Carboxylic Acids Accelerate Acidic Environment-Mediated Nanoceria Dissolution

Abstract

Ligands that accelerate nanoceria dissolution may greatly affect its fate and effects. This project assessed carboxylic acid contribution to nanoceria dissolution in aqueous, acidic environments. Nanoceria has commercial and potential therapeutic and energy storage applications. It biotransforms in vivo. Citric acid stabilizes nanoceria during synthesis and in aqueous dispersions. In this study, citrate-stabilized nanoceria dispersions (~ four nm average primary particle size) were loaded into dialysis cassettes whose membranes passed cerium salts but not nanoceria particles. The cassettes were immersed in iso-osmotic baths containing carboxylic acids at pH 4.5 and 37 °C, or other select agents. Cerium atom material balances were conducted for the cassette and bath by sampling of each chamber and cerium quantitation by ICP-MS. Samples were collected from the cassette for high-resolution transmission electron microscopy observation of nanoceria size. In carboxylic acid solutions, nanoceria dissolution increased bath cerium concentration to > 96% of the cerium introduced as nanoceria into the cassette and decreased nanoceria primary particle size in the cassette. In solutions of citric, malic, and lactic acids and the ammonium ion ~ 15 nm ceria agglomerates persisted. In solutions of other carboxylic acids, some select nanoceria agglomerates grew to ~ one micron. In carboxylic acid solutions dissolution half-lives were 800 to 4000 h; in water and horseradish peroxidase they were ≥ 55,000 h. Extending these findings to in vivo and environmental systems, one expects acidic environments containing carboxylic acids to degrade nanoceria by dissolution; two examples would be phagolysosomes and in the plant rhizosphere.

Introduction

Nanoceria (~ 1 – 100 nm cerium oxide, CeO₂) is auto-catalytically redox active, cycling between Ce³⁺ and Ce⁴⁺ (Reed et al., 2014). It is used as a diesel fuel additive (Dale et al., 2017), an abrasive in chemical mechanical planarization in integrated circuit manufacture (Speed et al., 2015), as a catalyst in storage batteries, and as a catalyst structural support (Senanayake et al., 2013). Nanoceria has therapeutic potential to treat conditions with an oxidative stress/inflammation component. It has been shown beneficial in models of many conditions, including cancer (Gao et al., 2014, Pesic et al., 2015), radiation damage (Madero-Visbal et al., 2012, Li et al., 2015), bacterial infection (Alpaslan et al., 2017) and sepsis (Selvaraj et al., 2015), wounds (Chigurupati et al., 2013), stroke-induced ischemia (Kim et al., 2012), neurodegenerative disease (Heckman et al., 2013), cardiovascular dysfunction (Minarchick et al., 2015), liver dysfunction (Oró et al., 2016), and retinal degeneration (Wong and McGinnis, 2014). It has been noted that “Good colloidal stability and narrow size distribution are essential for the successful biomedical application of ceria nanoparticles” (Kim et al., 2012). The above referenced studies used three to five nm ceria, often coated (with citrate/EDTA, dextran, PEG, or other agents) to prevent agglomeration.

Nanoceria is quite insoluble in circumneutral aqueous solutions. Solubility of five to 10, 40, and < 5000 nm ceria was extremely poor in water at neutral pH (Geraets et al., 2012). No dissolution of 18 nm ceria was detected after 30 minutes in MES buffers at pH 4.5 to 7.5 (Mirshafiee et al., 2018). Dissolution of seven nm ceria was reported to be insignificant after 72 h at 37 ^◦C in physiological saline (pH 7.0) and artificial phagolysosomal fluid (pH 4.5) (He et al., 2010). Solubility of five and 10 nm ceria was null at pH 7.4 and 0.2 and 0.3% after 24 h incubation in artificial lysosomal fluid (pH 5.5) at room temperature (Cho et al., 2012). Three and < 1% of 8 to 9 nm ceria dissolved over 28 days in pH 4 and 7 artificial soil solution, respectively (Cornelis et al., 2011). Solubility of 10 to 200 nm ceria in water and DMEM + 10% fetal calf serum (FCS) after 24 h at 25 °C was < 0.001% (Wohlleben et al., 2013). Solubility after incubation for 28 days in water was 0.002%; in DMEM/FCS, phagolysosomal simulant fluid, PBS, and fasted state simulated intestinal fluid < 0.001%; and 0.02% in 0.1 N HCl (Keller et al., 2014). Solubilization was not seen after 28 days in phosphate-buffered saline or synthetic phagolysosomal simulant fluid, or after seven days in fasted state simulated intestinal fluid. After 1 day in 0.1 N HCl, 0.24% of the nanoceria dissolved (Molina et al., 2014). The log K solubility product of ~ 5.6 nm ceria was found to be −59.3 in 0.01 M NaClO₄ (Plakhova et al., 2016).

Poorly soluble nanomaterials are of concern in biological environments where they may persist for months, e. g., in mammals, or accumulate with repeated exposure (Laux et al., 2017). The mass of cerium in the liver, spleen, and bone marrow (the sites that accumulate the greatest amount of nanoceria) accounted for 45% of a single intravenous dose of ~ 30 nm ceria 90 days later, compared to 60% of the dose after one day, demonstrating in vivo persistence (Yokel et al., 2012). Cerium concentrations in the liver and spleen five months after a single intravenous dose of 2.9 nm ceria were 12 and 116% of the concentrations after one day, respectively (Heckman et al., 2013). One hundred twenty days after an intravitreal injection of three to four nm ceria particles, 90% of the injected cerium remained in the eye and 70% was in the retina, resulting in estimated half-lives of 525 and 414 days, respectively (Wong et al., 2013). Nanoceria partially degraded in rat liver and spleen over 90 days after a single intravenous administration of ~ 30 nm ceria, resulting in formation of one to three nm, crystalline, cerium-containing particles that had more reduced (Ce³⁺) surface cerium. This is thought to have occurred through a dissolution/re-crystallization process (Graham et al., 2014, Graham et al., 2017, Graham et al., 2018).

Some dissolution of eight nm ceria was seen in pH 4, but not 7 or 9 artificial soil solution after 28 days (Cornelis et al., 2011), and of 32 and 78 nm ceria in pH 1.65, and less rapidly in 4.45, but not in pH 7.45 or 12.4 media over 120 h, evidenced as increased cerium in the supernatant (Dahle et al., 2015). Ascorbic acid and glutathione released cerium from four nm mesoporous silica CeO₂ nanoparticles (Muhammad et al., 2014). Twenty-five nm ceria incubated with citric and ascorbic acids or catechol as reducing agents at pH 5.5 released cerium into the supernatant over 21 days (Rui et al., 2015). Citric acid on the surface of four, nine, and 39 nm ceria stabilized it against sedimentation due to agglomeration (Siriwardane, 2012). Citric acid is used as a stabilizing agent for hydrothermal syntheses of nanoceria (e.g., Masui et al., 2002, Zhang et al., 2007, Muhammad et al., 2014), and forms various coordination complexes with cerium (Bobtelsky and Graus, 1955, Leal, 1959, Zhang et al., 2007, Zhou et al., 2008).

In some systems, free radical sources are known to degrade nanomaterials. For example, single-walled carbon nanotubes degraded in the presence of peroxidases and hydrogen peroxide (H₂O₂) (Allen et al., 2008, Kagan et al., 2010, Andón et al., 2013). Nanosilver dissolution rate correlated with the number of thiols per biomolecule (Marchioni et al., 2018). Other than the dissolution of nanoparticles that are considered quite soluble, such as manganese oxide, silver, and zinc oxide, the mechanisms of nanoparticle dissolution have not been well described.

To test the hypothesis that nanoceria must first be reduced to enable solubilization, nanoceria was incubated for 21 days at pH 5.5 in the presence of KH₂PO₄ or KH₂PO₄ and citric acid. Citric acid increased the biotransformation of nanoceria, with partial conversion to Ce³⁺, in the presence of phosphate and ascorbic acid, suggesting a role for reducing carboxylic acids in nanoceria dissolution (Zhang et al., 2012). A similar study was conducted in the presence of citric and ascorbic acid or catechol that resulted in increased nanoceria solubility (Rui et al., 2015). They did not study citric acid alone, to ascertain if it was sufficient to solubilize nanoceria. A study of single components of artificial plant root exudates, that included succinic and malic acids at a non-reported pH, showed that they facilitated nanoceria dissolution over 5 days (Zhang et al., 2017).

The present study was conducted to address the hypothesis that individual ligands are sufficient to mediate nanoceria dissolution at the pH of the phagolysosome. Carboxylic acids, via their ability to form complexes with cerium, the ammonium ion as a potential proton source, and horseradish peroxidase (HRP) + H₂O₂ as a peroxidase (peroxidases are known to mediate carbon nanomaterial dissolution (Vlasova et al., 2016)) were studied. To address this, an acellular system was used to model nanoceria’s chemical fate during in vivo bioprocessing. Specifically, a nanoceria dispersion (one ml) in a Slide-A-Lyzer™ dialysis cassette with 2 kD MWCO regenerated cellulose membranes was immersed in a 200 ml iso-osmotic bath containing carboxylic acids or select agents. In most cases, the bath pH was 4.5 (to model lysosomal pH), shown to mediate nanoceria dissolution (Dahle et al., 2015). Citric acid, known to complex cerium, and structural analogues of citric acid containing a carboxylic acid and hydroxyl group were included to identify the carboxylic acid chemistries that enhance nanoceria dissolution or stabilization. Adipic and pimelic acid were included because they appear to bind solely to the (100) cubic crystal face of nanoceria (Grulke et al., 2014). Bath and cassette samples were repeatedly collected for cerium analysis by inductively coupled plasma mass spectrometry (ICP-MS) and size and shape determination by high-resolution transmission electron microscopy (HRTEM). In some cases, electron energy loss spectroscopy (EELS) was used to determine valence of the dissolving nanoceria.

Materials and Methods

Materials

The chemicals, their sources, purity, and CAS #s were: adipic acid, TCI, ≥ 99%, 124-04-9; ammonium nitrate, Fisher, ACS grade, 6484-52-2; citric acid monohydrate, Fisher, ACS grade, 5949-29-1; DL-3-hydroxybutyric acid sodium salt, Chem Impex Int’l Inc., 100.30%, 150-83-4 & 306-31-0; DL-malic acid, Alfa Aesar, 98%, 6915-15-7; glutaric acid, Acros organics, 99%, 110-94-1; hydrogen peroxide 3% W/W, BDH chemicals, 7722-84-1; horseradish peroxide type II, Sigma, 150-250 U/mg, 9003-99-0; lactic acid, TCI, ≥ 85%, 50-21-5; pimelic acid, Alfa Aesar, 98+%, 111-16-0; sodium acetate, VWR, ACS grade, 127-09-3; sodium azide, Sigma, 99.8%, 26628-22-8; sodium nitrate, BDH chemicals, ACS grade, 7631-99-4; succinic acid, TCI America, ≥ 99%,110-15-6; and tricarballylic acid, Alfa Aesar, 98%, 99-14-9. For electron microscopy, hexagonal copper square grids with 200 mesh carbon support film from Electron Microscopy Sciences were used. Pierce Biotechnology’s 2 kD MWCO Slide-A-Lyzer™ dialysis cassettes were used.

Nanoceria synthesis and characterization

Polyhedral citrate-coated nanoceria crystallites were synthesized using a hydrothermal approach (Masui et al., 2002) and dialyzed five times, 12 h each, against iso-osmotic (110 mM) citric acid at pH 7.4 to remove any unreacted cerium salts and to citrate coat the nanoceria to stabilize the dispersion. The nanoceria was stored at room temperature in the dark. It was sterilized by autoclaving prior to introduction into the cassettes. A sample (500 μg in one ml) of the citrate-coated nanoceria was dialyzed against 200 ml water for 24 h with three changes of water, then dried, for HRTEM characterization.

Nanoceria primary and hydrodynamic particle sizes were determined by TEM using a 200-keV field emission analytical transmission electron microscope (JEOL JEM-2010F; Tokyo, Japan) and dynamic light scattering (DLS) in 110 mM pH 7.4 citric acid using a 90Plus Nanoparticle Size Distribution Analyzer (Brookhaven Instruments Corp., Holtsville, NY). Analytical hi-res TEM/STEM imaging was conducted to characterize its structure, surface nature, and d-spacing (distance between planes of atoms in the crystal structure). Surface and core Ce³⁺/Ce⁴⁺ ratios were determined from M4/M5 (Ce⁴⁺/Ce³⁺) peak heights after background subtraction and examined for eV shift of the Ce peaks and O peak. Similar EELS determinations were made of the nanoceria harvested from the cassette after its partial dissolution. Instruments included a JEOL 2010F field emission transmission and scanning transmission microscope (HRTEM/STEM) operated at 200 kV with an analytic pole piece. Images were recorded using a Gatan Ultrascan 4k × 4k CCD camera. Data analysis and processing used Gatan Digital Micrograph software and Digiscan II. HRSTEM imaging was performed in combination with energy dispersive spectroscopy (EDS) and EELS analysis of select particles. Dark field imaging used a Gatan HAADF detector and Gatan GIF Tridiem^® Filter. All HRSTEM images were acquired using an analytical probe with 0.17 nm resolution. An Oxford Aztec EDS system was used for select elemental mapping. The EELS measurements and trace lines were obtained using a 1 nm probe, an alpha of 12 mrad and beta of 6 mrad. Analysis was obtained from core edge intensity acquired after background subtraction for different locations on select particles (center and rim) using an integration window ranging from 10 to 30 eV.

Dialysis/dissolution system

In an initial experiment, dialysis cassettes were loaded with one ml of citrate-coated nanoceria in 110 mM citric acid or citrate-coated nanoceria in water. They were immersed in 400 ml beakers containing 200 ml of iso-osmotic 110 mM citric acid at pH 4.5 or iso-osmotic sodium nitrate at pH 4.5, respectively, plus 0.02% sodium azide as a bacteriostatic and fungistatic agent. Over 28 weeks dissolution was ~ 50% greater in the presence of citric acid in the bath (Figure S1). In the main study dialysis cassettes were loaded with citrate-coated nanoceria (containing ~ 500 μg cerium) in one ml of iso-osmotic citric acid and immersed in 400 ml beakers containing 200 ml of an aqueous solution containing the tested solutions plus 0.02% sodium azide. This nanoceria concentration was used by Dahle et al. (2015). The tested solutions were water (at ~ pH 6); horseradish peroxidase + hydrogen peroxide; ammonium (20 mM at pH 4.5); and acetic, adipic, citric, glutaric, DL-3-hydroxybutyric, lactic, DL-malic, pimelic, succinic, and tricarballylic acids (110 mM at pH 4.5). Horseradish peroxidase (HRP) 15 nmoles (0.67 mg) + hydrogen peroxide (40 μM added to the bath at the beginning of the experiment and each time the bathing medium was sampled) were at pH 6.1, the pH of HRP maximal activity. Bath solutions, except water, were adjusted to iso-osmotic strength by sodium nitrate addition, using a Fiske Model 110 osmometer. The carboxylic acid structures are shown in Figure S2. The experimental set-up and sampling (described below) are illustrated in Figure S3. Each condition, except water, was studied in duplicate. The concentration of most ligands (110 mM) was based on the concentration of citric acid to produce an iso-osmotic solution at pH 4.5, assuming total citric acid dissociation. Determination of the osmotic strength of 110 mM citric acid revealed that it did not produce an iso-osmotic solution, presumably due to the lack of complete ionization (non-adherence to van’t Hoff’s law at this concentration). The acetic acid concentration (20 mM) was used by Dahle et al (2015). The HRP-H₂O₂ condition was based on Allen et al (2008).

Cassette/beaker systems were sealed with Parafilm^® and aluminum foil and housed in a rotary shaking incubator at 37 °C rotated at 60 rpm. The HRP-H₂O₂ system was housed at 4 °C, the temperature shown to degrade single walled carbon nanotubes (Allen et al., 2008).

Samples (Table 1) were acidified by addition of trace metal grade concentrated nitric acid (5 μl to 75 μl samples, 10 μl to 1 ml samples). Samples were spotted onto EM grids for TEM/STEM imaging. The bath and the cassette volumes were measured at the end of the experiment. The cassette was disassembled and its membranes and gasket immersed in 25 ml of 5% nitric acid to determine residual cerium by ICP-MS.

Table 1.

Sampling protocol for cerium mass and nanoceria size.

CHAMBER	SAMPLE TYPE	TIME	VOLUME	PURPOSE	ANALYSIS
Cassette	Nanoceria dispersion	t = 0	75 μl	Ce concentration	ICP-MS
		8, 16, & 24 weeks	75 μl each time	Ce concentration	ICP-MS
			~ 10 or 25 μl each time	Nanoceria size	TEM/STEM
		Final	75 μl	Ce concentration	ICP-MS
			10 μl	Nanoceria size	TEM/STEM
Bath	Bath solution	t = 0	1 ml	Ce concentration	ICP-MS
		weekly	1 ml each time	Ce concentration	ICP-MS
		4, 8, 16, & 24 weeks	~ 5 μl each time	Nanoceria size	TEM/STEM
		Final	1 ml	Ce concentration	ICP-MS
			~ 5 μl	Nanoceria size	TEM/STEM

The carboxylic acid concentrations used for the dissolution experiments were significantly higher than those expected for in vitro or in vivo environments. To address the carboxylic acid concentration dependence of nanoceria dissolution, DLS was used to measure nanoceria size in systems containing carboxylic acid concentrations used in the bath study and one, two, and three orders of magnitude smaller. To assess the ability of representative carboxylic acids to accelerate nanoceria dissolution at these acid concentrations, nanoceria (1000 μg) was added to 2 ml of iso-osmotic solution at pH 4.5 containing 0.11, 1.1, 11, or 110 mM citric or malic acid, housed at room temperature. Nanoceria hydrodynamic particle size was determined immediately after its addition (time = 0), then biweekly for 6 weeks, then weekly for 10 weeks, followed by semi-weekly (to 28 weeks) by DLS, as described above. Samples were agitated before DLS measurement.

The diffusivity of cerium salts through the cassette membranes was measured to ensure that cerium salt membrane transport was much faster than the nanoceria dissolution rate. An aliquot of ~ 500 μg cerium ion in one mL water was placed in a cassette and dialyzed against 200 ml of 110 mM citric acid at pH 4.5 in two experiments. Bath samples were collected 1, 3, 6, and 24 h later (and longer in one experiment) for cerium quantitation. The data were fit to an unsteady-state model for diffusion through the membrane to compute cerium ion membrane diffusivity and calculate the diffusion process half-life.

Cerium quantitation

Samples containing nanoceria were digested with two:one HNO₃:H₂O₂ in Teflon vessels in a CEM MARS Express microwave digestion system. Terbium was added as an internal standard and analyzed compared to standards. Cerium was quantified by ICP-MS (Agilent 7500cx, Agilent Technologies, Inc., Santa Clara, CA) (Yokel et al., 2009). To determine the efficiency of nanoceria recovery, duplicate cassettes were loaded with ~ 7, 30, and 110 μg nanoceria and immersed in 200 ml DI water for four days. The identical amount of nanoceria was introduced into 200 ml DI water in similar beakers. The contents of the cassettes were removed and added to their respective baths, the cassettes disassembled and the membranes and gasket soaked in 25 ml 5% nitric acid, which was combined with its respective bath. The mass of cerium recovered from the cassettes and their baths was compared to that recovered from the water.

Modelling nanoceria dissolution

Apparent mechanisms of nanoceria dissolution

Spherical solid particle degradation kinetics are usually described by first order or other similar nonlinear rate laws. A set of these models (zero-, half-, first-, second-order in nanoceria concentration) were compared to the initial data. The results showed poor correspondence between data and any of the models. Rather, a surface-controlled dissolution model (Forryan et al., 2005) was found to provide good correspondence between experimentally-measured cerium mass in the bath and its prediction via the discrete material balance/dissolution rate model. This model gives the loss of atoms from a solid particle as directly proportional to its surface area. It was implemented using a population balance of nanoceria particles in the cassette plus the integrated kinetic rate equation that predicted the number of cerium atoms in a nanoparticle as a function of time. The number of atoms per particle was calculated from the bulk density, Avogadro’s number, and particle size. For this implementation of the surface-controlled model, an average particle size was assumed.

Discrete material and population balances

The sampling protocol (Table 1) resulted in removal of samples from the bath (one ml) and cassette (75 μl) that contained cerium ions in the bath and cassette, and nanoceria in the cassette. In addition, there were water evaporative losses from the bath and solution transfers between the bath and cassette; these transfers are inferred from the initial and final cassette and bath volumes plus knowledge of the volume lost from the sampling protocol. We devised discrete balances of water volumes, cerium ions, and cerium atoms in nanoceria particles, along with a population balance for nanoceria particles in the cassette. The transport mechanisms of sample withdrawal and evaporation are shown in the diagram of the experimental set-up (Figure S3).

Cassette phase:

The surface-controlled dissolution model predicted the number of cerium atoms in a nanoparticle as a function of time, which is independent of the number of nanoparticles present in the cassette. The number of cerium atoms in the nanoceria solid phase in the cassette was computed by multiplying the number of nanoceria particles in the cassette by the number of cerium atoms per nanoparticle at any time. For the discrete balances, the number of cerium ions/nanoparticle was computed from the model for the starting and ending interval between sampling. Ion and solution flow between the cassette and bath was assumed to be in equilibrium. Cerium ions removed by sampling were also accounted for, but were essentially negligible except for a few experiments in which most of the nanoceria dissolved.

Bath phase:

Water was lost by evaporation, which was averaged (as volume) across the entire experimental period. Cerium ions were lost due to sampling for ICP-MS and TEM/STEM analysis. This was accounted for at each sampling interval by the discrete volume and cerium ion balances for the bath.

Quantification of nanoceria agglomeration over time

The size and shape of nanoceria as it dissolved in two representative carboxylic acids was determined from HRTEM images: hydroxybutyric acid, representative of the carboxylic acids in which nanoceria formed superstructures, including adipic, pimelic, and glutaric acids; and lactic acid, representative of the carboxylic acids in which nanoceria did not form superstructures, including succinic and malic acids. Nanoceria size and shape were quantified after 4, 12, and 28 weeks exposure. ImageJ was used to analyse the images, from which nanoceria size and shape (linear, branched, ellipsoidal, and spherical) distributions were calculated (Grulke et al., 2017, Grulke et al., 2018a, Grulke et al., 2018b). Differences among the sizes and shapes were compared by a 3-way ANOVA.

Synthesized nanoceria

Seventeen batches of nanoceria were prepared, all yielding similar products. One batch of nanoceria was used for this study. The mean ± S.D. primary particle size by ImageJ analysis of 50 particles visualized by TEM was ~ 4.2 ± 1.2 nm. Twenty-nine measurements of its hydrodynamic diameter, taken 12 times over 9 months, showed that the nanoceria dispersion was stable, with a predominant agglomerate (which consists of closely arranged primary nanoceria crystallites) size of ~ 14 nm (Figure 1a). There was also a small secondary peak of the DLS results, which reflects the tendency of nanoceria agglomerates to self-associate. All twenty-nine measurements gave similar agglomerate distributions, showing that the nanoceria dispersion was stable under these storage conditions. Autoclaving did not significantly change the distribution mean sizes. Preparation and purification included dialysis against 110 mM citric acid to remove reaction components not incorporated into the synthesized nanoceria, a step not previously reported in most reports of nanoceria synthesis. This results in citrate-coated nanoceria. When subsequently dialyzed against water for 24 h, nine months after its synthesis and purification by dialysis against citric acid, only ~ 0.1% of the cerium in the nanoceria diffused into the water. Figure 1b shows a typical agglomerate consisting of a typical number and arrangement of nanoceria crystallites. Figure 1c shows a single nanoceria crystallite prior to dissolution. In profile it has sharp edges. The nanoceria surface has a predominance of Ce³⁺. The surface and core Ce³⁺/Ce⁴⁺ ratios were determined using EELS by comparing the M4/M5 (Ce⁴⁺/Ce³⁺) peak heights after background subtraction and by recording any satellite peak presence which is indicative of Ce⁴⁺ and not associated with Ce³⁺ (Graham et al., 2014). These findings are consistent with previous results that showed greater Ce³⁺ surface concentration (Graham et al., 2014). TEM/STEM data showed crystal morphology, lattice, and d-spacing characteristic for nanoceria (Figure 1d) (Kurian and Kunjachan, 2014).

Figure 1.

Hydrodynamic diameter and TEM/STEM analysis of nanoceria. (a) Hydrodynamic diameter as surface area. (b) HRTEM/STEM image of individual (primary) ~ 4 nm particles and their agglomerate. Dominant faces are (100) and (111). (c) Nanoceria primary crystallite. (d) Representative nanoceria crystallites (of 20 measured), showing d-spacings that ranged from 3.15 to 3.17 Å and lattice unit cells ranging from 5.40 to 5.41 Å, using HRTEM.

Nanoceria recovery

The mass of cerium recovered from the dialysis cassettes immersed in water averaged 95, 107, and 98% of the same amount of nanoceria introduced into the same volume of water, for an overall recovery of 100 ± 6%.

Modelling nanoceria dissolution

The dissolution process model relates the dissolution rate directly to the particle surface area. As nanoceria particle size decreases, surface Ce³⁺ increases, leading to lattice strain and increased solubility (Grulke et al., 2014). Ce³⁺ is relatively enriched on the nanoceria surfaces. The concentration is dependent on the particle size due to a larger number of oxygen vacancies in the surface lattice of the smaller grains, which is due to a higher defect concentration relative to the bulk cerium which is almost entirely in the Ce⁴⁺ oxidation state (Dutta et al., 2006). TEM/STEM images of nanoceria particles were used to verify that dissolution occurred, and to enable quantification of their size and shape. It was easier to detect dissolution by imaging larger particles, such as the one shown in Figure 2. This particle has a diameter of ~ 7.15 nm after 1344 hours of dissolution, but has lost the sharp edges and corners of the starting material. Using this particle and its size at 1344 hours of dissolution as an example, the dissolution rate model was used to estimate the original particle size (7.53 nm) and number of cerium atoms/particle (5645). At 1344 hours, the particle would have 4834 cerium atoms, having lost about 15% of its mass. Nanoceria particles of the average size, 4.2 nm, would have lost 25% of their mass over the same time.

Figure 2.

Partially degraded nanoceria, evidenced by rounding at the edges and smaller primary particle size after 1344 h exposure to citric acid at pH 4.5.

Diffusivity of cerium salts through the cassette membranes

Unsteady state diffusion experiments using cerium nitrate were conducted to determine the diffusivity of cerium and its half-life of diffusion through the membranes. Membrane cerium diffusivity was 6 and 7 × 10⁻¹¹ m²/s in the two experiments; half-life = 12.5 h. By direct half-life comparison, cerium ion diffusion through the membrane is > 60 times faster than any of the measured nanoceria rates (12.5 vs. ≥ 800 h), implying that cerium ion concentrations between the cassette and bath were at steady-state relative to the dissolution rate. The rate of cerium ion diffusion through the 2 kD membrane and establishment of equilibrium in the 200 ml bathing medium did not appreciably influence the appearance of cerium in the bathing medium as nanoceria dissolved.

Cerium mass in the bath

Mass balance calculations were conducted incorporating the results of the cerium concentration of the nanoceria loaded into the cassette and the mean of the lower, the higher, and the average of each of the seven pair of results (532 ± 106) obtained throughout the study. The results were 461, 613, and 532 μg, respectively. The overall mean (532 μg) provided the best overall material balance results when compared to the mass of cerium in the cassette solution and the cassette at the end of the experiment, and was used for the mass balance and rate constant determinations.

Figure 3 shows plots of the percentage of nanoceria dissolved as a function of time for 12 bath solution additives. The data are shown for two replicate experiments. The plots are ordered (left to right, top to bottom) to correspond with the dissolution half-lives (Table 2). Table 2 also lists the average diameters for the nanoceria particles after 28 weeks of dissolution, based on the surface-controlled dissolution predictions. The HRP and water conditions have similar, very large, half-lives (55,000 and 105,000 vs. 58,200 h, respectively), so the water results are not presented in Figure 3. The ammonium ion bath half-time for dissolution was a factor of two higher than those of any carboxylic acid. Lactic acid has the shortest dissolution half-life. The model predicts complete dissolution in 3024 h. The plot for the lactic acid system (Figure 3) is consistent with this prediction, as it reaches a maximum (100%) at ~ 3,000 h. The cerium material balance was most accurate for the lactic acid system: since dissolution was rapid, there were fewer cerium and nanoceria losses due to sampling.

Figure 3.

Percentage of nanoceria dissolved vs. time for 12 bath solution additives. Panels show the results of the two individual experiments (as the upper and lower boundaries of the shaded area) and the mean.

Table 2.

Model estimates of dissolution half-life (h) and average nanoceria diameter at 28 weeks from the replicate experiments.

Bath solution	Dissolution half-life (h)	Average diameter at 28 weeks (nm)
pH ~ 6
Water	58,200	4.17
HRP	55,200, 105,000	4.16, 4.19
pH 4.5
Ammonium	7490, 6300	3.69, 3.57
Tricarballylic acid	2310, 4400	2.46, 3.30
Glutaric acid	3030, 3300	2.88, 2.99
DL-3-hydroxybutyric acid	2310, 3300	2.46, 2.99
Pimelic acid	2310, 2620	2.46, 2.96
Acetic acid	2290, 2300	2.45, 2.83
Adipic acid	1990, 2310	2.18, 2.46
Citric acid	1950, 2270	2.13, 2.42
Succinic acid	1560, 2080	1.78, 2.47
DL-Malic acid	1420, 1960	1.70, 2.23
Lactic acid	794, 811	Dissolved at 18 weeks

Table 2 also shows the average nanoceria particle diameter at 28 weeks (4704 h) as predicted from the dissolution rate model. The starting value, 4.2 nm, was established from TEM particle size distributions (above). For reference, the expected nanoceria particle diameter was 3.4 nm at the half-life conditions.

Effects of lower carboxylic acid concentration on nanoceria dissolution

In the presence of both citric and malic acids (0.11, 1.1, 11, or 110 mM in 2 ml of iso-osmotic solution at pH 4.5 containing 1000 μg nanoceria) there was a reduction of nanoceria hydrodynamic diameter during the initial several hundred hours (Figure 4). Nanoceria approached an asymptotic hydrodynamic diameter of ~ 17.5 nm in the presence of 0.11 and 1.1 mM citric acid, whereas higher citric acid concentrations resulted in a lower asymptote. Similarly, the higher malic acid concentrations resulted in a smaller nanoceria hydrodynamic diameter. No particle settling was observed for either carboxylic acid at any of the concentrations. In the presence of two ml of 11 mM citric or malic acid, the molar ratio of carboxylic acid to 1000 μg cerium in nanoceria is three, whereas the ratio for 200 ml of 110 mM carboxylic acid to 500 μg cerium in nanoceria (as in the cassette experiments) is 6 × 10³. Because reduction of nanoceria hydrodynamic diameter (dissolution) occurred through a range of concentration ratios of carboxylic acid to nanoceria from three to 6000, these results suggest the dissolution of nanoceria in the dialysis system is not an artifact of the excess of carboxylic acid compared to nanoceria, and carboxylic acid-accelerated nanoceria dissolution extends down to equimolar carboxylic acid to nanoceria concentrations.

Figure 4.

Nanoparticle size over time. Effective nanoceria agglomerate hydrodynamic diameter over time during exposure to 0.11, 1.1, 11 or 110 mM citric (left panel) or malic acid (right panel).

Morphology, size, valence, and agglomeration of nanoceria in the cassettes

Transformation of nanoceria in the cassettes occurred over three phases. A summary of the changes based on HRTEM-obtained images during dissolution in two representative carboxylic acids (hydroxybutyric and lactic) is shown in Figure 5. Nanoceria developed up to micron sized agglomerates during exposure to hydroxybutyric acid. It did not form agglomerates > 100 nm as it dissolved with lactic acid exposure. A 3-way ANOVA showed statistically significant differences among the 3 phases for nanoceria in the presence of hydroxybutyric acid for area, perimeter, Feret, MinFeret, and aspect ratio, all at p < 10^{−13 to −16}.

Figure 5.

Time course of nanoceria dissolution/aggregation in representative carboxylic acid solutions. (a) Formation of nanoceria agglomerates and superstructures in hydroxybutyric acid after 4 weeks exposure (Phase 1) compared with the as-synthesized nanoceria, shown in Figure 1. Blue circles are the as-introduced nanoceria. Red circles are spheroidal agglomerates. Green open diamonds are ellipsoidal agglomerates. Purple + symbols are branched agglomerates. Black bars are linear agglomerates. (b) Nanoceria after 12 weeks (Phase 2) hydroxybutyric acid exposure. (c) Nanoceria after 28 weeks (Phase 3) hydroxybutyric acid exposure. (d, e and f) Nanoceria after 4, 12, and 28 weeks exposure to lactic acid.

Phase 1 findings four weeks after initiation of dialysis/dissolution were characterized by formation of superstructures (characterized by accumulation of many nanoceria agglomerates into architectures up to micron scale made up of teeming, interconnecting, agglomerates, associated in ligand-dependent arrangements) and evidence of pH-dependent nanoceria dissolution in the presence of some carboxylic acids, such as hydroxybutyric (Figure 6). The superstructures were susceptible to external forces and could be broken up into the agglomerate units. In contrast, the agglomerates, which consisted of closely arranged primary nanoceria crystallites, were structurally stable. The agglomerates formed during nanoceria synthesis; superstructures formed during Phase 1 of carboxylic acid medium-dependent nanoceria dissolution (Figure 6 left column). Superstructures were seen in the presence of most ligands (Figure 6), but the degree of superstructure formation was strongly ligand dependent, ranging up to micron size in the presence of tricarballylic, hydroxybutyric, pimelic, acetic, and adipic acids (Figure 6 j, m, p, s, and y), with very little superstructure formation with succinic, malic, and lactic acids (Figure 6 ab, ae, and ah). At pH ~ 6 (water) there was no formation of superstructures or reduction in primary or agglomerate particle size.

Figure 6.

HRTEM images of nanoceria withdrawn from cassettes. Primary, agglomerate, and superstructures of nanoceria-ligand complexes are shown for each ligand. A representative image of nanoceria in the presence of each ligand at three different time intervals is shown (Phase 1 at 4 weeks; Phase 2 at 12 weeks; and Phase 3 at 28 weeks). Details follow.

• a,b,c: HRP + H₂O₂. Nanoceria agglomerates associated to form aligned and dense superstructures in Phase 1; no reduction in primary or agglomerate particle size or rounding of nanoceria crystals in Phase 2 or Phase 3 or residual superstructures in Phase 3.

• d,e,f: ammonium. Several hundred nm wide superstructures formed in Phase 1; no recognizable size reduction due to dissolution of nanoceria in Phase 2; and similar superstructures and particle sizes in Phase 3.

• g,h,i: glutaric acid. Superstructures with unique long-range string of pearl arrangements in Phase 1; rounding of nanoceria in the agglomerates in Phase 2; skeletal agglomerates in Phase 3.

• j,k,l: tricarballylic acid. Uniquely arranged long-range superstructures in Phase 1; rounding of nanoceria in the agglomerate structures in Phase 2; skeletal agglomerates in Phase 3.

• m,n,o: hydroxybutyric acid. Short-range superstructures (only involving a select few aligned agglomerates) in Phase 1; rounding of nanoceria in the agglomerate structures in Phase 2; skeletal agglomerate structures in Phase 3 with some isolated nanoceria.

• p,q,r: pimelic acid. Short range superstructures in Phase 1; rounding of nanoceria in Phase 2 with some skeletal agglomerates; more skeletal agglomerates in Phase 3 with isolated nanoceria more common.

• s,t,u: acetic acid. Short range superstructures in Phase 1; rounding of nanoceria in Phase 2 with breakup of skeletal agglomerates and release of isolated nanoceria; Phase 3 skeletal agglomerates with very small and rounded nanoceria.

• v,w,x: citric acid. Minor formation of superstructure in Phase 1; rounding of nanoceria crystals in Phase 2; breakup of superstructures in Phase 3 and skeletal appearance of agglomerates.

• y,z,aa: adipic acid. Very few superstructures in Phase 1; extensive rounding of nanoceria in agglomerates in Phase 2; no superstructures in Phase 3 and greatly reduced amount of agglomerates with major skeletal development and more isolated nanoceria.

• ab,ac,ad: succinic acid. Isolated superstructures in Phase 1 with most agglomerates not attached to other agglomerates; major skeletal development in agglomerates in Phase 2; agglomerates are much smaller compared with the starting materials shown in Figure 1 and isolated nanoceria are reduced in size.

• ae,af,ag: malic acid. No superstructures and some rounding of nanoceria in agglomerates in Phase 1; strong rounding of nanoceria in Phase 2 with skeletal appearance of agglomerates; isolated small-sized agglomerates (some only harboring a few nanoceria) and overall quantities of nanoceria greatly reduced compared with Phase 1.

• ah,ai,aj: lactic acid. No superstructures and only individual agglomerates in Phase 1; Phase 2 has reduced nanoceria size and skeletal appearance of the agglomerates; Phase 3 shows only very few isolated very small agglomerates or individual nanoceria at greatly reduced size and quantities compared with Phase 1.

Phase 2 findings, 12 weeks after initiation of dialysis/dissolution, were characterized by lack of significant change in the superstructure from Phase 1, but obvious presence of nanoceria dissolution, resulting in rounding of nanoceria crystallite edges under all carboxylic acid conditions at pH 4.5. For example, in the presence of citric acid (Figure 6 w) nanoceria agglomerates persisted into Phase 2, but the primary crystal particles bound to each other in the agglomerates changed due to dissolution. This was evidenced by reduction of primary particle size, and to a lesser degree reduction of the agglomerate size, creating much larger voids between primary particles, that gave the agglomerates a skeletal appearance. This was observed for all ligands, with some variability among the ligands. Ligand type did not alter the crystallinity of the primary particles as they dissolved within the agglomerates, but led to smaller and more rounded nanoceria. In general, the agglomerates did not collapse or reorganize as a result of the initial dissolution process. However, some carboxylic acids (malic and lactic) caused a much greater skeletal formation in the agglomerates, due to more rapid nanoceria dissolution. This resulted in significant void formation between primary nanoceria particles. This was observed to a lesser extent for other carboxylic acids (Figure 6). The reduction of primary particle size was associated with the increasing concentration of cerium in the bath (Figure 3).

In Phase 3, which represents the late dissolution stage (Figure 6 column 3) at 28 weeks, most superstructures were gone and there was a significant reduction in the primary nanoceria and agglomerate size; HRP being the only exception (Figure 6 a-c). Particle size reduction correlated well with dissolution half-life (Table 2). The citric acid example shows isolated residual small porous agglomerates (Figure 6 x) while the lactic acid example shows minute crystallites (individual nanoceria with reduced size compared to the starting material), suggesting agglomerate disintegration (Figure 6 aj). Comparison among images (Figure 6, column 3) shows that agglomerates were no longer closely packed, but rather adopted a skeletal appearance due to the greatly increased void space caused by dissolution and size reduction of the primary nanoceria. In the presence of glutaric (Figure 6 i), tricarballylic (Figure 6 l), hydroxybutyric (Figure 6 o), pimelic (Figure 6 r), acetic (Figure 6 u), citric (Figure 6 x), adipic (Figure 6 aa), and succinic (Figure 6 ad) acids, dissolution observed in Phase 3 resulted in isolated occurrences of skeletal agglomerates, but superstructures were completely gone. In the presence of malic (Figure 6 ag) and lactic (Figure 6 aj) acids, skeletal agglomerates were further destroyed, such that only isolated primary nanoceria crystallites remained, with most having smaller than original size and rounding along edges due to ion shedding from the surface layers.

EELS analyses were performed on the as-synthesized and partially-dissolved nanoceria to determine if dissolution initiated and progressed along the nanoceria particle surface, affecting its surface valence (Figure 7). The nanoceria surfaces after citric acid exposure were relatively enriched in Ce³⁺, the antioxidant valence state, determined by the cerium M4/M5 peak heights in comparison with as-synthesized nanoceria, while nanoceria core regions were less affected (Figure 7). These results were observed for all ligands. Continued dissolution along the nanoceria surfaces caused defect formations in the crystallite surface layers, including oxygen vacancies which affect nanoceria’s electronic and chemical surface properties. This is reflected in the observed increased M5 vs. M4 peak height (Figure 7). Oxygen defect density increased along the destabilized dissolving nanoceria surfaces.

Figure 7.

EM images and EELS analysis for nanoceria as synthesized and after exposure to citric acid at pH 4.5 at Phase 1 (a) and Phase 2 (b). M4/M5 peaks for cerium are indicated.

Discussion and Conclusions

Nanoceria prepared as described appears to be stable in aqueous solution for at least 9 months. Carboxylic acids (Taguchi et al., 2009, Grulke et al., 2014), imine-containing polymers (Kitsou et al., 2017), and surfactants can stabilize nanoceria particles during synthesis. When exposed to pH 4.5 in the presence of the carboxylic acids of this study, citrate-coated nanoceria underwent dissolution over time. This is in contrast to nanoceria prepared in citric acid to efficiently coat its surface, cease particle growth during production, and prevent agglomeration when stored in iso-osmotic citrate for months at pH 7.4, in which it was stable. The results suggest the citric acid coating is susceptible to desorption in an acidic environment and demonstrate the role of the acidic environment in nanoceria dissolution, as shown by Dahle et al. (2015). The absence of nanoceria dissolution in HRP + H₂O₂ at pH 6 suggests a mechanism of nanoceria dissolution different from carbon-based nanomaterial biodegradation. Carbon-based nanomaterial biodegradation is mediated by enzymatic catalysis, whereas nanoceria biodegradation is pH dependent.

In the presence of carboxylic acids at pH 4.5, electron micrographic images showed initial dissolution at the corners of the polyhedral (predominantly (111) and (100) faces) polycrystalline nanoceria, the regions of highest instability, followed by primary particle size reduction and rounding over time. Based on TEM results, the absence/presence of nanoceria particles < 2 nm cannot be addressed, therefore the possibility that primary particles totally dissolved cannot be assured. Carboxylic acids may facilitate dissolution by providing a ligand to complex the cerium ion released during nanoceria dissolution, preventing agglomeration in some cases and promoting superstructure formation in others.

Lactic, malic, succinic, citric, and adipic acid produced the lowest dissolution half-times. Nanoceria hydrodynamic agglomerate diameter in lactic, malic, and citric acids did not increase with dissolution time. Considering the effect of citric, malic, and lactic acids on the relative rates of nanoceria dissolution and their ability to maintain nanoceria’s agglomerate structure and prevent formation of micron sized superstructures as the nanoceria dissolved, they were more effective than the other carboxylic acids to solubilize and stabilize nanoceria. The National Institute of Standards and Technology Critically Selected Stability Constants of Metal Complexes database includes reports of the log K₁ and β_n for Ce³⁺ and Ce⁴⁺. K1 is the first stepwise formation/stability/association constant of the cerium-ligand coordination complex and β_n is the cumulative constant of the n stepwise coordination complexes. Values from that source for the carboxylic acids of the present study are in Table S1. The carboxylic acids that have the greater ability to accelerate nanoceria dissolution generally better stabilize it (Figure S4). The ability to stabilize nanoceria presumably relates to the ability to form a stable cerium-carboxylic acid coordination complex. We addressed this by calculating coordination complex formation energies for one and two carboxylic acids per CeO₂ formula unit. The results showed positive formation energies for one: one complexes for all cases, and negative energies for two carboxylic acid: one CeO₂ complexes, indicating a thermodynamic driving force for nanoceria dissolution and formation of a bi-carboxylic acid: CeO₂ complex. However, this only partially explains the different dissolution rates (unpublished results).

The coordination bonds that form between citrate and lanthanide series metals (lanthanum, cerium, and gadolinium) (Bobtelsky and Graus, 1955, Baggio et al., 2005, Zhou et al., 2008, Chen et al., 2012, Heller et al., 2012) involve carboxylate and hydroxyl groups, presumably mediating the acceleration of nanoceria dissolution and maintenance of nanoceria stability at pH 4.5. Adipic and pimelic acids accelerated nanoceria dissolution, consistent with their selective binding to the (100) crystallite face of nanoceria and presence of two carboxyl groups that could form a coordination complex with cerium (Taguchi et al., 2009). Dicarboxylic acids with longer chain lengths adsorbed to nanoceria surfaces, but generated membrane-like protective coatings on the particles (Taguchi et al., 2009).

Small cerium-containing coordination complexes were able to diffuse through the ~ 2 nm pores of the cassette membrane into the bathing medium. At the levels of cerium ligands in these experiments, the cerium salts do not appear to be present above their solubility limits. Cerium citrate is quite insoluble (Table S1). We are aware of only one report of cerium citrate solubility (3.02 and 6.40 g/L in H₂O at 20 and 90°, respectively (Ezerskaya and Cherches, 1973). This is greater than the cerium concentration if all the nanoceria in the dialysis cassette equally distributed throughout the dialysis/dissolution system (~ 0.00250 g/L). Reported solubility products of the other ligands of this study are lower than the citric acid solubility product (Table S1).

The experimental system of this study does not totally model the complex in vivo environment that mediates nanoceria dissolution and reorganization. However, the results of this study clearly show the primary role of pH mediating nanoceria dissolution, and the release of cerium salts that could be a key step of nanoceria bioprocessing in vivo. The primary difference is the lack of phosphate-containing ligands. This is being investigated in ongoing work, as well as the biological response to the partially degraded, and probably ligand-coated, nanoceria.

The formation of carboxylic acid-cerium complexes after nanoceria dissolution may enable redistribution of cerium released from nanoceria dissolution within organisms and uptake into plant roots, similar to the role of citrate to release iron from the low-pH environment of endosomes (Arbab et al., 2005). This may explain the organ-specific bioprocessing of ceria (Graham et al., 2018). Ligands that enable nanoceria dissolution in acidic environments may greatly affect nanoceria’s fate (dissolution rate and size as well as transport phenomena through ion release).

The carboxylic acids that accelerate nanoceria dissolution are biologically relevant. Lactic acid is a product of anaerobic glycolysis and anaerobic metabolism. Citric, malic, and succinic acids are intermediates in the tricarboxylic acid cycle. Acetic acid is a product of free fatty acid and alcohol metabolism. The presence and concentration of these small carboxylic acids in conjunction with an acidic pH may influence nanoceria dissolution and stabilization of the released cerium ion in vivo. Although the concentration of most of these carboxylic acids in mammalian cells has apparently not been determined, the interstitial fluid total organic anion concentration is ~ 5 mEq/l, citric acid can reach 10 mM in some cells, and muscle lactate can reach ~ 30 mM during intense exercise (Legiša and Kidrič, 1989, Bangsbo et al., 1990). Their constant turnover provides a continual source of carboxylic acids to form complexes with nanoceria in vivo. Citric, succinic, malic, acetic, and other carboxylic acids are released by plant roots to chelate and acquire minerals (Cieslinski et al., 1997), can increase soil acidity, enhance nanoceria dissolution, and perhaps cerium uptake (Zhang et al., 2017).

Citric acid has been shown to accelerate the dissolution of other metal oxide nanomaterials. The rate of iron oxide nanoparticle dissolution in the presence of 20 mM citric acid was greater at pH 4.5 than 5.5 than 7.0 to 7.2 (Arbab et al., 2005, Soenen et al., 2010, Hoskins et al., 2012), but was not seen with acetate under the same condition (Arbab et al., 2005). Citric acid, 1.56 mM, at a starting pH of 5, greatly increased the dissolution of ZnO and CuO nanoparticles (Zabrieski et al., 2015). Nanotitania is another metal oxide nanomaterial generally considered to be quite inert. A sodium citrate pH 4.5 buffer, mimicking the lysosomal compartment in the present study, degraded anatase nanotitania over 96 h to a greater extent than the rutile form. Neither degraded in water or DMEM cell culture medium at pH 7 (De Matteis et al., 2016).

During dissolution nanoceria transformation occurred in three temporal phases, including self-association into superstructures and demonstrable reduction of the primary particle size. Phase 1 findings are characterized by nanoceria agglomerates interlinked to various degrees into superstructures, which was carboxylic acid dependent. In many experimental conditions these persisted through Phases 2 and 3. Superstructure formation provides insight into the interaction of nanoceria particles in different environmental conditions. Superstructures occur because the nanoceria agglomerates experience a surface leaching (ion shedding) effect during exposure to acidic environments containing carboxylic acids. The surface modification promotes the attraction of nanoceria agglomerates to link into superstructures. This interpretation is based on the assumption that surface leaching of ions promotes a shift in electronic states of surface species which either increases or decreases attractive or repulsive forces (Deng et al., 2016). HRTEM revealed that the type and degree (intensity) of superstructure formation is ligand dependent. The bonding type between these agglomerates in superstructures is assumed to be rather weak since they are susceptible to mechanical breakage or other external forces (e.g., the EM electron beam). However, the individual agglomerates that comprise the superstructures are structurally stable, whereas the association of the agglomerates in the superstructure is not stable. We saw agglomerate persistence in vivo (Graham et al., 2014, Graham et al., 2018). Variations in superstructure morphology were seen in different organs after nanoceria intravenous administration, such as in the liver vs. the spleen where nanoceria undergoes organ-specific bioprocessing (Graham et al., 2018), which may reflect certain environmental conditions. Organ-specific bioprocessing and assemblies of nanoceria may depend on available ligand types and concentrations that modify nanoceria’s surface functional properties after uptake, such as particle charge and ability to complex with available organ-specific molecules. Depending on the environmental concentrations and chemistries of ligands in different organs, their role in nanoceria dissolution may vary, which may have a strong influence on nanoceria or cerium ion transport mechanisms and ability to translocate to other regions in the body.

Dissolution occurred first at crystal corners and edges, resulting in particle rounding (Figure 2) and some decrease of primary particle size in pH 4.5, but not pH ~ 6, conditions. Nanoceria dissolution did not result in a phase transformation, e.g., recrystallization. In the presence of some ligands, nanoceria dissolution revealed fresh particle surface layers (due to ion shedding) that appeared to promote self-assemblies of nanoceria particles in superstructures (Figure 6).

In Phase 2 there was a clear difference in solubility and primary and agglomerate particle size among the experimental conditions from Phase 1, with ligand-selective superstructure destruction and increased distance among the superstructures. A result of the increased distance between primary particles in the agglomerates (Figure 6, particularly during Phase 2, which shows the ligand-dependent tendency of dissolving agglomerates to form skeletal structures), is a general surface area increase. This is due to the fact that the smaller nanoceria inside an agglomerate contribute to a higher surface area per volume and also help form more void spaces which reveal more nanoceria surface to the surrounding solvent medium. Nanoceria dissolution is controlled/driven by kinetic processes, quantified by the calculated rate constants that result in variable dissolution half-lives (Table 2).

In Phase 3 equilibrium was reached, evidenced by little reduction in the primary particle size from Phase 2, in which the agglomerates have a skeletal appearance, concurrent with little to no change in cerium concentration in the bathing medium. This is shown in Figure 3 for glutaric, citric, malic, and lactic acids after ~ 4000 h. For those ligands we also observed in HRTEM some remaining nanoceria crystallites, albeit with significantly reduced size and strong surface rounding effects (Figure 6 i, u, ad, and aj). There is no evidence from our observations that cerium ions contributed to nanoceria formation, e.g. it appears that nanoceria dissolution occurred in the absence of nanoceria precipitation. This is based on HRTEM investigation that indicated that all nanoceria correspond to the starting materials and the physical difference is due to surface shedding of ions and not recrystallization or reformation. In the current study the cassette experiments that resulted in the partial dissolution of the nanoceria (ligand-dependent process) did not generate conditions that provide supersaturation of cerium ions. No Ostwald ripening effects were observed in this study where original nanoceria crystals grow due to surface attachment of dissolved ions that come from dissolving nanoceria in the cassette. However, the shedding of ions from nanoceria that are concentrated in macrophages in vivo may provide such supersaturation conditions. This would explain why we observed a dissolution of nanoceria that is accompanied by regrowth of nanoceria or reprecipitation of other cerium phases (cerium phosphate) (Graham et al., 2018).

In summary, acidic environments, as found in phagolysosomes, may degrade nanoceria by dissolution, accelerated by carboxylic acids. Carboxylic acids may coat the nanoceria to form a “corona”, which would be expected to profoundly influence nanoceria’s fate and cellular response. Nanoceria coating, dissolution, and coordinate complex formation with carboxylic acids can profoundly influence nanoceria’s fate and effects.