VI) on goethite. Environ. Sci. Technol. 35 (19), 3849–3856. Xu, Y., Boonfueng, T., Axe, L., Maeng, S., Tyson, T., 2006a. Surface complexation of Pb(II) on amorphous iron oxide and manganese oxide: spectroscopic and time studies. J. Colloid Interface Sci. 299, 28–40.

605

Xu, Y., Axe, L., Yee, N., Dyer, J.A., 2006b. Bidentate complexation modeling of heavy metal adsorption and competition on goethite. Environ. Sci. Technol. 40 (7), 2213–2218.

Field methods for rapidly characterizing paint waste during bridge rehabilitation.

For Department of Transportation (DOT) agencies, bridge rehabilitation involving paint removal results in waste that is often managed as hazardous. He...
492KB Sizes 1 Downloads 5 Views