Formulation and Validation of a Functional Assay-Driven Model of

Subscriber access provided by Washington University | Libraries

Environmental Modeling

Formulation and Validation of a Functional AssayDriven Model of Nanoparticle Aquatic Transport Nicholas K. Geitner, nathan bossa, and Mark R. Wiesner Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.8b06283 • Publication Date (Web): 28 Feb 2019 Downloaded from http://pubs.acs.org on March 2, 2019

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 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 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 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.

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.

Page 1 of 20

Environmental Science & Technology

Formulation and Validation of a Functional Assay-Driven Model of Nanoparticle Aquatic Transport Nicholas K Geitnera,b, Nathan Bossaa,b, Mark R Wiesnera,b* aCenter

for the Environmental Implications of Nanotechnology, Duke University, Durham, NC, USA. bCivil and Environmental Engineering Department, Duke University, Durham, NC, USA. *Corresponding Author: [email protected]

ACS Paragon Plus Environment


1

Abstract

2

Here, we present a model for the prediction of nanoparticle fate in aquatic environments,

3

parameterized using functional assays that take into account conditions of the environmental media

4

and nanoparticle properties. The model was used to explore scenarios for 5 nanomaterials in a

5

freshwater wetland setting and compared with experimental results obtained in mesocosm studies.

6

Material characteristics used in the model were size, density, dissolution rate constants, and surface

7

attachment efficiencies. Model predictions and experimentally measured removal rate constants

8

from the water column were strongly correlated, with Pearson correlation coefficient 0.993.

9

Further, the model predicted removal rate constants quantitively very close to measured rates. Of

10

particular importance for accurate predictions were 2 key processes beyond the usual

11

heteroaggregation with suspended solids. These were homoaggregation of nanomaterials and

12

nanomaterial attachment to aquatic plant surfaces. These results highlight the importance of

13

including all relevant aggregation and deposition processes over short time scales for nanoparticle

14

transport, while demonstrating the utility of functional assays for surface attachment as model

15

inputs.

16

1. Introduction

17

An understanding of nanomaterial environmental transport is critical for natural, incidental, and

18

engineered materials. This is not only because of toxicity concerns, but because these nanoscale

19

materials may play a critical role in the transport of other environmental contaminants.1-3 Because

20

of the rapidly growing space of engineered materials and the discovery of diverse natural and

21

incidental particles, efficient and predictive models of particle transport are required. Challenges


Page 2 of 20

Page 3 of 20


22

in developing such models include parameterizing nanoparticle behavior in complex environments

23

and subsequently validating model predictions.

24

There have been several recent attempts at modeling nanoparticle environmental transport, most

25

of which utilize multimedia transport frameworks. Praetorius et al developed a model for transport

26

through flowing river systems;4 this model utilized a range of values in attachment efficiencies (

27

𝛼) and predicted particle concentrations as a function of distance from a source. Interactions in this

28

model focused on suspended solids in the water column, flow, sedimentation, and transformations

29

of the particles themselves. Taghavy and Abriola examined the effect of heterogeneous size

30

distributions on model predictions and utilized a random-walk particle-tracking framework.5

31

MendNano,6 also a multimedia mass transfer model, includes nanoparticle dissolution and flow

32

transport processes but assumed an “attachment factor” which defined a global fraction of

33

nanoparticles which were instantaneously and permanently attached to suspended solids. Previous

34

efforts have also been made in examining values of 𝛼 as predictors of nanoparticle transport, which

35

also found that these values and attachment mechanisms may be calculated from theory with

36

reasonable accuracy.7-8

37

There are three primary goals for the model presented here: to more completely capture short-term

38

transport processes for nanoparticles in wetlands and similar systems, to demonstrate the utility of

39

functional assay parameters for model inputs,9 and to experimentally validate model predictions.

40

A key process in this model not featured in those cited above is the kinetic attachment of

41

nanomaterials to immobile surfaces such as plants, which may dominate particle collision events.

42

Parameters for the model are determined using functional assays developed previously.10 These

43

key parameters are the surface attachment efficiency 𝛼 and dissolution rate constants in the

44

respective media compartments. Parameterizing the model using functional assays allows for the



45

facile modeling of kinetic transport processes for nanomaterials in complex matrices without

46

making over-simplifying assumptions of heteroaggregation or dissolution state.9 In a recent study,

47

we examined the removal kinetics of 5 different nanomaterials from the water column of wetland

48

mesocosms. These included 2 silver nanoparticles, 2 different metal oxides, and single walled

49

carbon nanotubes, also with a range of surface functionalities.11 Within these realistic simulated

50

wetlands, we subsequently demonstrated that 𝛼 was strongly correlated to this removal for some,

51

but not all, of the 5 materials. In this study, we formulate a model which is driven by experimental

52

functional assays including those for 𝛼 and for dissolution. We then compare transport predictions

53

for the same 5 nanomaterials with previously reported transport kinetics from wetland mesocosms

54

to validate the model.

55

2. Materials and methods

56

2.1 Materials

57

The five nanomaterials considered in this study were: TiO2 (Evonik Industries, Essen, Germany)

58

and CeO2 (Sigma-Aldrich, USA) without additional coating, gum Arabic-stabilized single-walled

59

carbon nanotubes (GA-SWCNT), and two silver nanoparticles coated with polyvinylpropaline

60

(PVP-AgNP) and gum arabic (GA-AgNP). Details concerning nanoparticle synthesis and

61

suspension preparation are detailed elsewhere.11 GA-SWCNT were long rods under 1 µm in

62

length, and all other materials were spheres of varying size. Previously performed characterization

63

results for each material are provided in Table S1.

64 65

2.2 System description

66

The wetland mesocosms are engineered ecosystems in open air, located in a clearing in the Durham

67

division of the Duke Forest in Durham, NC, USA. The structure of the mesocosms has been


Page 4 of 20

Page 5 of 20


68

previously described.12 Briefly, each mesocosm (H: 0.81 × W: 3.66 × D: 1.2 m) contained an

69

aquatic region (1 m) which was always submerged, then continues with an upward slope (13°

70

inclination), providing for zones with a gradient of humidity and redox conditions in the system

71

that are meant to simulate a freshwater wetland. The mesocosms contain 20 cm of natural soil

72

(64% sand, 28% silt, 13% clay, and 5% loss on ignition), and were filled with 250 L of untreated

73

groundwater. Water volumes subsequently fluctuated due to evapotranspiration and precipitation.

74

Mesocosms were filled and planted in March and allowed 5 months to grow to maturity and

75

establish a local ecosystem. The plants were Egeria densa in the aquatic zone and Lobelia

76

elongate, Carex Lurida, Panicum Virgatum, and Juncus effusus in a uniform grid in the transition

77

and terrestrial zones. At the end of the growth phase, Egeria densa plant mass density in the water

78

column was approximately constant and evenly distributed through the water column. Additional

79

details are available elsewhere.12-14 The resulting systems were self-sustaining for long periods of

80

time. They also displayed characteristics of natural aquatic systems including diurnal cycles in pH

81

and dissolved oxygen.

82 83

2.3 Model description

84

The present model utilizes a multimedia compartmental framework. Movement of nanomaterials

85

between these compartments is described by a series of linked mass balances. In the water column,

86

the rate of change in number concentration of nanoparticles, n, is expressed as

87

𝑑𝑛 = ―𝑘𝐻𝑒𝑛 ― 𝑘𝐻𝑜𝑛2 ― 𝑘𝑠𝑒𝑡𝑡𝑙𝑒𝑛 ― 𝑘𝑑𝑖𝑠𝑠𝑛#1 𝑑𝑡

88

where, kHe, kHo, ksettle, and kdiss are rates of heteroaggregation/deposition (heteroattachment),

89

homoaggregation, settling, and dissolution, respectively. These rate constants depend on

90

parameters determined using functional assays for particle attachment and dissolution in the



91

respective compartment media. The use of functional assay to parameterize the model allows for

92

integration of system complexity in the description of kinetic transport processes for nanomaterials

93

without making over-simplifying assumptions concerning environmental media. For the specific

94

wetland aquatic system of this study, the heteroattachment rate constant includes contributions

95

from heteroaggregation with suspended solids and deposition on aquatic plant surfaces (Equation

96

2) that each depend on transport and attachment terms. Here, 𝛼𝐻𝑒 and 𝛽𝐻𝑒 are the attachment

97

efficiency and collision rate kernel with suspended solids, respectively, and 𝐵𝑆𝑆 is the number

98

concentration of those solids. Values of 𝛽𝐻𝑒 are approximated using the rectilinear model for

99

particle transport, and include terms for the simultaneous processes of diffusion, shear, and

100

differential settling.15-16 The rectilinear approximation for the collision rate kernel was used since

101

a priori knowledge of the suspended particle/ aggregate geometry is not available, but is inherently

102

integrated into the estimates of 𝛼𝐻𝑒 in the functional assay.

103

𝑘𝐻𝑒 ≡ 𝛼𝐻𝑒𝛽𝐻𝑒𝐵𝑆𝑆 + 𝛼𝑝𝑎 ∗ 𝑝𝑣𝑝#2

104

The rate of homoaggregation is similarly described as the product of a collision rate kernel and an

105

attachment efficiency (Equation 3).

106

𝑘𝐻𝑜 = 𝛼𝐻𝑜𝛽𝐻𝑜#3

107

In Equation 2, 𝛼𝑝 is the attachment efficiency for nanoparticles on aquatic plant surfaces, 𝑎 ∗ 𝑝 is

108

the effective specific surface area of aquatic plant leaves, 𝑎𝑝 is the total measured plant specific

109

surface area, and 𝑣𝑝 is the rate of transport of particles to plant surfaces. Encounters with plant

110

surfaces may occur through two distinct processes: settling onto the tops of leaves, and diffusion-

111

mediated collisions on top and bottom of the leaves. Therefore, the total collision rate with plant

112

surfaces can be described by Equation 4.


Page 6 of 20

Page 7 of 20


(

𝑎 ∗ 𝑝𝑣𝑝 ≡ 𝑎𝑝 𝑣𝑠𝑒𝑡𝑡𝑙𝑒 ∗

113

)

𝑘𝐵𝑇 𝑐𝑜𝑠𝜃 + 𝛿 #4 2 12𝜋𝜇𝑑2

114

In Equation 4, 𝜃 is the average angle plant leaves form with the water surface, 𝑣𝑠𝑒𝑡𝑡𝑙𝑒 is the particle

115

settling velocity, 𝑘𝐵 is Boltzmann’s constant, T the temperature, 𝜇 is the solution viscosity, 𝑑 is

116

the particle diameter, and 𝛿 is the thickness of the diffusion boundary layer surrounding the plant

117

leaf, through which nanoparticles must move prior to attachment. In the present model with no

118

flow in the system, the value of 𝛿 was approximated as the radius of a plant leaf.17 These processes

119

are summarized in in Figure 1.

120

A

Homoaggregation

 Ho Ho

Dissolution

kdis Plant Attachment

 pap* v p

Heteroaggregation

 He  He B

Settling

v set

B 

121 122

Figure 1. Schematic of the key processes in the mesocosm water column included in the model formulation.



123 124

Number concentrations in the remaining compartments are formulated in a similar fashion below

125

for homoaggregates (Ho) and heteroaggregates (He) in the water column, in aquatic sediments (S),

126

and attached to aquatic plant surfaces (P):

127

128

𝑑𝐻𝑒 = 𝛼𝐻𝑒𝛽𝐻𝑒𝐵𝑆𝑆𝑛 ― 𝑘𝑠𝑒𝑡𝑡𝑙𝑒,𝐻𝑒𝐻𝑒 ― 𝑘𝑑𝑖𝑠𝑠,𝐻𝑒𝐻𝑒#5 𝑑𝑡 𝑑𝐻𝑜 1 = 𝑘 𝑛2 ― 𝑘𝑠𝑒𝑡𝑡𝑙𝑒,𝐻𝑜𝐻𝑜 ― 𝑘𝑑𝑖𝑠𝑠,𝐻𝑜𝐻𝑜#6 𝑑𝑡 𝑐 𝐻𝑜

129

𝑑𝑃 = 𝛼𝑝𝑎 ∗ 𝑝𝑣𝑝𝑛 ― 𝑘𝑑𝑖𝑠𝑠,𝑃𝑃#7 𝑑𝑡

130

𝑑𝑆 = 𝑘𝑠𝑒𝑡𝑡𝑙𝑒𝑛 + 𝑘𝑠𝑒𝑡𝑡𝑙𝑒,𝐻𝑒𝐻𝑒 + 𝑐 ∗ 𝑘𝑠𝑒𝑡𝑡𝑙𝑒,𝐻𝑜𝐻𝑜 ― 𝑘𝑑𝑖𝑠𝑠,𝑆𝑆#8 𝑑𝑡

131 132

where c is the average number of nanoparticles expected to comprise a single homoaggregated

133

agglomerate, as estimated by DLS measurements in mesocosm water (Malvern Zetasizer ZS).

134

Further, note that each compartment has an independent dissolution rate. This is because previous

135

studies have confirmed that differences in local chemistry and biological activity sometimes results

136

in very different nanoparticle dissolution rates, such as in water compared to plant surfaces or

137

sediment.18-19 Equations were solved analytically using Wolfram Mathematica v11.0.

138

Comparisons to experimental results was done by providing each of the relevant nanoparticle

139

parameters and fitting model results for removal rates from the water column over the same 1-day

140

time period as was used in experimental data fits of the initial removal phase.11 This includes all

141

relevant forms of nanoparticles and aggregates as would have been collected at 10 cm, just as were

142

collected in mesocosm experiments.

143


Page 8 of 20

Page 9 of 20


144

2.4 Parameter quantification

145

Attachment efficiency (α): Mixing studies to determine the relative αHe of each nanoparticle were

146

performed as reported previously.10, 20 A suspension of glass beads previously homogenized with

147

mesocosm water to obtain a surface coated by organics and other residual material from mesocosm

148

water12 were used as reference surface for αHe measurements. The kinetics of nanoparticle removal

149

were then used to calculate values of α for each nanoparticle type according to Equation 9:

150

ln

𝑛0

() 𝑛

= 𝛼𝛽𝐵𝑡#9

151

where, 𝑛0 is the initial nanoparticle number concentration and 𝑛 the freely disperse nanoparticle

152

number concentration at time t, β is the collision rate kernel describing contacts between

153

nanoparticles and background particles (i.e., glass beads), and B is the concentration of those

154

collector particles. Values of 𝑛 were obtained using UV-vis absorbance at a peak wavelength

155

selected for each nanomaterial. Homoaggregation was neglected in these cases due to the very

156

high collector particle concentrations (approximately 16 g/L glass beads). Assay conditions were

157

controlled to ensure that the product βB was constant across all experiments by keeping glass bead

158

concentrations and mixing speeds constant. Values of α are obtained by performing the assay under

159

conditions that ensure that α → 1 and then normalizing to this case. 10,15-16 Values of 𝛼𝐻𝑜 were

160

estimated at lab-scale by noting homoaggregation in raw mesocosm water during DLS

161

measurements.

162 163

Settling velocity (𝜈 𝑠𝑒𝑡 ): Stokes Law was used to estimate 𝜈𝑠𝑒𝑡 (Equation 10) for each material

164

using the TEM size input for free nanomaterials, the DLS size for homo-aggregates and

165

background particles size for hetero-agglomerates (Malvern ZetaSizer). All size measurements

166

from DLS measurements were taken from number-weighted results, converted using the Malvern



167

Zetasizer software, in order to avoid the size bias from intensity weighted measurements. The

168

model assumed that homoaggregates reach and maintain a single average size as measured by

169

DLS. Details on nanoparticle size and other characteristics were previously reported and may be

170

found elsewhere.11

171 172

𝑣𝑠𝑒𝑡 =

(

𝛾𝑠 ― 𝛾𝑤 18𝜇

)

𝑑2#10

173 174

In Equation 10, 𝛾𝑠 is unit weight of the particles, 𝛾𝑤 the unit weight of liquid, 𝜇 the viscosity of

175

the liquid and d the diameter of the particle or aggregate. For heteroaggregation, we assume a

176

constant background particle size distribution and that low nanoparticle concentrations will not

177

appreciably change the background particle size distribution.

178 179

Dissolution rate (𝑘 𝑑𝑖𝑠 ): Dissolution rates were obtained from previous results within mesocosm

180

studies or elsewhere in the literature. For example, a previous mesocosm experiment revealed that

181

AgNP did not dissolve within detection limits due to rapid sulfidation at the surface in the

182

mesocosms.13, 21 Additional previously-reported results found no dissolution in mesocosm water

183

or sediment of CeO2 nanoparticles.12 We further assume no dissolution of GA-SWCNT or TiO2.

184 185

2.5 System parameters

186

2.5.1 Plants

187

Water plants were composed of Egeria Densa. The specific surface area of aquatic plants was

188

assessed by quantifying the biomass concentration, the number of leaves per mass plant, and the

189

surface area per leaf (SI Figure S1) The surface area of a single leaf was obtained using ImageJ


Page 10 of 20

Page 11 of 20


190

analysis on photographs by averaging 24 leaves, 50% of which were obtained from near the tip of

191

the plant, and 50% from halfway down the stem as a representation of new and old growth leaves.

192

In these measurements, leaves were assumed to be rectangular.

193 194

2.5.2 Background particles

195

Background particles concentration was assessed by monitoring both turbidity (2100Q Portable

196

Turbidimeter, Hach) and TOC-L analyzer (Shimadzu) with an ASI-L autosampler. Water for these

197

analyses was sampled from mesocosms at a depth of 10 cm, mirroring the nanoparticle sampling depth.

198

Particulate organic concentration was calculated by subtracting the TOC in filtered mesocosm water at

199

0.45 microns from the total TOC measurement. Turbidity measurements collected over time in the

200

mesocosm water were correlated to particle mass concentrations using a turbidity calibration curve. This

201

was obtained using a known mass of mesocosm sediment suspended in filtered mesocosm water (SI

202

Figure S2). The size of the background particles was measured by static light scattering laser

203

diffractometry (Malvern Mastersizer 3000), and the size of suspended soil particles for calibration

204

closely matched the average size of mesocosm suspended solids.

205 206

3. Results and Discussion

207

Two key system parameters required to model nanoparticles attachment were aquatic plant leaf

208

surface area and characteristics of the suspended solids. The total plant dry biomass was measured

209

at the end of the experiment and was an average of 1.38 ± 0.11 g/L E. densa across 5 mesocosms.

210

A subsample of 12 plants yielded an estimate of 237 ± 14 leaves per gram of plant mass from

211

which it was calculated that each mesocosm contained approximately 327 ± 32 leaves/L of water

212

in the aquatic compartment of the mesocosms. Measurements of individual leaf surface area



213

further resulted in a total leaf specific surface area of 232.8 ± 21.1 cm2/L. To calculate 𝑎𝑝∗ , we

214

further assumed an average leaf orientation such that 𝜃~2 .

215

Unfiltered mesocosm water had a total organic carbon content of 37.3 ± 9.2 mg/L. The same

216

unfiltered mesocosm water had an average turbidity of 3.2 ± 0.6 NTU near the time of dosing.

217

After calibrating turbidity across a range of mesocosm soil concentrations in filtered mesocosm

218

water (SI Figure S2), we found this corresponds to a suspended solids concentration of 80 ± 12

219

mg/L of background particles. Static light scattering measurements yield a number average

220

diameter of 1.5 ± 0.1 µm for the background particles. This results in approximately 3100 cm2/L

221

suspended solids.

222

Nanoparticles were dosed into separate mesocosms, each with an initial mass concentration of 2.5

223

mg/L. Due to differences in particle size and material density, this resulted in a range of initial

224

number concentrations. Initial number concentrations (n0) were calculated from mass

225

concentrations using each primary or initially homoaggregated particle diameter (d) from DLS

226

measurements and material density (𝜌). These values and other key nanoparticle parameters are

227

summarized in Table 1. For modeling purposes, we also assumed that values of 𝛼𝐻𝑒 obtained from

228

experimental mixing studies were constant for a given material and were equal for

229

heteroaggregation with suspended solids and plant surfaces. This is to highlight the utility of

230

utilizing a model surface for rapid measurement of 𝛼𝐻𝑒. However, independent values could

231

feasibly be obtained for each surface should a system contain multiple, separable surface types.

232

Nanoparticle diameters were extracted from previously reported size characterization by TEM and

233

DLS measurements.11 Values of 𝛼𝐻𝑜 were estimated at lab-scale by noting homoaggregation in

234

raw mesocosm water during DLS measurements. Both GA-AgNP and PVP-AgNP displayed no

235

measurable homoaggregation after 1 hour, and so were assigned 𝛼𝐻𝑜 = 0.0. GA-SWCNT

𝜋


Page 12 of 20

Page 13 of 20


236

homoaggregated rapidly (𝛼𝐻𝑜 = 0.9) and formed large homoagglomerates, and CeO2 and TiO2

237

each exhibited moderate, slow homoaggregation (𝛼𝐻𝑜 = 0.1).

238 239

Table 1. Key nanoparticle parameters used in model calculations n0 (#/L)

𝜌 Material

d (nm)

12

(g/cm3)

× 10

𝛼𝐻𝑒

𝛼𝐻𝑜

GA-AgNP

12.0

10.5

263

0.11

0

PVP-AgNP

44.3

10.5

5.23

0.15

0

CeO2

50

7.22

5.29

0.75

0.1

TiO2

100

4.23

1.13

0.30

0.1

GA-SWCNT

200

1.9

0.314

0.15

0.9

240 241

The focus on benchmarking the model was on removal rates of each nanomaterial from the

242

mesocosm water column. We first examined the removal rates at time 0 (immediately after dosing)

243

due to 3 major processes: heteroaggregation with suspended solids, attachment to aquatic plant

244

leaves, and homoaggregation. Aggregation influences settling, and rapid homo- or hetero-

245

aggregation at time 0 will result in greater settling from the water column at later times. Because

246

these rates depend in part on the number concentration of each particle, n0, we normalized each

247

initial rate by the particle’s respective initial concentration, n0, resulting in an initial removal rate

248

fraction, n’(0)/n0, as shown in Figure 2.

249



0.35 0.30

Suspended Solids

Homoaggrega�on

Plants

n'(0)/n0

0.25 0.20

0.15 0.10 0.05 0.00

250

GA-AgNP PVP-AgNP

TiO2

GA-SWCNT

CeO2

251

Figure 2. Model results for the initial removal rates (time 0) of each nanomaterial for initial heteroaggregation with

252

suspended solids, deposition to plant surfaces and settling following initial homoaggregation. These were normalized

253

by the initial number concentrations of each particle type.

254 255

In general, attachment to plant surfaces appears to dominate heteroaggregation to suspended

256

solids. As the surface area of suspended solids was approximately 10 times that of available plant

257

surface area, this result is largely due to the differences in mechanisms of collision and attachment

258

between small mobile background particles and large, fixed plant leaves. Further, in all but one

259

case, homoaggregation was also small compared to plant attachment rates. In the case of GA-

260

SWCNT, the rate of homoaggregation dominated all other processes at time 0. We can therefore

261

expect the overall removal of GA-SWCNT to be dominated by settling of CNT aggregates, while

262

all others will be removed primarily through attachment to plant surfaces. It is noteworthy

263

however, that homoaggregation appears to be a significant process for both the GA-SWCNT and

264

CeO2 nanoparticles. Long-term transport studies in mesocosms found that both large and small

265

CeO2 nanoparticles were eventually found with greater than 90% of original concentrations in the


Page 14 of 20

Page 15 of 20


266

sediment.12 This is in agreement with a number of studies of other nanoparticles.4,

267

However, the path taken to reach the sediment is expected to significantly impact the

268

transformation and speciation of nanomaterials,18 and thus their potential impacts. Therefore, close

269

examination of the mechanisms of this transport on various time scales is critical. In this case, we

270

expect a small portion of transport to sediments to occur through settling of homo- and

271

heteroaggregates, while the bulk is likely to occur through the cycling of plant mass along with

272

attached nanomaterials. From these initial rates, we expect overall removal rates from the water

273

column to be lowest for both AgNP materials, followed by TiO2, GA-SWCNT, and CeO2.

274

The total removal rates as predicted by the model and as measured in mesocosm experiments are

275

shown in Figure 3A. The modeled removal rates were calculated from the concentration of each

276

nanomaterial remaining in the water column as a function of time. Specifically, the concentration

277

as would have been measured in the corresponding unfiltered mesocosm samples was calculated,

278

including free nanoparticles and those which had agglomerated but not yet settled. This was done

279

to ensure comparable results between predictions and experiments. In both cases, removal was

280

measured over the first 1 day after particle dosing, as this allowed for at least 90% removal for the

281

most quickly removed particles. Qualitative agreement between measured and modeled removal

282

rates is evident from Figure 3A, with the same removal order as predicted from initial removal

283

rates, GA-AgNP = PVP-AgNP < TiO2 < GA-SWCNT < CeO2. The correlation between predicted

284

and measured removal rates is shown in Figure 3B. The Pearson correlation coefficient was 0.993,

285

suggesting a very strong correlation between predictions and measurements. However, the slope

286

of 0.65 ± 0.04 suggests that removal rates were underestimated for the most quickly removed

287

particles. It is not immediately clear what the cause of this underestimation may be, as GA-

288

SWCNT and CeO2, the most quickly removed particles, are primarily removed by different


6, 14, 22-24


289

processes, homoaggregation and plants attachment, respectively. One possibility to explain this

290

underestimation is the formation of secondary, larger heteroaggregates for those particles with

291

high values of 𝛼𝐻𝑒, as observed previously.12 The correlation here between prediction and

292

measurement are similar to the previous correlation between these measurements and 𝛼𝐻𝑒 (0.997),

293

though that correlation neglected GA-SWCNT. Including GA-SWCNT, the Pearson correlation

294

coefficient between 𝛼𝐻𝑒 and measured removal rates was just 0.850. Therefore, the current model

295

more completely and accurately captures the processes governing removal of nanoparticles from

296

aquatic systems than heteroaggregation alone, although heteroaggregation was the dominant

297

process for most materials. Specifically, deposition onto plant surfaces far outpaced that with

298

suspended solids with significant implications in material speciation, transformation, and impacts.

299

The kinetic model also allowed differentiating the initial processes driving water column removal

300

(i.e. plant leaves vs suspended solids). The model further provided quantitative predictions for

301

absolute removal rates with reasonable accuracy across a range of material properties. This result

302

highlights the importance of including all relevant kinetic processes in modeling nanomaterial fate

303

and transport without assuming a single process is instant or dominant. This was especially

304

important when homoaggregation rates were significant. Further, it requires only simple inputs

305

from functional assays in order to parameterize each nanoparticle. The model therefore appears

306

both simple to implement and highly predictive of nanoparticle transport in complex aquatic

307

systems. The resulting removal rate constants and long-term solutions may be used in the future

308

for rapid exposure assessment of these materials.

309


Page 16 of 20

Page 17 of 20


5.0

Removal Rate (days -1)

4.5

A

Model

Measured

4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0

5.0

Removal Rate (Model)

4.5

AgNP-GA AgNP-PVP

TiO2

GA-SWCNT

CeO2

B

4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

Removal Rate (Measured)

310 311

Figure 3. The modeled and measured total removal rate constants (A), and the correlation between them (B) for all 5

312

nanomaterials.

313 314

Acknowledgements

315

This material is based upon work supported by the National Science Foundation (NSF) and the

316

Environmental Protection Agency (EPA) under NSF Cooperative Agreement EF-0830093 and

317

DBI-1266252, Center for the Environmental Implications of NanoTechnology (CEINT). Any



318

opinions, findings, conclusions or recommendations expressed in this material are those of the

319

author(s) and do not necessarily reflect the views of the NSF or the EPA. This work has not been

320

subjected to EPA review and no official endorsement should be inferred.

321

Supporting Information

322

Supporting information is available in 3 pages, 1 table, and 2 figures. Nanoparticle

323

characterization, leaf images for surface area measurements, water turbidity calibration.

324 325

References

326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354

1. Deng, Y. Q.; Eitzer, B.; White, J. C.; Xing, B. S., Impact of multiwall carbon nanotubes on the accumulation and distribution of carbamazepine in collard greens (Brassica oleracea). Environmental Science-Nano 2017, 4 (1), 149-159. 2. De La Torre-Roche, R.; Hawthorne, J.; Deng, Y.; Xing, B.; Cai, W.; Newman, L. A.; Wang, Q.; Ma, X.; Hamdi, H.; White, J. C., Multiwalled carbon nanotubes and c60 fullerenes differentially impact the accumulation of weathered pesticides in four agricultural plants. Environ Sci Technol 2013, 47 (21), 12539-47. 3. Geitner, N. K.; Zhao, W.; Ding, F.; Chen, W.; Wiesner, M. R., Mechanistic Insights from Discrete Molecular Dynamics Simulations of Pesticide-Nanoparticle Interactions. Environ Sci Technol 2017, 51 (15), 8396-8404. 4. Praetorius, A.; Scheringer, M.; Hungerbuhler, K., Development of environmental fate models for engineered nanoparticles--a case study of TiO2 nanoparticles in the Rhine River. Environ Sci Technol 2012, 46 (12), 6705-13. 5. Taghavy, A.; Abriola, L. M., Modeling reactive transport of polydisperse nanoparticles: assessment of the representative particle approach. Environmental Science-Nano 2018, 5 (10), 2293-2303. 6. Liu, H. H.; Cohen, Y., Multimedia environmental distribution of engineered nanomaterials. Environ Sci Technol 2014, 48 (6), 3281-92. 7. Wu, L.; Gao, B.; Munoz-Carpena, R.; Pachepsky, Y. A., Single collector attachment efficiency of colloid capture by a cylindrical collector in laminar overland flow. Environ Sci Technol 2012, 46 (16), 8878-86. 8. Hotze, E. M.; Phenrat, T.; Lowry, G. V., Nanoparticle Aggregation: Challenges to Understanding Transport and Reactivity in the Environment. Journal of Environment Quality 2010, 39 (6). 9. Hendren, C. O.; Lowry, G. V.; Unrine, J. M.; Wiesner, M. R., A functional assay-based strategy for nanomaterial risk forecasting. Sci Total Environ 2015, 536, 1029-1037. 10. Geitner, N. K.; O'Brien, N. J.; Turner, A. A.; Cummins, E. J.; Wiesner, M. R., Measuring Nanoparticle Attachment Efficiency in Complex Systems. Environ Sci Technol 2017, 51 (22), 13288-13294.


Page 18 of 20

Page 19 of 20

355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400


11. Espinasse, B. P.; Geitner, N. K.; Schierz, A.; Therezien, M.; Richardson, C. J.; Lowry, G. V.; Ferguson, L.; Wiesner, M. R., Comparative Persistence of Engineered Nanoparticles in a Complex Aquatic Ecosystem. Environ Sci Technol 2018, 52 (7), 4072-4078. 12. Geitner, N. K.; Cooper, J. L.; Avellan, A.; Castellon, B. T.; Perrotta, B. G.; Bossa, N.; Simonin, M.; Anderson, S. M.; Inoue, S.; Hochella, M. F., Jr.; Richardson, C. J.; Bernhardt, E. S.; Lowry, G. V.; Ferguson, P. L.; Matson, C. W.; King, R. S.; Unrine, J. M.; Wiesner, M. R.; HsuKim, H., Size-Based Differential Transport, Uptake, and Mass Distribution of Ceria (CeO2) Nanoparticles in Wetland Mesocosms. Environ Sci Technol 2018, 52 (17), 9768-9776. 13. Lowry, G. V.; Espinasse, B. P.; Badireddy, A. R.; Richardson, C. J.; Reinsch, B. C.; Bryant, L. D.; Bone, A. J.; Deonarine, A.; Chae, S.; Therezien, M.; Colman, B. P.; Hsu-Kim, H.; Bernhardt, E. S.; Matson, C. W.; Wiesner, M. R., Long-term transformation and fate of manufactured ag nanoparticles in a simulated large scale freshwater emergent wetland. Environ Sci Technol 2012, 46 (13), 7027-36. 14. Schierz, A.; Espinasse, B.; Wiesner, M. R.; Bisesi, J. H.; Sabo-Attwood, T.; Ferguson, P. L., Fate of single walled carbon nanotubes in wetland ecosystems. Environmental Science-Nano 2014, 1 (6), 574-583. 15. Therezien, M.; Thill, A.; Wiesner, M. R., Importance of heterogeneous aggregation for NP fate in natural and engineered systems. Sci Total Environ 2014, 485-486, 309-318. 16. Barton, L. E.; Therezien, M.; Auffan, M.; Bottero, J. Y.; Wiesner, M. R., Theory and Methodology for Determining Nanoparticle Affinity for Heteroaggregation in Environmental Matrices Using Batch Measurements. Environmental Engineering Science 2014, 31 (7), 421-427. 17. Ahn, H. S.; Bard, A. J., Single-Nanoparticle Collision Events: Tunneling Electron Transfer on a Titanium Dioxide Passivated n-Silicon Electrode. Angew Chem Int Ed Engl 2015, 54 (46), 13753-7. 18. Avellan, A.; Simonin, M.; McGivney, E.; Bossa, N.; Spielman-Sun, E.; Rocca, J. D.; Bernhardt, E. S.; Geitner, N. K.; Unrine, J. M.; Wiesner, M. R.; Lowry, G. V., Gold nanoparticle biodissolution by a freshwater macrophyte and its associated microbiome. Nat Nanotechnol 2018, 13 (11), 1072-1077. 19. Schwabe, F.; Schulin, R.; Rupper, P.; Rotzetter, A.; Stark, W.; Nowack, B., Dissolution and transformation of cerium oxide nanoparticles in plant growth media. Journal of Nanoparticle Research 2014, 16 (10). 20. Geitner, N. K.; Marinakos, S. M.; Guo, C.; O'Brien, N.; Wiesner, M. R., Nanoparticle Surface Affinity as a Predictor of Trophic Transfer. Environ Sci Technol 2016, 50 (13), 6663-9. 21. Levard, C.; Hotze, E. M.; Colman, B. P.; Dale, A. L.; Truong, L.; Yang, X. Y.; Bone, A. J.; Brown, G. E., Jr.; Tanguay, R. L.; Di Giulio, R. T.; Bernhardt, E. S.; Meyer, J. N.; Wiesner, M. R.; Lowry, G. V., Sulfidation of silver nanoparticles: natural antidote to their toxicity. Environ Sci Technol 2013, 47 (23), 13440-8. 22. Besseling, E.; Quik, J. T. K.; Sun, M.; Koelmans, A. A., Fate of nano- and microplastic in freshwater systems: A modeling study. Environ Pollut 2017, 220 (Pt A), 540-548. 23. Alimi, O. S.; Farner Budarz, J.; Hernandez, L. M.; Tufenkji, N., Microplastics and Nanoplastics in Aquatic Environments: Aggregation, Deposition, and Enhanced Contaminant Transport. Environ Sci Technol 2018, 52 (4), 1704-1724. 24. Dale, A. L.; Casman, E. A.; Lowry, G. V.; Lead, J. R.; Viparelli, E.; Baalousha, M., Modeling nanomaterial environmental fate in aquatic systems. Environ Sci Technol 2015, 49 (5), 2587-93.



401 402 403 404 405

TOC Art

406


Page 20 of 20

Formulation and Validation of a Functional Assay-Driven Model of

Recommend Documents