Cross Validation of Two Partitioning-Based ... - ACS Publications

Aug 2, 2017 - This cross validation on two levels would make the two approaches truly .... loading time, as calculated from the MATLAB mass transfer m...
0 downloads 0 Views 708KB Size
Subscriber access provided by Warwick University Library

Article

Cross validation of two partitioning-based sampling approaches in mesocosms containing PCB contaminated field sediment, biota, and activated carbon amendment Stine Nørgaard Schmidt, Alice P. Wang, Philip T. Gidley, Allyson H. Wooley, Guilherme R. Lotufo, Robert M. Burgess, Upal Ghosh, Loretta A. Fernandez, and Philipp Mayer Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.7b01909 • Publication Date (Web): 02 Aug 2017 Downloaded from http://pubs.acs.org on August 3, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers

Environmental Science & Technology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Subscriber access provided by Warwick University Library

and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Environmental Science & Technology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Environmental Science & Technology

In situ sampling

Page 1 of 35

Ex situ equilibrium sampling

ACS Paragon Plus Environment

Environmental Science & Technology

Page 2 of 35

1

TITLE:

2

Cross validation of two partitioning-based sampling approaches in mesocosms containing

3

PCB contaminated field sediment, biota, and activated carbon amendment

4 5

AUTHORS:

6

Stine N. Schmidt,*1 Alice P. Wang,2 Philip T. Gidley,3 Allyson H. Wooley,3 Guilherme R.

7

Lotufo,3 Robert M. Burgess,4 Upal Ghosh,5 Loretta A. Fernandez,*2 and Philipp Mayer1

8 9

AFFILIATIONS:

10

1

11

Lyngby, Denmark; 2Northeastern University, Department of Civil and Environmental

12

Engineering, Boston, MA, USA; 3US Army Corps of Engineers, Engineer Research and

13

Development Center, Vicksburg, MS, USA; 4US Environmental Protection Agency,

14

NHEERL/Atlantic Ecology Division, Narragansett, RI, USA; 5University of Maryland

15

Baltimore County, Department of Chemical, Biochemical, and Environmental Engineering,

16

Baltimore, MD, USA

Technical University of Denmark, Department of Environmental Engineering, Kgs.

17 18

CORRESPONDING AUTHORS (*):

19

Stine N. Schmidt, Technical University of Denmark, Department of Environmental

20

Engineering, Kgs. Lyngby, Denmark. Phone: (+45) 45251425. E-mail: [email protected]

21

and Loretta A. Fernandez, Northeastern University, Department of Civil and Environmental

22

Engineering, Boston, MA, USA. Phone: (+1) 617 373 5461. E-mail: [email protected]

1 ACS Paragon Plus Environment

Page 3 of 35

23 24

Environmental Science & Technology

ABSTRACT The Gold Standard for determining freely dissolved concentrations (Cfree) of

25

hydrophobic organic compounds in sediment interstitial water would be in situ deployment

26

combined with equilibrium sampling, which is generally difficult to achieve. In the present

27

study, ex situ equilibrium sampling with multiple thicknesses of silicone and in situ pre-

28

equilibrium sampling with low density polyethylene (LDPE) loaded with performance

29

reference compounds were applied independently to measure polychlorinated biphenyls

30

(PCBs) in mesocosms with (1) New Bedford Harbor sediment (MA, USA), (2) sediment

31

and biota, and (3) activated carbon amended sediment and biota. The aim was to cross

32

validate the two different sampling approaches. Around 100 PCB congeners were

33

quantified in the two sampling polymers, and the results confirmed the good precision of

34

both methods and were in overall good agreement with recently published silicone to

35

LDPE partition ratios. Further, the methods yielded Cfree in good agreement for all three

36

experiments. The average ratio between Cfree determined by the two methods was factor

37

1.4±0.3 (range: 0.6-2.0), and the results thus cross-validated the two sampling

38

approaches. For future investigations, specific aims and requirements in terms of

39

application, data treatment, and data quality requirements should dictate the selection of

40

the most appropriate partitioning-based sampling approach.

2 ACS Paragon Plus Environment

Environmental Science & Technology

41 42

Page 4 of 35

INTRODUCTION Worldwide, enormous amounts of sediment contaminated with hydrophobic organic

43

compounds (HOCs) require assessment and remediation.1,2 The first step is often to

44

measure the extent of the contamination in terms of concentration and spatial distribution.

45

Historically, exhaustive extraction methods have been applied to determine total

46

concentrations (Ctotal). However, Ctotal is poorly related to the actual exposure and thereby

47

the potential risk for bioaccumulation and adverse effects caused by sediment associated

48

contaminants.3,4 Therefore, partitioning-based sampling methods are increasingly used for

49

determining freely dissolved concentrations (Cfree),5,6 which quantify effective

50

concentrations for diffusive mass transfer and partitioning7 and thus provide a basis for

51

quantitative thermodynamic exposure assessments.

52

Within the last two decades, several partitioning-based sampling methods have been

53

developed, tested, and applied for determining the Cfree of HOCs in sediment interstitial

54

water.6 They fall into two categories: (1) The equilibrium sampling approach is to

55

equilibrate a polymer with the sediment, which generally requires ex situ incubation of very

56

thin polymers under agitated conditions in the laboratory.8,9,10 The advantage of this

57

approach is that equilibrium partitioning is a simple and well-defined regime, whereas its

58

limitation is that it does not necessarily provide in situ levels of Cfree.7 (2) The in situ pre-

59

equilibrium sampling approach is to place a polymer within the sediment on site and then

60

infer equilibrium concentrations through the use of performance reference compounds

61

(PRCs) or time series measurements.11,12,13 While equilibration is still achievable, it

62

typically requires sampling times of months, and for larger HOCs even years, which is

63

often impractical and can lead to considerable delays in project conduct and decision

64

making. The advantage of this approach is its potential for capturing in situ levels of Cfree.

3 ACS Paragon Plus Environment

Page 5 of 35

Environmental Science & Technology

65

However, this is a more complicated approach as it requires additional steps such as mass

66

transfer modeling and initial estimates of contaminant levels and partitioning behavior in

67

sediment to determine appropriate sampler sizes and quantities of PRCs.14 The use of

68

PRCs and mass transfer models can also introduce error and uncertainty.7

69

The Gold Standard for determining Cfree of HOCs in sediment interstitial water would

70

be in situ deployment combined with equilibrium sampling (i.e., “in situ equilibrium

71

sampling”), which is generally difficult to achieve. However, two promising but still

72

developing approaches may make in situ equilibrium sampling feasible and practical: in

73

situ vibration of the sampler15 and in situ sampling with multiple thicknesses of silicone.16,17

74

During the interim, the strategy of the present study was to compare ex situ equilibrium

75

sampling with in situ pre-equilibrium sampling in mesocosm experiments with sediment

76

contaminated with polychlorinated biphenyls (PCBs). The aims were (1) to quantify the

77

magnitude, precision, and relationship of equilibrium concentrations in two polymers that

78

were deployed in two different ways and (2) then to compare Cfree determined by the two

79

methods. Ex situ equilibrium sampling with multiple thicknesses of silicone and in situ

80

sampling with low density polyethylene (LDPE) loaded with PRCs were applied. Despite

81

differences in their working principles, the null hypothesis was that the two sampling

82

approaches would yield (1) equilibrium concentrations in silicone and LDPE in good

83

agreement with recently reported silicone to LDPE partition ratios and (2) similar Cfree

84

values. This cross validation on two levels would make the two approaches truly

85

compatible, complementary, and well aligned.

86

The present study is part of a project that combines quantitative thermodynamic

87

exposure assessments and bioaccumulation studies to assess sediment restoration

88

techniques based on amendment with activated carbon (AC). A range of mesocosm

4 ACS Paragon Plus Environment

Environmental Science & Technology

Page 6 of 35

89

experiments were prepared with PCB contaminated sediment from New Bedford Harbor

90

(NBH, MA, USA), which contains total PCB concentrations in the hundreds of mg kg-1 (dw)

91

primarily from contamination by Aroclors 1242 and 1254.18 The mesocosms were used for

92

a systematic comparison of the two sampling approaches under controlled conditions to

93

avoid boundary effects and lateral transport commonly noted in field plot studies.19 The

94

two approaches were applied in experimental setups with biota, AC amendment, and

95

ongoing PCB contamination of the sediment. The masses of PCBs accumulated in both

96

sampling polymers were measured with the same precise analytical method, providing the

97

analysis of up to 130 congeners. In this way, the basis for comparing the two sampling

98

approaches was free of biases caused by differing analytical laboratories.

99 100 101

EXPERIMENTAL SECTION This paper includes results from three mesocosm experiments with sediment

102

(Experiment 1), sediment and biota (Experiment 2), and AC-amended sediment and biota

103

(Experiment 3). Sediment was collected from NBH on two occasions and stored in steel

104

drums at 2.8-4.0°C for about a year. Sediment from three drums was combined,

105

homogenized, and stored for days to months before use. Prior to starting Experiment 3,

106

the experimental sediment was amended with 4.3% AC (dw) during careful mixing for a

107

month (see page S2-3, SI). Experimental mesocosms were prepared in 52-L glass aquaria

108

measuring 51×25×41 cm (length×width×height, Glasscages.com, USA), and 13-14 kg of

109

wet sediment was added to each of six aquaria for each of the three experiments.

110

Reconstituted seawater (30‰) was prepared by dissolving Instant Ocean® in water

111

treated by reverse osmosis filtration and then slowly added to fill the aquaria. The

112

mesocosms were kept at 20°C using a water bath, covered by acrylic lids, and left to

5 ACS Paragon Plus Environment

Page 7 of 35

Environmental Science & Technology

113

consolidate for 8-14 d before the experiments started. After consolidation, the sediment

114

layer was approximately 6 cm thick. In Experiments 2 and 3, five worms (Nereis virens,

115

~3.5 g ww each), 15 clams (Mercenaria mercenaria, ~10.6 g ww each), and five fish

116

(Cyprinodon variegatus, ~1.4 g ww each) were added to each mesocosm, with the fish

117

being held in mesh, stainless steel cages to prevent them from directly contacting and

118

bioturbating the sediment. During the 90-d experiments, diffused air was continuously

119

supplied (via an aquarium air stone) to aerate the surface water. Gravel bed treatment

120

systems were operated to maintain low ammonia levels, and the surface water was re-

121

circulated with an overall flow rate of 17.3±9.5 L h-1 (range: 7.9-46.1 L h-1, n=53) in the 18

122

mesocosms. Temperature, pH, dissolved oxygen, and salinity were monitored daily, and

123

ammonia levels were monitored at least once a week. The mesocosm experiments were

124

conducted with a photoperiod of 16 h light and 8 h dark.

125

In each experiment, three replicate mesocosms received PCB-spiked sediment input

126

three times a week for the full period of 90 d to simulate ongoing PCB contamination of the

127

sediment. Correspondingly, three replicate mesocosms received un-spiked sediment. The

128

sediment used for inputs was collected from a relatively clean area (Bayou Lafourche near

129

Leeville, LA, USA, see page S3). Part of the sediment was spiked with the “input” PCB

130

congeners 13, 54, and 173 (Ultra Scientific, USA). The spiking was performed following

131

the shell coating method described by Northcott and Jones (2000) and Lotufo et al

132

(2001).20,21 For each of the three congeners, an acetone solution was prepared and

133

pipetted into the same 2-L glass jar. The jar was rolled horizontally (5.5 rpm) without lid

134

until complete evaporation of the acetone, leaving the compounds coated on the inner jar

135

surface. Then, 700 g of wet sediment was added, the jar sealed and left to roll horizontally

136

at 5.5 rpm for two weeks before use as PCB-spiked input sediment. Four batches of PCB-

6 ACS Paragon Plus Environment

Environmental Science & Technology

Page 8 of 35

137

spiked sediment were prepared and used in Experiments 1-3. The weighted average

138

concentrations of PCBs 13, 54, and 173 were 9.3, 12.5, and 2.8 mg kg-1 (dw), respectively

139

(see page S4). In the same way, part of the sediment was rolled for two weeks in the

140

absence of added PCBs before use as un-spiked input sediment (the three congeners

141

were not detected in this material). The sediment inputs were introduced to the 18

142

mesocosms manually by pre-mixing 5 mL sediment (containing approximately 1.35 g dry

143

sediment particles) with water from the respective mesocosm to form a uniform and

144

dispersible plume in the mesocosm surface water. Water circulation through the treatment

145

systems was stopped just before adding the input material and switched off for 7 h to allow

146

the introduced sediment to settle.

147

In situ sampling with LDPE. Strips were prepared from a LDPE sheet with a

148

thickness of 25.4 µm (ACE Hardware Corp., USA) and measured 2.5×15.2 cm. First, the

149

strips were pre-cleaned by two overnight methylene chloride rinses, followed by two

150

methanol rinses, and two additional rinses with laboratory water treated by reverse

151

osmosis filtration, ion exchange, and AC filtration (EMD Millipore, Germany). Then, the

152

strips were loaded with four PRCs (rare PCB congeners 14, 35, 73, and 122, Ultra

153

Scientific, USA). PCB standard solutions of 100 mg L-1 in hexane (Ultra Scientific, USA)

154

were combined, and the majority of hexane evaporated under nitrogen. The PRCs were

155

then re-dissolved in methanol (J.T. Baker, USA) to reach a final concentration of 5 mg L-1

156

for each of the compounds in the mixture. Each strip was loaded separately by adding it to

157

a 500-mL amber jar filled with laboratory water spiked with 200 µL of the PRC mixture. The

158

jars were sealed with aluminum foil lined caps and placed on an orbital shaker for six

159

weeks (80 rpm) in accordance with a necessary loading time, as calculated from the

160

MATLAB mass transfer model.22 After loading, the nominal mass of each PRC was 1000

7 ACS Paragon Plus Environment

Page 9 of 35

Environmental Science & Technology

161

ng per strip and the precise pre-exposure concentrations (t0) were measured at the time of

162

sampler analysis, as described below.

163

At day 0, one loaded strip was inserted into the sediment bed in each of the 18

164

mesocosms (supported by a stainless steel frame). The strips were deployed in the

165

sediment for the full period of 90 d and then recovered for analysis. Each strip was rinsed

166

with deionized water to remove adhering sediment particles, dried with lint free tissue, and

167

cut into five approximately equally-sized horizontal segments. One segment, the part

168

sampling the interstitial water in the top 2-3 cm of sediment, was analyzed per mesocosm.

169

The segments weighed 18.6±2.4 mg (n=18) and were extracted individually in 5 mL

170

hexane (95%, Fisher Scientific, USA) overnight before chemical analysis, as described

171

below. T0 samples (PRC-loaded, unexposed samplers, n=4) were rinsed and extracted,

172

correspondingly.

173

Cfree (ng L-1) were calculated from the PCB concentrations in LDPE when

174

extrapolated to equilibrium (CLDPE∞, ng kg-1) and compound specific LDPE to water

175

partition ratios (KLDPE:water, L kg-1):

176

C

177

(1)

= :

178 179

CLDPE∞ were determined from the concentrations measured in LDPE after deployment

180

(CLDPE(t), ng kg-1) and an adjustment for the fractional equilibration of the target PCBs (feq,

181

unitless):

182 183

C

=

()

(2)

184

8 ACS Paragon Plus Environment

Environmental Science & Technology

Page 10 of 35

185

feq values for the target PCBs were calculated from the fractional equilibration of the PRCs,

186

feq PRC (unitless):

187

f

188



=



! ( " )#



!(

)

! ( " )

(3)

189 190

where CLDPE PRC (t0) is the initial PRC concentration in the sampler (ng kg-1) and

191

CLDPE PRC(t) is the PRC concentration in the sampler after deployment (ng kg-1). PRC

192

calculation software is available through the Environmental Strategic Technology

193

Certification Program.23 This MATLAB graphical user interface runs a Laplace-space

194

mathematical model of Fickian diffusion through polymeric and porous media. After

195

entering feq PRC values for a given LDPE strip segment, compound specific KLDPE:water

196

values, deployment time (90 d), the default sediment porosity (0.7), and thickness of the

197

LDPE (25.4 µm), the feq for each target PCB was calculated.23 One PRC, PCB 14, was not

198

used in the calculation of feq as it was completely depleted from the samplers. KLDPE:water (L

199

kg-1) values were calculated for each PCB congener from its octanol to water partition ratio

200

(Kow)24 using a previously published linear free energy relationship based on compiled data

201

for 22 polycyclic aromatic hydrocarbons (PAHs) and 110 PCBs (r2=0.89, RSME=0.29,

202

Table S1):22

203 204

log K

:()

= 0.97 × log K /( − 0.07

(4)

205 206

Ex situ equilibrium sampling with silicone. Non-depletive equilibrium sampling

207

was conducted in silicone coated jars with multiple coating thicknesses.9 These equilibrium

208

sampling devices (ESDs) consisted of 120-mL, clear glass jars coated with µm-thin layers

9 ACS Paragon Plus Environment

Page 11 of 35

Environmental Science & Technology

209

of silicone (Dow Corning® 1-2577 conformal coating, Diatom A/S, Denmark) on their inner

210

vertical surface. The ESDs were coated and pre-cleaned in the following manner: Silicone

211

stock solution, as received from the supplier, was weighed into a disposable bottle and

212

diluted in pentane. The bottle was sealed and the silicone solution homogenized at 500

213

rpm on an orbital shaker for at least 1 h. To produce different coating thicknesses, the

214

concentration of silicone was adjusted in the pentane solution, taking into consideration

215

the percentage of silicone in the silicone stock solution, the density of the silicone, and the

216

area of the inner vertical surface of the glass jars. Then, the glass jars (without lids) were

217

placed horizontally on a roller mixer, and a 2-mL aliquot of the well-mixed silicone solution

218

was gently added to each jar while rolling (60 rpm). The silicone solution quickly dispersed

219

evenly to cover the full inner height of each jar (i.e., from the bottom to the rim). After

220

application, the jars were left to roll for 30 min until all organic solvent had evaporated,

221

leaving the silicone coated on the inner vertical surface. The jars were left at room

222

temperature overnight for the silicone to cure, before final curing at 100°C for 2 h. The jars

223

were then pre-cleaned in three steps with 5-mL aliquots of ethyl acetate, acetone, and

224

ethanol during horizontal rolling of the sealed ESDs for at least 30 min (60 rpm). After

225

removal of the ethanol and evaporation of residual solvent, the mass of silicone in each jar

226

was determined gravimetrically by weighing before and after coating and pre-cleaning. For

227

this study, ESDs were prepared with three coating thicknesses: (1) 0.76±0.04 µm (6.5±0.3

228

mg silicone, n=19), (2) 1.62±0.05 µm (13.7±0.4 mg silicone, n=19), and (3) 3.13±0.13 µm

229

(26.6±1.1 mg silicone, n=18). Just before use, the silicone was pre-cleaned with hexane.

230

At termination of the mesocosm experiments (day 90), most of the surface water was

231

siphoned off and sediment was collected for equilibrium sampling. Three sediment cores

232

(with a diameter of 7.2 cm) were taken from each of the 18 mesocosms, and the top 2-3

10 ACS Paragon Plus Environment

Environmental Science & Technology

Page 12 of 35

233

cm of sediment was transferred to ESDs coated with 0.76, 1.62, and 3.13 µm silicone,

234

respectively. Approximately 80 g wet sediment was added to each jar. Small amounts of

235

surface water from the given mesocosm was added to the nearly filled ESDs to obtain

236

viscous sediment slurries, which allowed good contact between the sediment and silicone

237

during sampling, while at the same time, keeping the water content at a minimum to

238

ensure fast PCB mass transfer from sediment to silicone. The ESDs were sealed with

239

PTFE-lined plastic lids and rolled horizontally at 19 rpm for 14 d (at room temperature).

240

After equilibration, the coated jars were emptied and rinsed with several portions of Milli-Q

241

water to remove any sediment particles adhering to the silicone, and the silicone was

242

gently dried using lint free tissue. The silicone was then extracted with 2 mL hexane (95%,

243

Fisher Scientific, USA) during horizontal rolling (12-14 rpm) of the ESD for at least 30 min.

244

Again, the ESDs were sealed with clean PTFE-lined lids. The hexane was collected, and a

245

fresh portion of 2 mL hexane added to each jar. The extraction was repeated, the two

246

extracts combined, and the volume adjusted to exactly 4 mL before chemical analysis, as

247

described below.

248

Cfree (ng L-1) were determined via concentrations in silicone at equilibrium with the

249

sediment (Csilicone⇌sediment, ng kg-1) and experimentally derived silicone to water partition

250

ratios (Ksilicone:water, L kg-1):25

251 252

C

=

2343567 ⇌2 839 7

(5)

2343567 :

253 254

The Csilicone⇌sediment was determined in the following manner: The mass of PCB (ng) was

255

plotted against silicone mass (g) in GraphPad Prism 5.0 software (GraphPad Software,

256

Inc., USA). Best estimate of the PCB concentration in silicone was determined via linear

11 ACS Paragon Plus Environment

Page 13 of 35

Environmental Science & Technology

257

regression through the origin (i.e., the slope, ng g-1) and supplied with a r2 value and

258

standard error of the mean (SEM, ng g-1) by the software. For each concentration, the

259

relative standard error (RSE, %) was calculated as SEM over Csilicone⇌sediment × 100. For

260

PCBs with 8-10 chlorine atoms (i.e., PCBs 194-209), data was excluded for jars with the

261

thickest silicone coating due to disequilibrium. Concentrations were deemed valid when

262

the regression described more than 60% of the variation in the data set (r2>0.60) and the

263

RSE on the concentration was less than 20%. Concentrations were categorized as highly

264

precise when r2>0.70 and RSELOD was included in the data treatments.

Page 14 of 35

282

Before the experiments started, NBH sediment was screened in order to select

283

appropriate internal standards and PCB congeners to be used as PRCs and ongoing input

284

congeners. These compounds were selected to avoid interference with native PCBs in the

285

sediment in terms of presence and analytical separation.

286 287 288

RESULTS AND DISCUSSION Results from the two sampling approaches were used to quantify the magnitude,

289

precision, and relationship of equilibrium concentrations in the two polymers and to

290

compare Cfree determined by the two methods.

291

Equilibrium polymer concentrations. A total of 111, 110, and 76 congeners were

292

quantified in silicone for the experiments with sediment only, sediment and biota (+biota),

293

and AC amended sediment and biota (+AC), respectively. Of these, 97% (sediment), 93%

294

(+biota), and 83% (+AC) were deemed valid with respectively 94%, 92%, and 90% of the

295

valid concentrations fulfilling the extended validity criteria (r2>0.70 and RSE