Colloidal stability and aggregation kinetics of nanocrystal CdSe/ZnS quantum dots in aqueous systems: Effects of ionic strength, electrolyte type, and natural organic matter

Understanding the stability and aggregation of nanoparticles in aqueous milieu is critical for assessing their behavior in the natural and engineered environmental systems and establishing their threat to human and ecosystems health. In this study, the colloidal stability and aggregation kinetics of nanocrystal quantum dots (QDs) —CdSe/ZnS QDs—were thoroughly explored under a wide range of aqueous environmental conditions. The z-average hydrodynamic diameters (z-avg. HDs) and zeta potential (ξ potential) of CdSe/ZnS QDs were measured in monovalent electrolyte (NaCl) and divalent electrolyte (CaCl2) solutions in both the absence and presence of natural organic matter (NOM)—Suwannee River natural organic matter, SRNOM to assess the dynamic growth of these nanoaggregate-QD-complexes, and the evaluation of their colloidal stability. Results show that CaCl2 was more effective to destabilize the QDs compared to NaCl at similar concentrations. An increase in NaCl concentration from 0.01 to 3.5 M increased the z-avg. HD of QD aggregates from 61.4 nm to 107.2 nm. The aggregation rates of QDs increased from 0.007 to 0.042 nm·s−1 with an increase in ionic strength from 0.5 to 3.5 M NaCl solutions, respectively. In the presence of Na+ cations, the aggregation of QDs was limited as steric forces generated by the original surface coating of QDs prevailed. In the presence of CaCl2, the aggregation of QDs was observed at a low concentration of CaCl2 (0.0001 M) with a z-avg. HD of 74.2 nm that significantly increased when the CaCl2 was higher than 0.002 M. Larger sizes of QD aggregates were observed at each level of CaCl2 concentration in suspensions of 0.002–0.1 M, as the z-avg. HDs of QDs increased from 125.1 to 560.4 nm, respectively. In the case of CaCl2, an increase in aggregation rates occurred from 0.035 to 0.865 nm·s−1 with an increase in ionic strength from 0.0001 M to 0.004 M, respectively. With Ca2+ cations, the aggregation of QDs was enhanced due to the bridging effects from the formation of complexes between Ca2+ cations in solution and the carboxyl group located on the surface coating of QDs. In the presence of SRNOM, the aggregation of QDs was enhanced in both monovalent and divalent electrolyte solutions. The degree of aggregation formation between QDs through cation-NOM bridges was superior for Ca2+ cations compared to Na+ cations. The presence of SRNOM resulted in a small increase in the size of the QD aggregates for each of NaCl concentrations tested (i.e., 0.01 to 3.5 M, except 0.1 M), and induced a monodispersed and narrower size distribution of QDs suspended in the monovalent electrolyte NaCl concentrations. In the presence of SRNOM, the aggregation rates of QDs increased from 0.01 to 0.024 nm 1 with the increase of NaCl concentrations from 0.01 to 2 M, respectively. The presence of SRNOM in QDs suspended in divalent electrolyte CaCl2 solutions enhanced the aggregation of QDs, resulting in the increase of z-avg. HDs of QDs by approximately 19.3%, 42.1%, 13.8%, 1.5%, and 24.8%, at CaCl2 concentrations of 0.002, 0.003, 0.005, 0.01, and 0.1 M, respectively. In the case of CaCl2, an increase in aggregation rates occurred from 0.035 to 0.865 nm·s−1 with an increase in ionic strength from 0.0001 to 0.004 M, respectively. Our findings demonstrated the colloidal stability of QDs and cations-NOM-QD nanoparticle complexes under a broad spectrum of conditions encountered in the natural and engineered environment, indicating and the potential risks from these nanoparticles in terms of human and ecosystem health.

The large amount of production and diversity of applications of nanoparticles greatly augments the probability of their discharge into the natural and engineered environmental systems. The concurrent risk to environmental health and public safety from such releases has generated much concern worldwide about their fate and transport in the environment after release and potential toxicity following exposure [26][27][28][29][30][31][32].
At present, a consistent body of studies on the fate, behavior, and effects of nanoparticles in the environment indicates a clear correlation between the mobility and retention of nanoparticles in various media, including soil and water, and their physical and chemical properties and the physical, chemical, and biological factors and processes in the environment. The physical and chemical properties of nanoparticles and the physical, chemical, and biological conditions and processes in the environment that affect the nanoparticles' transport behavior and fate encompass adsorption, aggregation/stabilization, dissolution, and transformation [33]. The mobility of nanoparticles in soil and water is governed by their aggregation, deposition kinetics, and straining. Although nanoparticles may aggregate to form larger colloidal size particles [34], these nanoaggregates have fractal structures [35] and impact the surface interaction processes; consequently the nanoparticles size and geometry are of environmental significance [36]. Concerning the QD nanomaterials, the presence of a shell, surface modification, and resistance to aggregation significantly enhances their mobility in sand columns [37][38][39] and in other media [26,40,41]. Environmental factors influencing the fate and transport characteristics of nanoparticles, including QDs, include: water chemistry (e.g., pH, electrolyte type, ionic strength, organic matter, and surfactant) [39,[42][43][44][45][46][47][48]; natural and engineered porous media materials (e.g., sand and soil) [47,49]; degree of saturation (e.g., unsaturated and saturated) [45,50,51]; hydraulic properties of porous media materials (e.g., flow velocity and hydraulic conductivity); fluid flow in porous media materials (e.g., saturated flow, unsaturated flow, and preferential flow phenomena such as unstable fingered flow) [42,50,52,53]; and porous media geometrical characteristics (e.g., grain size distribution and grain shape irregularity) [45].
Like transport, the bioavailability and toxicity of nanoparticles are influenced by their physico-chemical characteristics, particularly the size and surface properties that determine the nanoparticle aggregation state and stability behavior. Stability (i.e., the resistance to aggregation) has also been evaluated and is believed to play a pivotal role in determining toxicity [26,54,55]. Research on the toxicity of QD nanoparticles on the ecosystem has involved investigations of microorganisms, fungi, aquatic and terrestrial plants (e.g., algae, Arabidopsis thaliana), invertebrates (e.g., mussels), and vertebrates (e.g., crustaceans, fish, mice, and primates) [29,[56][57][58][59][60] at the cellular, subcellular, and molecular scales [61]. Moreover, with an accumulation of evidence, the adverse effect of QD nanoparticles on the ecosystem is now beyond dispute. For instance, growth inhibition in bacteria [62] and protozoa [63]; physical changes in cellular structure decreasing catalase activities of algae [28,64]; DNA damage-causing immunotoxin effects mediated in mussel cells [65,66]; damage to the hematopoietic organ and hematocytes of invertebrate (e.g., Daphnia magna) organisms [67][68][69][70]; reduced the phagocytic activity of phagocytes of fish [71]; and aberrations including malformations of the placenta and fetus in mice [72,73] have all been observed following exposure to QD nanoparticles.
The potential exposure routes for human to QDs are contaminated environmental media, workplace, and therapeutic or diagnostic administration for medicinal purposes [56]. The adverse effect of QDs on human cells (e.g., lung, kidney, liver, skin, breast, intestine, vein, and prostate) [72,[74][75][76][77][78][79] is also now well recognized. Cytotoxicity in human cells has been observed following exposure to QDs in many in vitro studies. For instance, mercapto-undecanoic acid (MUA) modified CdSe QDs caused bronchial epithelial cell apoptosis, change in gene level response, and generation of reactive oxygen species [77].
Considering the significant influence of stability on the QD nanoparticles' fate and transport [49,80,81], bioavailability [82][83][84], and toxicity [26,61,[85][86][87], it is of immense importance to understand their stability and aggregation behavior in the natural and engineered environmental systems, and thus to evaluate the interaction mechanisms and to model and forecast the fate, transport, transformation, and toxicity of nanoparticles in these environmental systems [26,88].
The role of pH in the stability behavior and aggregation kinetics of nanoparticles is well known in terms of modifying the surface charge of nanoparticles and disrupting the electrostatic energy barriers [26,42,96,97]. The variation of surface charge of the nanoparticles at different pH levels is caused by the change in the protonation-deprotonation equilibria of the functional groups of capping agent located at the surface of the nanoparticles. Rapid aggregation of nanoparticles occurs at the point zero charge (pH PZC ) because of the decrease of repulsive electrostatic interactions between the surfaces of nanoparticles, whereas stabilization of nanoparticles occurs at both higher and lower pH conditions due to the increasing repulsive electrostatic interactions between the surfaces of nanoparticles [26,40,97,113,114].
Among the factors that may affect nanoparticles' stability, the effects of ionic strength and electrolyte types have been the subject of substantive studies. Electrolytes in the environment can cause the compression of electric double layers (EDLs) of nanoparticles, resulting in their destabilization as a result of the decrease of the electrostatic repulsion energy barrier [50,85,96,97,103,115]. Compared to the monovalent electrolytes (e.g., NaCl and KCl), divalent electrolytes (e.g., CaCl 2 and MgCl 2 ) have a more pronounced effect on the destabilization of the nanoparticles in suspension by cause of the high efficiency in screening the EDLs and the bridging effect through the complexation of divalent cations [105,111,[115][116][117].
NOM existing in the natural and engineered environmental systems can alter the chemical processes and affect the aggregation of nanoparticles through various mechanisms depending upon the properties of nanoparticle and the environmental chemistry [100,105,106,118,119]. The various mechanisms involved in the interactions between nanoparticles and NOM include hydrophobic interactions, electrostatic interactions, steric interactions, hydrogen bonding, and bridging. In monovalent electrolyte and low concentrations of divalent electrolyte solutions at higher pH well above the pK a values of NOM, the presence of NOM has been found to enhance the stability of nanoparticles. The enhanced stability is caused by the electrostatic repulsions from the strongly negative surface of nanoparticles due to either the completely deprotonated functional groups or steric repulsion. In a higher concentration divalent electrolyte suspension, NOM is thought to enhance the aggregation of nanoparticles due to the bridging interactions [111,115].
The potential for synergistic interaction between electrolytes and NOM is thus important for understanding the mechanisms of QD nanoparticle aggregation for the purpose of predicting their environmental fate, transport, and toxicity. However, the interaction among QD nanoparticles and electrolytes and NOM that results in a variation of colloidal stability and aggregation kinetics is complex. Indeed, only a few studies have sought to explore the reasons that tie aggregation heterogeneity to the complex interactions of electrostatic and steric forces among particles, electrolytes, and NOM in natural surface and subsurface environments [120]. Therefore, it is critical to study the interactions between nanoparticles, the interactions between nanoparticles and electrolytes, the interactions between nanoparticles and electrolyte-NOM systems and their resulting structures for purposes of elucidating their effect on the aggregation and stability of QD nanoparticles in aqueous systems.
The objective of this study is to systematically investigate the effects of ionic strength, monovalent and divalent electrolytes, and NOM upon the aggregation behavior and the mechanisms that control the stability of QD nanoparticles in aqueous systems. To this end, we investigated the dynamic stability of QD nanoparticles as a function of monovalent (NaCl) and divalent (CaCl 2 ) cations concentration in both the absence and presence of NOM. The aggregation kinetics of QD nanoparticles were studied using time-resolved dynamic light scattering (DLS) and electrophoretic light scattering (ELS) for the various nanoparticle-aqueous systems. Linking experimental measurements and theoretical calculations, the critical coagulation concentration (CCC) values for the aggregation and the DLVO energy profiles between QD nanoparticles were defined. The results obtained through this systematic study of QD nanoparticles in aqueous systems contribute to advance the understanding of the water chemistry factors and mechanisms that influence the stability of QD nanoparticle. Such an understanding is critical to assess the environmental and toxicological risks to QD nanoparticles exposure.
The novelty of this research consists of the analysis of the mechanisms governing the interfacial interactions between the surface of QDs in the presence of NOM and different electrolytes (monovalent electrolyte NaCl and divalent electrolyte CaCl 2 ), the description and measurement of the dynamic growth of these nanoaggregate-QDcomplexes, and the evaluation of their colloidal stability in aqueous environment.

CdSe/ZnS quantum dots synthesis
The synthesis of water-dispersible CdSe/ZnS QDs that include a core of cadmium selenium (CdSe) and a protecting shell of zinc sulfide (ZnS) was achieved following fabrication processes reported in the research literature [50,[121][122][123]. The QD ligand shell is made of tetradecylphosphonic acid (TDPA) and is covered by 40% octylamineadjusted poly(acrylic acid) (OA-PAA) [124,125]. The synthesized CdSe/ZnS core-shell nanoscale crystals consist of the following elements and dimensions: the core of CdSe/ ZnS QD has a diameter of 2.3 nm and is encapsulated by a shell with a layer thickness of 1.5 nm, yielding CdSe/ZnS QDs that have a particle size of 5.3 nm in diameter. These synthesized QDs have a quantum yield that is evaluated at 30% [122]. These QDs formed have an emission wavelength that is measured at 630 nm. Transmission Electron Microscopy (TEM) characterization of the CdSe core of QDs are displayed in micrographs (Fig. 1). Knowledge of the properties of nanoscale crystal structured materials, such as QDs, at the atomic level can be accomplished by utilizing atomic number contrast scanning transmission electron microscopy (Z-STEM) [126,127]. The Z-STEM imaging of the structural details of CdSe/ZnS core/shell nanoscale crystals are presented in the research of McBride et al.

Electrolyte-Suwannee River natural organic matter-Quantum dots-based suspensions preparation
Sodium chloride (BDH®, ACS Grade) and calcium chloride dihydrate (Fisher Chemical®, ACS Grade) were used for the ionic strength and electrolyte type dependent QD aggregation experiments. Different concentrations of background electrolyte solutions were prepared by dissolving electrolytes in ultrapure diH 2 O and then stirred for 1 h. All the electrolyte-QD suspensions were prepared by mixing 1.8 mL background electrolyte solutions and 0.2 mL QDs stock suspension in a BI-SCP plastic sample cuvette (10 W × 10L × 52H mm, 3.5 mL), resulting in a QD concentration of 5 × 10 -8 M. Before incorporation, both the QD stock suspension and background solution were subjected to filtration utilizing a syringe filter unit (Millex®, Millipore-Sigma, Burlington, MA) with a 100 nm pore size membrane to eliminate contaminations. The total ionic strengths of the QD suspensions were 0.001-3.5 M for sodium chloride and 0.0003-0.3 M for calcium chloride as calculated using Eq. (1) [128,129]: where I is the ionic strength of solution (M), c i is i th species molar concentration (M), and Z i is the charge number of the i th species.
For the experiments involving electrolyte-NOM-QDbased suspensions, a fixed amount of SRNOM stock solution was added to ionic strength-modified electrolyte solutions to attain SRNOM concentrations of 10 mg/L. All background electrolyte-NOM solutions were titrated with 0.1 M sodium hydroxide (Sigma-Aldrich, Reagent Grade) or 0.1 M hydrochloric acid (BDH®, ACS Grade) to pH of 7.00 ± 0.01. The measurements of the pH values of the solutions were performed using an Orion 720A pH/Ion Meter (Thermo Fisher Scientific, Waltham, MA). A total of 42 solutions were created for experimentation including diH 2 O, two types of electrolytes -monovalent electrolyte NaCl at concentrations ranging from 0.001 to 3.5 M, and divalent electrolyte CaCl 2 at concentrations ranging from 0.0001 to 0.1 M-in both the absence and presence of SRNOM at 10 mg/L at pH 7 (Table 1).

Particle size and zeta potential measurements
The z-avg. HD of the QD nanoaggregates suspended in a wide range of ionic strengths and electrolytes in both the absence and presence of SRNOM were quantified by the time-resolved dynamic light scattering (DLS) method using a NanoBrook 90Plus Zeta Particle Size Analyzer (Brookhaven Instruments Corporation, Holtsville, NY). Immediately after mixing, the electrolyte-NOM-QD suspensions were agitated manually for 10 s to achieve a well-mixed and homogeneous suspension for z-avg. HD measurements at a 90° scattering angle. The z-avg. HD measurements data were obtained at 10 s intervals for approximately ~ 11 min to obtain the aggregation kinetics of QDs. The z-avg. HD were reported as average values and included standard deviations. All measurements were conducted at 25 °C.
The NanoBrook 90Plus Zeta Particle Size Analyzer was also utilized to measure the zeta potential (ξ potential) of QDs under various ionic strengths and electrolytes in both the absence and presence of SRNOM. Regarding particle size analysis, immediately after mixing, the electrolyte-NOM-QD suspensions placed in a BI-SCP plastic cell were agitated manually for 10 s, and the ξ potential was measured using electrophoretic light scattering (ELS). Three ELS measurements (10 cycles per measurement) were made for each sample within ~ 11 min. ξ potentials were reported as average values and included standard deviations. The temperature was maintained at 25 °C during all

Aggregation kinetics
The quantitative analysis of the aggregation kinetics of QDs was determined by the aggregation rate constant(k 11 ) , that was correlated to the rate of increase of the z-avg. HDs (dd z−h (t)) of the aggregates with time t measured by DLS, and k 11 , which is expressed as follows [105,[131][132][133]: where, N o is the initial particle concentration. The aggregation rate was computed using a linear leastsquares (LLS) regression analysis of the measured z-avg. HD data during a ~ 11 min aggregation stage [111]. The aggregation kinetics of the colloidal systems were characterized by the aggregation attachment efficiencies ( ) as a function of the experimental electrolyte types and concentrations, and in both the absence and presence of SRNOM. The attachment efficiency was then computed by normalizing the aggregation rate constant (k 11 ) acquired under reaction-limited aggregation (RLA) regimes to the aggregation rate constant ((k 11 ) fast ) taken under the diffusion limited aggregation (DLA) regimes and , which is expressed as follows [105,131,132]: The RLA and DLA regimes take place at ion concentrations below and above the CCC, respectively. Two lines of attachment efficiencies dependent upon the ion concentrations under RLA and DLA regimes were

Potential energy of interaction between the particles
Classical DLVO theory can be utilized to forecast the stability of (nano)particles in aqueous solutions by studying the interaction energy between particles. As stated in this theory, the two main forces between particles are the van der Waals attraction and the EDL repulsion [136,137]. The sum of these two forces is then used to determine either a repulsive or attractive net interaction between the particles [106,138]. The EDL repulsion (∅ EDL ) energy can be calculated and expressed as follows: In the case of r > 5 , the Hogg-Healy-Fuerstenau (HHF) formula is used [139]: In the case of r < 5 , the linear superposition approximation (LSA) expression is used [140]: The van der Waals attraction ( ∅ vdW ) energy can be calculated using the following expression [141]: The sum of these two energies, known as the total potential energy, is used to determine if the net interaction Table 1 The 42 solutions created for experimentation on the stability and aggregation of nanocrystal CdSe/ZnS QDs in aqueous systems as a function of both electrolyte type and concentration and the absence and presence of SRNOM at 10 mg/L. These solutions include diH2O, two types of electrolytes -monovalent cation Na + at concentrations ranging from 0.001 to 3.5 M, and divalent cation Ca 2+ at concentrations ranging from 0.0001 to 0. 25  where r is the radius of particles (m), is the zeta potential (V), is the permittivity of water, which is 6.95 × 10 −10 (C 2 J −1 m −1 ) [98], H is the separation distance between two particles (m), and A is the Hamaker constant of particles in water [142]. A value of 6.216 × 10 −20 J was used for A [143,144]. Further, e is the charge of an electron 1.60 × 10 −19 C , k B is the Boltzmann constant 1.38 × 10 −23 JK −1 , T is the absolute temperature, and k B T is thermal energy (at room temperature 25 • C , Furthermore, is the inverse of the Debye length (m −1 ) [145,146] and is expressed as: where, N A is the Avogadro number 6.02 × 10 23 mol −1 , Z i is the valency of the i th ion, and c ∞ i is the bulk concentration of ion species i , expressed in molL −1 . At 25 °C, is calculated as a function of the electrolyte concentration C with units of M [147][148][149][150], and is expressed as follows: For 1:1 electrolyte, Finally, Y is the scaled effective surface potential (or the asymptotic constant) [140] and is expressed as follows:

Characterization of quantum dot nanoparticles at pH 7 in diH 2 O
The z-avg. HD of QDs dispersed in diH 2 O at the neutral pH of 7 was measured by DLS prior to investigating the effect of the electrolyte type and concentration, in both the absence and presence of SRNOM on their aggregation kinetics. The QDs were well dispersed, with a z-avg.

1∕2
HD of approximately 57.5 nm. The aggregates' size distribution indicated that 9%, 48%, 36%, and 7% of the QD aggregates were in the ranges of 21-40, 41-60, 61-80, and 81-100 nm, respectively ( Figure S1). The surface charge measurements indicated that the QDs with a ZnS shell and OA-PAA surface modification used in this study were negatively charged in a diH 2 O suspension at a pH of 7. The average ξ potential of QDs in this condition was approximately − 3.99 mV (Table 2).

Effect of monovalent and divalent electrolytes on quantum dot nanoparticles
NaCl as a monovalent electrolyte and CaCl 2 as a divalent electrolyte were added to QD samples, both with suspensions at pH 7.0 to determine their effect on QD stability and aggregation. No significant change in QDs aggregate size was observed in NaCl concentrations from 0.001 to 0.875 M (except for 0.1 M). However, the effects of the monovalent electrolyte on the QD nanoparticle aggregation significantly increased when the NaCl concentration was higher than 1 M (Fig. 2a). Larger sizes of QD aggregates were observed at each level of ionic strength in suspensions of 1-3.5 M, with the z-avg. HDs of QDs increasing from 57.5 nm (diH 2 O, pH 7) to approximately 102.3 nm (3.5 M), respectively ( Table 2). Monodispersed and narrow size distribution of QDs were in suspensions with NaCl concentrations varying between 0.001 and 3.5 M (Fig. 3a, b).
In the presence of divalent electrolyte CaCl 2 , the aggregation of QD nanoparticles was detected at a low concentration of CaCl 2 (0.0001 M) with a z-avg. HD of 74.2 nm that significantly increased when the CaCl 2 was higher than 0.002 M. Larger sizes of QD aggregates were observed at each level of CaCl 2 concentration in suspensions of 0.002-0.1 M, as the z-avg. HDs of QDs increased from 125.1 nm to 560.4 nm, respectively (Fig. 2b). Monodispersed and narrow size distribution of QDs were observed in suspensions with CaCl 2 concentrations varying between 0.0001 and 0.002 M (Fig. 4a, b). Both poly-dispersed and larger size distribution of QDs were present in suspensions with CaCl 2 concentrations varying between 0.002 and 0.1 M (Fig. 4a, b).
The measured ξ potentials of QDs at different ionic strength solutions for monovalent and divalent electrolytes are shown in Table 2, and in Figs. 5a and c. These data indicated both an increase of the ξ potential (i.e., becoming less negative and sometimes slightly positive) and a decrease in magnitude of the ξ potential of QDs, with increasing ionic strength of both monovalent and divalent electrolyte concentrations. Significant differences were observed for the concentrations of monovalent and divalent electrolytes at which a reduction in the magnitude of the ξ potential was observed. The values of the ξ potential of QDs were approximately − 15.03 mV at a NaCl concentration of 0.001 M, which increased rapidly to approximately − 1.89 mV once the ionic strength increased beyond 0.875 M. The effect of the divalent electrolyte CaCl 2 on the ξ potential of QD nanoparticles was similar to that of the monovalent electrolyte. The values of the ξ potential of QDs were approximately − 10.77 mV at a CaCl 2 concentration of 0.0001 M and increased rapidly to values near 0 mV and − 7 mV once the CaCl 2 concentration increased beyond 0.03 M. However, compared to the monovalent electrolyte concentrations, at lower concentrations the presence of divalent electrolyte in suspensions enhanced both the increase of the ξ potential of the QD nanoparticles (i.e., becoming less negative) and also the decrease in the magnitude of the ξ potential of those particles. In addition, as the NaCl concentration increases in the NaCl-QDs system, the ξ potential of QDs decreased, as illustrated by the trend (linear) regression analysis (Fig. 5a), although large standard deviations are observed for ξ potential values for high concentrations of NaCl at about and above 1 M. Similarly, as the CaCl 2 concentration increases in the CaCl 2 -QDs system, the ξ potential of QDs decreased, as illustrated by the trend (linear) regression analysis (Fig. 5c)

Effect of Suwannee River natural organic matter on quantum dot nanoparticles in monovalent and divalent electrolyte solutions
The stability of QDs was also investigated as a function of both the monovalent and divalent electrolyte concentrations in the presence SRNOM at 10 mg/L. Here, an increase in the monovalent electrolyte NaCl concentration from 0.01 to 3.5 M increased the z-avg. HD of QD aggregates from 61.4 nm to 107.2 nm. The presence of SRNOM resulted in a small increase in the size of the QD aggregates for each of NaCl concentrations tested (i.e., 0.01 to 3.5 M, except 0.1 M), and induced a monodispersed and narrower size distribution of QDs suspended in the monovalent electrolyte NaCl concentrations (Fig. 3c, d).
Also in the presence of SRNOM, the effect of the divalent electrolyte CaCl 2 on the z-avg. HDs of QDs in suspensions was concentration-dependent (Fig. 2b), and an increase in the divalent electrolyte CaCl 2 concentration from 0.001 to 0.25 M increased the z-avg. HD of QD aggregates from 51.9 nm to 692.9 nm. The presence of SRNOM in QDs suspended in divalent electrolyte CaCl 2 solutions enhanced the aggregation of QDs, resulting in the increase of z-avg. HDs of QDs by approximately 19.3%, 42.1%, 13.8%, 1.5%, and 24.8%, at CaCl 2 concentrations of 0.002, 0.003, 0.005, 0.01, and 0.1 M (Table 3), respectively. A monodispersed and narrow size distribution of QDs was observed in suspensions in the presence of 10 mg/L SRNOM with CaCl 2 concentrations varying between 0.001 and 0.002 M (Fig. 4c, d). However, a poly-dispersed and larger size distribution of QDs was observed in suspensions with CaCl 2 concentrations, varying between 0.0025 and 0.25 M (Fig. 4c, d).
The influence of electrolyte type and concentration on the surface charge properties of QDs in the presence of 10 mg/L SRNOM was characterized as shown in Figs. 5b and d. The presence of SRNOM did not show a measurable difference in ξ potentials of QDs at each electrolyte concentration (both for monovalent electrolyte NaCl and divalent electrolyte CaCl 2 ) compared to the absence of SRNOM due to the overlap of the standard deviations. However, as the NaCl concentration increases in the NaCl-SRNOM-QDs system, the ξ potential of QDs decreased, as illustrated by the trend (linear) regression analysis (Fig. 5b), although large standard deviations are observed for ξ potential values for high concentrations of NaCl at about and above 1 M. Similarly, as the CaCl 2 concentration increases in the CaCl 2 -SRNOM-QDs system, the ξ potential of QDs decreased, as illustrated by the trend (linear) regression analysis (Fig. 5d)

Effect of monovalent and divalent electrolytes on aggregation kinetics of quantum dot nanoparticles
To study the effects of electrolytes type and ionic strength on the aggregation kinetics of QDs, the z-avg. HDs of QD nanoparticles were measured at different electrolyte concentrations over a period of ~ 11 min (Figs. 6 and 7). The type and concentration of the electrolytes significantly affected both the aggregation kinetics and the z-avg.
HDs. An increase in the electrolyte concentration resulted in a significant increase in the aggregation rates in most of the systems analyzed (Fig. 8). The CaCl 2 systems produced a higher set of aggregation rates than NaCl within the ranges of electrolyte concentrations analyzed. The aggregation rates of the NaCl increased from 0.007 to 0.042 nm·s −1 with an increase in ionic strength from 0.5 to 3.5 M, respectively. The size of the QD aggregates in NaCl was observed as nearly constant with time at low electrolyte concentrations (0.001 to 0.875 M), with a slight increase observed at higher electrolyte concentrations (2.5 to 3.5 M) ( Table 2 and Fig. 6). QD aggregation was observed in all suspensions with the addition of CaCl 2 , as the z-avg. HD of QDs increased with time (Table 2 and Fig. 7). Higher aggregation rates of QD nanoparticles were measured in the presence of the divalent electrolyte compared to the monovalent electrolyte for the same ionic strength. In the case of CaCl 2 , an increase in aggregation rates occurred from 0.035 to 0.865 nm·s −1 with an increase in ionic strength from 0.0001 to 0.004 M, respectively. The aggregation process was almost constant and independent of the CaCl 2 concentration again when the CaCl 2 concentration was higher than 0.004 M (Fig. 8).

Effect of Suwannee River natural organic matter on aggregation kinetics of quantum dot nanoparticles in monovalent and divalent electrolytes solutions
To examine the effects of 10 mg/L of SRNOM on the aggregation kinetics of QDs in the presence of different types of electrolytes and ionic strengths, the z-avg. HDs of QDs were measured over a period of ~ 11 min ( Fig. 6 and 7). The presence of SRNOM had significant effects on the aggregation rates of QDs in suspensions with different concentrations of NaCl and CaCl 2 (Fig. 8). The aggregation profiles In the presence of SRNOM, the size of the QD aggregates in suspensions under low NaCl concentration conditions (i.e., 0.01 to 2 M) increased slightly with time ( Fig. 6). Therefore, in the presence of SRNOM, the aggregation rates of QDs increased from 0.01 to 0.024 nm s −1 with the increase of NaCl concentrations from 0.01 to 2 M, respectively (Fig. 8a). The presence of 10 mg/L SRNOM enhanced the aggregation rates of QDs significantly compared to the absence of SRNOM in the monovalent electrolyte solutions under low NaCl concentration conditions (i.e., 0.01 to 2 M). Under high NaCl concentration conditions (i.e., 2.5 to 3.5 M), the aggregation rates of QDs in suspensions in the presence of SRNOM increased significantly with the increase of the NaCl concentrations.
In the presence of SRNOM, the z-avg. HDs of QDs increased rapidly with time once the concentration of CaCl 2 in suspension was above 0.001 M, which resulted in a rapid aggregation of QDs (Fig. 7). The aggregation rates of QDs were highly correlated with the concentration of CaCl 2 in the regimes of 0.001 to 0.003 M (Fig. 8b).
QD aggregation rates increased from 0.012 to 0.800 nm•s −1 as CaCl 2 concentration increased from 0.001 to 0.003 M (Fig. 8), respectively. The aggregation rate was then almost constant and independent of the CaCl 2 concentration when the CaCl 2 concentrations were higher than 0.003 M (Fig. 8b). As also observed in the absence of SRNOM, the CaCl 2 systems produced higher aggregation rates as the NaCl at the same electrolyte concentration in the presence of 10 mg/L SRNOM. As an overall trend, the presence of SRNOM in the divalent electrolyte CaCl 2 solutions enhanced the aggregation rates of QDs under all CaCl 2 concentration conditions examined in this study.

Stability of quantum dot nanoparticles in diH 2 O at pH 7
The water-soluble QDs with a concentration of 5 × 10 -8 M diluted by diH 2 O at pH 7 utilized in this study were well dispersed, with a z-avg. HD of 57.5 nm. Compared to the synthesized water-soluble QDs at concentration of 1.38 × 10 -5 M with a z-avg. HD of 46.8 nm, the slight   [153], respectively. Therefore, the spontaneous aggregation and the formation of QD nanoaggregates did not occur when the QDs were diluted in diH 2 O, which differed from other studies on metal-based nanoparticles, such as ZnO, in which nanoparticles easily formed large aggregates when diluted in water [154]. The stability of the CdSe QDs in water is determined by both the surface modification and the surface charge [93,153]. The measurements of the ξ potential of the nanoparticles suspended in aqueous solutions are used to assess changes in the surface charges. When the ξ potential values are higher than ± 30 mV, the charged particles may provide sufficient electrostatic repulsive forces to stabilize the particles in suspension by forming an EDL [134,155]. In this study, the QDs suspended in diH 2 O with a neutral pH had a negative charge because of the deprotonation of the poly(acrylic acid) which contained carboxyl functionalities and which was used in the QD surface coatings. This finding is in agreement with previous reports indicating a negatively charged nanocrystal CdSe/ZnS QD coated with MUA (i.e., QD-COOH) [156]. Regarding the negative sign of the charged surface, the ξ potential of QDs was lower (ca. − 3.99 mV) than the ξ potential that was reported previously. For example, the ξ potentials of poly(acrylic acid) modified CdSe/ZnS QDs in suspensions at pH 7.0 were − 45.8 mV [38] and − 53.9 mV [123]. Therefore, the limited surface charge may prove insufficient for generating an electrostatic repulsive barrier between the QDs to prevent aggregation.
To examine the interaction between particles, the total potential energy was calculated based on the classical DLVO theory of the stability of colloids. The results of the DLVO interaction net potential energy profile are presented in Fig. 9. The total potential energy is described essentially by the van der Waals attractive interaction between particles at distance less than 20 nm. Though the net total potential energy was dominated by the electrostatic repulsive interaction between particles at distances of more than 20 nm, the energy barrier was negligible with potential energies as low as 0.25 k B T . This limited repulsive energy barrier was not sufficient for maintaining QD dispersal in water as compared with the required minimum repulsive energy of 15 k B T [157]. Although the z-avg. HDs of QDs remained constant with time in diH 2 O at pH 7.0, the contribution of the electrostatic interactions was minor, thus inferring a dominant role of the steric repulsive force in ensuring the stability of QDs under this condition.

Stability and aggregation kinetics of quantum dot nanoparticles as a function of both ionic strength and electrolyte type
In this study, divalent (Ca 2+ ) electrolyte was more effective in destabilizing the QDs compared to monovalent (Na + ) electrolyte. The decrease in magnitude of the ξ potentials   [47]. When QDs were suspended in 0.0001-0.1 M divalent electrolyte CaCl 2 solutions, the QD diameters ranged from 52.5 to 685.0 nm, findings which again are corroborated by previous results on the influence of CaCl 2 on QD diameters ranging from 86 to 674 nm [47,158]. As modelled by the classical DLVO theory, the QDs were less stable when suspended in monovalent and divalent electrolyte solutions compared to diH 2 O, the results of which are corroborated by observed increases in the diameters of QD nanoaggregates and the enhanced aggregation rates. The attachment efficiency which is a function of particle size was also computed for different monovalent electrolyte NaCl and divalent electrolyte CaCl 2 concentrations (Fig. 10). At low NaCl concentrations (0.001 to 2 M), the increase in the electrolyte concentrations did not affect the aggregation rate of QDs which resulted in an attachment efficiency (α) equal to zero. At high NaCl concentrations (2.5, 3, and 3.5 M), however, each of these electrolyte concentrations induced a high aggregation rate and a high attachment efficiency. The attachment efficiency profile of QDs for the electrolyte NaCl with concentrations ranging from 0.001 to 3.5 M was atypical for the two regimes-RLA and DLA regimes-that are consistent with the DLVO colloidal theory. The attachment efficiency profile of QDs indicated the absence of distinct DLA regimes for the QDs dispersed in suspensions with monovalent electrolyte NaCl, thus compounding the difficulty of determining the CCC for NaCl. The lack of CCC in the electrolyte NaCl concentration ranges from 0.001 to 3.5 M may indicate the absence of a sufficient level of electrolytes needed to eliminate the existing energy barrier preventing aggregation between the negatively charged QDs. Similar findings detailing the failure to develop CCCs in nanoparticles suspended in electrolyte solutions are also available for both polyvinylpyrrolidone (PVP)-Ag nanoparticles and branched polyethyleneimine (BPEI)-Ag nanoparticles suspended in NaCl solutions [159]. The aggregation kinetics of PVP-Ag and BPEI-Ag nanoparticles were not conforming to the DLVO theory, and the nanoparticles did not undergo aggregation in spite of being exposed to high electrolyte concentrations. Further, an increase in the NaCl electrolyte concentrations to 8 M for PVP-Ag nanoparticles and 1 M for BPEI-Ag nanoparticles failed to create the RLA or fast aggregation, indicating the absence of CCC in both the PVP-Ag nanoparticles and the BPEI-Ag nanoparticles at the investigated ionic strength values.
The aggregation kinetics of QDs in the divalent electrolyte CaCl 2 revealed the existence of both RLA and DLA regimes. Within the RLA regime at low CaCl 2 concentrations (i.e., 0.0001 to 0.004 M), the attachment efficiency increased with an increase in CaCl 2 concentrations. Under these conditions, the increase in the electrolyte concentration elevated the degree of charge screening of QDs (i.e., becoming less negative) and hence reduced the energy barrier to aggregation, consequently causing a faster rate of aggregation. Once the electrolyte concentration reached and exceeded the CCC, the sufficient presence of electrolyte in the solution, induced a complete compression of the EDL of QDs to eliminate the energy barrier between the particles. Under such DLA conditions, QDs undergo a random walk due to Brownian motion to cluster and form QD nanoaggregates. The attachment efficiencies reached a maximum value of 1 at the CCC and were independent of the CaCl 2 concentration once past the CCC. As determined from the attachment efficiency profile, the CCC for CaCl 2 was 0.004 M in this study.
The results of the aggregation kinetics of QDs indicated that the divalent electrolyte CaCl 2 was more efficient in destabilizing the QDs than the monovalent electrolyte NaCl, a finding that is consistent with the results reported previously for other nanoparticles. Specifically, the CCC of NaCl and CaCl 2 were 0.12 M and 0.0048 M, respectively, for fullerene nanoparticles [131], 0.04 M and 0.002 M, respectively, for silver nanoparticles at pH 7.0 [133], and 0.122 M and 0.0022 M, respectively, for citrate coated silver nanoparticles [111]. In the divalent electrolyte systems, the CaCl 2 was reported to cause aggregation from the bridging effects between the particles [160,161] and a screening of the surface charge of the particles [116,162,163]. The adsorption of Ca 2+ is hypothesized as the catalyst behind the formation of interparticle bridging flocculation. As previously mentioned, QDs have a negatively charged surface. Thus, when using CaCl 2 as the electrolyte, 2 mol of Ca 2+ is added, which was double that of the per mol electrolyte used in NaCl. The divalent ions in the electrolyte were observed to promote a greater aggregation of QDs than their monovalent counterparts as they placed a The DLVO theory is useful in analyzing the interaction energies of QD nanoaggregates suspended in an electrolyte solution. The effects of type and concentration of electrolytes on the DLVO calculations for QD nanoaggregates are shown in the resulting potential energy profiles in Figs. 11 and 12. Here, a demonstrable decrease of the energy barrier is evident with a diminishing electrostatic repulsion and a more significant van der Waals attraction with an increase in the concentration of both monovalent and divalent electrolytes.
As indicated in the potential energy profile, the introduction of a low concentration of NaCl (0.001 M) resulted in a significant compression of the EDL such that the electrostatic repulsive energy was rather small (< 10 K B T ) and that the net energy between particles was attractive when the distance between particles was less than 5 nm. An increase in NaCl concentration caused a rapid decrease in the energy barriers between particles. For instance, the repulsive energy was less than 5 K B T in suspension with a NaCl concentration of 0.01 M and an EDL that was completely compressed, which eliminated the repulsive energy

Stability and aggregation kinetics of quantum dot nanoparticles as a function of ionic strength, electrolyte type, and Suwannee River natural organic matter
The presence of SRNOM destabilized the QDs at all NaCl concentrations examined (i.e., 0.01 to 3.5 M). The addition of SRNOM failed to produce appreciable effects on the calculated net potential energy (Fig. 11), and to influence the aggregation extent and rates by changing the energy barriers between QDs. The results suggested a governing role of the steric interactions between the nanoparticles, the electrolytes, and the SRNOM in determining the colloidal stability and aggregation of QDs under monovalent electrolyte conditions. Monovalent cations are known to form bridges between molecules [164][165][166] with the subsequent bridging interactions resulting in nanoparticle aggregation. For example, AuNPs coated with MUA (11-mercaptoundecanoic acid) strengthened the bridging interaction between monovalent cations and the carboxylate groups on two interacting nanoparticles. This interaction, which was stronger than the van der Waals attractions, accounted for the aggregation of like-charged nanoparticles [166]. In this study, the formation of single rod-like complexes [29,166] occurred between the SRNOM molecules and the monovalent cations, which brought QDs together by adsorbing concurrently onto at least two QDs. This bridging effect may be rational explanation for the destabilization of QDs by the SRNOM in the monovalent electrolyte. As shown in Fig. 10, the enhanced aggregation of QDs by the SRNOM was also observed in the divalent electrolyte system with the van der Waals attraction predominant in the interaction between QDs (Fig. 12). The increase in the magnitude of the net potential energy between QDs was obtained by the SRNOM in the divalent electrolyte system. Specifically, CaCl 2 concentrations of 0.003 and 0.005 M respectively caused an increase in the net potential energy between QDs at a separation distance of 10 nm from 13.30 to 20.41 K B T and from 26.63 to 30.32 K B T in the presence of SRNOM. The increased net potential energy may account for the observed enhancement of aggregation of QDs by the SRNOM. Furthermore, the CaCl 2 may enhance the aggregation of QDs through the mechanism of divalent cation bridging. The interaction between divalent cation and SRNOM resulted in the formation of macromolecule aggregates of SRNOM through calcium complexation, aggregates that subsequently bridge the QDs and cause the enhanced destabilization. These bridging phenomena are also proposed to explain the enhanced aggregation of fullerene nanoparticles by HA [105], hematite nanoparticles by alginate [167], zinc oxide by NOM [34], citrate-coated silver nanoparticles by HA [111], graphene oxide nanomaterials by NOM [168], titanium oxide by EPS [169], and copper nanoparticles by DOC [170]. The degree of aggregation formation between QDs through cation-SRNOM bridges was superior for Ca 2+ cations compared to Na + cations due to more effective and strong interactions of Ca 2+ cations with SRNOM than Na + cations.

Conclusion
The environmental behavior and toxicological risks of nanoparticles are related to their physico-chemical properties, particularly their surface coating properties, and the chemistry of the environment encountered. Although it is impossible to characterize the behavior of nanoparticles in all possible chemical scenarios found in the environment, our study aimed at contributing to the understanding of QDs in the presence of salts and organic matter commonly encountered in the terrestrial and aquatic environment. Results demonstrate that both the types and concentrations of monovalent (Na + ) and divalent (Ca 2+ ) electrolytes will significantly influence the particle size distribution, the stability behavior, and aggregation kinetics of QD nanoparticles suspended in these aqueous systems. The formation of QD nanoaggregates was controlled by the valency and concentration of cations. The increase in the cation concentrations induced a decrease in the electrostatic repulsion as a result of the enhancement of the compression of the EDL. Divalent cations generated larger QD nanoaggregates than monovalent cations at the same concentrations. Under monovalent (Na + ) solutions, the aggregation of QDs was limited in that the steric forces generated by the original surface coating of QDs prevailed, whereas under divalent (Ca 2+ ) solutions, the aggregation of QDs was enhanced due to the bridging formations between the Ca 2+ in solution and the carboxyl group located in the surface coating of QDs. The QD suspended in divalent (Ca 2+ ) solutions remained stable at low Ca 2+ concentrations (< 0.002 M) or formed nanoaggregates of smaller sizes compared to QDs in high Ca 2+ concentrations ranging from 0.004 M to 0.1 M.
The changes in solution chemistry exhibited a substantial effect upon the interactions between QDs and SRNOM. In the presence of SRNOM, QDs suspended in the monovalent (NaCl) and divalent (CaCl 2 ) electrolyte solutions formed electrolyte-SRNOM-QDs' complexes with higher aggregation rates compared to the absence of SRNOM. Despite similarities in the effect of the SRNOM on the general aggregation trends of QDs between the monovalent and divalent electrolytes, each behaved differently in terms of the mechanisms that accounted for the enhanced aggregation. In the monovalent electrolyte solutions, the destabilization induced by the SRNOM was attributed to the bridging effect caused by the single rod-like complexes of monovalent (Na + ) cation and the SRNOM. In the divalent electrolyte solutions, the destabilization by the SRNOM was imputed to the bridging effect of the SRNOM aggregates produced through the calcium complexation between Ca 2+ cations in solution and the carboxyl group on the SRNOM. The resulting aggregation between QDs through cation-NOM bridges was superior for Ca 2+ cations compared to Na + cations due to more effective and strong interactions of Ca 2+ cations with NOM than Na + cations.
Equations predicting the attachment efficiency indicated the presence of both the RLA and DLA regimes only for the divalent (Ca 2+ ) electrolyte solution in both the absence and presence of SRNOM. The experimental CCCs of the QDs at pH 7.0 were 0.004 M and 0.003 for the divalent (Ca 2+ ) electrolytes in both the absence and presence of SRNOM, respectively. The absence of a distinct DLA for the monovalent (Na + ) electrolyte solution did not allow to determine the CCC of the systems in both the absence and presence of SRNOM. It was also determined that although classical DLVO theory could simulate the QD nanoparticles behavior, certain physico-chemical properties of QD nanoparticles manifested in non-DLVO behavior in systems involving nanoparticles, organic ligands coating, NOM, and salts. Examples comprise the utilization of polymer and organic ligand coatings to alter the surfaces of QD nanoparticles in the presence of monovalent electrolytes, which caused steric stabilization and a reduction in nanoparticle aggregation. Surface chemistry was also determined as most important in controlling the stability of QD nanoparticles in aqueous systems.
Although SRNOM was selected to represent NOM in aquatic systems in our study, it prevented an assessment Calculation of DLVO theory includes the contributions of the van der Waals and electric double layer (EDL) interactions. The net energy (Net) shown is the sum of EDL repulsion and van der Waals attraction between two approaching particles ◂ of the effects of the fractions of NOM such as humic acids and fulvic acids. Therefore, the influence of these NOM fractionations on the interfacial interactions between metal ions and QD nanoparticles, and the mechanism of aggregation of QD nanoparticles should be further evaluated in future studies.
Thus, elucidating the stabilization/aggregation, interaction, and transformation of QD nanoparticles in aqueous systems of different chemistries will allow the forecasting of exact QD nanoparticle characteristics, aggregation states, and stability behavior under a broad range of environmental settings. Aggregation and sedimentation of nanoparticles that are released in the natural and engineered environmental systems have critical effects on the fate and transport, reactivity, bioavailability, and toxicity of these nanoparticles. Consequently, although the risk from QD nanoparticles suspended in water with salts and NOM under relevant environmental conditions may be lessened as nanoparticles are inclined to aggregation, sedimentation and/or adsorption onto the surfaces of media encountered, QDs released in water with high salt concentrations and NOM may threaten ecosystems and public health as they may settle in the sediments of aquatic systems. Such knowledge is essential for evaluating the environmental and toxicological risks of exposure to QD nanoparticles.

Conflict of interest
On behalf of all authors, the corresponding author states that there is no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.

Fig. 12
Total potential energy versus inter-particle distance according to DLVO theory, showing the influence of representative divalent electrolyte CaCl 2 concentrations and the absence and presence of the SRNOM at 10 mg/L for QDs suspended in aqueous chloride salt solutions: a 0.001 M, b 0.002 M, c 0.003 M, d 0.005 M, e 0.01 M, and f 0.05 M. Calculation of DLVO theory includes the contributions of the van der Waals and electric double layer (EDL) interactions. The net energy (Net) shown is the sum of EDL repulsion and van der Waals attraction between two approaching particles