Heavy metal contamination in soils of urban highways (comparision between runoff and soil concentration

Heavy metal contamination in soils of urban highways (comparision between runoff and soil concentration

COMPARISON BETWEEN RUNOFF AND SOIL CONCENTRATIONS AT

CINCINNATI, OHIO

DILEK TURER1, J BARRY MAYNARD1∗and J JOHN SANSALONE2

3510 CEBA Bldg, Baton Rouge, LA 70803-6405, U.S.A.

(Received 17 March 2000; accepted 15 November 2000)

Abstract Rainfall runoff from urban roadways often contains elevated amounts of heavy metals

in both particulate and dissolved forms (Sansalone and Buchberger, 1997) Because metals do not degrade naturally, high concentrations of them in runoff can result in accumulation in the roadside soil at levels that are toxic to organisms in surrounding environments This study investigated the accumulation of metals in roadside soils at a site for which extensive runoff data were also available For this study, 58 soil samples, collected from I-75 near Cincinnati, Ohio, were examined using X-ray fluorescence, C-S analyzer, inductively coupled plasma spectroscopy, atomic absorption spec- trometry and X-ray diffraction The results demonstrated that heavy metal contamination in the top

15 cm of the soil samples is very high compared to local background levels The maximum measured amount for Pb is 1980 ppm (at 10–15 cm depth) and for Zn is 1430 ppm (at 0–1 cm depth) Metal content in the soil falls off rapidly with depth, and metal content decreases as organic C decreases The correlation to organic C is stronger than the correlation to depth The results of sequential soil extraction, however, showed lower amounts of Pb and Zn associated with organic matter than was expected based on the correlation of metals to % organic C in the whole soil Measurement of organic

C in the residues of the sequential extraction steps revealed that much of the carbon was not removed and hence is of a more refractory nature than is usual in uncontaminated soils Cluster analysis of the heavy metal data showed that Pb, Zn and Cu are closely associated to one another, but that Ni and

Cr do not show an association with each other or with either organic C or depth ICP spectroscopy

of exchanged cations showed that only 4.5% of Pb, 8.3% of Zn, 6.9% of Cu and 3.7% of Cr in the soil is exchangeable Combined with the small amounts of metals bound to soluble organic matter, this result shows that it is unlikely that these contaminants can be remobilized into water At this site, clays are not an important agent in holding the metals in place because of low amounts of swelling clays Instead, insoluble organic matter is more important Mass balance calculations for Pb in soil showed that most of the Pb came from exhausts of vehicles when leaded gasoline was in use, and that about 40% of this Pb is retained in the soil This study shows that, highway environments being

a relatively constant source of anthropogenic organic matter as well as heavy metals, heavy metals will continue to remain bound to organic matter in-situ unless they are re-mobilized mechanically Removal of these heavy metals as wind-blown dust is the most likely mechanism Another possibility

is surface run-off carrying the metals into surface drainages, bypassing the soil This study also shows that for those countries still using leaded gasoline, important reductions in Pb contamination of soils can be achieved by restricting the use of Pb additives.

Keywords: copper, flux, highway soils, lead, organic carbon, pavement runoff, zinc

Water, Air, and Soil Pollution 132: 293–314, 2001.

© 2001 Kluwer Academic Publishers Printed in the Netherlands.

Abbreviations: AAS, Atomic Absorption Spectrometry; CEC, Cation Exchange Capacity; EMC,

Event Mean Concentration; ICP, Inductively Coupled Plasma; LECO, C-S Analyzer; meq, equivalent; XRD, X-ray diffraction; XRF, X-ray fluorescence.

at 10 cm as unusually high and they stated that although contamination is limited

to a narrow zone along highways, it is not limited to surface soil They furtherstated that Pb in soil can be leached and mobilized by solutions containing NaCl,for example from road salting

Ward and others (1975) investigated the lead content of soil and vegetationalong a part of a state highway passing through an uninhabited area of New Zea-land They observed an inverse relationship between Pb content of vegetation anddistance from the road, as has been reported from other areas Their analysis showedthat washed vegetation samples contained 70–80% of the Pb levels of unwashedsamples, indicating that the majority of the Pb is relatively immobile They foundthe same fall-off of Pb levels in soil samples with distance from the road Thehighest levels of soil Pb, reaching 160 ppm, were obtained from the top 5 cm ofthe soil (the background level of Pb was 40 ppm) To calculate total excess Pb inthe soil, they plotted the values of excess lead for 1-m by 1-m by 6-cm volumeincrements as a function of distance and found an integrable function which fitthe data: M(x) = M(0) exp [–k(x)1/2] (where M(x) is the excess mass of lead inthe increment at distance x) They estimated the total emitted lead from vehiclesfor that area using the known traffic flow of 6.0±1.0 × 106 vehicles since 1960.When they compared the total amount of emitted Pb (240 g along each meter ofthe road) with the calculated excess in the soil (140 g along each meter of theroad) they concluded that the elevated levels of Pb in the top 6 cm of soil wereprimarily sourced from leaded gasoline Their results suggest that about 60% ofthe Pb emitted is retained by the soils close to the highway

Wheeler and Rolfe (1979) found that lead from automotive sources in roadsidesoil and vegetation follows a double exponential function of the following form: Pb

= A1e−k1D + A2e−k2D The terms A1and A2are linear functions of average dailytraffic volume and the exponents represent different particle sizes Their studiesshowed that larger particles are deposited within 5 m of the roadside and are inert

in the soil whereas small particles are deposited more slowly and are depositedwithin 100 m of the roadside Also they suggested that 72–76% of historical lead

deposited on the soil has been lost from the surface 10 cm of soil The highestamount of lead found was 1225 ppm in soil and 196 ppm in vegetation at 0.3 maway from highways of central Illinois with 8100 vehicles day−1 traffic density Inthis area, background levels, which were 16 ppm for soil and 10 ppm for vegetation,were reached at 50 m from the highways.

Similar results come from Onyari and others (1992), who worked on roadsidesoils in Kenya where lead is still used as a gasoline additive They found that leadconcentrations within Nairobi City varied from 137 to 2196 ppm with a mean of

659 ppm The highest value was measured in the Nairobi hill region, which theyexplained by acceleration of motor vehicles because of the steep nature of the hill.The amount of Pb emitted as a percentage of Pb consumed increases as vehiclespeed increases

Gratani and others (1992) studied the accumulation of Pb in agricultural soiland vegetation along the Fiano-San Cesareo highway in Italy They documented

an increase of Pb values in the soil within the few years that had passed sincethe highway was opened Agricultural soils were found to accumulate more Pb,because the organic matter causes it to be bound to organic exchange sites, reducingits availability for root uptake (Albasel and Cottenie, 1985) They also looked atoak leaves, which showed similar increases in Pb concentration with time

Teichman and others (1993) sampled yards within 1 mile of Interstate 880 inAlameda County, California Surface samples contained an average of 570 ppm Pbwith a maximum of 2030 ppm Subsurface samples from the same sites showed

an average of 620 and a maximum of 1400 ppm Pb 63% of the subsurface Pbconcentrations exceeded corresponding surface concentrations They interpretedthis pattern as indicating that as the use of leaded gasoline decreased, the Pb content

of the upper layers of soil also decreased

There have been a few studies that included other heavy metals like zinc, mium and copper with measurements of lead Gibson and Farmer (1984) applied

cad-a six-step sequenticad-al lecad-aching procedure to soil cad-and street dirt in order to stand environmental mobility and bioavailability of Pb, Zn, Cu and Cd The results

under-of this study revealed that the exchangeable fraction was under-of significantly greaterrelative importance in street dust than in soil, especially for Pb, Zn and Cu Theyreported exchangeable percentages of Pbdust: 13%, Pbsoil: 2%; Zndust: 10%, Znsoil:3%; Cudust: 11%, Cusoil: 2%; Cddust: 27%, Cdsoil: 19% Hamilton and others (1984)investigated levels of Cd, Cu, Pb and Zn in road dust at three sites with differenttraffic usage and surface textures The results showed that amount of contamina-tion increases as traffic density increases They also applied sequential extractionprocedure on size-fractionated dust samples Cd is found as the highest proportion

of total metal in the exchangeable fraction whereas Cu is mainly in the stronglybound organic and residual phases Hewitt and Candy (1990), examined levels of

Pb, Cd and Zn in soil and dust samples collected in and around the city of Cuenca,Ecuador The metal concentrations for the urban environment were considerablyelevated (Pb: 77–970 ppm, Cd: 0.23–0.42 ppm, Zn: 155–1018 ppm) The dominant

TABLE I Event mean concentration data for I-75 experimental site with EPA criteria (Sansalone and Buchberger, 1996)

Violations of EPA discharge criteria in bold.

∗Not EPA priority pollutants.

source for the Pb in urban street dust was shown to be emission of Pb aerosol fromgasoline vehicles Tyre rubber was shown to be the main source for Zn and alsofor Cd, plus some from metal platings on car parts It was also suggested that thepoor condition of road surfaces in Cuenca might have been enhancing tyre wear.Suburban samples taken from 5.5 km away from the city center had lower values

of metals (Pb: 54–109 ppm, Cd: 0.20–0.27 ppm, Zn: 44–120 ppm) Samples takenfrom close to a rural track used by only 100 vehicles per day, had lower values

of Pb: 0.6–15 ppm but not Cd and Zn: (0.29 ppm; 52–541 ppm) The background

Pb obtained from Rio Mazan valley was very low: 0.02–9 ppm The Cd levelshowever were not significantly different from those found in the other areas (0.05–0.5 ppm) They suggested that the influence of vehicular emission of Cd was muchmore localized than it was for Pb, probably due to the emission of Cd as very largeparticles that are transported only short distances

In these studies, the main source for Pb in the soils was shown to be leaded oline in highway vehicles Also all the analyses showed that the amount of heavymetal contamination decreased with depth and with distance from the highway.Although some countries like the U.S prohibit the use of leaded gasoline there aremany other countries that continue using leaded gasoline in their transportation.Even for the U.S the problem of heavy metal contamination has not been elimin-ated Sansalone and Buchberger (1997) sampled lateral pavement sheet flow from

gas-a study gas-aregas-a with gas-an gas-aregas-a of 15× 20 m on I-75 in Cincinnati, during five rainfallevents in 1995 Their results showed that the event mean concentrations (EMC)

of Zn and Cu exceeded surface quality discharge standards for all rainfall eventsand that Pb had two and Cd had three excedences (Table I) They also investigatedpartitioning of metals and solids in storm water Their results indicated that Zn, Cd,

Cu were mostly in dissolved form whereas Pb, Fe and Al were particulate-bound

in storm water

In the current study we use the same site as Sansalone and Buchberger (1997) toexamine how much of this runoff of heavy metals gets transferred to the soil Wehave also attempted to determine the mechanisms that control both the retentionand remobilization of metals To do so, we have included information about claymineralogy and organic carbon content of the soil samples taken from the same sitewhere Sansalone and Buchberger carried out their work Also the work presentedhere goes one step further than previous studies in that it makes mass balancecalculations for Zn, Cu, Ni and Cr in addition to Pb, calculations made possible bythe availability of runoff data for the same site

2 Methods

2.1 SAMPLING LOCATION

The samples were collected along I-75, a heavily traveled north-south interstate

in Cincinnati (Figure 1) 156 670 vehicles were counted per day in 1994 (ODOT,1999) The soils are clay-rich and are visually uniform both laterally and withdepth Some 1961–1990 climate characteristics of Cincinnati (Climate DiagnosticCenter, 1999) are

Mean annual temperature 54◦F

Minimum temperature –15◦F

Mean annual precipitation 39.7 inches

Mean annual snowfall 18.3 inches

Winter conditions are such that road salting is commonly practiced The soil sampleswere taken with Shelby tubes, one set (BH) in a N–S direction (parallel to thehighway) and another (XS) in an E–W direction Next, the soil samples were di-vided into sections with 5 cm increments down to 15 cm; at greater depths largerincrements were used

2.2 ANALYTICAL METHODS

In this study five different types of analysis have been applied for different pects of the research: X-ray fluorescence (XRF, Rigaku 3070 spectrometer), C-Sanalyzer (LECO), Inductively coupled plasma spectroscopy (ICP, Perkin-ElmerOptima 3000), Atomic absorption spectrometry (AAS, Perkin-Elmer 3110) andX-ray diffraction (XRD, Siemens D-500)

The first step in this research has been application of XRF to determine the bulkchemistry of the samples For this, samples were dried at 100◦C and ground using asteel ball mill Pressed pellets for XRF were prepared with 5–6 g of sample pressedunder 18 tons for 4–5 min Samples were run against a set of U.S GeologicalSurvey rock standards combined with a set of roadside soil samples previouslyanalyzed by XRAL, Inc of Toronto, Canada by neutron activation.

LECO analysis was applied in order to find percentages of organic C, total Cand total S in the soils Total C and total S were run on dried powders Organic

C was measured on acidified samples The acidification was done using 50 mL of

1 N HCl to 0.5 g of sample on a hot plate at 60◦C for 12 hr Fifty mL of distilledwater was then added to stop reaction The solution was filtered through a glassfiber filter and the residue rinsed with distilled water to remove all acid Sampleswere then dried at least four hours

ICP was used to determine the nature of the exchangeable ions Fifteen mL

of 1 molar NH4 acetate at pH 7 was added to 0.2 g of sample The suspensionwas left overnight and centrifuged the next day Five mL of nitric acid was added

to the liquid taken out from the centrifuged tubes to maintain metals in solution(Ulmschneider, 1977)

Atomic absorption (AAS) was used to monitor sequential extraction of metals

2 g of dry soil sample was placed into a labeled centrifuge tube The extraction

steps then are (Sposito et al., 1982):

• 25 mL of 0.5 M KNO3was added and shaken for 16 hr (exchangeable fraction);

• 25 mL of distilled H2O was added and shaken for 2 hr (absorbed fraction);

• 25 mL of 0.5 M NaOH was added and shaken for 16–21 hr (organically boundfraction);

• 25 mL of 0.05 M Na2EDTA was added and shaken for 6 hr (carbonate boundfraction);

• 25 mL 4 M HNO3 added and heated (70–80 ◦C oven) for 16–21 hr (residualfraction)

After each step the sample was centrifuged and filtered through a Whatman # 42filter into a nalgene bottle The solutions were refrigerated and saved for AtomicAbsorption Spectrometric Analysis

XRD was used to determine sample mineralogy, especially the clay mineraltypes Sample preparation started by putting 2–3 g of sample in a beaker filledwith 200 mL of water After stirring, the suspension was left for 45 min The clayminerals, which were floating close the surface, were caught with a pipette andtransferred onto a glass slide The sample was left to air dry For some samples itwas not possible to collect the necessary amount of clay minerals by pipette Inthat case, the top part of the water in the beakers was taken into centrifuge tubes.After centrifugation, the clay minerals separated at the bottom of the tubes wereapplied as a paste on glass slides One set of slides was left air dried, a duplicate

set was glycolated, and another set was heated to 350 and to 550◦C in order todifferentiate clay minerals.

3 Results

3.1 BULK SOIL COMPOSITION

XRF data showed that heavy metal content is very high in the top 15 cm of the soil(Table II) The maximum measured amount for Pb is 1980 ppm, which was takenfrom 10–15 cm depth in core BH9 The highest value for Zn is 1426 ppm at XS1from 0–1 cm depth For comparison, background values, calculated as weightedaverages of concentrations in samples taken from below 30 cm, were Pb 60 ppm;

Zn 85 ppm; Cu 35 ppm; Ni 40 ppm; and Cr 35 ppm Metal values decrease withdepth (Figure 2) This inverse relationship is stronger for Zn and Cu (R2Zn: 0.53 and

R2Cu: 0.53) than for Pb, Ni and Cr (R2Pb: 0.33, R2Ni: 0.22, R2Cr: 0.16)

From the LECO analysis, average organic C percent for these soil samples is3.8 and total C is 6.8% There is a positive correlation between organic C contentand metal values: as the amount of organic C increases, the amount of heavy metalcontamination also increases (Figure 3) In addition the correlation is stronger fororganic C and metal content than for depth vs metal content for each of the metals(R2Zn: 0.59, R2Cu: 0.77, R2Pb: 0.62, R2Ni: 0.40, R2Cr: 0.24) Note that the correlationcoefficient for Pb is much higher for the Pb vs organic C relationship than for the

Pb vs depth relationship, whereas both Ni and Cr show very weak relationships toboth depth and to organic C

Cluster analysis was used to further illustrate which metals have close ciations with each other, with depth and with organic C amount in the soil Theresult showed that Pb, Zn and Cu are acting together and they are more closelyassociated with the amount of organic C in the soil than with depth Ni and Cr,however, did not show any association with other metals, with organic C or withdepth (Figure 4)

asso-The yields of exchangeable metals using NH+4 as the exchange ion were low(except for Ca, which probably comes from dissolution of calcite as discussed by

Tessier et al., 1979) For 12 samples analyzed by ICP (Table III), average

exchange-able Pb was only 4.5%, Zn 8.3%, Cu 6.9%, and Cr 3.7% of the amount in the wholesoil based on XRF

3.2 SEQUENTIAL EXTRACTION

The sequential soil extraction procedure was applied to 5 soil samples The ults confirmed that the metal amounts in the exchangeable fraction are very low(Table IV) On average only 1.6% of Pb, 0.4% of Zn, 1.7% of Cu, 4% of Niand 5% of Cr are exchanged with KNO Adsorbed metals were also very low

res-TABLE II Results of LECO and XRF

BH5-01 0–1 17.98 15.52 0.11 72 353 64 89 169 1207 148 942 BH5-15 1–5 4.98 10.53 0.08 82 389 64 68 192 578 218 1073 BH5-510 5–10 5.06 9.32 0.05 65 134 56 62 250 288 173 957

BH9-1015 10–15 6.60 11.77 0.10 73 275 54 49 195 619 266 1980 BH9-612 15–30 1.41 4.08 0.06 66 73 54 103 156 167 218 407 BH9-1218 30–46 0.77 2.69 0.10 68 30 59 130 132 71 198 27

Pb and Zn Pb and Zn were found to be released dominantly in the carbonate step

or to remain in the residual fraction

Because the low values of Pb and Zn in the organically bound fraction arecontrary to the results of the correlation analysis, which indicated that these metalsare strongly associated with organic C, we applied additional tests to this fraction.The first three steps of the procedure were reapplied to 5 g each of two sampleswith 100 mL of extractant solutions, with the objective of checking for any organiccarbon left in the sample after application of NaOH For XS3-1015 organic C after