In the Search for Nanospecific Effects of ... - ACS Publications

Mar 25, 2019 - Eva Blomberg,. †,‡ and Inger Odnevall Wallinder. †. † ...... (all else assumed equal). This was indeed observed in a study by H...
1 downloads 0 Views 760KB Size
Subscriber access provided by Drexel University Libraries

Critical Review

In the search for nano-specific effects of dissolution of metallic nanoparticles at freshwater-like conditions – a critical review Jonas Hedberg, Eva Blomberg, and Inger Odnevall Wallinder Environ. Sci. Technol., Just Accepted Manuscript • Publication Date (Web): 25 Mar 2019 Downloaded from http://pubs.acs.org on March 25, 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 39

1 2 3

Environmental Science & Technology

In the search for nano-specific effects of dissolution of metallic nanoparticles at freshwater-like conditions – a critical review

4

Jonas Hedberg,*,a Eva Blomberg,a,b Inger Odnevall Wallindera

5

a

6

and Health, Department of Chemistry, Division of Surface and Corrosion Science, Stockholm, Sweden

7

b

8

* corresponding author, e-mail: [email protected]

KTH Royal Institute of Technology, School of Engineering Sciences in Chemistry, Biotechnology

RISE Research Institutes of Sweden, Division Bioscience and Materials, Stockholm, Sweden

Abstract

9 10

Knowledge on relations between particle properties and dissolution/transformation characteristics of

11

metal and metal oxide nanoparticles (NPs) in freshwater is important for risk assessment and product

12

development. This critical review aims to elucidate nano-specific effects on dissolution of metallic

13

NPs in freshwater and similar media. Dissolution rate constants are compiled and analyzed for NPs of

14

silver (Ag), copper (Cu), copper oxide/hydroxide (CuO, Cu(OH)2), zinc oxide (ZnO), manganese

15

(Mn), and aluminum (Al), showing largely varying (orders of magnitude) constants when modelled

16

using first order kinetics. An effect of small primary sizes (>100 mg/L). For the

230

yellow and orange boxes, the ionic metal solubility is intermediate (0.01-5 mg/L) and the solubility

231

can also be sensitive to changes in experimental conditions (e.g. DOM and Cl- content).

232

Table 1 clearly shows that dissolution experiments of Al and Cu NPs as a function of DOM

233

concentration will be influenced by the solubility of the metal ions if the concentrations are close to

234

their respective saturation level. Several investigations on Cu metal and Cu oxide NPs have been

235

performed at oversaturated conditions at different NOM concentrations.69, 78 Such conditions can result

236

in largely lower predicted dissolution rate constants and half-lives compared with unsaturated

237

conditions.49 Very slow release (months) of Cu ions from CuO NPs has been observed at

238

undersaturated conditions.49 Some dissolution rate constants in the literature may also not be

239

representative for realistic environmental conditions (NP concentrations in the order of µg/L or

240

lower1) due to too high NP concentrations investigated at laboratory conditions. Environmental

241

settings further likely act as sinks at which dissolved metal ions from NPs become immobilized via

242

different metal complexation and settling processes.76

10 ACS Paragon Plus Environment

Page 11 of 39

Environmental Science & Technology

243

As illustrated in Table 1, the presence of Cl- in solution influences the solubility of Ag+. The release of

244

Ag in Cl--containing solutions may hence, depending on the Ag+/Cl- ratio, form insoluble AgCl-

245

complexes.79 This formation has been thoroughly discussed in terms of fate and dissolution of Ag

246

NPs.53, 80-82

247

A schematic depiction of the connection between AgCl speciation and the dissolution rate of Ag NPs

248

is shown in Figure 2 based on data from Levard et al.79 The figure shows that the formation of soluble

249

AgCl species increases the dissolution rate, while the formation of insoluble AgCl-species results in

250

decreased dissolution rates (oversaturated conditions with respect to soluble Ag+ species).

Soluble Ag+ species (%)

100 Formation of soluble AgCl: Higher dissolution rate of Ag NPs compared with pure water conditions

80 60 40 Formation of insoluble AgCl: lower dissolution rate of Ag NPs compared with pure water conditions

20 0 1

10

100 1000 + Cl /Ag ratio

10000

100000

251 252

Figure 2. Fraction of soluble Ag+ species as a function of the Cl-/Ag+ ratio in pure water

253

complemented with information on the connection between the Cl-/Ag+ ratio and the dissolution rate

254

of Ag NPs. Data based on the work of Levard et al.79

255 256

The results presented in Figure 2 show that it is very likely that investigations of Ag NPs at different

257

Cl-/Ag+ ratios result in largely different dissolution rates as the formation of soluble or insoluble AgCl

258

species may differ.79 As a consequence, the toxic response will be non-linear with respect to added Ag

259

NP concentrations since ionic Ag in general has a large impact on toxicity.83, 84 Knowledge on solution

11 ACS Paragon Plus Environment

Environmental Science & Technology

Page 12 of 39

260

speciation of released Ag is thus essential when interpreting toxicity and dissolution results of Ag NPs

261

in chloride-containing solutions.79, 85

262

The role of Cl- is moreover manifold as it also interacts with the particle surface.84 This interaction

263

often reduces the protective ability of the surface oxide, sometimes seen as the occurrence of local

264

corrosion events on for example stainless steel and Al metal.86 The presence of Cl- has furthermore

265

shown to enhance the dissolution of Ag NPs at undersaturated Ag+ conditions.87 This has been

266

deduced using an experimental design using nanosphere lithography where Ag NPs were attached to a

267

substrate and thereby hindered to agglomerate.87 This approach is appealing as an increased Cl-

268

concentration will increase the extent of particle agglomeration and further confound any comparison

269

between different Cl- concentrations. These aspects will be discussed below.

270 271

First-order rate dissolution constants at freshwater-like conditions and principal component

272

analysis to visualize experimental conditions

273

Dissolution rates constants obtained using first-order rate dissolution kinetics (eq. 2) are presented in

274

Figure 3A for NPs of Cu, CuO/Cu(OH)2, Ag, ZnO, Mn, and Al.19, 35, 36, 51, 54, 58, 59, 70, 71, 78, 88-90 Complete

275

data and references are given in supporting information (Table S2). The dataset comprises freshwater-

276

relevant media with a pH varying between 4.5 and 8.6 (mean value=6.9), an ionic strength varying

277

from 0 to 100 mM (mean value=49 mM), and a NOM concentration between 0 and 100 mg/L (Table

278

2). Some NPs were coated with capping agents such as citrate or PVP (Table S2). All results reflect a

279

test temperature varying between 20 and 25 °C. The dissolution rate constants have been normalized

280

to the specific geometrical surface area, which is based on the primary particle size assuming spherical

281

particles. The applicability of such normalization depends on several approximations including the

282

effect of agglomeration that will be discussed below.

283 284

Table 2. Dissolution rate constants of metal and metal oxide NPs in freshwater-like media calculated

285

based on first-order kinetics (eq. 2) and normalized to the specific geometrical area (more details given

286

in supporting information in Table S2).

287 12 ACS Paragon Plus Environment

Page 13 of 39

Environmental Science & Technology

NPs

Mean

Range of dissolution rate,

Number of

dissolution rate

10-90% range (mole m-2 h-1)

unique data

pH range

k (mole m-2 h-1) CuO/Cu(OH)2

1.9·10-3

4.4·10-6- 7.6·10-4

10

5.8-7.7

Cu

4.8·10-3

4.8·10-4- 1.0·10-2

25

6.2-7

Ag

9.0·10-6

1.3·10-6- 2.9·10-5

43

4.5-7.7

ZnO

1.4·10-4

8.1·10-5- 2.6·10-4

36

6.4-8.6

Mn

4.5·10-6

4.5·10-6

4

6.2

Al

4.9·10-6

3.9 - 7.0·10-6

4

6.2

288

289 13 ACS Paragon Plus Environment

Environmental Science & Technology

290

Figure 3. A: First-order dissolution rates for NPs of CuO, Cu, Ag, ZnO, and Al at freshwater-like

291

conditions. B: PCA plot of observed dissolution rates shown in A, highlighting differences in

292

temperature, NOM concentration, loading (NP concentration), primary particle size, and pH.

Page 14 of 39

293 294

The PCA plot for all reported dissolution rates are displayed in Figure 3B and describes differences in

295

experimental settings including NOM concentration, primary particle size, temperature, loading (NP

296

concentration), and pH. The dotted lines in Figure 3B represent the experimental parameters in terms

297

of the principal components. For example, the pH of the majority of investigations on ZnO dissolution

298

is higher than in the other investigations and line hence up far out on the “pH” dotted line. The first

299

two principal components in this analysis account for 55% of the variance of the dataset. This means

300

that there are variations between the parameters that are not included in Figure 3B.

301

In line with previous analysis,9, 23 the dataset shows Ag and Al NPs to dissolve among the slowest

302

compared to more rapid kinetics reported for Cu and ZnO NPs. However, the scattering in dissolution

303

rate constants is large, and largely connected to varying exposure settings, seen in Figure 3B. For

304

example, ZnO NPs were predominantly investigated at different experimental conditions compared

305

with the other NPs including differences in primary size, pH, and NP concentration. These differences

306

in conditions may influence dissolution and disable direct comparisons,91 aspects discussed below.

307

Metal oxide NPs such as Fe-oxides, CeO2, and Al2O3 are not included in Figure 3 due to lack of

308

detailed dissolution rate constants in the literature. Fe-oxides and CeO2 show slow dissolution rates at

309

freshwater-like conditions with half-lives in the order of months, while corresponding rates for Al2O3

310

are in the range of weeks.9 Such slow rates would position these NPs in the lower region of Figure 3A.

311

TiO2 and Au NPs are not included in the compiled dataset due to their very low solubility at near

312

neutral pH conditions such as freshwater, which makes any determination of dissolution rate constants

313

very difficult.92-94 Literature findings show however that Au NPs can dissolve even at non-acid

314

conditions as a result of interaction with macrophages and in cell medium.95, 96

315

Other equations apart from the first-order equation (eq. 2) have been proposed to describe NP

316

dissolution, for instance taking into account effects of particle size, pH, and oxygen partial pressure.51,

14 ACS Paragon Plus Environment

Page 15 of 39

Environmental Science & Technology

317

63, 81, 90, 97

318

phenomena clustered in Figure 1.

This will be discussed next in connection with the different dissolution processes and

319 320

Influence of particle size on dissolution rate constants and solubility

321

A reduction in particle size will in general result in an increased dissolution rate due to increased

322

surface area.25, 98, 99 An increased surface area with decreased NP size promotes in turn dissolution due

323

to an increased number of surface sites that can take part in the dissolution process.25 A reduced

324

particle size can also result in a surface oxide with more defects and edges that further can promote

325

dissolution, especially for nano-sized particles ( 20 nm (kbulk, Cbulk) are collected from the studies depicted in Table 4. The

354

dashed line represents bulk-like conditions in terms of solubility or dissolution rate constant.

355 356

Table 4. Compilation of reported investigations used to compare NP solubility and dissolution rate

357

constants. Nanoparticle ZnO

Primary particle size (nm) 50, 100

Hydrodynamic radius (nm) N/A

Medium

Observation

Reference

pH 7, 0.01 M ammonium

Higher dissolution rate constant for bulk Zn

Avramescu et al.110

16 ACS Paragon Plus Environment

Page 17 of 39

Environmental Science & Technology

acetate ZnO

30

>178

0.01 M Ca(NO3)2 solution in Milli-Q water, pH 7.5 HEPES buffer, pH 7.5 1 mM NaNO3, pH 7

ZnO

4-130

2000-3000

Ag

5-20

13-24

Ag

6-70

10-72

Ultrapure water, pH 7

Ag

5-38

61-156

1 mM NaHCO3, pH 8

when normalized to specific geometric surface area Similar dissolution rate constants (not surface area normalized) and solubility for bulk ZnO and ZnO NPs Higher solubility for smaller-sized NPs Higher dissolution rate constant for 5 nm compared with 20 nmsized particles, normalized to specific geometric surface area Higher dissolution rate constant, normalized to specific geometric surface area, for smallsized NPs Higher solubility for smaller sized NPs

Franklin et al.37

Mudunkotowa et al.104 Mollerman et al.34

Peretyazko et al. 35

Ma et al.46

358 359

Figure 4 shows nano-specific effects of small primary NPs with significant changes taking place for

360

particles sizes less than approximately 15 nm.103 The trend qualitatively follows the Gibbs free energy

361

and Kelvin equation (eq. 4). The Kelvin equation assumes that the surface free energy is independent

362

on particle size. However, large variations in surface energy values have been reported for ZnO (ca.

363

0.06-1.31 J/m2).104 These differences in surface energy may indicate nano-specific effects as may

364

reflect a size-dependence.28

365

Despite forming agglomerates, small primary NPs can still show nano-specific size effects. This can

366

be seen from increased hydrodynamic sizes compared with primary particle sizes, Table 4. Nano-

367

specific effects of the curvature (eq. 4) or increased amount of defects can hence be maintained also

368

upon agglomeration in the sense that the smaller sized NPs retain at least some of their more reactive

369

dissolution behavior also when present in an agglomerated state. Even though the normalization of

370

dissolution rate constants to specific geometric surface areas in Figure 4 is a crude approximation, the

371

results indicate nano-specific effects for NPs sized less than 15 nm with higher dissolution rate

372

constants for smaller NPs when normalizing to the surface area.

373

There are also examples of a lack of effect on dissolution rate constants for primary NPs sized less

374

than 15 nm,38, 80 but the reason is unresolved. One explanation could be the use of far higher ionic 17 ACS Paragon Plus Environment

Environmental Science & Technology

Page 18 of 39

375

strength (seawater38) than in the studies compiled in Figure 4. This could imply a stronger binding

376

between the particles, which would reduce any nano-specific effect. This explanation is supported by

377

the work of Allen et al. whom showed higher oxidation potentials of monodisperse and agglomerated

378

Au NPs sized 50 nm compared with smaller sized Au NPs (4, 15 nm), and a shift in potential for the

379

smaller NPs when agglomerating to similar values as the 50 nm-sized NPs.

380

The primary particle size will moreover change over the course of the experiment as the NPs dissolve,

381

which in turn will affect the Gibbs free energy and the surface area and hence also the dissolution rate

382

constant. Vogelsberger et al. have performed extensive modelling and experimental work on

383

dissolution and solubility of SiO2 NPs.44, 45, 111-113 The reduction of particle size due to dissolution was

384

taken into account when constructing models able to estimate changes in surface charge and surface

385

potential for NPs of different size. The dissolution process was moreover shown to be highly

386

dependent on the NP mass to solvent ratio.113 This is not the same effect that was discussed in the

387

previous section on the solution complexation capacity, but rather a nano-specific effect that for

388

example will influence the nucleation of new NPs (Ostwald ripening).113 Ostwald ripening has also

389

been observed for dissolution of Ag NPs sized below 20 nm.35 The reported modelling work of SiO2

390

NPs44, 45, 111-113 is very encouraging and the application of these models to more reactive NPs would be

391

a very important contribution to an improved understanding of the dissolution of metallic NPs.

392

Models for dissolution rates have in some cases been constructed without considering the primary

393

particle size.97 The size of agglomerates formed in solution has instead been included as illustrated in a

394

model for ZnO NP dissolution.97 The influence of agglomeration and surface area on dissolution of

395

metal and metal NPs is discussed next.

396 397

The influence of agglomeration and aggregation on dissolution of metallic NPs

398

As already discussed in several reviews,42, 114 NPs occur in freshwater-like media as agglomerates

399

(particles held together by relatively weak physical interactions), aggregates (particles fused together,

400

a largely irreversible process), heteroagglomerates (NPs attached to other particles such as naturally

401

occurring colloids), or as free NPs.17 This means that the description of size and surface area in

402

solution is a distinctly complex assembly of different particle states, each with different dynamic 18 ACS Paragon Plus Environment

Page 19 of 39

Environmental Science & Technology

403

behavior.42, 115, 116 Adsorbed NOM plays an important role for the extent of agglomeration of metal and

404

metal oxide NPs in natural waters.42, 43 NOM can provide stabilization through steric repulsion or

405

electrostatic forces upon its adsorption,117, 118 but there are also examples where NOM induces

406

agglomeration due to charge neutralization and bridging.78, 119-121

407

Agglomeration of NPs has in several cases been shown to reduce the rate of dissolution due to a

408

reduced surface area.25 It has also been shown that agglomeration will reduce diffusion of species

409

involved in the dissolution reactions and thereby further reduce the dissolution rate.122, 123 This

410

reduction has for example been shown for PbS NPs that were more rapidly dissolved when freely

411

exposed in solution compared when exposed in more confined (nm scale) conditions.124

412

The effect of agglomeration on the actual surface area of NPs available for dissolution is however

413

poorly understood.63 In some cases, agglomeration does not lead to a large reduction of the dissolution

414

rate constant,68 also shown in Figure 4 for NPs sized less than 15 nm in media of low ionic strength.

415

Since agglomeration processes play an important role for the dissolution process, it is desirable to

416

normalize dissolution rate constants to surface areas and also to find relationships between

417

agglomerate structures and rate constants to obtain mechanistic insights on NP dissolution.

418

Unfortunately, no analytical method currently exists that can measure the reactive surface area in

419

solution. The BET specific surface area is therefore often used for normalization. However, since this

420

method measures the surface area at dry conditions, its relevance for solution conditions with

421

agglomeration processes taking place is questionable.111 Another approach sometimes used to estimate

422

the surface area is the average hydrodynamic particle size in solution, determined by means of laser

423

scattering techniques. However, such estimates also have drawbacks as the calculations for example

424

assume spherical particles and do not account for agglomerate porosity. Another option is to use the

425

size distribution to estimate the surface area, however still with the uncertainty connected to the

426

reduction in area due to possible strong binding between particles and dissolution effects (as described

427

above).63

428

Another way to approach the complexity of agglomeration is to consider the fractal dimension (DF) of

429

the agglomerate.125 DF is the power that an equivalent radius of a NP agglomerate scales to in order to

430

scale to its mass. A DF of 3 hence corresponds to a solid sphere (no porosity) and a value of 1 19 ACS Paragon Plus Environment

Environmental Science & Technology

Page 20 of 39

431

describes a rod. NP agglomerates can be divided into diffusion limited agglomerates (DLCA) with DF

432

values of approximately 1.8, and reaction limited cluster agglomerates (RCLA) with a DF value

433

between 2.1 and 2.5.115 The DF value can be determined via experimental methods such as small-angle

434

x-ray scattering,115 static light scattering,126 or the volumetric centrifugal method (VCM).127 A high

435

fractal dimension should result in a lower dissolution rate constant compared with a lower DF value as

436

more compact agglomerates have a lower surface area (all else assumed equal). This was indeed

437

observed in a study by He et al.,47 whom formulated a model to account for changes in surface area

438

due to agglomeration. The model showed that an increased DF and size of the agglomerates resulted in

439

lower dissolution rate constants.47 Observed relationships between the normalized apparent dissolution

440

rate constant (kd*, see definition in47) of a given metal NP (Ag NPs, primary particle size 50 nm) in

441

NaCl and its agglomerate size as a function of DF are displayed in Figure 5, using the model of He et

442

al.47

Normalized apparent dissolution rate constant, kd*

1,0 DF 1.7 DF 1.9 DF 2.1 DF 2.3 DF 2.5

0,8 0,6 0,4

More compact agglomerates 0,2 0,0 50

100 150 200 250 Agglomerate size (nm)

300

443 444

Figure 5. Modelling of apparent dissolution rate constants of Ag NPs (primary particle size set to 50

445

nm) in NaCl (5-500 mM) as a function of agglomerate size for different DFs. The calculations are

20 ACS Paragon Plus Environment

Page 21 of 39

Environmental Science & Technology

446

based on the work of He et al.47 The results are presented as dissolution rate constant normalized (kd*)

447

to the dissolution rate of primary 50 nm-sized Ag NPs.

448 449

The model of He et al. introduced the parameter α that reflects the proportion of accessible reactive

450

sites on primary particles compared to the total surface area. This parameter was approximated to 0.1,

451

as deduced from curve fitting.47 The impact of different Cl- concentrations on the dissolution rate was

452

taken into account since Cl- influences the dissolution rate via surface interactions (as previously

453

discussed) and if non-considered, would confound any conclusions based on agglomerate size and

454

fractal dimension. DLCA-agglomerates with a DF value of approximately 1.7 were observed in

455

solutions of high ionic strengths (500 mM NaCl) and RCLA-agglomerates at lower ionic strength (50

456

mM NaCl).47 Other effects can be observed for different Cl-/Ag ratios as Chambers et al. observed that

457

the formation of solid AgCl in solution induced RCLA.128 Reported DF values should however be

458

used with caution since the multidimensional nature of the agglomerates has not always been taken

459

into account in the analysis of DF values.129

460

Other modelling methods exist to estimate changes in surface area upon NP agglomeration.63 Jiang et

461

al. modelled as an example a reduced surface area of ZnO NPs with time due to dissolution,68 and

462

David et al. constructed a model that with information on the radius of the agglomerates and the NP

463

concentration could predict the dissolution rate of ZnO NPs.97 The structure of the agglomerates was

464

most probably similar for these exposure conditions as the dissolution rate otherwise would have been

465

influenced by different surface areas of agglomerates of different DFs. Estimates of the surface area

466

from theoretical predictions coupled with dissolution experiments of SiO2 NPs have been done by

467

Vogelsberger et al.44

468

Existing models are promising but in need of further experimental data and validation,63 including

469

parameters for fractal modelling.130 Second order dissolution rate equations have been employed for

470

Ag NPs to capture the initially fast release of Ag ions followed by slower kinetics due to effects such

471

as particle agglomeration.50 Such second-order equations have in several cases been shown to fit

472

experimental data better compared with first-order equations,50 but need further validation with respect

473

to experimental data for different kind of NPs. 21 ACS Paragon Plus Environment

Environmental Science & Technology

Page 22 of 39

474

The surface characteristics of the NPs influence their agglomeration behavior and hence the trend of

475

reduced surface area for dissolution reactions. The DLVO theory describes the resulting surface forces

476

between particles, classically taking into account electrostatics and van der Waals forces.131 Revised

477

versions of DLVO also consider steric interactions and solvation forces.132 Hammes et al. have

478

modelled the electrostatic component for different freshwaters representative of Europe,133 showing

479

variations due to differences in ionic strength.133 That modelling did however not take into account the

480

presence of NOM. The van der Waals forces depend on intrinsic NP properties and are particularly

481

large for metal NPs.75 This results in strong attractive inter-particle forces (and hence lower DFs) that

482

reduce the surface area compared with metal oxide NPs (assuming that other surface forces are

483

equal).134

484

Heteroagglomeration of NPs has been described as the most important process for the environmental

485

dispersion of NPs as it likely dominates over homoagglomeration.10, 135 Heteroagglomeration can be

486

related to an attachment coefficient of the NPs to other particles or materials.136, 137 However, there are

487

no experimental investigations on relationships between heteroagglomeration parameters such as the

488

attachment efficiency and the dissolution of NPs. These are important aspects to consider in future NP

489

dissolution studies.

490

More efforts are needed in order to predict and determine the actual surface area to more accurately

491

quantify dissolution and understand the importance of agglomeration. Information on agglomeration

492

sizes was unfortunately missing in several of the datasets of Figure 2, which prohibited the use of this

493

parameter in the PCA plot in Figure 3B. Still, with a single value of hydrodynamic size there is still an

494

uncertainty of several orders of magnitude of connecting this size to a surface area due to effects of

495

particle size distribution and particle fusing or lack of the same.63 A method to standardize

496

measurements of NP stability in simulated environmental media has recently been proposed by

497

OECD.138 This experimental protocol may in the future be further developed to include estimates of

498

fractal dimensions and surface area. The generation of dissolution rate data normalized to surface area

499

would simplify the deduction of possible nano-specific mechanisms and further facilitate comparison

500

with bulk dissolution data.80, 139

22 ACS Paragon Plus Environment

Page 23 of 39

Environmental Science & Technology

501 502

NP-interactions with ligands such as natural organic matter

503

It is expected that dissolution processes can have nano-specific attributes due to changes in NP

504

characteristics with size (e.g. defects, surface charge, curvature)43 140 that will influence the adsorption

505

of ligands (Figure 1).27, 103 The number of surface kinks and defects increases with decreasing NP size,

506

an effect observed to influence the adsorption geometry of oxalic acid on 5 versus 32 nm-sized TiO2

507

NPs.27 Investigations of ZnO NPs in the presence of citrate revealed enhanced dissolution (normalized

508

with the BET surface area) only for the smallest sized particles (4 nm) compared with small-sized NPs

509

(5-20 nm).104 This was explained by ligand-induced dissolution due to the adsorption of citrate to flat

510

terraces that masked any increased dissolution at kinks and defects for the small-sized NPs.104

511

NOM is the most important group of ligands in freshwater and is mainly composed of humic and

512

fulvic acids. It is difficult to draw general conclusions on the effect of NOM on the underlying

513

dissolution mechanisms of metal and metal oxide NPs compiled in Figure 3 due to the large variety in

514

NPs and NOM characteristics. Humic acid is the most common NOM in the dataset. The NOM was

515

however not characterized in some studies as it, for example, was derived from natural sources such as

516

river water.

517

Increased dissolution rates of metallic NPs have been reported for ZnO NPs,90, 141 (only at high pH105),

518

Al NPs,78 Cu NPs,41, 69, 78, 91, 142 and CuO (in amino acids) in the presence of NOM.143 This effect is the

519

same as expected for larger sized particles and bulk material as an increased dissolution is connected

520

to a relatively high affinity of the functional groups of NOM to the metallic surface. 144 In general

521

(valid also for bulk material), it has been observed that the adsorption of covalent (inner-sphere)

522

monodentate complexes results in increased dissolution rates.139 Multinuclear complexes are

523

conversely less prone to promote the extent of dissolution since their adsorption to multiple surface

524

sites makes the detachment of the ligand-metal complex less likely.145

525

A change in the interaction between NOM and NPs due to nano-specific attributes mentioned above

526

can influence the dissolution process. This depends for example on the binding geometry between

527

functional groups of NOM and the metal (discussed above) in addition on the thickness and surface 23 ACS Paragon Plus Environment

Environmental Science & Technology

Page 24 of 39

528

coverage of adsorbed NOM. Observations of such effects are unfortunately so far largely lacking. A

529

study on Cu NPs and humic acid suggests interactions between carboxylate groups and the oxidized

530

copper surface followed by a relatively slow build-up and gradual change in the structure of the

531

adsorbed NOM layer (hours). This adsorption process resulted in enhanced dissolution.78 Surface

532

interactions with NOM can hence enhance the dissolution of metallic NPs via destabilization of

533

adsorbed metal-ligand complexes. This process is caused by the breakage of metal-oxygen bonds due

534

to localization of electrons to the adsorbed-ligand metal complex,146 a process that is proportional to

535

the concentration of adsorbed ligands. The stability of metal-oxide bonds can, as mentioned previously

536

(Kelvin effect), be reduced for small NPs (