Chemical composition of individual aerosol particles in workplace air during production of manganese alloys

Abstract

Aerosol particle samples were collected at ELKEM ASA ferromanganese (FeMn) and silicomanganese (SiMn) smelters at Porsgrunn, Norway, during different production steps: raw material mixing, welding of protective steel casings, tapping of FeMn and slag, crane operation moving the ladles with molten metal, operation of the Metal Oxygen Refinement (MOR) reactor and casting of SiMn. Aerosol fractions were assessed for the analysis of the bulk elemental composition as well as for individual particle analysis. The bulk elemental composition was determined by inductively coupled plasma atomic emission spectrometry. For individual particle analysis, an electron microprobe was used in combination with wavelength-dispersive techniques. Most particles show a complex composition and cannot be attributed to a single phase. Therefore, the particles were divided into six groups according to their chemical composition: Group I, particles containing mainly metallic Fe and/or Mn; Group II, slag particles containing mainly Fe and/or Mn oxides; Group III, slag particles consisting predominantly of oxidized flux components such as Si, Al, Mg, Ca, Na and K; Group IV, particles consisting mainly of carbon; Group V, mixtures of particles from Groups II, III and IV; Group VI, mixtures of particles from Groups II and III. In raw material mixing, particles originating from the Mn ores were mostly found. In the welding of steel casings, most particles were assigned to Group II, Mn and Fe oxides. During the tapping of slag and metal, mostly slag particles from Group III were found (oxides of the flux components). During movement of the ladles, most particles came from Group II. At the MOR reactor, most of the particles belonged to the slag phase consisting of the flux components (Group III). The particles collected during the casting of SiMn were mainly attributed to the slag phase (Groups III and V). Due to the compositional complexity of the particles, toxicological investigations on the kinetics of pure compounds may not be easily associated with the results of this study.

---

Introduction

During the production of ferro- and silicomanganese alloys, workers are exposed to a variety of manganese oxides, primarily MnO₂, and metallic forms of manganese. Particles are the most common physical form of manganese in human respiratory exposure and, from a health-related perspective, interest is limited to those particles that can at least penetrate into the nose or mouth, and particularly to particles that deposit on respiratory tract surfaces. In human exposure by inhalation, three size-dependent particular fractions are defined: the inhalable fraction (aerodynamic diameter, d_ae ≤ 100 µm), which is the mass fraction of total airborne particles that enters the body through the nose and/or mouth during breathing; the thoracic fraction (50% mass fraction of total particles at d_ae = 10 µm); and the respirable fraction (50% mass fraction of total particles at d_ae = 4 µm).¹ Rappaport² has pointed out that health-related aerosol measurements need to be carried out with regard not only to the above particle size-selective criteria, but also to the kinetics of the processes by which the particles themselves can cause harm after inhalation. In understanding the overall picture of the risk associated with exposure to deposited particles in inhalation, both the particle size distribution of the aerosol, as well as the chemical and morphological states, are important determinants of reactivity recognized in human toxicology.³

Workers absorb manganese mainly through the lungs, although additional absorption via the gastrointestinal tract may occur to some extent.⁴ Manganese crosses the blood–brain barrier and, in high concentrations, Mn causes irreversible brain disease with prominent psychological and neurological disturbances.⁵ At present, it is still unclear whether industral exposure to manganese during the production of manganese alloys poses such health risks for workers at moderate and low exposure levels.^5–7 Since very little is known about the morphology and chemical composition of individual particles, as well as the chemical form of manganese in the manganese industry, such information is important in understanding the biological role of aerosol exposure.

Besides the conventional bulk methods of particle characterization used extensively and routinely, such as atomic absorption spectrophotometry (AAS), inductively coupled plasma atomic emission and mass spectrometry (ICP-AES/ICP-MS) and X-ray fluorescence (XRF) spectrometry, there are a number of topochemical methods which can contribute especially to single-particle characterization.^8–10 The most important and widely used method for morphological and compositional particle characterization is scanning electron microscopy in combination with energy-dispersive X-ray emission (SEM-EDX) analysis. Due to the inherent disadvantages of the latter method, wavelength-dispersive X-ray (WDX) analysis using crystal spectrometers is also applied in electron microprobes (EMP). WDX analysis has a better spectral resolution and lower detection limits than EDX analysis.

In the present study, electron probe microanalysis (WDX-EPMA) has been applied to the characterization of 1474 individual workroom aerosol particles collected at different working areas during the production of ferro- and silicomanganese.

Experimental

Site description

In Norway, ferromanganese and silicomanganese smelters are located at Porsgrunn (PEA), Kvinesdal and Sauda. Samples collected at the smelter at Porsgrunn were the focus of the present study. Air sampling was conducted in both the raw material and product sites of this plant. A typical plant consists of a site where the main raw materials, manganese and iron ores, coke, quartz and dolomite, are mixed and transported either by trucks or conveyor belts to the electrical furnaces. Two main alloys are produced: ferromanganese and silicomanganese. The ferromanganese product has a high carbon content [called standard FeMn, or high-carbon FeMn (HC-FeMn)]. The HC-FeMn product is either cast or brought as a fluid to the Metal Oxygen Refining (MOR) reactor, where carbon is removed for the production of medium-carbon (MC-) and low-carbon (LC-) FeMn. Finally, the alloys are crushed, screened and packaged for sale. The slag from the HC-FeMn process is rich in MnO and, after cooling, this slag is crushed and used as raw material in the production of SiMn of different qualities. Usually, the plants have only one SiMn furnace in operation. A more detailed description of the production steps can be found elsewhere.^11,12

Samples collected from various stages of the PEA smelter during the production of HC-FeMn, MC-FeMn and SiMn were investigated (sample number in parentheses).

PEA_mixing (21): mixing of raw materials.
PEA_welding (9, 11): welding of protective steel casing around Søderberg electrodes of the arc furnace.
PEA_tapping (14): tapping of HC-FeMn and slag.
PEA_ladles (3): transport of ladles with alloy and slag from the arc furnace to the MOR reactor.
PEA_MOR (4): operation of the MOR reactor.
PEA_castingSiMn (8): casting of SiMn.

Air sampling

Aerosol fractions for the measurement of the bulk elemental composition were assessed by employing 37 mm diameter air sampling cassettes equipped with 0.8 µm pore size cellulose-ester membrane filters (37 mm; Millipore, Bedford, MA, USA, AAWP 03700).

Aerosols for individual particle analysis were collected on 25 mm diameter polycarbonate filters with an effective pore size of 0.8 µm using open-faced graphite-filled polypropylene filter holders with a 50 mm extension tube (Gelman 25 mm Air Monitoring Cassette, Gelman Sciences, Ann Arbor, MI, USA). The polycarbonate filters were coated before exposure with a 100 nm layer of platinum using a Balzers SCD 050 sputter coater (Balzers, Liechtenstein) in order to avoid a contribution from the substrate to the carbon and oxygen X-ray count rates.

The pumps employed were the in-house, pulsation-free personal units constructed and used routinely at NIOH (National Institute of Occupational Health, Oslo, Norway). They were operated at a constant flow rate of 2.0 L min⁻¹, and the flow rate was measured with a calibrated rotameter, type Brooks 2-65 MM (Emerson Electric Co., Hatfield, PA, USA).

The Institute of Occupational Medicine, Edinburg (IOM) personal inhalable dust spectrometer (PIDS) was used for particle size distribution measurements (commercially available from SKC Ltd., Eighty Four, PA, USA). The manganese mass collected by each of the 10 impactor stages was recovered by carefully wiping with a wet PVC filter. These filters were treated in the same way as the air filters.

Analysis

Aerosol filters for bulk elemental measurements were dissolved in a mixture of aqua regia and hydrofluoric acid in Teflon autoclaves with microwave-assisted digestion. A Perkin-Elmer Optima Model 3000 inductively coupled plasma atomic emission spectrometer (Perkin-Elmer, Norwalk, CT, USA) was used for the measurements. A more detailed procedure is described elsewhere.¹³

The size and morphology of individual aerosol particles were studied by secondary electron (SE) and back-scattered electron (BSE) imaging on a Cameca Camebax SX50 electron microprobe. The chemical composition of the particles was measured quantitatively with the same instrument using wavelength-dispersive techniques. All measurements were carried out at an accelerating voltage of 15 kV and a beam current of 10 nA. The X-ray count rates for each individual particle and for standard samples were obtained from elemental distribution maps. The image resolution was 256 × 256 pixels in all cases, and counting times of 75 ms per pixel for the aerosol particles and 40 ms per pixel for standard samples were chosen. In order to detect particles with geometric diameters from approximately 0.3 to 10 µm, the magnification was varied between 1600× and 4000×. All electron images and element distribution images were recorded in the beam-scan mode, i.e. the electron beam was scanned across the sample. Spectrometer defocusing resulting from the scanning of the electron beam was corrected by techniques outlined in Kluckner et al.¹⁴ Experimental k-ratios were corrected for matrix and geometric effects using the CITZAF algorithm¹⁵ (version 3.02). For a detailed description of the procedures applied in this study, refer to Weinbruch et al.¹⁶ and Wentzel.¹⁷

We have restricted ourselves to measuring the elements C, O, Na, Mg, Al, Si, P, S, Cl, K, Ca, Mn, Fe, Co, Ni, Cu and As only. These elements were selected from preliminary studies and from the bulk composition of workroom aerosols, raw materials and products. Other elements may, however, be present in these particles as minor or trace constituents. Additional experimental parameters for the elemental measurements are summarized in Table 1.

Table 1 The X-ray lines and detector crystals used for elemental analysis by wavelength-dispersive EPMA

Element	X-ray line	Detector crystal^a
^aPC1, W/Si multilayer crystal; TAP, thallium acid phthalate; PET, penta-erythritol; LiF, lithium fluoride.
C, O	Kα	PC1
Al, Mg, Na, Si	Kα	TAP
Ca, Cl, K, P, S	Kα	PET
Co, Cu, Fe, Mn, Ni	Kα	LiF
As	Lα	TAP

---

The accuracy of the elemental composition can be estimated from the analysis of particles with known composition. According to Weinbruch et al.,¹⁶ the accuracy of the applied technique is better than 7.5% (relative) and the repeatability is on the order of 5–7%.¹⁷

It is assumed that these errors are also representative of the samples analysed in this work, except for carbon and oxygen which suffer from a contribution of the substrate.

Results and discussion

The typical bulk chemical composition of the aerosols determined by ICP-AES is summarized in Table 2 by working area. Clearly, the main components present in all samples, as expected, are manganese, iron, silicon, calcium, magnesium, potassium and sodium. During tapping, the volatile elements potassium, sodium, zinc and lead are enriched in the workroom aerosol.

Table 2 Typical aerosol composition (mass%) as measured with ICP-AES

	Sampling site
Element	Mixing of raw material	Welding of protective steel casing	Tapping of FeMn and slag	Crane operator, FeMn	Operator, MOR reactor	Casting of SiMn
Al	1.6	1.2	0.5	2.1	0.7	1.6
Ba	0.5	0.2	0.04	0.2	0.06	0.2
Ca	4.8	2.68	2.86	7.8	2.2	3.4
Co	0.07	0.02	<0.01	0.02	<0.01	<0.01
Cr	0.02	0.04	0.06	0.1	0.1	0.2
Cu	<0.04	0.17	<0.04	<0.04	<0.04	<0.04
Fe	5.5	12.3	2.0	4.7	2.6	0.5
K	4.0	5.8	51.0	2.7	0.7	2.5
Mg	1.9	2.4	3.7	13.2	6.0	5.8
Mn	44.8	23.0	17.9	29.9	7.8	17.9
Na	0.5	0.8	3.0	1.0	0.4	0.9
Ni	0.02	0.03	0.3	0.07	0.05	0.06
Pb	0.3	0.07	1.6	0.05	<0.01	<0.01
Si	4.6	9.0	8.9	12.9	6.5	18.0
Sr	0.04	0.02	<0.01	0.03	<0.01	0.03
Ti	0.09	0.1	0.05	0.2	0.05	0.2
Zn	0.6	0.5	6.4	0.5	0.1	0.2
Elements not detected above detection limit/mass%
Ag	<0.0004	Hg	<0.008	Sn	<0.001
As	<0.001	La	<0.0005	Ta	<0.009
B	<0.05	Li	<0.0005	V	<0.0004
Be	<0.0004	Mo	<0.0009	W	<0.008
Bi	<0.002	Sb	<0.001	Y	<0.0004
Cd	<0.002	Se	<0.009	Zr	<0.005

---

The mass distribution of manganese measured by PIDS shows that the largest portion of the inhalable manganese fraction is extrathoracic for all working areas as presented in Fig. 1. Therefore, relatively little manganese mass was found in the respirable and thoracic fractions. This general finding, that manganese was present in the form of larger particles, is also illustrated in Fig. 2, in which raw data from a PIDS sample, collected during the casting of ferromanganese, are depicted. The aerosol impaction and distribution of iron were also similar to those described for manganese.

Fig. 1 Aerosol mass fraction distribution of manganese measured by PIDS in different departments (in mass% of total).

Fig. 2 Particle size distribution curves for total airborne manganese: the inhalable fraction, the thoracic fraction and the respirable fraction for the casting operation. The fitted distribution function is denoted by f(d). It should be noted that the PIDS software calculates the total mass including the mass of large particles which are not collected in the inhalable fraction.

The size distribution, morphology and composition of 1474 particles with diameters greater than 0.3 µm from various stages of the production process have also been studied. In Fig. 3, the size distribution (equivalent projected area diameter) of all analysed particles is shown. The equivalent projected area diameter is the diameter of a sphere which projects, in two dimensions, the same area as the particle under consideration. If we assume that the average density of the particles under study is four and the particles are spherical in shape, the aerodynamic diameter of a given particle is twice the equivalent projected area diameter.

Fig. 3 Distribution of the equivalent projected area diameter, determined by secondary electron imaging with an electron microprobe.

Correlations between the chemical composition and the equivalent projected area diameter were not observed. Correlations between the particle shape and composition were not found either. Most particles analysed have complex compositions that cannot be attributed to simple stoichiometric compounds.

In many particles, an excess of oxygen was observed. In this case, the measured oxygen content was higher than the calculated value, assuming a stoichiometric composition and the highest oxidation state for all cations (i.e. Mn⁴⁺, Fe³⁺, Cu²⁺, As⁵⁺). The abundance of particles with excess oxygen values varies in a given sample between 15% and 75% of all particles. In samples PEA_mixing, PEA_tapping and PEA_castingSiMn, the excess oxygen content may reach values up to 66 at.%, and in sample PEA_MOR even up to 79 at.%. In samples PEA_welding (9, 11), the maximum excess value is 32 at.%. Small excess values of oxygen (of the order of a few at.%) may be explained by counting statistics. Higher values may be due to either a contribution of the substrate or partial oxidation of metallic particles leading to an oxide surface layer. High carbon contents of some particles may be caused either by the influence of the substrate or by the presence of soot.

It is concluded that the particles studied consist of a mixture of different phases. Despite their complex compositions, the particles can be divided into the following six groups.

Group I. Particles dominated by metallic phases and predominantly consisting of the elements Mn and Fe. The majority of these particles have higher Mn (38–77 at.%) than Fe (0–23 at.%) contents. The presence of oxygen (up to 34 at.%) indicates the presence of oxides in addition to the metal phases. The amount of Mn and Fe is always significantly higher than that of O. Therefore, it can be concluded that metal phases are present in all particles of this group. Carbon (up to 34 at.%) was also detected in many particles.

Group II. Particles consisting of Mn (1–44 at.%), Fe (0.5–38 at.%) and a high amount of O (25–79 at.%). In contrast with Group I, all particles contain more O than Mn and Fe, so that obviously no or only a small amount of metal phases are present. The samples presumably consist of a mixture of Mn and Fe oxides or, alternatively, of mixed FeMn oxides. However, the oxide phases present are not known. The Mn/Fe ratio is highly variable. The major fraction of particles in this group consists predominantly of Mn compounds, except for samples PEA_welding and PEA_MOR, where an equal number of particles consist predominantly of Fe or Mn.

Group III. Particles dominated by the elements Si, Al, Mg, Ca, Na and K. Most particles are completely oxidized (silicates, carbonates, oxides). The Mn content in these particles is usually small (below 6 at.%). Si, Al, Mg and Ca are added to the blend of raw materials and act as fluxes and slag formers. Na and K enter the process mainly as minor or trace constituents in various raw materials (coke, fluxes, ores).

Group IV. Particles predominantly consisting of C (≥60 at.%). In addition to carbon, all other elements measured were found in variable amounts without any systematic dependence on sampling site. The particles in this group may be considered as coke or soot particles with adsorbed or incorporated impurities, or as agglomerates of small particles that cannot be resolved in our instrument.

Group V. Particles with C contents between 35 and 60 at.%. As the ratios of metallic elements are similar to those within Groups II and III, Group V can be regarded as a mixture of Groups II, III and IV. In addition, a few particles can be considered as mixtures of Groups IV and VI.

Group VI. Mixture of Groups II and III. The concentration ratio of Group II components to Group III components varies between 1∶5 and 2∶1, but is in most cases around 1∶1. The difference between the abundance of Group II and Group III elements in these particles does not exceed 10 at.%. Most of the particles contain less than 10 at.% of Mn.

In addition to the elements used for classification, most particles contain variable amounts of P, S, Cl and the metals Co, Ni, Cu and As. In very few cases, the concentrations of the non-metallic elements listed above can reach 20 at.%, but in most particles they do not exceed 5 at.%. The same is true for the elements Co, Ni, Cu and As. Arsenic, especially, was found in only minor concentrations. However, the classification of the particles into the groups of metal, slag, coke and mixtures is not affected by the abundance of these elements. Furthermore, three particles consisting of equal proportions of Mn and Si were not included in the classification scheme. Typical examples of particle compositions for each group are given in Table 3 (all data taken from sample PEA11). It should be considered here that Mn is oxidized quite easily. As the samples were stored under atmospheric conditions, it cannot be excluded that a few metallic particles oxidized after collection and prior to chemical analysis. Examples of particles from the groups defined above are shown in Figs. 4(a)–(f) (all images from sample PEA9). The images were obtained using a Philips XL 30 FEG high-resolution scanning electron microscope with beam currents in the pA range. Due to the better lateral resolution of this instrument, more details of the morphology of the particles can be seen compared with the electron images from the electron microprobe. Due to the limited lateral resolution of the electron microprobe, the small particles in Figs. 4(a)–(f) could not be analysed. The variable composition of the particles can be seen in Table 4, where the median, the lower quartile and the upper quartile for each group are presented (50% of all measured values lie between the lower and upper quartile).

Fig. 4 Scanning electron images of typical particles from sample PEA9: (a) particle consisting predominantly of Mn and Fe with low O content (Group I); (b) manganese oxide particle (Group II); (c) slag particle (Group III); (d) carbon dominated particle (Group IV); (e) mixed particle (Group V); (f) mixed particle (Group VI).

Table 3 Typical particle compositions (at.% normalized to a total of 100 at.%)

	Group
Element	I	II	III	IV	V	VI
^a<d.l., concentration below detection limit (approximately 0.01 at.%).
C	5.2	3.0	12.0	91.4	40.7	20.8
O	16.9	65.0	57.2	5.5	33.1	40.2
Na	0.7	0.4	0.2	0.1	4.5	11.0
Mg	0.6	0.1	<d.l.^a	<d.l.	0.1	1.3
Al	0.4	0.5	0.2	0.1	<d.l.	0.5
Si	0.6	5.1	23.1	0.3	0.1	1.9
P	0.3	0.2	0.1	0.1	0.6	0.2
S	0.8	0.5	0.3	0.5	0.6	2.0
Cl	0.7	0.4	0.5	0.2	0.8	1.8
K	1.1	0.3	0.5	0.1	0.1	1.9
Ca	0.7	0.1	0.1	0.1	0.1	2.2
Mn	60.0	20.8	3.8	0.4	14.0	4.0
Fe	9.5	2.7	0.7	0.4	1.7	10.7
Co	0.6	0.5	0.5	0.3	0.8	0.3
Ni	0.6	0.1	0.5	0.3	0.8	0.4
Cu	1.0	0.3	0.2	0.1	2.0	0.6
As	0.2	<d.l.	<d.l.	<d.l.	<d.l.	<d.l.

---

Table 4 Median, lower and upper quartile (respectively) of the main element concentrations for different particle groups

Group	Mn	Fe	O	C	Na	Mg	Al	Si	K	Ca
I	54.3	5.9	10.7	12.5	0	0	0	1.1	0.6	0
	46.6	1.5	4.8	4.8	0	0	0	0	0	0
	62.6	10.2	16	22.2	0.7	0.9	0	5	1.1	0.6
II	17.1	2.7	52.9	11.8	0.7	0	0	0.9	1.1	0
	9.2	1.2	45.2	4.7	0	0	0	0	0.5	0
	24.1	10.4	58.2	21	1.1	0	0.9	1.9	2.1	0.6
III	1.2	1.1	48.2	16.2	1.8	0	0	0	12	0
	0	0.7	38.7	8.5	0.7	0	0	0	0.8	0
	2.7	1.8	56.1	25.1	2.6	1	0.9	4.1	18.4	0.7
IV	1	1.4	13.1	68.1	0.7	0	0	0	1.2	0
	0	0.8	3.2	63.6	0	0	0	0	0.3	0
	3.2	2.1	18.8	75.9	1.25	0	0.6	0.6	3.2	0.5
V	3.4	1.7	28.2	42.2	1.3	0	0	0.6	4.1	0.4
	1.5	1.1	20.2	38.5	0.7	0	0	0	1.6	0
	7.9	3.1	33.5	48.1	2	0.6	0.7	1.3	8.1	0
VI	6.8	2	49.8	19.3	1.3	0	0.6	1.2	1.5	0
	2.6	1.3	42.2	9.8	0.6	0	0	0	0.8	0
	10.7	4.1	60.5	26.9	2.7	0.8	1.7	3.3	5.5	1

---

The relative abundances of the particle groups in the samples from different production steps are shown in Table 5. It can be seen that the abundance of particle groups is highly variable.

Table 5 Abundance (%) of particle groups in samples from different production steps

Sample	Group I	Group II	Group III	Group IV	Group V	Group VI	Total number of particles
PEAmixing (21)	4.9	27.2	11.1	0	28.4	28.4	81
PEAwelding (9)	2.3	54.0	8.4	9.7	16.8	8.7	298
PEAwelding (11)	2.2	43.9	13.7	9.0	14.4	13.7	277
PEAtapping (14)	1.2	2.1	69.3	2.1	21.9	3.3	329
PEAladles (3)	13.7	34.6	23.9	7.3	11.5	9.0	232
PEAMOR (4)	0	19.6	47.6	1.4	5.6	25.9	143
PEAcastingSiMn (8)	0	3.6	32.4	9.0	36.9	18.0	111

---

PEA_mixing (21)

Most of the particles found at this site can be assigned to the Mn ores used in the process. The particles consist of mixed Mn/Fe oxides with variable amounts of C (Group IV) and/or K (Group VI). The elements Si, Ca and Mg, which are added with the slag formers in this process, were only found in very small amounts.

PEA_welding (9, 11)

As could be expected, the percentages for most particle groups are similar. The highest number of particles belongs to Group II, i.e. Mn, Fe oxides. Soot or coke particles are also found here more frequently than at other sampling sites, except for PEA_castingSiMn (8).

PEA_tapping (14)

This site contains by far the most slag particles (Group III), which is to be expected. The abundances of particles of other groups are small. Due to instrument limitations, this study was restricted to particle diameters above 0.3 µm. In a recent transmission electron microscopy (TEM) study,¹⁸ however, a large number of particles ranging from a few nanometres to about 0.2 µm were observed. These particles were predominantly of one type and were identified from electron diffraction patterns as crystalline MnO.

PEA_ladles (3)

All main oxide containing particles are abundant at this site with Mn, Fe oxides present at the highest percentage. It is remarkable that the highest abundance of metallic particles (13.7%) was also observed at PEA3.

PEA_MOR (4)

Most interesting, almost half of the particles analysed (diameters >0.3 µm) are slag particles, followed by Group VI particles which are made up of Group II and III particles (MnFe oxides and slag particles). Agglomerated, very fine Mn₃O₄ particles were expected, but not found, presumably due to the use of a very effective ventilation system above the MOR reactor.

PEA_castingSiMn (8)

Most of the particles found at this sampling site show mixed compositions, as 55% of all particles belong to Groups V and VI. The dominating cation in the particles in Groups III, V and VI is K; Si and Mg are found only in a few particles. Mn and Fe compounds are also found to a lower extent. Particles containing metallic phases of both elements could not be identified at this sampling site.

General observations

The aerosol samplers used in the present study measure a non-specific subfraction of the total aerosol. Under the given sampling conditions, the largest particle that theoretically can be aspirated is approximately 50 µm in aerodynamic diameter¹⁹ given a particle density of 4 g cm⁻³. This corresponds approximately to the thoracic fraction (mass fractions of total aerosol of 50% at d_ae = 10 µm and of 1% at d_ae = 28 µm), which is the inhaled particle component which penetrates into the lung and is significant in asthma, bronchitis and lung cancer.³ Although Kenny et al.²⁰ have shown that the sampling efficiency of the 37 mm sampler, with similar aspiration characteristics to the 25 mm sampler used, is approximately 20% for particle aerodynamic diameters higher than d_ae = 30 µm, very few particles with d_ae > 30 µm were found in the samples collected for individual particle characterization.

A small number of partially oxidized metallic particles are found in most samples [except PEA_MOR (4) and PEA_castingSiMn (8)]. High abundances of Fe and Mn oxides were encountered in samples from production steps prior to ore melting [PEA_mixing (21) and PEA_welding (9, 11)]. These samples contain a relatively low abundance of particles consisting of slag-forming elements (Group III). The opposite trend, i.e. a relatively low abundance of Fe and Mn oxides and a high abundance of Group III particles, was observed in samples from production steps following ore melting [PEA_tapping (14), PEA_MOR (4), PEA_castingSiMn (8)]. Soot or coke particles are present in most samples [except PEA_mixing (21)] at levels between approximately 1 and 10%. Mixtures of Groups II, III and IV are observed in all samples.

The most important general observation is that none of the particles found consists just of one single phase or compound. Hence, many toxicological investigations on the kinetics of such pure compounds may not be easily associated with the remarkable compositional complexity of the particles observed here. Whether these particles originating from raw materials, process intermediates or final products have the same fate after deposition in the respiratory tract, compared with those of homogeneous composition normally studied in animal exposure, is presently not well known. The possible diversity in biological responses induced by homogeneous/non-homogeneous manganese containing particles is a warning that the physicochemical properties of manganese compounds used in laboratory toxicological assessments must be well defined for the meaningful interpretation of experimental results.³ The observations in this study of the complexity of the chemical composition of individual particles naturally occurring in the workroom atmosphere during the production of manganese alloys provide further rationale for the need to generally characterize air particulates in exposure assessments.

Acknowledgements

Financial support is gratefully acknowledged from the Confederation of Norwegian Business and Industry and ELKEM ASA.

References

1. EN481. Workplace Atmospheres—Size Fraction Definitions for the Measurements of Airborne Particles.European Committee for Standardization, Brussels, Belgium, 1993..
2. S. M. Rappaport, Ann. Occup. Hyg., 1991, 35, 61.
3. E. Nieboer, G. G. Fletcher and Y. Thomassen, J. Environ. Monit., 1999, 1, 1.
4. H. Roels, R. Lauwerys, P. Genet, M. J. Sarhan, M. De Fays, I. Hanotiau and J.-P. Buchet, Am. J. Ind. Med., 1987, 11, 297.
5. M. Aschner and J. L. Aschner, Neurosci. Biobehav. Rev., 1991, 15, 333.
6. R. Lucchini, L. Selis, D. Folli, P. Apostoli, A. Mutti, O. Vanoni, A. Iregren and L. Alessio, Scand. J. Work Environ. Health, 1995, 21, 143.
7. D. Mergler, G. Huel, R. Bowler, A. Iregren, B. Belanger, M. Baldwin, R. Tardif, A. Smargiassi and L. Martin, Environ. Res., 1994, 64, 151.
8. H. M. Ortner, P. Hoffman, F. J. Stadermann, S. Weinbruch and M. Wentzel, Analyst, 1998, 123, 833.
9. R. Van Grieken and C. Xhoffer, J. Anal. At. Spectrom., 1992, 7, 81.
10. W. Jambers, L. De Bock and R. Van Grieken, Analyst, 1995, 120, 681.
11. D. B. WellbelovedP. M. CravenJ. W. Wandby, in Manganese and Manganese Alloys, Ullmann's Encyclopedia of Industrial Chemistry, ed. B. Elvers, S. Hawkins and G. Schulz, VCH Verlagsgesellschaft GmbH, Weinheim, 5th edn., 1990, vol. A16, 77..
12. S. O. Wasbø, Ferromanganese Furnace Modelling Using Object-Oriented Principles, PhD Thesis (in English), Norwegian University of Science and Technology, Trondheim, Norway, 1996..
13. Y. Thomassen, E. Nieboer, D. Eillingsen, S. Hetland, T. Norseth, J. Ø. Odland, N. Romanova, S. Chernova and V. P. Tchatchtchine, J. Environ. Monit., 1999, 1, 15.
14. M. Kluckner, O. Brandl, S. Weinbruch, F. J. Stadermann and H. M. Ortner, Mikrochim. Acta, 1997, 125, 229.
15. J. T. Armstrong, Microbeam Anal., 1995, 4, 177.
16. S. Weinbruch, M. Wentzel, M. Kluckner, P. Hoffmann and H. M. Ortner, Mikrochim. Acta, 1997, 125, 137.
17. M. Wentzel, PhD Thesis, Chemistry Department, Technical University of Darmstadt, Germany, 1999..
18. K. GjønnesA. SkogstadS. Hetland, to be published..
19. L. SilvermanC. E. BillingsM. W. First, Particle Size Analysis in Industrial Hygiene, Academic Press, New York and London, 1971..
20. L. C. Kenny, R. Aitken, C. Chalmers, J.-F. Fabris, E. Gonzalez-Fernandez, H. Kromhout, G. Lidén, D. Mark, G. Riediger and V. Prodi, Ann. Occup. Hyg., 1996, 41, 135.

---