Primary particle size analysis by scanning electron microscopy (SEM) indicated size ranges between 288 and 549 nm for the MCs, while transmission electron microscopy (TEM) portrayed size ranges between 7 and 50 nm for the NCs (Table 1). Stability settling characteristics of the colloid suspensions as affected by size and contaminant load are shown in Figure 1 and Table 2. A particle is considered stabile if it tends to remain dispersed in solution [2] . Based on size alone, the settling experiments showed NCs to be significantly more stable in suspension (α = 0.05) after 48 hours than the MCs in the presence and absence of contaminants (Figure 1). After 48 hours, in the absence of contaminants mineral MCs had an average of 80% colloids in suspension while NCs had an average of 96% (Table 2). In comparison, the bio-MCs and bio-NCs had 36% and 58%, respectively (Table 2). This highlights two things: first, the enhanced mobility and contaminant transport potential of the NCs over their corresponding MCs, and second, the greater stability of the mineral colloids over the bio-colloids without contaminants present. Upon the addition of 2 mg/L contaminants, the mineral NCs had an average of 35% more colloids in suspension than corresponding MCs (65% versus 30%, respectively) (Table 2). Despite greater stability in the mineral colloids at 2 mg/L contaminant additions, the bio-NCs had as much as 18% colloids in suspension in the presence of 10 mg/L contaminants, compared to only 1% on average for the mineral colloids (Table 2) [26] [27] .

The average intensity weighted (Z-average) hydrodynamic diameters (d_h) can be used to indicate stability as well (Figure 2). Since d_h is related to the size of aggregates formed by the colloids, the smaller values would

^‡SD = Standard Deviation from duplicate or triplicate measurements. ^§SEM = Scanning Electron Microscopy, TEM= Transmission Electron Microscopy data represent the average smallest dimension of 300 representative particles from three to eight images of each size fraction as measured until the average and standard deviation values had less than 10% variation.

indicate smaller aggregate size, thus more dispersive colloids. In the absence of contaminants the d_h values complemented the stability trends, indicating greater stability in the mineral NCs over that of the bio-NCs (Figure 1 and Figure 2). For the MCs, the anomaly was the kaolinitic MC fraction, which showed the smallest d_h (Figure 2), yet in the stability settling experiments (Figure 1) portrayed lower stability than the smectitic and mixed MC fraction. Other than the MC kaolinitic fraction, the smectitic colloids showed consistently small d_h values and high stability in the settling studies compared to other colloid compositions (Figure 1 and Figure 2). Since d_h also reflects on the size of aggregates formed by the colloids, the smaller values indicate smaller aggregate size thus more dispersive colloids. The d_h values in the absence of contaminants suggested that the MCs were nearly four-times larger than the NCs, with an average d_h of 842 nm for the MCs and 224 nm for the NCs (Figure 2). The d_h also showed compositional trends, with mean comparisons showing the bio-colloids to be significantly larger than the mineral colloids (α = 0.05). While the larger d_h values (as measured by DLS) compared to the crystallite diameters (as measured by TEM/SEM) may indicate some aggregation in both size fractions, the larger DLS size range may also be partially explained by the disproportionate influence of the larger particles on the overall average diameter compared to the smaller particles (Figure 1), or by potential surface coatings on the particles [8] [16] .

The d_h values of the kaolinitic MCs, the smectitic and mixed NCs were unaffected by the addition of contaminants, but increased significantly for the smectitic MCs, and the kaolinitic, and biosolid NCs. Surprisingly, the biosolid MCs experienced a considerable reduction in size in the presence of contaminants (Figure 2). The higher d_h values in the presence of contaminants were probably caused by enhanced aggregation due to increased ionic strength from the addition of polyvalent ions (Figure 2) [8] [16] . MCs showed greater aggregation potentials in the presence of contaminants than NCs through larger shifts in d_h values, especially in the biosolid and mixed MCs (Figure 2). The largest d_h shifts with contaminant additions occurred in the kaolinitic and biosolid NCs and the mixed and biosolid MCs (Figure 2). Greater contaminant loading potential of cation contaminants onto larger surface areas could cause aggregation of other negatively charged colloids, thus creating pseudo-aggregates that increase the intensity weighted averages measured (Figure 2) [8] [16] . In spite of a size increase of about 230 nm, the smectitic MC fraction showed a slight reduction in stability comparable to that of the NC fraction. In contrast, the stability of the kaolinitic, mixed, and biosolid MC fraction decreased significantly regardless of the d_h size changes (Figure 2) following the mixed > kaolinitic > biosolid trend. The addition of the 2 mg·L⁻¹ contaminant load caused a consistent stability reduction across all NCs following the sequence smectitic = kaolinitic > mixed > biosolid. However, the stability of the NCs was generally higher than that of the MCs under the 2 mg·L⁻¹ contaminant load, with the exception of the smectitic MCs which showed similar stability to the smectitic NCs.

Mineralogy appears to have played a partial role in the stability processes of the MCs and NCs (Table 3). In the absence of contaminants, the smectitic MCs had greater stability than the kaolinitic and biosolid MCs, but equal stability to the mixed MCs (Figure 1). In the absence of contaminants, the smectitic, mixed and kaolinitic NCs were more stable than the biosolid NCs (Figure 1). After additions of 2 mg·L⁻¹ contaminants the mineral colloids showed more surface repulsion than the biosolid colloids. The smectitic and kaolinitic NCs were more stable than the mixed and biosolid colloids, while within the MC fraction, the smectitic and mixed MCs were more stable than the kaolinitic and biosolid colloids (Figure 1). With the exception of the bio-NCs, the addition of 10 mg·L⁻¹ contaminants overwhelmed the available surface area of the colloids and induced significant flocculation (Figure 1). The greater stability (Figure 1) of the smectitic colloids over that of other mineralogies is likely due to higher surface charge potential, with the greater charge densities in smaller sized colloids resulting in greater repulsion (Table 1, Figure 1) [1] [28] . Lower stability in colloids containing kaolinite are likely due to the tendency of kaolinite to remain flocculated at pH less than 7.5 [1] [9] . Additionally, smaller sized fractions that are high in kaolinite, Al and Fe hydroxides may have enhanced surface charge density contributing to destabilization potentials [2] [8] . This is evidenced by the attachment of smaller nanoparticles to several MC surfaces (Table 1). Additionally, the presence of hydroxy-interlayered vermiculite may have released Al and promoted flocculation processes in the mixed and kaolinitic colloid fractions (Figure 1 and Figure 2; Table 3) [2] . In the stability kinetics experiments, mixed MCs were more stable than kaolinitic MCs after the addition of 2 mg·L⁻¹, with a reversed stability sequence for the NCs (Figure 1). The mixed NCs were less stable than the kaolinitic NCs in the presence of 2 mg·L⁻¹ contaminants, probably due to their higher Fe-hydroxide content (Figure 1,

^*K = Kaolinite, Ge = Goethite, Gi = Gibbsite, Q = Quartz, M = Mica, Sm = Smectite, MVI = Mica-Vermiculite Interstratified, HIV = Hydroxy- interlayered Vermiculite, ND = Not Detected.

Table 3). One confounding observation is that there are higher quantities of flocculating agents (Fe and Al hydroxides, goethite, and gibbsite) in the NCs, yet they remain more stable than their corresponding MCs (Table 3, Figure 1). The answer to this dilemma may lie in the EDS data which indicated greater (Al + Fe):Si ratios in the NCs than in their corresponding MCs. Increased (Al + Fe):Si ratios have shown to enhance humic acid sorption capacities and stabilize colloid suspensions [16] [29] .

The smectitic MCs showed typical montmorillonitic honeycomb shapes with edge-face interactions (Table 1). Montmorillonite can be more dispersive than other minerals because weak Van der Waals’ forces between oriented tactoids allow penetration between layers by water and exchangeable cations, which may inhibit stable aggregate formation [9] . The smectitic NCs indicated less particle to particle attachment than the MCs (Table 1). Morphologies displayed by the mixed mineralogy MCs showed tightly aggregated honeycomb-like plates, while the mixed NCs had various shapes encompassing tubes, rods, plates and hexagons within a tightly aggregated mineral matrix (Table 1). The greater morphological variation in the shapes of the mixed NCs in TEM images may indicate some surface deformation (Table 1). Similar HIV images were shown with associated iron minerals [21] . Kaolinitic MC SEM images had the typical hexagonal shapes displayed by kaolinitic minerals, while TEM images of the NCs showed much smaller hexagonal shapes with a noted increase in iron minerals embedded within and binding the hexagonal aggregates (Table 1) [22] . Kaolinitic minerals tend to flocculate below pH 7, which is representative of the colloid suspension pH range [9] (Table 4).

The morphological compositions of the biosolid MCs, as shown in their SEM images, displayed various aggregations of organic materials with some quartz grains present, while the NC TEM images lacked distinct particle shapes or displays of individual particle morphology, indicating less integrity of sample and potential dissolution (Table 1). Overall, within the mineral colloids, SEM images displayed more tightly held aggregates within the MCs (Table 1), indicating greater flocculation potential (Figure 1) as opposed to the more loosely held aggregates displayed by their corresponding NCs in TEM images (Table 1).

Additionally, TEM images suggested potential particle morphology alterations from surface disorder, which may have modified NC stability behavior (Table 1). TEM images of the smectitic and mixed NC showed similar disorder and potential structural abnormalities (Table 1), with particles adhering to each other to display extended and oblong morphologies. The cause of such abnormalities has been suggested to be impurities or oriented aggregation resulting from forced structural incorporation of previously sorbed species that could alter surface sites and influence stability and reactivity characteristics [5] [30] - [32] .

With the exception of the biosolids, NCs had greater surface area than the MCs despite the SEM indications of nanoparticle adhesion to MC surfaces (Table 1) [5] [6] [30] . Surface area showed trends by composition, with the largest coming from the biosolid materials, followed by the smectitic, mixed, and kaolinitic colloids, respectively (Table 1). In addition to size effects, increased surface area values measured in the NCs could be the result of organic surface coatings or the presence of iron hydroxides maximizing sorption of the EGME [33] [34] . The larger surface areas in the NCs also indicate greater potential for contaminant sorption and increased repulsion of similarly charged particles, thus enhancing stability potentials.

Unadjusted pH values ranged from 4.9 to 5.4 for both the NC and MC fractions (Table 4). Such a narrow pH range evinces that pH values may not have promoted differences in stability between the differing colloid compositions or sizes, but may have contributed to a greater overall flocculation tendency (Table 4, Figure 1) [2] . However, in spite of this narrow pH range, there were significant differences in stabilization potentials (Figure 1) across most of the studied colloid fractions. The same colloids in higher pH environments would likely be even more dispersive, as evidenced by the zeta potential titrations above pH 6 (Figure 3).

The electrical conductivity (EC) and ionic strength (IS) of the colloids were relatively low with the NCs showing higher overall values than their corresponding MCs (Table 4). Additionally, the bio-colloids had greater EC and ionic strength values than most mineral colloids except for the mixed NCs (Table 4). Higher ionic strength values for the NCs could decrease stability [35] and may have contributed to lower fractional recoveries as compared to MCs (Table 1 and Table 4). However, at the natural range these higher EC and ionic strength values of the NC suspensions did not appear to significantly deter the overall stability of the NCs as shown in Figure 1. Increased ionic strength effects from contaminant additions resulted in lower NC and MC stability, likely due to flocculation in the presence of contaminants, with greater flocculation at higher contaminant concentrations (Figure 1) [6] [16] . Increased ionic strength has been shown to reduce the effectiveness of Coulomb repulsion by shielding the charge of two approaching particles [6] . This effect on Coulomb repulsion may explain the stability findings after contaminant additions of 10 mg·L⁻¹, where significantly lower colloid suspension concentrations indicated flocculation in all but the bio-NC fractions (Figure 1). Coulomb repulsions appeared to overcome ionic strength effects at 2 mg·L⁻¹ contaminant concentrations.

The higher SAR values of the NCs over that of the MCs are more consistent with their stability trends than the IS values, indicating greater dispersion potentials in the NCs over their corresponding MCs (Table 4). The higher CEC values of the NCs over the MCs are also complimentary to their larger and likely more reactive surface areas (Table 4). As expected, the smectitic colloids had the greatest CEC within the mineral colloids, followed by the mixed mineralogy and kaolinitic colloids (Table 4). The higher OC content of the bio-colloids

^‡SD = Standard Deviation from duplicate or triplicate measurements. ^§Ionic Strength (IS) = Estimated using Griffin and Jurinak’s equation where IS (mol·L⁻¹) = 0.0127 × Electrical Conductivity (millimhos·cm⁻¹) (1973). ^#CEC = Cation Exchange Capacity by sum of cations. ^††SAR = Sodium Adsorption Ratio. ^‡‡OC = Organic Carbon (Total Carbon-Dissolved Organic Carbon).

may have contributed to their higher CEC over that of the mineral colloids (Table 4).There were higher exchangeable Ca²⁺ and Mg²⁺ concentrations in the NCs than the corresponding MC fractions (Table 4). Divalent cations can promote flocculation and may explain why NCs were found attached to MC surfaces likely due to bridging effects (Table 1 and Table 4) [16] .

NCs also contained higher OC content than their corresponding MCs which may have enhanced their stability over corresponding MCs (Figure 1, Table 4) [2] [8] [26] . While some studies indicated enhanced dispersibility and stability with increased OC content due to both charge and stearic stabilization [4] [16] [26] [27] [36] , other findings suggested weak correlations [1] . It is likely, that the presence of differing OC functional groups caused a different behavior. Referring back to the d_h findings (Figure 2), the greatest aggregation potentials with contaminant additions occurred in the biosolid colloids that had a combination of greater OC content and surface area (Table 1 and Table 4). Correlations between amount of C measured per unit surface area (SA) showed that the greatest aggregation potentials occurred in the colloids with the highest C:SA ratios (kaolinitic and biosolid NCs, respectively) (Figure 2, Table 1 and Table 4). The difference in OC functional group as well as C:SA ratio may also explain differences in stability findings (Table 1 and Table 4; Figure 1). It is also possible that OC bridges together micro-nano-aggregates that are more stable in suspension than individual particles. Despite higher OC and potential formations of micro-nano-aggregates, NCs were generally still more stable in the absence and presence of contaminants than were MCs (Figure 1).

NC zeta potentials were more negative than that of their corresponding MCs, with both sizes showing increasingly negative zeta potentials with increased pH (Figure 3). Mineral NCs exhibited more negative zeta potentials than their corresponding MC counterparts and the bio-colloids, especially above pH 6 (Figure 3). Based on the assumption that zeta potentials approximate the charge residing in the diffuse layer of the electrical double layer, the negative zeta potential values suggest that all colloids bear a net negative surface charge. The increased negative charge with increasing pH is ascribed to the deprotonation of phyllosilicate edge sites, and indicates repulsion energies of the particles, suggesting increased stability potential of NCs over corresponding MCs (Figure 3), especially in negatively charged subsurface environments [37] .

In the mineral colloids, there was never a point where the zeta potential reached zero (Figure 3), suggesting isoelectric points (the pH of zero mobility) of <4 for all colloids [38] . This was especially surprising in fractions where mineral compositions were dominated by kaolinite, which has an isoelectric point of 4.25 [38] . Apparently, the presence of other minerals and organic carbon surface coatings have considerably altered the isoelectric behavior of the mineral colloids [39] .

The biosolid colloids had differing trends from the mineral colloids, where the bio-NCs became less negative with increasing pH as opposed to bio-MCs that showed an increased negative charge with increasing pH (Figure 3). This may suggest a lower stability potential in the bio-NCs than in the bio-MCs, which is the opposite of what was seen in stability kinetics experiments (Figure 1). The less negative zeta potentials exhibited by the bio-colloids than the mineral colloids may be due to the greater ionic strength of the bio-colloids, especially those of the bio-NCs (Table 3), whose higher ionic strength likely overwhelmed potential negative surface charges generated by carboxyl groups above pH’s 2.5 and 6 [38] .

In the single point zeta potential measurements, additions of 2 and 10 mg·L⁻¹ mixed contaminants resulted in NC zeta potentials which were still more negative than those of their corresponding MCs, with both sizes showing trends based on composition (Figure 4). This display of greater stability in the NCs over the MCs complements the stability kinetics studies, suggesting that their stabilization potentials overcame their supposed high surface energy, even in the presence of up to 2 mg·L⁻¹ contaminants [6] . It is noted here that in order to mimic natural conditions, the pH of the zeta potential measurements varied, so the change in zeta potential is likely a reference to colloidal stability potential (Figure 3 and Figure 4). Overall, the mineral colloids indicated greater stability over bio-colloids through higher negative zeta potentials in the presence of 0, 2 and even 10 mg·L⁻¹ contaminant concentrations, although evidence of flocculation was more prevalent after 10 mg·L⁻¹ contaminant additions (Figure 1 and Figure 4). While some aggregation was implied with increasing ionic strengths of contaminants by slight shifts towards less negative zeta potentials (Figure 4), corresponding to increased d_h values in the presence of contaminants (Figure 2), the NCs indicated significantly greater stability potentials over their corresponding MCs (Figure 1). Colloids with the most negative zeta potential usually have the largest resistance to flocculation [16] , which was true for the smectitic, mixed, kaolinitic and biosolid NCs in the absence and presence of 2 and 10 mg·L⁻¹ contaminants (Figure 3). The MCs did not show the same stability trends based on composition, and surprisingly, the smectitic MCs were the least stable upon contaminant additions as compared to the other mineral (and even the biosolid) MCs (Figure 1 and Figure 4). The lack of stability indicated by lower negative zeta potentials in the smectitic MCs upon addition of contaminants (despite their higher initial zeta potential) may be a function of their smaller C:SA ratio (Table 1 and Table 4). The greater shifts in their zeta potentials (v27 mV without contaminants shifting to −18, then −11 mV with 2 and 10 mg·L⁻¹ contaminant additions, respectively) were most likely materialized by greater cation contaminant attraction to their high initial negative charge and to larger surface area availability (Table 1, Figure 4). Despite this large shift in the smectitic MC zeta potential, the stability kinetics studies indicated that smectitic MCs were more stable than any of the other mineral and biosolid MCs (Figure 1). The disparity of the zeta potential data to that of the stability kinetics data suggests that while the zeta potentials are consistent in predicting colloid stability of minerals versus biosolids, they maybe inconsistent in predicting stability trends based on mineralogy.

The implications may be that colloid stability may not be accurately predicted from zeta potential and d_h measurements in the presence of contaminants. Also, it emphasizes placing more weight on stability settling experiments than chemical data alone for transport predictions (although chemical characterization should still be considered important for explaining portions of the overall model). Finally, this showcases how irrelevant the average surface charge of colloids may become in contaminant transport behavior because of spatial heterogeneity and the importance of considering additional multiple relevant factors to achieve reliable assessments of environmental risks [17] .