Compiled information on particle characteristics in terms of particle shape and morphology (scanning electron microscopy, SEM), specific surface area (Brunauer-Elmer-Teller method, BET), and particle size distribution by means of laser diffraction (LD) in an aqueous medium of high ionic strength, PBS (phosphate buffered saline) is presented in Figure 1 and in Table 1 for all particles investigated (iron metal-Fe; chromium metal-Cr; ferrochromium-FeCr; FeCr dust; ferrosiliconchromium-FeSiCr; FeSiCr dust; stainless steel-316L; chromium(III)oxide-Cr₂O₃). More detailed information of coarser sized particles (based on BET-area) of Fe, Cr, FeCr, FeSiCr, coarse and fine particles of 316L and Cr₂O₃ is given elsewhere [9,10,18,19]. Particles of Cr, FeCr, FeSiCr, FeCr dust and FeSiCr dust revealed sharp edges and brittle visible cracks. These features were introduced during the crushing, sieving and re-crushing pre-process necessary to generate particles small enough to represent inhalation of respirable particles (90% of the number of fine particles were smaller than 5 μm (diameter in solution, including possible agglomerates), with the exception of fine chromium particles, see Table 1) but still have a chemical composition representative for the commercial products. The particle shapes of 316L particles were spherical and more irregular for particles of Fe. As expected, the specific surface area was significantly larger for the finer sized particles compared to the coarser sized particles, in particular evident for 316L, Fe, and Cr, Table 1. Despite the sieving process, SEM investigations still revealed the presence of some larger sized particles (> 20 μm).

The specific BET-surface area (surface area per unit mass) increased according the following sequence;

Cr-coarse ≈ Fe-coarse ≈ FeCr-dust ≈ 316L-coarse < FeCr-coarse < FeSiCr-dust < < FeCr-fine ≈ FeSiCr-coarse < 316L-fine < Fe-fine < Cr-fine < FeSiCr-fine < < Cr₂O₃.

According to size distribution measurements, Table 1, agglomeration occurred for all fine sized particles, in particular pronounced for the fine sized particles of Cr and Cr₂O₃. Tabulated size distribution numbers correspond to primary particles and/or agglomerates formed in solution, a distribution that may be different at dry conditions.

According to surface compositional analysis by means of x-ray photoelectron spectroscopy, XPS, chromium, iron, and silicon were the main components of the outermost surface (< 5 nm) oxide on particles of FeCr, FeSiCr, FeCr-dust and FeSiCr-dust, in non-proportional relations to their corresponding bulk compositions, Table 2. The relative mass surface content of oxidized silicon was significantly higher for dust particles of FeCr (Si_ox/(Fe_ox+Cr_ox+Si_ox): 0.46 ± 0.04) compared to both coarse (0.07 ± 0.03) and fine sized particles (0.08) at the same time as the relative amount of chromium was lower (Cr_ox/(Fe_ox+Cr_ox+Si_ox): 0.41 ± 0.02-dust; 0.65-fine; 0.70 ± 0.01-coarse). A similar effect, although less pronounced, was observed for FeSiCr with a slightly higher relative mass surface content of oxidized silicon for its dust particles (Si_ox/(Fe_ox+Cr_ox+Si_ox):0.41 ± 0.01 compared to 0.37 ± 0.06), and a corresponding lower oxidized Cr content (0.42 ± 0.03 compared to 0.50 ± 0.03). XPS findings imply slightly thinner surface oxides on dust particles on both alloys compared with their corresponding surface oxides on coarse and fine particles.

Unexpectedly, fine metal particles of Cr revealed oxidized iron in the outermost surface oxide (Fe_ox/(Fe_ox+Cr_ox): 0.19), but not for the coarser sized particles of Cr.

The surface oxide of fine particles of 316L (Cr_ox/(Fe_ox+Cr_ox): 0.26) was more enriched in oxidized chromium compared to the coarse particles (0.18). Oxidized manganese was detected in the surface oxide of both particle fractions. The results imply higher manganese content in the surface oxide on the finer sized particles. However due to the significant overlap between the Ni-LMM Auger and the Mn 2p peaks, no quantitative estimate of the manganese surface content was made.

Released amounts of iron and chromium per amount of particles loaded and corresponding release rates normalized to the particle BET-surface area and immersion period in artificial lysosomal fluid (ALF) of pH 4.5, simulating inflammatory conditions at 37°C, are presented for all particles investigated in Figure 2 after 168 hours (1 week) of immersion.

Except for particles of Cr₂O₃ (no measurable release of iron) and FeSiCr-fine particles (about the same amount of iron and chromium is released), significantly more iron was released compared with chromium, also for particles of pure fine-sized Cr metal (> 99.76 wt% chromium). Compared with chromium, the released amount of iron was approximately 20 times higher for 316L-fine particles, 16 times higher for FeCr-fine particles, and 6 times higher for Cr-fine particles.

The release of iron from all alloys and alloy-dust particles was significantly lower compared with particles of Fe metal (Figure 2). After one week of immersion in ALF, particles of Fe-fine and Fe-coarse were completely (≈1 μgFe/μgFe), and to a large extent (0.076 μgFe/μgFe, i.e. 7.6%), dissolved/released. For the alloys, only a very small amount of iron was released (< 0.022 μgFe/μgFe - 316L-fine). Similar or lower amounts of iron were released from dust particles of FeCr and FeSiCr compared with coarse and fine particles of FeCr and FeSiCr.

The released amount of chromium from particles of Cr metal was similar or lower compared to findings for particles of the alloys, despite more oxidized chromium present in the surface oxide of pure chromium compared with the alloys. In addition, when normalized to the BET surface area, it was evident that the release rate of chromium after one week of immersion in ALF was lower for chromium metal particles (0.000064 μg/cm²/h-Cr-fine) compared to all alloy particles (FeCr-fine: 0.0005 μg/cm²/h; FeSiCr-fine: 0.00025 μg/cm²/h; 316L-fine: 0.0006 μg/cm²/h), and the Fe metal particles (Fe-fine: 0.00024 μg/cm²/h). The highest release rate of chromium (0.0009 μg/cm²/h) was determined for Fe-coarse metal particles (> 99.96 wt% iron) and the lowest rate for particles of Cr₂O₃ (0.0000024 μg/cm²/h - 0.00003 μgCr/μgCr). Both when normalized to the specific surface area (Figure 2, bottom) or to the bulk iron and chromium content of the particles (figure 2, top), similar or lower amounts of both chromium and iron were released from dust particles of FeCr and FeSiCr compared with coarse and fine particles of FeCr and FeSiCr.

Nickel was released to a very low extent from alloy particles of FeCr-coarse (< 0.0002 μg/cm²/h, below limit of detection), FeSiCr-coarse (< 0.0002 μg/cm²/h, below limit of detection) and 316L-fine (< 0.0004 μg/cm²/h) after one week of immersion in ALF [9,10]. For the alloy particles, when only considering the iron, chromium, and nickel release, the remaining mass fraction of particles after one week of exposure to ALF corresponded to 99.96% (FeCr - coarse), 99.35% (FeCr - fine), 99.95% (FeSiCr - coarse), 99.90% (FeSiCr - fine), 99.98% (316L - coarse), 98.40% (316L - fine), 99.996% (FeCr dust), and 99.98% (FeSiCr dust).

When comparing metal release rates, which take into account differences in surface area, it is evident that all fine alloy particles investigated revealed higher rates of both chromium and iron compared to their coarser particle sizes, whereas the opposite situation was true for the pure metals. These findings were statistically significant for both chromium and iron released from FeCr and 316L (p < 0.01) and for chromium released from FeSiCr (p < 0.001) but not for iron released from FeSiCr (p > 0.05).

To investigate if these findings were also evident in other biological fluids, release rates of iron from fine and coarse particles of 316L are presented in Figure 3 after one week of immersion in each fluid. Higher release rates of iron from finer compared to coarse-sized particles were only statistically significant (p < 0.001) in the acidic medium of ALF (pH 4.5) of complex chemical composition, and not in the less complex, pH neutral and weakly alkaline media of artificial sweat, ASW (pH 6.5), Gamble's solution, GMB (pH 7.4) and artificial tear fluid, ATF (pH 8.0). The chemical composition of each medium is presented in Table 3. Similar trends were seen for determined release rates of chromium.

The fact that alloys and pure metals behave differently from a metal release perspective is illustrated in Figure 4 for fine and coarse particles of 316L and the pure metals Cr, Ni and Fe, immersed for one week in ALF. Significantly larger amounts of both iron and nickel were released from the pure metals compared with the alloy of both size fractions. Very low amounts of chromium were released from both the pure chromium metal and the alloys. Kinetic information (relatively high initial release rate (< 24 hours), strongly decreasing with time) on the release of alloy constituents from stainless steel as massive and particles, respectively, in ALF is presented elsewhere [13,18,20].

Chemical speciation measurements of released chromium by means of stripping voltammetry (DPAdCSV) on selected samples (Cr-coarse, FeCr-coarse, and 316L-coarse) after one week of exposure in ALF revealed chromium to be released as Cr(III). No evidence of any Cr(VI) in solution was perceived (< 0.01 μg/L). To assess the complexation capacity of ALF for Cr, standard addition of Cr(VI) was conducted. This is illustrated in Figure 5 for a given ALF solution with a chromium concentration of 15.8 μg/L. The gradual addition of Cr(VI) (in total 1.5 μg/L) revealed a complexing capacity of the solution of 17.3 μg/L when considering strong ligands only, as concluded from the lack of Cr(VI) in solution (no Cr peak). Further addition of Cr(VI) into the same solution revealed a slight increase of the Cr peak, indicative of Cr(VI) ions in solution. A significantly lower slope of the added Cr(VI) to peak height signal compared with the slope obtained from the same solution being UV treated to destroy complexing organic ligands clearly revealed a high complexing capacity also to weak ligands. The results show that organic ligands of the ALF medium have a high capacity to complex released chromium, at least if released at ppb-concentrations (μg/L).

Due to the fact that fine-sized particles have a higher probability of reaching the conducting airways or the alveoli, fine-sized particles of the alloys (FeCr, FeSiCr, 316L) and the pure metals (Fe, Cr) were subjected to different toxicological investigations of relevance for an inhalation scenario. Cr₂O₃ and NiO were investigated for comparative reasons, and single experiments were performed on larger dust particles (FeCr dust and FeSiCr dust) to exclude that these did not show unexpectedly higher toxicity than the fine sized fraction.

Surface reactivity was analyzed as hemolysis of red blood cells (erythrocytes) in three different concentrations, 0.67, 1.33 and 2.67 mg/mL. Figure 6A shows the release of hemoglobin, measured as absorbance intensity at 540 nm, for the dose 2.67 mg/mL. Each particle suspension was measured against a saline control and 1% Triton-X 100 as positive control causing total hemolysis. In addition, SiO₂ particles were used as a positive particle control. As seen in the figure, Cr₂O₃ particles caused hemolysis whereas none of the alloy particles or dust particles did. Calculated as % hemolysis, the Cr₂O₃ particles caused 1.4%, 3.1% and 22.5% hemolysis (mean value of three independent experiments) in the different doses. This can be compared to the SiO₂ particles causing 43.1%, 62.0% and 78.5% in the different doses.

The alkaline version of the comet assay, measuring mainly DNA single strand breaks and alkaline labile sites, was used to assess DNA damage after exposure of A549 human lung cells to a concentration of 40 μg/cm² (80 μg/mL) for 4 hours. Significantly higher amount of DNA damage was seen following exposure to 316L (16%, p < 0.05) when compared to the control (6%), see Figure 6B. A minor but statistical significantly effect (p < 0.05) was also observed for FeSiCr fine particles (8.0%). NiO nanoparticles, tested for comparison, showed approximately 4 times higher amount of DNA damage (25.3% tail) when compared to controls. None of the other particles caused any significant DNA damage.

Cytotoxic effects in terms of cell death were investigated after exposure of A549 human lung cells for 24 hours to 40 μg/cm² (80 μg/mL) particles. A slight, although non-significant, increase in cell death was observed after exposure to 316L particles. None of the other particles showed any effect (Figure 6C).

Alloy particles of a high-carbon ferrochromium alloy, FeCr, and ferrosiliconchromium, FeSiCr, of commercial relevance were supplied by Vargön Alloys AB, Sweden and Kazchrome, Republic of Kazakhstan, respectively [9]. These alloys are used as master alloys for further production of more advanced steel alloys. As the commercial products are different solid forms (lumps) and particle sizes, the materials were pre-treated (crushed, sieved, and re-crushed) to generate particles with a size distribution less than 60 μm. Dust particles of FeCr and FeSiCr were created during standardized dustiness tests using DIN 55992-1 [46]. Coarse (< 45 μm) and fine (< 4 μm) particles of stainless steel (gas-atomized, AISI 316L, used for powder metallurgy molding) were supplied by Arcam AB, Sweden, and Sandvik Osprey Limited, UK, respectively [18]. The nominal bulk composition of particles of FeCr, FeSiCr, and 316L is given in Table 2 based on supplier information. For comparative reasons, particles of Fe metal (99.96 wt%, Höganäs, Sweden), Cr metal (99.76 wt%, Delachaux, France), and chromium(III)oxide (99.1 wt%, Lanxess Deutschland GmbH, Germany) were investigated in parallel.

Particles of FeCr, FeSiCr, Fe, and Cr were further milled and sieved (20 μm sieve) by Elektrowerk Weisweiler GmbH, Germany, to obtain a smaller size fraction (< 20 μm). Particles sized less than 100 μm (aerodynamic diameter) are referred to as inhalable, particles less than 11 μm are defined as thoracic particles able to pass the larynx, and particles sized less than 5 μm are defined as respirable particles, able to reach the alveolar region of the deep lung (ISO 7708). It should be noticed that the aerodynamic diameter is larger than the physical diameter, with a relationship of approximately aerodynamical diameter = physical diameter * √density_particle (see e.g. [47]). In this study, the particle size distribution was measured in solution, accounting for agglomeration and hence particle cut-off points are larger compared with physical diameters measured in dry air.

The specific surface area (m²/g) per mass unit was determined by means of BET-analysis (adsorption of nitrogen at cryogenic condition) using a Micromeritics Gemini V instrument. Nitrogen adsorption measurements were performed at five different partial pressures (p/p₀ 0.10-0.25) with a standard deviation between replicas of less than 1% (except for Fe coarse particles: 8%). The adsorbent used was nitrogen, with a cross-sectional diameter of 0.162 nm² as input parameter. The sample was dried with nitrogen and flushed in a sample tube for 30 minutes at 150°C. After the measurement, several criteria (e.g. comparison to references) were verified, and if not sufficient, the measurement was repeated. The total surface area was about or larger than 1 m² to ensure accurate measurements using nitrogen gas. The measurement was conducted by Kanthal AB, Sweden.

Triplicate measurements of particle size distribution in solution (PBS, phosphate buffered saline, 8.77 g/L NaCl, 1.28 g/L Na₂HPO₄ and 1.36 g/L KH₂PO₄, pH 7.2) were conducted by means of laser diffraction using a Malvern Mastersizer 2000 equipment with a Hydro SM dispersion unit. Refractive indexes of Fe (2.81), Cr (3.51), Fe₃O₄ (2.42), Cr₂O₃ (2.5), and water (1.33) (since it was the solvent of the test medium), were used as input parameters applying standard operational conditions.

Particle shape and morphology were investigated by means of scanning electron microscopy (SEM), with a PHILIPS XL30ESEM instrument, and a field emission gun scanning electron microscopy (FEG-SEM) using a LEO 1530 instrument and a Gemini column.

Compositional analysis of surface oxides was performed by means of x-ray photoelectron spectroscopy, XPS (UltraDLD spectrometer from Kratos Analytical, Manchester, UK) using a monochromatic Al x-ray source (150 W) on areas approximately sized 700 × 300 μm. Wide spectra were run to detect elements present in the outermost surface oxide (information depth of a few nanometers) at five different locations, and detailed high resolution spectra (20 eV pass energy) were acquired for the main compositional elements. Experimental details are given in [9].

Metal release studies were conducted by immersing a specific particle loading (100 mg/L: 5 mg/50 mL) in artificial lysosomal fluid (ALF) of pH 4.5 (37°C) for exposure of 168 hours (1 week). A loading of 200 mg/L was used in the case of the fine 316L particles [18]. The loading of 100 or 200 mg/L has previously been shown to be low enough to avoid additional particle agglomeration in solution due to a too high loading [19]. The loading of 100 mg/L is furthermore recommended by the OECD transformation/dissolution protocol for aquatic acute tests [48]. ALF simulates in a simple way intracellular inflammatory conditions in lung cells following phagocytosis [49]. The composition of ALF is given in Table 3. The pH (4.5 ± 0.1) was adjusted by using 1.7 mL 50% NaOH to 1L solution. Chemical compositions of artificial sweat (ASW, pH 6.5), Gamble's fluid (GMB, pH 7.2), and artificial tear fluid (ATF, pH 8.0) are given in Table 3, and detailed information for these fluids given in [9,10]. All immersion studies were conducted at dark conditions in acid-cleaned polymethylpentene (PMP) Nalgene^® jars using Merck mini incubators standing on a shaking table with bi-linear shaking (30 cycles per minute, inclination 12°). Time-dependent studies were conducted using triplicate particle samples and one blank sample (without any particles). After exposure, the upper part of the test solution was poured into tubes and the particles were separated from the solution by centrifugation at 704 rcf (relative centrifugal force) for 10 minutes. This separation technique has previously been shown to be the most suitable procedure for bioaccessibility studies of similar metal particles [19]. To ensure sufficient separation of particles in the supernatant sampled from the centrifuged test medium, dynamic light scattering (Malvern Zetasizer nano ZS instrument) was performed. The supernatant, free of particles, was then poured into 25 mL polyethylene storage flasks and acidified to a pH less than 2 prior to metal analysis.

Selected samples of coarse sized Cr, FeCr, and 316L particles were immersed for 24 hours into the ALF solution to assess the chemical speciation (oxidation state and medium complexation capacity) of chromium. Non-acidified samples for these speciation analyses were immediately frozen. In order to avoid any metal contamination, all test vessels and experimental equipment were acid-cleaned with pure 10% HNO₃ for at least 24 hours and rinsed at least four times with ultra pure water (18.2 MΩ/cm). All experimental equipment and vessels were dried in ambient laboratory air before use. Experimental details are given in [9,10].

Total iron, chromium, and nickel (coarse FeCr and FeSiCr, coarse and fine 316L particles) concentrations were analyzed by means of graphite furnace atomic absorption spectroscopy, GF-AAS (Perkin Elmer AAnalyst 800), or flame atomic absorption spectroscopy (AAS) for higher (mg/L) concentrations (Fe and Ni particles). All measurements are based on three replicate readings of each sample, and quality control samples of known concentration were analyzed consecutively. Calibration was done with at least three standards of known concentration, e.g. 50, 100, and 500 μg/L for iron. All results presented are based on measured metal concentrations in the supernatants (particles separated) with the contribution from blank reference samples (matrix effects), if any, subtracted. The limits of detection (based on 3 times the standard deviation of blank samples) were 1, 1, and 0.5 μg/L for iron, chromium, and nickel, respectively (GF-AAS); and 0.1 and 0.5 mg/L for iron and nickel, respectively (AAS).

Concentrations of active Cr(VI) and total chromium were determined by means of differential pulse adsorptive cathodic stripping voltammetry (DPAdCSV) using a Metrohm 797 VA Computrace (with a hanging drop mercury electrode working electrode, an Ag/AgCl sat. KCl reference electrode and Pt auxiliary electrode) instrument and Metrohm 705 UV digester (high pressure mercury lamp, 500 W, 90°C). Calibration was conducted individually for each sample by standard addition of Cr(VI) (sufficient to double the peak height). The detection limit was 0.04 μg/L for both total chromium and Cr(VI). All blank concentrations measured were significantly below the limit of detection. More detailed information on the methodology is given in [9].

Cells from the human alveolar type II-like epithelial cell line, A549 (originally obtained from the American Type Culture Collection, ATCC) were grown and exposed to particles in Dulbecco's Minimal Essential Medium (DMEM) supplemented with 10% heat inactivated foetal bovine serum (FBS), 100 U/mL penicillin-streptomycin, and 1 mM sodium pyruvate in a humidified atmosphere at 37°C and 5% CO₂.

For the cytotoxicity assay, 0.08 million cells were seeded in wells of a 24-well plate (Becton Dickinson, Franklin Lakes, US) and grown for 24 h to obtain a 50% covering layer. After the additional exposure of particles for 24 h, a more than 90% confluent layer was obtained. In the comet assay 0.16 million cells were seeded in the wells and grown for 24 h to obtain a more than 90% confluent layer.

As a worst case inhalation scenario, the finer sized particles of FeCr, FeCr-dust, FeSiCr, FeSiCr-dust, Cr, Fe, and Cr₂O₃ were subjected for the toxicological testing. Before particle exposure of the cells, the particles were suspended in supplemented DMEM to a concentration of 1 mg/mL. The suspensions were vortexed for 20 s and sonicated using a probe (approximate output of 14 W in a 2 mL suspension) for 2 × 20 s with a 20 s break in between, to minimize particle agglomeration, and further diluted to 80 μg/mL (equal to 40 μg/cm² in the well plate) in 37°C sterile DMEM medium. Control cells were exposed to DMEM medium.

The cells were exposed to 40 μg/cm² (80 μg/mL) of particles for 24 h. After exposure, the cells were mixed with trypan blue, incubated for 3 min and the percentage of stained cells counted in a Bürker chamber, as a measure of cytotoxicity. The detailed experimental procedure is presented elsewhere [39]. A minimum of 100 cells was counted for each particle type in three separate experiments.

The cells were exposed to 40 μg/cm² (80 μg/mL) of particles for 4 h. To investigate DNA damage, mainly single strand breaks (SSB), alkali-abile sites (ALS) and double strand breaks (DSB), the alkaline version of the comet assay (single cell gel electrophoresis) was performed, previously described in [50]. The electrophoresis was however performed at ~17 V in a Comet Electrophoresis tank (SCIE-PLAS, Comet -20) where the negatively charged DNA-fragments migrate out from the nucleus, creating a "comet" with a tail. Fixation and then staining of DNA with ethidium bromide (1 μg/mL in Tris acetate-EDTA (TAE)) was performed the next day and the results were evaluated by computerized image analysis with a fluorescence microscope (Olympus BH2 with a 20x-apochromatic objective) using the program Komet 4.0 (Kinetic Imaging Ltd.). For each sample, 35 cells were examined in duplicates (i.e. 70 cells) and a mean value of the DNA damage as % tail of the comets for three independent experiments was calculated.

Fresh venous blood, from healthy blood donors (Blood Donor Center, Karolinska University Hospital, Stockholm), was collected in 10 mL EDTA tubes. The samples were gently mixed by inversion, added onto histopaque, centrifuged at 400 g (1370 rpm) for 30 min (20°C) and erythrocytes were collected. The cells were washed 3 times with PBS and then suspended in saline. 100 μL of suspended erythrocytes were mixed with 200 μL of particles in saline solution, pure saline or 0.1% Triton (negative and positive controls, respectively). The particles were prior to exposure of the erythrocytes suspended in saline to a concentration of 4 mg/mL, vortexed for 20 s and then sonicated in an ultrasonic bath for 15 min, to minimize agglomeration. The particles were then further diluted to 2 and 1 mg/mL (final concentration with erythrocytes; 2.67, 1.33 and 0.67 mg/mL). All exposures were conducted at dark conditions on a shaking table for 30 min (mixed every 10 min) and then centrifuged for 5 min at 10,000 rpm (15°C). Hemoglobin levels of the supernatant were determined from the optical density (OD), a measure of lysed erythrocytes, by using a microplate scanning spectrophotometer (PowerWave x, Bio-Tek Instruments, Inc. USA) at 540 nm (reference 620 nm) in two separate wells of the well plate. Three independent experiments were conducted for each particle. The % hemolysis was then calculated as presented in [37].

Un-paired two-tailed Student's t-test with unequal variance was used for statistical analysis of all bioaccessibility and toxicity results.